摘要
分析燃气轮机的 8种典型常见故障 ,建立了基于支持向量机的故障诊断模型 ,用实例计算证明其有效性。同时和神经网络方法对比后发现 :在小样本情况下 ,支持向量机方法的计算结果比神经网络要好 ,推广能力更强 ,而且效率高于神经网络。本方法针对故障诊断样本少的特点 ,为建立智能化的燃气轮机状态监控和故障诊断提供了一种新的途径 。
With respect to eight kinds of commonly seen typical faults a fault diagnosis model is set up based on a support vector machine. Specific sample calculations have demonstrated the effectiveness of such a model. A comparison with a neural network method has shown that under the condition of a small quantity of samples the support vector machine-based method is superior to the neural network method in terms of calculation results, generalization ability and efficiency. When a relatively small number of diagnosis samples is involved, the above method may provide a new approach for creating an intelligent system of highly practical value for the condition monitoring and fault diagnosis of gas turbines.
出处
《热能动力工程》
EI
CAS
CSCD
北大核心
2004年第4期354-357,共4页
Journal of Engineering for Thermal Energy and Power
关键词
燃气轮机
支持向量机
故障诊断
gas turbine, support vector machine, fault diagnosis system.
