【作者】 杨雄飞；

【导师】 沈艳霞；

【作者基本信息】 江南大学 ， 电气工程， 2016， 硕士

【摘要】 风能作为一种重要的自然能源,本身具有清洁环保、能量巨大、可持续供应的特点,相比自然界的煤炭、石油、天然气等其资源优势更加明显,目前在缓解世界能源危机中占有重要地位。风力发电设备通常建在高山或远离海岸的偏远地方,气候变化不可预测,在这样高度恶劣、复杂的工作环境中,传感器、执行器故障发生频繁,再加上风机本身具有非线性、多变量、强耦合等特点,对系统控制将会更加复杂。因此对风力发电系统进行实时故障诊断,实施有效的容错控制是确保风力发电系统可靠运行的重要手段。论文对控制系统现有的故障诊断技术、容错控制策略以及近年来风力发电系统的容错控制方法进行了详细调研,从风力机的空气动力学效应出发介绍了风力机的最大风能利用效率及各子系统的数学模型,讨论了系统在三种不同风速区域内的运行状态。通过对性能参数叶尖速比、风能利用系数最优值的计算及二者曲线图的分析阐述了最大风能捕获控制原理,并进行了仿真分析。针对风力发电系统的非线性、变量之间强耦合特点,利用结构简单、逼近能力强的T-S模糊算法,建立系统全局T-S模糊模型。对滑模控制理论分析给出等效控制量的求解方法并证明滑动模态对外界不确定因素的鲁棒性。选取合适滑模面,利用系统的输出信号作为控制器的输入,基于LMI方法设计模糊滑模控制器用于风力发电系统闭环反馈控制,确保系统性能参数叶尖速比和风能利用系数维持在最优值附近,实现部分负荷区的风能最大捕获。考虑滑模观测器在对非线性系统进行故障诊断时,能够保持对外界扰动不敏感,具有强的鲁棒性,论文将T-S模糊算法与滑模观测器理论相结合,设计模糊T-S系统滑模观测器对传感器故障进行重构。然后对传感器输出信号进行校正,以校正后的传感器输出信号代替控制器输入,实现风力发电系统主动容错控制的目的。针对风力发电系统中执行器故障与传感器故障并存情形,利用传感器硬件冗余技术结合状态观测器,建立残差逻辑判断表,实现多故障检测;其次通过引入一个简单的滤波器,将传感器故障转化为执行器故障,建立一个由原有的执行器故障和传感器故障组成的虚拟执行器故障,通过对虚拟执行器故障的重构来实现两种故障同时重构;将滑模算法与自适应技术结合用于风力发电系统的执行器故障容错控制中,保证故障系统能够准确跟踪所设计的期望状态和输出轨迹,达到容错控制目的。更多还原

【Abstract】 As one of the most important natural resources, the wind energy has become an important role to solve the problems of word energy. It has an obvious advantage over other resources because the characteristics of clean, enormous energy, sustainable supply. Wind energy equipments are usually built in the mountains or remote areas away from the coast, and the climate change can’t be predicted. So the sensor and actuator fault often occur in this terrible and complex environment. In addition the wind turbine system is a high order, nonlinear, multivariable, strong coupling system, it will become more complex to control subject to uncertain parameters. Thereby, the real-time fault diagnosis and effective faulttolerant control to ensure the reliability of wind turbine system will become more and more important.The paper describes the existing fault diagnosis technology, fault-tolerant scheme of control system and fault-tolerant approaches of wind turbine system in detail. The maximum efficiency of wind energy is introduced based on wind turbine aerodynamic characteristics and the dynamic model of wind turbine system is constructed. Besides, the system operating state is discussed in three different wind speeds. Finally the largest wind energy capture control theory is introduced by calculating the performance parameters and analyzing the curve graph between tip speed ratio and power coefficient.According to the nonlinearity and the strong coupling characteristics among variables of wind turbine system, the global T-S fuzzy model of the system can be easily built because T-S fuzzy theory has simple structure and powerful approximation capacity. Besides, the equivalence output is solved and the sliding mode has strong robust to uncertainty by analyzing the sliding mode control theory. The paper chooses suitable sliding surface and utilizes the output signal of high speed shaft speed and electromagnetic torque as the input of controller. So the global fuzzy sliding mode controller can be designed by LMI approach to achieve the closed-loop of wind turbine system. The performance parameters of tip speed ratio and power coefficient are maintained around optimal value to capture largest wind energy in partial load area of wind turbine system.The sliding-mode observer is a nonlinear observer, it can use the inherent robustness to overcome model uncertainties or nonlinearity in fault diagnosis of nonlinear system. So the paper combines the T-S fuzzy theory and sliding-mode observer theory to design T-S fuzzy sliding-mode fault observer, and the fault information can be directly obtained by the equivalence output control concept to reconstruct the sensor fault. Then the input of sliding-mode controller can be replaced by reconstructing the output signal of sensor to contact the fault-tolerant control of wind turbine system.In senor fault and actuator fault system, a residual error logical table is constructed by combing the sensor hardware redundancy technology with state observer to achieve multiple-fault diagnosis. Then the sensor fault can be transformed into actuator fault by a first-order filter to build a virtual actuator faults. Thereby the two faults simultaneous reconstruction can be achieved by constructing the virtual actuator faults. Finally, the sliding-mode theory is combined with adaptive technology to design the actuator fault-tolerant controller to guarantee the system accurately track the expected state and output trajectory.更多还原