基于改进深度森林算法的风电机组故障诊断技术研究

Research on fault diagnosis of wind turbine based on improved deep forest algorithm

为了及时有效地检测出风电机组发生的具体故障,同时克服传统故障诊断方法的局限性,文章提出一种基于改进深度森林算法的风电机组故障诊断方法。首先,利用有效的数据预处理方法处理SCADA原始数据并提取故障特征;然后,基于深度森林算法对风电机组具体故障进行诊断,同时,针对深度森林算法在故障诊断领域存在的缺陷,对算法提出改进;最后,利用河北某风场1.5 MW风电机组实际运行数据对文章提出的故障诊断算法进行验证,通过正确率、AUC等指标验证了所提故障诊断算法相比传统机器学习算法的有效性和优越性。该研究为风电机组运行和维修提供了依据,同时也为故障诊断领域提供了新的方法和思路。

In order to diagnose the wind turbine fault quickly and effectively and avoid the limitations of the traditional fault diagnosis method, this paper proposes a wind turbine fault diagnosis method based on improved deep forest algorithm. Firstly, the original SCADA data are processed using an effective data pre-processing method and get the features of the fault. Then,deep forest algorithm is used to fault diagnosis. In the meantime, an improved deep forest is put forward for the defects of deep forest algorithm in the field of fault diagnosis. The algorithm is used to diagnosis two specific faults of a 1.5 MW wind turbine in Hebei wind farm. Compared with traditional machine learning algorithm, the validity and superiority of the proposed fault diagnosis algorithm are proved by the indices such as accuracy,AUC. The research provides a basis for the operation and maintenance of wind turbines, and also provides a new method and ideas for the field of fault diagnosis.