论文部分内容阅读
柔性支持向量回归方法(F-SVR)的优势是在回归过程中无需对传统的三个参数进行设定,但是该方法的不利因素是要引入一个被称作“区间数”的数据分类参数.首先,根据提出的改进型支持向量回归方法,采用线性判别分析方法(LDA)计算“区间数”,将故障数据集有效划分,建立每一种故障的SVR模型,设定故障阈值.当需要判断新信号正常与否时,直接通过阈值,进行故障诊断.然后,利用该方法对一个典型快变系统—风力发电系统的故障模式进行了测试,结果表明:改进型SVR算法对2 MW的双馈风力发电机组并网系统中的三相短路故障和逆变器开路故障的故障漏诊率低于任意设置“区间数”的方法.“,”In this study, the flexible support vector regression (FSVR) and linear discriminate analysis (LDA) are combined to perform the fault diagnosis. The method F-SVR divides the training sample dataset into several “domains” according to the distribution complexity, and generates different parameter sets for each domain. In the study, the number of “domains” is calculated by LDA automatically, in which the Fisher criterion is used to realize the division of the data sets. In the case study, a 2 MW grid-connected doubly-fed wind turbine system is established to simulate the short-circuit faults of grid side and the inverter open-circuit faults of rotor side. Compared to the random “domains” selection method, the proposed method has lower missing faults ratio.