海上风力发电机组智能健康诊断与机会维护技术

Wind power development has been shifting from onshore to offshore and offshore wind power is one of the most important renewable energy sources in the world. Due to the terrible environment at sea, the maintenance of offshore wind turbines is very difficult and the failure rate is very high. Along with the increase of the electricity market share proportion of offshore wind power and individual generating unit, the safety, reliability and stability of system operation is now becoming more and more important. As a cross discipline, health diagnosis and opportunistic maintenance for offshore wind turbines is one of the important research fields. The research springs up the gradual diagnosis method when the fault has occurred and focuses on intelligent health diagnosis of offshore wind turbines based on data, fault mechanism analysis and experiential knowledge and opportunistic maintenance technique based on intelligent state evaluation. Allowing for the characteristic of offshore wind turbines, several problems are studied on emphases, and the contents are as following: data pre-processing under complicated operating conditions, operating characteristic description under variable speed conditions, intelligent state evaluation of offshore wind turbines based on operational condition recognition and preventive opportunistic maintenance technique considering accessibility . Thus advanced practicable key technology of health diagnosis and opportunistic maintenance are developed. It can improve the reliability of wind turbines and reduce failure rates and maintenance costs. Allowing for the characteristic of offshore wind turbines, several problems are studied on emphases, and the contents are as following: big data pre-processing under complicated operating conditions, operating characteristic description under variable speed conditions, intelligent state evaluation of offshore wind turbines based on operational condition recognition and preventive opportunistic maintenance technique considering accessibilty. Thus advanced practicable key technology of health diagnosis and opportunistic maintenance are developed. It can improve the reliability of wind turbines and reduce failure rates and maintenance costs.

风电开发正逐渐由陆地向海上延伸，海上风电已成为世界可再生能源发展的热点。海上气候环境恶劣，机组维护困难，故障率高，随着海上风电比例的上升和单机容量的增大，对系统安全、可靠、稳定运行的要求越来越高。健康诊断和机会维护技术是近年来风力发电与系统工程交叉学科中发展迅速的重要研究领域之一。本项目突破以往机组故障发生后再诊断的处理方法，着重研究海上风电机组运行中基于数据、故障机理分析和经验知识的智能健康诊断以及基于运行状态健康评价的机会维护技术。为此，针对海上风电机组的特点，集成多学科理论方法和技术，在面向海上风电机组复杂工况数据预处理、变速环境下运行特征信息的提取与识别、基于运行工况辨识的海上风电机组健康状态智能评价以及考虑可及性的海上风电机组预防性机会维护技术等方面探索新方法，研发先进、实用的海上风电机组健康诊断和机会维护关键技术，在提高风机安全性的同时，有效降低海上风电的成本。

本项目紧密结合海上风电场的运行维护现状，从提高海上风力发电机组的安全性、降低其运行维护成本的需求出发，综合应用人工智能、信号处理、建模与模式分类识别等技术与方法，研究开发海上风电机组智能健康诊断和机会维护等关键技术。具体内容包括：（1）研究了面向复杂工况的海上风电机组数据预处理，构造了基于多结构元素的多尺度形态滤波方法，考虑结构元素形状和尺度的影响，用信噪比和偏斜度构建判别消噪效果的指标。将核主元分析法与正交化局部敏感判别分析相结合进行海上风电机组大数据降维，并引入差分优化算法对KPCA中的核参数进行优化处理。（2）研究了基于改进微分经验模式分解和独立分量分析的海上风电机组早期故障特征提取，利用改进的微分经验模式算法将原始振动信号分解成若干个独立的IMF信号，结合ICA进行原始振动信号故障特征分量的提取。基于变分模态分解和排列熵进行信号特征提取，组成多尺度的复杂性度量特征向量，并将高维特征向量输入基于支持向量基建立的分类器进行故障识别分类。（3）结合类别可分性，利用KPCA进行特征提取后，应用模糊C-均值聚类建立运行工况分类模型，针对每一个工况子空间建立相应的基于高斯混合模型的机组运行健康状态评价模型，定义健康度为评价指标，量化运行健康状态的评价结果，并设定自适应报警阈值。（4）采用小波变换提取表征不同风速规律的特征信号，并分别利用自回归滑动平均模型和用果蝇算法进行改进的支持向量机对风速进行预测。结合海上风电机组的风速数据，通过粒子群算法对可及性松弛因子进行优化，得到各部件的机会维护阈值。. 通过本项目的研发，发表和录用了19篇学术论文，申请了2项国家发明专利，培养了5名硕士研究生。本项目已按要求完成了相关研究内容，达到了预期的研究目标。