【摘要】 该文利用Bladed软件模拟叶片覆冰和破损故障，通过对比和分析风电机组叶片故障与正常时的运行数据，发现叶片故障状态时的机组运行参数变化特征；然后利用叶片振动信号，基于叶片工作模态分析理论识别出叶片模态参数，根据模态参数的变化揭示了两种故障对叶片振动影响的区别；最后将所识别的叶片模态参数与风电机组的运行参数组成多源数据，采用LightGBM框架下的分类决策树算法实现了对叶片故障状态的有效判断和识别。更多还原

【Abstract】 Blade icing and damage faults were simulated by the software Bladed in this paper. By comparing and analyzing the operation data of wind turbine under blades normal and fault conditions,the change characteristics of operation data under the condition of blade failure was found. Then,the blade vibration signals were used to identify the blade modal parameters based on the blade operational modal analysis theory. According to the variation of blade modal parameters,the fault difference between blade icing and damage on blade vibration was revealed. Finally,the identified blade modal parameters and wind turbine operation parameters are combined into multi-source data. The classification decision tree algorithm under the framework of LightGBM(Light Gradient Boosting Machine)is used to identify the blade fault state effectively.更多还原