摘要
为提高航空发动机滑油系统的故障诊断有效性,提出了一种改进的DBN航空发动机滑油系统故障诊断方法,首先对航空发动机滑油系统参数数据进行预处理,利用DBN的特征提取能力和ELM的快速学习优势创建DBN与ELM结合的故障诊断模型。为减少人为调节网络参数随机性对诊断结果造成的影响,采用粒子群算法优化DBN-ELM的网络参数,得到最优的网络结构,创建改进的DBN故障诊断模型。最后,对所创建的改进的DBN故障诊断模型进行了试验验证技术研究,结果表明,所提出的改进的DBN故障诊断方法能有效提升航空发动机故障诊断准确率,诊断效果明显优于DBN-ELM,具有很好的应用前景。
In order to improve the effectiveness of fault diagnosis for aero-engine lubricating oil system,an improved DBN fault diagnosis method of aero-engine lubricant oil system is proposed.First,pre-process the parameter data of aero-engine lubricating system,and use the feature extraction capabilities of DBN and the rapid learning advantages of ELM to create a fault diagnosis model.In order to reduce the impacts of the randomness of artificial adjustment of network parameters on the diagnosis results,use the particle swarm algorithm to optimize the network parameters of DBN-ELM,obtain the optimal network structure,and create an improved DBN fault diagnosis model.Finally,the experimental verification technology of the created improved DBN fault diagnosis model is studied.The results show that the proposed improved DBN fault diagnosis method can effectively improve the accuracy of aero-engine fault diagnosis,and the diagnosis effect is significantly better than DBN-ELM.It has good application prospects.
作者
崔建国
李勇
崔霄
王景霖
蒋丽英
于明月
CUI Jian-guo;LI Yong;CUI Xiao;WANG Jing-lin;JIANG Li-ying;YU Ming-yue(School of Automation,Shenyang Aerospace University,Shenyang,110136,China;Wind Tunnel Equipment Research and Development Department,AVIC Aerodynamics Research Institute,Shenyang 110034,China;Aviation Key Laboratory of Science and Technology on Fault Diagnosis and Health Management,Shanghai 201601,China)
出处
《沈阳航空航天大学学报》
2020年第6期49-54,共6页
Journal of Shenyang Aerospace University
基金
国家自然科学基金(项目编号:51605309)
航空科学基金(项目编号:201933054002)
航空科学基金(项目编号:20163354004)。
关键词
航空发动机
深度置信网络
极限学习机
粒子群算法
故障诊断
aero-engine
deep belief network
extreme learning machine
particle swarm optimization
fault diagnosis
