风力发电机主传动链的无传感器故障诊断方法研究

At present, there is a technology gap in china in the field of wind turbine condition monitoring and fault diagnosis. The reasons of it are the existing problems in classical vibration diagnosis such as large quantity of sensors, inconvenient in install, difficult in data transmission and acquisition, etc. It is not a system-wide and simple comprehensive diagnostic method. Therefore, based on analysis of generator current, a sensorless fault diagnosis method has been proposed. This mainly based on dynamic torque transfer model of drive chain, via building a dynamics and electrical mathematics model about wind wheel, drive chain and generator, researches electromechanical coupling response and fault diagnosis principle of fault information in generators current signal of the main component of generators drive system, analyzes cross-coupling effect of mechanical fault, electrical fault and control information in current signal, and verifies by simulation. On this basis, using the instantaneous admittance and EMD methods research feature extraction, identification and diagnosis methods of weak mechanical fault in current signals, explore the quantitative evaluation and prediction model of the integrated fault. The research of this project can develop and extend the theory and application of sensorless fault diagnosis method, and also can break through the bottlenecks of vibration diagnosis in wind turbine system. Thus, the study of this project has the important academic value and expansive application prospect.

目前，我国风力发电机组的状态检测与故障技术应用基本上还是空白，原因是常规的振动诊断方法存在着传感器数量多、安装不便、数据传输与采集困难等问题，缺乏一个面向整机的、简便易行的综合性诊断方法，为此，提出了一种基于发电机电流信号分析的诊断方法─无传感器故障诊断方法。以传动链的动态扭矩传递模型为主线，建立从风轮、传动链部件到发电机的动力学和电气数学模型，研究机组传动系统主要部件故障信息在电机电流信号中的机电耦合响应与故障检测原理，分析电流信号中机械故障信息、电气故障和控制信息的交叉和耦合作用，并进行模拟试验台验证。在此基础上，采用瞬时导纳法和EMD等方法，研究电流信号中的微弱机械故障特征提取、识别与诊断方法，探讨综合故障定量评价与预测模型。项目研究既可丰富和扩展无传感器故障诊断方法的理论及应用，也能解决振动检测方法在风电系统应用中遇到的技术瓶颈，因此具有重要的学术研究价值和广泛的应用前景。

项目采用无传感器诊断方法，针对DFIG风力发电机传动系统主要部件进行故障诊断研究。根据Hertz弹性体碰撞理论和齿轮动力学方程建立轴承和齿轮扭矩计算模型，并以扭矩产生的转速波动联立DFIG发电机电气模块，建立机电联合仿真模型并进行各种条件下的仿真分析；采用普通电动机和通用变频器，建立一个风力发电机组故障模拟实验台，对齿轮和滚动轴承等传动链部件的典型故障进行各种工况下的故障模拟试验与验证。仿真和实验表明：发电机定子电信号可以有效地响应机械故障的高低频信息，但其中耦合有大量的发电机转子励磁谐波频率成分，不易识别。提出了一种瞬时解调功率的预处理信号方法，抑制励磁谐波，突出故障频率成分；提出了调制单边带移频解调的故障特征提取技术，抑制轴承等高频冲击扭矩信号产生的电信号过调制现象。另外，对EMD和人工神经网络故障分类等高级信号处理方法也做了深入研究。在此基础上，研制了一台基于DSP的“电机电压、电流信号采集与诊断仪”，用于现场验证试验。申请发明专利2项(目前处于公示阶段)，已授权实用新型专利8项。发表期刊EI收录论文4篇；国际会议EI检索论文6篇，ISTP检索 4篇；已发表待检索IEEE国际会议论文3篇。培养硕士研究生8名。