推进技术 ›› 2003, Vol. 24 ›› Issue (5): 414-416.

• • 上一篇    下一篇

航空发动机故障的支持矢量机智能诊断

朱家元,张喜斌,张恒喜,裴静   

  1. 空军工程大学工程学院;陕西西安710038;空军工程大学工程学院;陕西西安710038;空军工程大学工程学院;陕西西安710038;空军驻武汉滨湖机械厂军代表室 湖北武汉430077
  • 发布日期:2021-08-15
  • 基金资助:
    国防预研基金资助 (98J1 9 3 2 JB3 2 0 1 )

Aeroengine intelligent fault diagnosis using support vector machines

  1. Engineering Inst., Air Force Engineering Univ., Xi’an 710038, China;Engineering Inst., Air Force Engineering Univ., Xi’an 710038, China;Engineering Inst., Air Force Engineering Univ., Xi’an 710038, China;Air Force Military Agent in Binhu Mechanical Factory, Wuhan 430077, China
  • Published:2021-08-15

摘要: 引入支持矢量机和多元分类算法到航空发动机故障诊断当中。通过设计的多元分类支持矢量机构建了小样本多参数航空发动机故障智能诊断模型,然后通过发动机故障仿真器对典型发动机气路故障进行了诊断。结果表明,支持矢量机具有优秀的故障学习能力,采用它进行航空发动机故障诊断是可行、有效的。

关键词: 支持矢量机+;机器学习;航空发动机;故障诊断;人工神经元网络

Abstract: Most of artificial intelligent methods are based on theory of empirical risk minimization principle which may cause low aeroengine fault diagnosis rate even with large learning fault samples. In this paper, we present a new aeroengine intelligent fault diagnosis system model using support vector machines (SVM) which is based on an novel theory - structural risk minimization (SRM) inductive principle. Due to the binary classifier of common Support Vector Machines, we first develop a multi-class support vector machines for faults pattern recognition as described detailed in section 2. We then construct the diagnosis model via the multi-class SVM in section 3. Non-linear mapping relationship between faults patterns and faults performance can be induced by the model. Also the model can control learning generalization ability with SRM. In section 4, we use the model for turbofan engine gas fault diagnosis. The results show that the model is efficient for accurate and robust fault diagnosis, and only a few fault samples are necessary to be trained for new faults diagnosis.

Key words: Support vector machines~+;Aircraft engine;Fault diagnosis;Artifical neural network