大功率风力发电机多层复合绝缘在重复脉冲电压下树枝化击穿的机理与绝缘缺陷的诊断方法

The insulation systems of large capacity wind turbine generator are usually designed to be a multi-layer structure combining variety of materials，which withstand repetitive impulse voltage during operation. The premature electrical breakdown and damage is closely related to the electrical treeing phenomenon initiated by insulation defects. In this research project, insulation defects in newly manufactured and failed windings of wind turbine generator will be analysed and summarized. A series of laboratory tests on modeled specimens will be carried out in order to investigate the causes and microscopic morphology of the electrical tree in multi-layer composite insulation system under repetitive impulse voltage. Numerical simulation of electrical tree propagation will be executed with combined Cellular Automata and Finite Element approach. A comparative study will be started aiming to find out how different factors, such as insulation materials and structures, electrode forms, types of defects, voltage levels and recurring frequencies of the applied repetitive impulse, will influence the electrical tree initiation and growth in multi-layer composite insulation system. To realize the effective diagnosis, identification and location of typical inter-turn insulation defects, a set of method based on external impulse excitation and electromagnetic induction is to be developed and established. To be specific, a kind of non-contact electromagnetic induction sensor suitable for discharge detection due to insulation defects in wind turbine generator coil should be designed, wavelet analysis of the obtained signal is to be applied for feature extraction and the pattern recognition will be fulfilled by artificial neural networks. The research achievement of this project will enrich the electrical tree breakdown theory in composite dielectric under repetitive impulse voltage and make a difference on improving the quality of wind turbine generators, and ensuring the safety and reliability of their operation.

大功率风力发电机绝缘系统具有多种材料复合和承受重复脉冲电压的特点，其绝缘过早电击穿破坏与绝缘缺陷及其引发的电树枝化密切相关。本项目首先剖析现有制造的和发生故障的风力发电机绝缘缺陷，实验室模拟试验研究多层材料复合绝缘体系在重复脉冲电压下的电树枝化的致因及破坏通道的微观形态，并应用元胞自动机方法数值模拟电树枝的微观演化过程及规律，对比研究不同绝缘材料和结构、电极形式、缺陷类型、脉冲电压水平和重复率等因素对复合绝缘体系中电树枝引发和生长的影响，研究基于外部激励和电磁感应实现匝间绝缘缺陷诊断和检测的方法，设计适宜于风力发电机绕组缺陷放电等信号的非接触式电磁感应传感器；应用小波分析对检测的缺陷特征信号进行提取分析，并采用人工神经网络进行模式识别，以实现风力发电机典型绝缘缺陷的有效诊断、辨识和定位。研究成果将丰富复合电介质的电树枝击穿理论，对提高风力发电机的品质，保证其运行安全及可靠性有重要的意义。

本项目分析了大功率风力发电机绝缘的故障特征，建立了在重复脉冲电压和工频电压下电树枝化显微实时观测试验平台，制备了多种具有层状屏障式结构复合绝缘体系的试样。观测了各种复合绝缘试样在不同幅值的工频和不同重复率及波形的重复脉冲电压下电树枝引发、生长、界面延伸和击穿的微观特征，对比研究电压幅值、电压波形、脉冲重复率、复合材料构成等条件因素对电树枝形态及发展的影响。研究表明，随着工频电压幅值的升高，电树枝的枝状分支减少、引发时间缩短、漆基中径向和绝缘带界面横向的生长率加快，相比于工频电压，在重复脉冲电压下电树枝更易于引发和生长，脉冲电压的重复率升高会促进电树枝的引发和生长，重复脉冲电压波形中的陡上升沿尖峰幅值会影响电树枝的引发和生长，而均值则会影响试样的击穿时间。确立了感应脉冲法以及基于此方法设计的绕组匝间绝缘局部放电检测的试验装置，研究了匝间局部放电信号在电机绕组中的传输衰减特性，试验对比研究了不同元件对匝间感应脉冲电压幅值的影响。研制了适宜风力发电机绕组匝间缺陷或故障检测的高频电流传感器，通过对其进行三维有限元仿真以及实例分析，验证了其良好的传变特性。进行了人工匝间绝缘缺陷的设计和真机绕组上绝缘缺陷的验证试验，研究了电机绕组匝间典型气隙型和电导型绝缘缺陷的识别和诊断方法，并建立了连接变频器的风力发电机匝间缺陷诊断流程。研究结果表明基于小波分解的诊断方法可以有效识别电导型匝间缺陷和匝间气隙型放电缺陷。分析了双馈型风力发电机现场运行中的绝缘缺陷特征，研究了适宜于风电场现场的发电机绝缘缺陷检测技术，开发了相应的绝缘缺陷现场检测装置。研究成果有助于风力发电机在生产和运行中绝缘缺陷的诊断，对保证其安全稳定运行具有重要意义。