摘要
针对基于单层感知器模型的发动机故障进行分类器设计,研究了故障信号的学习样本容量和分类误判率之间的关系。结果表明,随着学习样本量的增加,该分类器的误判率越来越小,最终在一定范围内呈现平稳波动变化。
Based on the single-layer perceptron model, a relation between sample size and error classifying for the design of fault classifier for the AFR engine is given. Experiments on engine show that, with the increasing of sample size, error classifying decreases continuously and finally, it fluctuates smoothly in some ranges.
出处
《长春工业大学学报》
CAS
2006年第2期118-119,共2页
Journal of Changchun University of Technology
关键词
分类器设计
故障诊断
感知器
误判率
classifier design
fault diagnosis
perceptron
error rate.
