摘要
设计了基于最小二乘回归型支持向量机(LSSVR)的系统残差生成器,并引入遗传算法(GA)对LSSVR的参数进行优化,通过残差生成器生成的残差可以对系统状态进行有效识别。仿真实验表明:基于遗传优化的LSSVR可以高精度地模拟系统的动态特性,进而生成高精度的故障残差,有效地保证了故障诊断的准确性。
The residual error generator of system is designed based on least square support vector regression （LSSVR） ,and the parameters of LSSVR are optimized by genetic algorithm （GA）. The states of system can be identified effectively through the residual error generated by residual error generator. The results of simulative experiments show that the dynamic character of system can be simulated accurately by LSSVR based on genetic optimization,and accurate fault residual error can be generated. The veracity of fault diagnosis is guaranted effectively.
出处
《传感器与微系统》
CSCD
北大核心
2008年第12期94-96,105,共4页
Transducer and Microsystem Technologies
关键词
故障诊断
残差生成器
最小二乘回归型支持向量机
遗传算法
fault diagnosis
residual error generator
least square support vector regression （LSSVR）
genetic algorithm（ GA ）
