模拟退火算法与支持向量机在机械故障诊断中的应用

The Application on the Simulated Annealing Algorithm and the Support Vector Machines in Mechanical Fault Diagnosis

为提高支持向量机在机械故障诊断测试中的分类正确率,将模拟退火算法与支持向量机相结合,用模拟退火算法优化支持向量机核函数及其参数,再将故障特征输入支持向量机进行故障识别.诊断实例表明,该方法与传统支持向量机方法相比能得到较高的诊断精度.

To improve the classification accuracy of the support vector machine （SVM） in mechanical fault diagnosis, the simulated annealing algorithm and the support vector machine （SVM） is combined. First, the kernel parameters for SVM are optimized by using the simulated annealing algorithm. Then the fault feature is inputted into the support vector machines with the best kernel parameters for fault identification. The experimental result shows that the method, compared with the traditional support vector machine, can obtain higher diagnosis accuracy with fewer features.