近年来，风电产业迅猛发展，风机的额定功率、机舱重量、塔架高度和叶片长度不断增加，风机的风致振动问题更为突出。本文对1.5MW风机塔架进行了长期监测，对不同工况下风机结构的振动响应和疲劳损伤进行研究。论文的主要内容如下： (1)开发了一套振动监测系统，基于风机的振动监测数据和SCADA(Supervisory Control and Data Acquisition)数据，对风机所在地的实测风特性和风机的运行状态、振动特征进行统计分析，掌握了运行状态参数之间的关系，及不同运行工况下风机的振动规律，并详细分析了风机结构在典型工况下的振动机理与变化规律。 (2)基于数据驱动的随机子空间法对风机结构的模态参数进行识别，研究了不同工况下风机结构自振频率和阻尼比的规律；绘制了风机结构不同阶次自振频率与叶轮转频及过桨频率关系的Campell图，本文风机的基频虽然避开了 3f频率带，但没有满足DNVGL规范规定的预留10%安全度的要求；当风机启动经过并网转速时，振幅明显加强，风机停机经过并网转速时，振幅变化不明显，基于模态特性和Sommerfeld效应对该现象进行解释。 (3)基于风机的实测应变数据，研究了不同风速下塔顶等效风荷载及叶轮推力系数的变化规律；建立了风机结构的数值模型，利用指数模型模拟平均风速，基于Davenport谱模拟脉动风速，根据推力系数法计算了作用在风机上的风荷载，进行了动力响应分析，将其与实测值进行对比，验证数值模型的合理性。 (4)利用不稳定的风剖面计算了台风的平均风速，基于J.H〓jstrup谱及谐波叠加法模拟了台风的脉动风速，计算了台风作用下风机的动力响应，台风风向的偏转增加了风机的迎风面积，风机的动力响应也随之增加，风向偏转90°是最为不利的工况；利用M-P谱描述降雨粒子的谱分布，基于动量理论，模拟了作用在风机上的雨荷载，分析了不同降雨强度对风机动力响应的影响，同等风况下，风机的动力响应随着降雨强度的增加而增加；在台风环境下，建议改进风机的控制策略，并选用S型抗台风风机。 (5)提出了基于实测数据的疲劳寿命评估流程。采用精细化局部有限元模型，计算了法兰焊缝处的应力集中系数，将监测应力转化为焊缝处的热点应力；采用雨流计数法和S-N曲线，计算了监测期间每lh的损伤，根据年风速-风向的联合分布，建立了lh最大和平均损伤矩阵；基于P-M损伤准则，推算出法兰焊缝处的年最大和平均损伤矩阵，对疲劳寿命进行分析；同时，研究了不同运行工况对疲劳损伤的影响，疲劳损伤随着风速和叶轮转速的增大而增大，50%以上的损伤都是在风速大于额定风速、转速为额定转速时发生的。 关鍵词：风机；振动监测；模态分析；动力响应；疲劳分析

 In recent years, the wind power industry has been growing rapidly worldwide. Alongside, with the increasing of the rated power, nacelle weight, tower height and blade length of wind turbine, the wind-induced vibration problem is more prominent. In this paper, the vibration of a 1.5 MW turbine tower was monitored for over one year. Based on the vibration data and SCADA data, vibration response and fatigue damage of the wind turbine structure under different operating conditions were studied. The primary research results of this dissertation are as follows: (1)A vibration monitoring system has been developed. Based on the vibration monitoring data and SCADA data, statistical analyses were performed on the wind characteristics of the area where the turbine is located, the operating states as well as vibration characteristics of the wind turbine. Relationship among the operating state parameters, and vibration levels under different operating conditions of the turbine were studied. The vibration mechanisms and variation rules of the turbine structure under typical operating conditions were analyzed in detail. (2)The modal parameters of the wind turbine were identified utilizing a data-driven stochastic subspace identification method. The rules of natural frequencies and damping ratios of the structure under different operating conditions were studied. Campbell diagrams were plotted to illustrate relationship between different orders of natural frequencies and blade passage frequencies. Although the fundamental frequencies of the turbine in this paper avoided the 3f frequency band, they didn't meet the requirement of 10% safety allowance set by DNVGL. Obvious vibration amplitude at turbine startup was noted as it surpassed the rotation speed of grid connection, but was not apparent at turbine shutdown. The characteristics of different vibration phenomena in the turbine during startup and shutdown processes were explained with modal characteristics and the Sommerfeld effect. (3)Based on the measured strain data of the turbine, equivalent wind loads on the tower top and thrust coefficients of the rotor under different wind speeds were studied. A numerical model of the wind turbine was established. The exponential model was used to simulate the average wind velocity, and the fluctuating wind velocity was simulated based on the Davenport spectrum. The wind loads on the turbine were calculated according to the thrust coefficient method. The dynamic analyses were carried out, and the results were compared with the measured data to verify the rationality of the numerical model. (4)The typhoon average wind velocity was simulated according to the unstable wind profile, and the fluctuating wind velocity was simulated based on the J. H〓jstrup spectrum and harmony superposition method. Dynamic responses of the turbine under the action of typhoon were calculated. The deflection of typhoon wind direction has increased the windward area of the blades, and the dynamic responses of the turbine have also increased. The wind direction deflects 90° after the blades feathering was the most unfavorable condition. The raindrop size distribution was simulated by M-P spectrum, and the rain loads on the turbine were calculated according to the momentum theorem. The dynamic responses of turbines under typhoon-rain loads were also analyzed. Under the same wind condition, the effect of rain loads on the dynamic responses of the turbine increase with the rainfall intensities increasing. Under typhoon environment, it is suggested to improve the control strategy of the wind turbine, and use an S type turbine. (5)A fatigue assessment method for wind turbine based on measured strain data was developed. The stress concentration factor of flange weld was calculated by a local refinement finite-element model, and transformed the monitoring stress to hot spot stress at the weld toe. Rain-flow counting method and S-N curve were used to calculate the damage from the 1-hour recorded time series during monitoring period. According to joint distribution of wind speed and wind direction, the maximum and average damage matrices were established. Based on the P-M damage criterion, the total damage was calculated and the fatigue life was analyzed. At the same time, the influence of different operating conditions on fatigue damage was studied. The fatigue damage increases with the increase of wind speed and blade rotation speed. Over 50% of the damage occurs at normal operation above the rated speed. Key words: Wind turbine; Vibration monitoring; Modal analysis; Dynamic Response; Fatigue analysis