风力发电机组齿轮箱混杂故障智能综合辨识与复合诊断研究

The working conditions of wind turbines are often complex, and its fault diagnosis is still lack of effective techniques. The primary focus of this research project is bearings of gearbox in wind turbines. Combing with its unique operation and vibration characteristics, we strive to achieve the following breakthroughs and innovations about intelligent fault diagnosis of bearings, and establish a series of fault diagnosis technology: To solve the problems of the number of IMF iterations and mode mixing. We proposed a method which is used to improve EMD and extract characteristics of incipient fault; A fault classification and severity diagnosis of bearings based on EMD and divergence is proposed; A method which is applied to diagnose compound fault is also presented; The prospective result of this project is a integrated technology system which is used to diagnose different style and severity fault of bearings. It is significance to study fault diagnosis of gearbox bearings of wind turbine which can insure the safe and normal running of wind farm.

风力发电机组工作环境恶劣且工况复杂多变，其故障诊断尚缺乏有效的理论支撑和技术手段。本项目以风力发电机组齿轮箱为研究对象，结合其特有故障规律和振动机理，力求在齿轮箱智能故障辨识和复合诊断方面实现以下突破和创新，并建立一套相应的理论和技术体系：结合EMD及其优化方法，解决其迭代次数对经验值的依赖和模态混叠问题，实现轴承早期故障振动信号本质特征的高效精确提取；通过将其与散度指标相结合，提出一种轴承故障部位、类型及程度的精确综合辨识方法；针对齿轮箱存在多种故障混杂的情形，研究具有自适应性的核独立成分分析方法，将故障信号进行有效分解，实现复合故障的分析与诊断；针对齿轮箱润滑系统，采用研究基于D-S证据多元数据融合的方法实现风力发电机齿轮箱润滑系统的高效故障诊断。预期成果将形成较为系统的风电机齿轮箱故障辨识与诊断理论体系和方法，对保障我国风电场的安全、稳定、经济运行具有重要的理论研究价值和应用前景。

我国风力发电机组工作环境恶劣且工况复杂多变，其故障诊断尚缺乏有效的技术手段。本项目以风力发电机组齿轮箱为研究对象，结合其特有的运行规律和振动机理，力求在风力发电机组齿轮箱智能故障辨识和复合诊断方面实现以下突破和创新，并建立一套相应的技术体系：结合EMD及其优化方法，解决其迭代次数对经验值的依赖和模态混叠问题，实现齿轮箱早期故障振动信号本质特征的高效精确提取；通过将EMD（经验模态分解）方法与散度指标相结合，提出一种轴承故障部位、类型及程度的准确辨识方法；针对齿轮箱复合故障情形，提出了两种有效的盲源分离算法——基于量子粒子群优化的EMD-KICA（核独立成分分析）算法和EMD-SCA（稀疏分量分析）算法，这两种算法均能有效解决盲源分离中故障源数目未知的问题，以及欠定盲源分离这一技术难题。以上解决方案均被仿真信号、实验平台测试数据以及风场实测数据验证其有效性。本项目成果形成较为完整的风力发电机组齿轮箱轴承故障诊断技术体系和方法，符合该领域的理论研究趋势和工程实际，对保障我国风电场的安全、稳定、经济运行具有重要的现实意义。