摘要
为了解决车辆状态远程故障诊断系统中的不确定性问题,提出了一种基于贝叶斯网络模型的故障诊断方法.这种故障诊断方法可在输入数据不完备,甚至含噪的情况下,充分利用贝叶斯网络的先验知识以及建模学习能力和概率推理算法来应对不确定性问题的表示和推理,完成系统的故障诊断.实验结果表明,贝叶斯网络方法在车辆故障诊断速度、准确性方面优于传统的基于BP算法或RBF算法的诊断方法,并且提高了故障诊断系统的鲁棒性.
To resolve the uncertainty issues in remote fault diagnosis system for vehicle sta- tus, a fault diagnosis approach based on Bayesian network（BN） model is proposed. This method handles with the uncertain representation and reasoning by exploiting the prior knowledge, learning and probabilistic inference abilities of BN. A fault diagnosis can be a- chieved even with incomplete or noisy data. Experimental results demonstrate that the pres- ented BN approach may provide faster and more accurate fault diagnosis results than tradi- tional Back-propagation（BP） or Radial Basis Function（RBF） method for vehicles. The pres- ented approach can also improve the robustness performance for the fault diagnosis system.
出处
《陕西科技大学学报（自然科学版）》
2013年第1期121-125,共5页
Journal of Shaanxi University of Science & Technology
基金
陕西省教育厅科研计划项目(11JK0995)
陕西科技大学自然科学基金项目(ZX11-29)
关键词
远程故障诊断
车辆状态
不确定性
贝叶斯网络
推理
remote fault diagnosis
vehicle status
uncertainty
Bayesian network
inference
