中文摘要：

针对风电机组齿轮箱传统故障诊断方法以全局误诊断率最小化为目标,忽略了误分类型之间的差别的问题,提出基于代价敏感最小二乘支持向量机（Cost-sensitive Least Squares Support Vector Machine,CLSSVM）的风电机组齿轮箱故障诊断方法。该方法在最小二乘支持向量机原始最优化问题中二次损失函数中嵌入不同样本的误分类代价,建立以误分类代价最小化为目标的CLSSVM故障诊断模型,并同最小二乘支持向量机和代价敏感支持向量机比较。实验结果表明,该方法能提高误分类代价高的故障类样本的诊断正确率,具有代价敏感性,其训练速度也足以满足风电机组齿轮箱故障诊断实时性的需求。

英文摘要：

Aiming at problems of the traditional fault diagnosis method, a novel fault diagnosis method of wind turbine gearbox based on cost-sensitive least squares support vector machine（CLSSVM） is proposed in the paper, which takes the minimizing global error diagnosis rate as the goal and ignores the differences between misclassification types. In this method, the misclassification costs of different samples are embedded into quadratic loss function of primal optimization problem of least squares support vector machine. CLSSVM fault diagnosis model, which is aimed at minimizing misclassification costs, is set up. Compared support vector machine with CLSSVM, experimental results show that CLSSVM tends to improve the gearbox fault recognition rate of high misclassification cost, which has cost sensitivity, meanwhile the diagnosis speed meets the real-time requirements of wind turbine gearbox fault diagnosis.