摘要
为进一步提高柴油机智能化诊断系统的诊断准确率,针对不同工况柴油机缸盖振动信号的特点,利用4层小波包分解提取缸盖振动信号的特征向量,采用基于信息熵最小化的多区间离散化分析确定特征向量的离散区间,利用改进的超1-依赖贝叶斯分类器对模拟故障状态下的离散数值进行分类.实验结果表明,该方法对不同载荷、转速及混合工况下柴油机故障的诊断准确率均较高,为超1-依赖贝叶斯分类器在柴油机智能化故障诊断系统的进一步应用作出有益的尝试.
In order to improve the diagnosis accuracy of intelligent diesel engine diagnosis system,based on the characteristics of cylinder-head vibration signal in different conditions,the four-layer wavelet packet decomposition is used to extract characteristic vectors of cylinder-head vibration signal,and multi-interval discretization analysis based on the minimization of information entropy is used to determine discretization intervals of the characteristic vectors,and then discrete data under the simulated fault state are classified by the improved super parent one-dependence Bayesian classifier.The experiments show that the rate of accuracy of fault diagnosis for diesel engine is high under different load,speed and mixed conditions.It makes a helpful attempt for further application of super parent one-dependence Bayesian classifier on intelligent diesel engine fault diagnosis system.
出处
《上海海事大学学报》
北大核心
2011年第3期49-53,共5页
Journal of Shanghai Maritime University
基金
辽宁省自然科学基金(20072144)
中央高校基本科研业务费资助项目(2009QN027)
关键词
柴油机
振动信号
贝叶斯分类器
故障诊断
1-依赖分类器
diesel engine
vibration signal
Bayesian classifier
fault diagnosis
one-dependence classifier
