基于VMD-HT-ResNet的风电机组塔筒振动状态监测及预警  

VMD-HT-ResNet-BASED VIBRATION CONDITION MONITORING AND EARLY WARNING FOR WIND TURBINE TOWER

作　　者：胡阳 李博 胡耀宗 付道一 胡号朋 Hu Yang;Li Bo;Hu Yaozong;Fu Daoyi;Hu Haopeng(State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources(NCEPU),Beijing 102206,China;China Shipbuilding Group Haizhuang Wind Power Co.,Ltd,Chongqing 401123,China)

机构地区：[1]新能源电力系统全国重点实验室(华北电力大学),北京102206 [2]中国船舶集团海装风电股份有限公司,重庆401123

出　　处：《太阳能学报》2024年第12期388-396,共9页Acta Energiae Solaris Sinica

基　　金：国家海上风力发电工程技术研究中心开放基金(HSFD22002)。

摘　　要：针对大型风电机组塔筒多模态振动难以可靠监测及预警的问题,提出一种变分模态分解(VMD)-希尔伯特变换(HT)-残差神经网络(ResNet)的风电机组塔筒振动非线性动态多输入多输出(MIMO)建模方法,分解并提取塔筒固有模态振动,定义精准表征塔筒振动特性的差分动态回归向量,并进行全工况均衡采样和MIMO建模。然后,建立指数加权移动平均(EWMA)指标的塔筒健康度评价机制,用于其状态监测和预警。仿真结果表明,所提方法可实现风电机组塔筒全工况振动特性的高精度监测和预警,可为机组高安全运行提供保障。Aiming at the problem that it is difficult to reliably monitor and warn the multimodal vibration of large wind turbine tower,a multi-input and multi-output(MIMO)nonlinear dynamic modelling method combining variational modal decomposition(VMD)-Hilbert transform(HT)-residual network(RestNet)is proposed to model the nonlinear dynamic characteristics of tower vibration of wind turbines.The inherent modal vibration of the tower is decomposed and extracted,and the differential dynamic regression vectors that accurately characterize the vibration characteristics of the tower are defined,and balanced sampling and MIMO modelling are carried out for the whole working conditions.Then,an exponential moving average(EWMA)index is established to evaluate the health of the tower,which is used for its condition monitoring and early warning.Simulation results show that the proposed method achieves high-precision monitoring and early warning of the tower vibration characteristics of wind turbines under all operating conditions,which provides a guarantee for the high safety operation of the turbines.

关 键 词：风电机组 塔筒 变分模态分解 差分动态建模 残差神经网络 健康度评价 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]