基于多尺度排列熵和极限学习机的风机叶片覆冰故障检测方法*

针对风机叶片结冰故障检测中状态数据维度高和检测率低的问题，提出1种使用功率数据驱动的多尺度排列熵（multiscale permutation entropy,MPE）和极限学习机（extreme learning machine,ELM）的风机叶片结冰故障检测方法。首先，使用多尺度排列熵提取功率数据的多重尺度特征，得到特征向量；随后，采用极限学习机，结合环境温度，对结冰故障进行检测；最后，通过使用某风电场的数据采集与监视控制系统（supervisory control and data acquisition,SCADA）对数据进行仿真。研究结果表明:所提方法的故障检测率达到100％，同时虚警率仅有0.14％，表明所提方法在风机叶片的覆冰故障检测中的有效性。研究结果可为风机叶片覆冰故障检测提供1种有效方法。

Detection method on blade icing fault of wind turbine based on multi-scale permutation entropy and extreme learning machine

Aiming at the problems of high dimension of state data and low detection rate in the detection on the blade icing fault of wind turbine,a detection method on the blade icing fault of wind turbine based on multi-scale permutation entropy (MPE) and extreme learning machine (ELM) driven by power data was proposed.Firstly,the multi-scale features of power data were extracted by MPE to obtain the feature vectors.Then,ELM was used to detect the icing fault in combination with ambient temperature.Finally,the simulation was conducted by using the data from the supervisory control and data acquisition (SCADA) of a wind power plant.The results showed that the fault detection rate of the proposed method reached 100%,and the false alarm rate was only 0.14%,which verified the effectiveness of the proposed method in detecting icing fault of wind turbine blades.The results provide an effective method for the icing fault detection of wind turbine blades.