基于核化局部全局一致性学习的提升机故障诊断

Hoister Fault Diagnosis Based on Kernel Learning with Local and Global Consistency

典型的局部全局一致性学习(Learning with Local and Global Consistency,LLGC)是基于图的半监督学习算法,虽然可以对样本进行有效标注,但对非线性数据却无能为力,且会出现维数灾难现象。为此,在LLGC的基础上引入核函数,提出核化局部全局一致性学习(KLLGC)解决上述问题。提升机故障诊断的实验结果表明KLLGC的有效性和可行性。

The typical LLGC （Learning with Local and Global Consistency ） is a semi-supervisory learning algorithm based on the charts. Although, it is able to effectively label the samples, but helpless for nonlinear data and even may fall into the dimension disaster. Thus, the Kernel function is introduced into LLGC to solve the problem. The test results of hoister fault diagnosis show that the KLLGC （ Kernel Learning with Local and Global Consistency） algorithm is available and feasible.