摘要
本文中根据汽油机空燃比故障的特点,提出了一种基于数据流趋势异常和相关性分析的故障实时检测方法。为了提高准确率和运行效率,通过衰减和时间判断法在误差累积和算法CUSUM基础上进行改进,对数据流之间的相关性进行快速估计,使整个系统能运行在对资源比较敏感的车载平台上。在长安福特发动机上的实验结果表明,相比传统的基于SVM和神经网络的故障检测方法,该方法能以较低的资源消耗获得更好的故障检测效果。
According to the features of air fuel ratio fault in gasoline engine,a real-time fault detection method based on the abnormality of data flow trend and correlation analysis is proposed in this paper. For enhancing accuracy and operation efficiency,modifications are made based on the algorithm of cumulative sum of error,with rapid estimation on the correlation between data flows,to enable the whole system operates on on-board platform relatively sensitive to resources. The results of test on the engine of Chang-an vehicle show that compared with traditional fault detection method based on support vector machine and neural network,the method proposed can obtain better results of fault detection with lower resource consumption.
出处
《汽车工程》
EI
CSCD
北大核心
2017年第4期462-470,485,共10页
Automotive Engineering
基金
国家自然科学基金(61305134)
博士点基金(20133219120035)资助
关键词
汽油机
空燃比故障
实时故障检测
数据流趋势
累积和算法
相关性
gasoline engine
air fuel ratio fault
real-time fault detection
data flow trend
CUSUM algorithm
correlation
