支持向量回归参数估计在风电机组故障模式分析中的应用
Application of Support Vector Regression Parameter Estimation to Fault Mode Analysis in Wind Turbines

风轮系统的可靠与否直接影响着风电机组的安全运行，有必要对其建立合理的可靠性模型并准确估计模型参数，以反映其真实可靠性情况。威布尔分布模型被广泛应用于各领域的可靠性建模，支持向量回归机（SVR）保持了支持向量机适用于小样本的特性，可用于小样本数据可靠性模型的参数估计。以某个风电场投运以来的风轮系统故障数据为基础，建立其威布尔分布模型，采用SVR估计模型参数，并与传统的基于最小二乘法的参数估计结果对比，结果表明，采用SVR估计模型参数具有更高的准确性，更适合于小样本数据可靠性模型的参数估计。
Reliability of rotor systems directly affects safe operation of the whole wind turbines. Therefore, it's necessary to establish reliability model reasonably and estimate its parameter accurately, to reflect the real reliability of rotor systems. Weibull distribution is widely used to reliability modeling in various fields. Support vector regression (SVR) maintaining the feature, suiting for small samples, the support vectors machine can be used to estimate reliability model parameters under small sample data. Based on rotor system fault data from a certain field since putting into operation, the reliability model of which is established, parameters of model are estimated by SVR, the estimating result is compared with that of traditional least square parameter estimation, and the final result indicates that the parameter estimation method based on SVR is more accurate, and more suitable to small sample data