双馈式风电机组同轴度误差故障机理与诊断技术研究

Fault diagnosis and prediction with analysis of failure mechanisms play an essential role to ensure safe and economic operation of wind turbines. The accumulation of eccentric error between the high speed shaft of gearbox and rotor shaft of generator has been found to be an important factor, which causes damage to some key elements in gearbox and generator. However, there is not a systematic research in this area so far. Therefore, by considering the coupling relationship between gearbox and generator, this research will use three-dimensional drawing, finite element analysis, Adams simulation to study the eccentric error fault, analyze the failure mechanisms caused and explore the evolution rules. It will create hybrid simulation model for wind turbines combined with Simulink software to simulate fault evolution process from initiation to small and large scratch. It will analyze the changes of the generator operating characteristics. It will adopt intelligent signal analysis and processing methods, use the homologous information fusion techniques to effectively separate the potential fault characteristic signals from the electrical signals.This research will carry out an in-depth study of new potential fault diagnosis methods and combined prediction methods using the support vector machines associated with other intelligent methods. Finally, this research will design the eccentric error simulation platform for wind turbines with doubly-fed induction generator (DFIG), to validate the models and algorithms developed for diagnostics and prediction by adding temperature and other relevant factors in order to provide useful information including models and algorithms for condition based maintenance and operation of wind turbines.

风电机组故障机理分析、故障预测和诊断方法的研究是保证机组安全经济运行的关键。齿轮箱高速轴与发电机转子轴的同轴度误差累积是造成齿轮箱和发电机两大主要部件损坏的重要因素。但目前对这方面的研究还不是很系统。因此，本课题将综合考虑各部件之间的耦合关系，使用三维制图、有限元分析、Adams等软件研究风电机组同轴度误差故障机理，探索其演化规律；结合Simulink建立风电机组的潜伏性故障联合仿真模型，模拟故障从无到有，从小到大的演变过程，分析发电机运行特性的变化规律；采用智能的信号处理方法，运用同源信息融合技术从各电气信息中有效分离潜伏性故障特征信号；再以支持向量机为工具，并结合其他智能方法，有针对性地探索新的潜伏性故障诊断和组合预测方法；最后，设计双馈风电机组同轴度误差仿真实验平台，验证预测和诊断算法，并增加温度等信息，进一步改进模型，为风电机组基于状态的维修和运行奠定基础。

随着传统能源的减少以及环境污染日益严重，风能作为一种清洁的可再生能源得到了快速发展。目前，双馈式风力发电机组是主力机型，当其传动系统的同轴度误差超过一定范围时就会造成不对中故障，导致主要部件损坏。因此，对风力发电机组传动系统同轴度误差进行故障机理分析、故障预测和诊断方法的研究是保证机组安全经济运行的关键。本项目采用仿真模拟和实验分析相结合的方法来研究双馈式风电机组同轴度误差。首先使用SolidWorks和ADAMS 软件建立了正常的风电机组仿真模型，通过修改Marker点的方式来实现各种不对中故障的模拟，并结合Ansys Workbench，研究了风电机组同轴度误差的故障机理，分析了速度、加速度、温度场和应力等随着不对中程度的演化规律；其次结合Simulink实现了风电机组的联合仿真，模拟了潜伏性故障的演变过程，探索了发电机的定子电流在不同风速下按同轴度误差由小到大逐渐变化的规律，并结合Ansoft Maxwell，对不同情况下的电磁场变化规律进行了分析；然后分别基于振动信号、电流信号以及它们与温度的融合信号进行了时域、频域和时频域特征提取和故障诊断算法的研究，采用最小二乘支持向量机进行了齿轮箱高速输出轴的温度预测，采用组合预测模型进行了风电机组振动和电流特征的预测；最后搭建了模拟双馈风电机组的同轴度误差实验平台，通过采集的振动和电流数据验证了前面提出的潜伏性故障诊断和预测方法的准确性，为风电机组的状态维修提供了一定的理论依据。