风电齿轮箱微弱故障自适应多尺度随机共振增强检测方法研究

The wind turbine gearbox, as the core component for energy transfer of the widely-used wind turbines, possesses unique and complex structural compositions, resulting in critically complex vibrations which easily submerge the weak fault feature generated from the damaged key components. Thus, the fault feature characterization and detection are significantly important for system health monitoring, while still challenging in industries. This study starts with the systemic dynamic modeling for fault features characterization of the wind turbine gearbox. By considering the comprehensive effects of various factors such as multiple gear pair meshing vibration, time-varying transfer path and direction of meshing force, a precise dynamic model of the wind turbine gearbox is first established. Then, the damage mechanism of various typical faults of the wind turbine gearbox is estimated, which uncovers the mapping relationship between different fault states and corresponding dynamic characteristics. On this basis, the techniques for weak fault features extraction from the comprehensive vibrations measured on the turbine base are investigated. To address the shortcomings of traditional signal processing methods that are mostly based on filter operation or noise cancellation, the adaptive multiscale stochastic resonance enhancement method is proposed. Moreover, considering the nonlinear amplification in the stochastic resonance, the high-precision recovery method from the resonant outputs is studied to obtain magnitude-based fault diagnosis and realize adaptive and precious identification of weak faults in the wind turbine gearbox. This research could offer effective means for weak fault detection of the wind turbine gearbox, and provide important theoretical significance and engineering application guidelines.

风电齿轮箱作为风电机组能量传递的核心部件，其独特的结构组成和特殊的运动形式，导致其系统动态响应异常复杂且信息特征比较微弱，增加了风电齿轮箱故障诊断的难度。本项目首先从正问题风电齿轮箱传动系统的动力学建模入手，通过考虑多对齿轮啮合振动、时变传递路径及啮合力方向等因素的综合影响，拟构建精确的风电齿轮箱传动系统动力学模型，研究系统故障信号的产生机理，揭示系统不同故障状态与振动响应特征之间的映射关系。在此基础上，开展反问题风电齿轮箱微弱故障诊断研究，为克服基于消噪思想的传统信号处理方法在提取微弱故障特征时存在有损特征信息的不足，拟研究自适应多尺度随机共振增强提取方法；为解决随机共振导致系统输出响应非线性放大的问题，拟研究高精度随机共振输出信号恢复方法，从而实现强噪声下风电齿轮箱微弱故障的自适应精准识别。研究成果可为风电齿轮箱故障诊断提供有效的分析手段，具有重要的理论意义和工程应用价值。

风电齿轮箱作为风电机组能量传递的核心部件，其独特的结构组成和特殊的运动形式，导致其系统动态响应异常复杂且信息特征比较微弱，增加了风电齿轮箱故障诊断的难度。本项目通过正反问题相结合的方式，开展了风电齿轮箱传动系统动态特性分析和微弱故障增强提取方法研究。经过三年研究，取得以下成果：（1）建立了风电齿轮箱单级平行级传动系统动力学模型，揭示了系统裂纹故障信号的产生机理，发展和丰富了风电齿轮箱传动系统建模体系。（2）结合风电齿轮箱传动系统振动响应特性，提出了基于经验小波变换的多尺度噪声调节周期势随机共振，实现了信号中多尺度噪声信息的精确调整；构建了修正峭度指标，克服了多尺度噪声调节过程中对先验知识的依赖；提出了基于灰狼优化算法的自适应周期势随机共振，实现了风电齿轮箱传动系统微弱故障的快速提取。（3）提出了基于余弦拟合的随机共振恢复方法，实现了风电齿轮箱传动系统故障响应特征信息的精确获取。（4）开展了传动系统关键零部件实验室故障模拟实验和现场测试实验，对所提算法进行了相应的实验验证和工程应用。基于上述研究工作，发表国内外重要学术期刊和会议论文8篇，其中SCI论文6篇，申请发明专利2项。