摘要
提出了一种基于主成分分析、遗传算法和支持向量机的失火故障诊断方法,以用于建立发动机失火故障诊断模型。采集不同失火故障模式下瞬时转速信号,通过归一化和主成分分析对数据集进行降维预处理,提取样本特征;随机选取1/3的样本训练支持向量机失火诊断模型,并结合网格搜索和遗传算法优化模型参数;剩余样本作为测试数据,对12种失火模式进行辨别,准确率为98.33%。因此,该方法有效。
A misfire diagnosis method was proposed based on Principal Component Analysis (PCA), Genetic Algorithm (GA) and Support Vector Machines (SVM), to build engine misfire diagnosis model. Instantaneous speed signals were collected under different misfire modes. Firstly, normalization and PCA were used to reduce the dimensions of dataset, to extract the sample features; subsequently, 1/3 of the data samples selected randomly was trained to support the SVM misfire diagnosis model; a grid search and GA were then applied as the combination strategy to optimize the model parameters; the remaining 2/3 data was used as test data to identify 12 misfire fault modes, the accuracy of which reached 98.33%. The proposed method can effectively solve the problem of engine misfire faults recognition with small sample and high dimensions.
出处
《汽车技术》
CSCD
北大核心
2017年第1期38-42,57,共6页
Automobile Technology
基金
国家自然科学基金项目(51406140)
武汉理工大学自主创新研究基金项目(155207006)
关键词
汽油机失火
主成分分析
遗传算法
支持向量机
Gasoline engine misfire, Principal component analysis, Genetic algorithm, Supportvector machine