基于模型去伪的风能转换系统鲁棒容错控制策略研究

Focus on improving the safety and reducing the cost of wind energy conversion systems(WECS), the robust fault tolerant control strategies with system uncertainties and exogenous disturbances for wind energy conversion systems is studied in this application. Considering the disturbance inputs, parameter uncertainties of output and system matrix, the dynamics of the WECS is analyzed, then the uncertain linear parameter variable(LPV) model with the wind speed as the scheduling parameter is built. Based on the typical faults in WECS and their characteristics, the corresponding set valued observers(SVO) are designed, which stem from the set membership theory. Model falsification technology is introduced in this study for fault detection and diagnosis with the SVOs designed, and only the input and output data are used. With the wind speed estimated, the fault detection information from the SVOs, the controller designing method will be studied, and some appropriate rules for robust fault tolerant controllers' scheduling will also be consideraed . After this study, some conclusions are drawn for the future reliablity improvements of WECS.

围绕提高风能转换系统安全性、降低其运行成本的主题，考虑由于风速随机性导致的系统不确定性和外部干扰等因素，研究风能转换系统鲁棒容错控制策略。在干扰输入、输出及系统矩阵参数不确定情况下，分析风能转换系统动态特性，建立以风速为调度参数的系统不确定线性参数变化（LPV）模型；分析系统中各类典型故障特性，采用集员估计理论设计故障集值观测器（SVO），基于模型去伪技术的思想，进行故障的检测与隔离；根据预测风速及SVO故障估计，按照离线设计-在线选择的方案，设计合理的鲁棒控制器调度算法，确保时变不确定性风能转换系统的闭环稳定性和动态性能。

为提高风力发电系统可靠性，降低成本，项目基于集值观测器，进行风力发电系统的故障诊断和容错控制策略研究。提出一种基于信息反馈和改进适应度评价的人工蜂群算法，用于优化小波神经网络的参数，实现风电功率及风速的区间预测。以风速为调度参数，建立系统的LPV模型；基于该模型，分析了各故障诊断的特性，实现了独立故障诊断；利用模型去伪的思想，设计SVO进行多故障的诊断与识别，并设计滑模鲁棒控制器实现了容错控制。项目资助下，相关研究成果共录用和发表学术论文18篇，其中CSCD刊物以上14篇（包括SCI 2篇，EI 6篇），会议论文4篇。申请发明专利5项，培养硕士研究生7名。