为解决船舶柴油机故障诊断复杂多样、难于辨识的问题,有效提高分类的准确率,研究一种基于遗传算法优化合成核支持向量机的故障诊断方法.首先，将全局核(多项式核)与局部核(径向基核)通过凸组合的方式构成合成核；然后，利用二进制遗传算法对合成核支持向量机的核参数、权系数、惩罚因子寻优，并应用该方法对采集的多类船舶柴油机故障进行诊断，获得了较高的分类准确率.
Abstract
To solve the complex, diverse and difficult to identify problems in marine diesel engine fault diagnosis while effectively improving the classification accuracy , a novel method of marine diesel engine fault diagnosis is proposed based on genetic algorithm (GA) optimized composite kernel support vector machine (SVM). Firstly, a composite kernel is constituted by combining global kernel (polynomial kernel) with local kernel (RBF kernel).Then binary coded genetic algorithm is used for parameter optimization of composite kernel SVM. Multiclasses of marine diesel engine faults are diagnosed. Experiment results show that the proposed method has the higher classification accuracy.
关键词
船舶柴油机 /
故障诊断 /
遗传算法（GA） /
合成核 /
支持向量机（SVM）
Key words
marine diesel engine /
fault diagnosis /
genetic algorithm (GA) /
composite kernel /
support vector machine (SVM)
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
［1］ VAPNIK V N. The nature of statistical learning theory［M］. New York: SpringerVerlag, 1995.
［2］ 汪洪桥，孙富春，蔡艳宁，等. 多核学习方法［J］.自动化学报，2010，36(8): 1037-1050.
WANG Hong-qiao, SUN Fu-chun, CAI Yan-ning, et al. On multiple kernel learning methods［J］. Acta Automatica Sinica , 2010，36(8): 1037-1050.(in Chinese)
［3］邬啸，魏延，吴瑕. 基于混合核函数的支持向量机［J］.重庆理工大学学报:自然科学，2011，25（10）：66-70.
WU Xiao，WEI Yan，WU Xia. Support vector machine based on hybrid kernel function［J］. Journal of Chongqing University of Technology: Natural Science，2011，25（10）：66-70.(in Chinese)
［4］ JIANG T J, WANG S Z, WEI R X. Support vector machine with composite kernels for time series prediction［C］// Proceedings of the 4th International Symposium on Neural Networks: Advances in Neural Networks. Nanjing:Springer, 2007: 350-356.
［5］ HUANG Jian, HU Xiao-guang, YANG Fan. Support vector machine with genetic algorithm for machinery fault diagnosis of high voltage circuit breaker［J］. Measurement,2011,44(6): 1018-1027.
PDF(441 KB)
