【作者】 任海军；

【导师】 何玉林；

【作者基本信息】 重庆大学 ， 机械制造及其自动化， 2011， 博士

【摘要】 本文以风力机控制系统为研究对象,将功率、载荷控制作为目标,提出了相应的控制策略。在分析建立风力机模型的基础上,对风力机在额定风速以下的最优功率控制、在额定风速以上的变桨控制、功率预测控制以及整机的载荷减小等策略进行了深入的研究。运用提出的控制策略和方法,对风力机控制系统进行了仿真设计。全文的主要研究内容为:第1章:介绍了课题的研究背景,阐述了研究意义,对风力发电机组研究现状进行了归纳分析,尤其是对控制系统的策略和算法进行了详细的分析,在此基础上介绍了论文的主要内容和组织结构。第2章:根据风力机动态入流理论建立非定常入流模型,详细分析了采用动态入流理论建立风力机模型的合理性,指出了尾流对风力机动态特性的影响,对非定常流中的加速势方法进行了调整。以von Karman理论建立湍流风模型,得到系统需要的三维风场。将Matlab仿真模型和Bladed软件设计分析结果进行对比,验证本文所建风力机空气动力学模型的正确性。第3章:根据风力机运行特点,将其控制阶段分为三个区域,按照区域的不同设计相应的控制策略,重点研究了全负荷最优控制阶段的控制策略,实现控制环的解耦,确保控制平滑过渡。由于风力机功率控制系统具有延迟特性,以系统特性为基础,结合预测控制方法,减小系统延时,提高系统控制性能。分析了线性状态空间模型预测控制算法,结合风力机特征,设计系统模型预测控制器;将模型预测控制器分为干扰模型和估计器、目标计算以及动态最优等三个模块;分析了各个模块的特点并建立数学模型。采用仿真的方式验证了模型预测控制在风力机功率控制方面的优越性。第4章:分析了风力机线性化方法,根据本章的研究重点,选择3状态风力机线性模型作为控制对象。对单神经元的特点进行分析,推导了单神经元数学模型,并对算法进行改进,得到改进的单神经元数学模型,采用改进单神经元控制算法对常规PID进行在线参数调整,得到基于改进单神经元的自适应PID控制器,并将其应用到变桨系统的控制中,比较分析了改进单神经元自适应PID控制器和常规PID控制器的控制性能。第5章:建立了风力机塔架/转子运动学方程,设计了塔架前后向和侧向阻尼滤波器,增大其阻尼,减小振动。增大传动链主动阻尼并屏蔽桨叶穿越频率,避免塔架和转子发生共振;分析了多叶片坐标转换理论,结合独立变桨控制方法,将旋转坐标系运动方程转化到固定坐标系,简化控制设计过程,再利用多叶片坐标逆变换将结果转换为旋转坐标,此物理量与功率控制环输出量一起构成风力机变桨控制的输入量,可以有效减小湍流引起的不确定载荷。第6章:总结本文的主要研究内容和成果,并给出了今后有待进一步研究的工作和方向。更多还原

【Abstract】 Based on the research on the wind turbine system, this dissertation sets the power and loads control as objects and further puts forward corresponding control strategies. On the basis of analyzing and establishing the model of wind turbine, the dissertation makes in-depth research on the control strategies about the optimum power control below rated wind,pitch control above rated wind , predictive control of power, and the reduction of the loads. Then, with the employment of those proposed control strategies and methods, it makes simulation design for the control system of wind turbine. The following are those main points of this dissertation.In chapter 1, the research background was introduced, the significance of the research was expounded and those research situation of wind turbine were summarized. Besides, the control strategies and control algorithms were especially emphasized. Then, the main content and structure of dissertation were recommended.In chapter 2, according to the wind turbine dynamic inflow theory, the unsteady inflow model was created. Then, the rationality of the establishment of wind turbine model by employing the dynamic inflow theory was analyzed. Furthermore, the influence of wake on the dynamic features of wind was indicated. And the acceleration potential method of unsteady flow was modified. Besides, based on the theory of von Karman, the turbulent wind model was set up and the three-dimensional turbulent wind field was gained. Finally, by comparing the analysis results between Matlab simulation model and Bladed design model, the correctness of aerodynamical model of wind turbine that has been proposed was validated.In chapter 3, on the basis of analyzing the features of performance of wind turbine, three districts were divided in the process. Then the corresponding control strategies of the three districts were designed owing to the difference between different districts. Focusing on the research on those control strategies in full loads period, the smooth transition of control by decoupling the control loop was realized. Then, based on the characteristic of the system, the performance of control system would be improved by combining predict control method to reduce time lag of system. Besides, with the combination of characteristics of wind turbine, by analyzing the model predictive control algorithm of linear state space, the model predicative controller of the system was designed. Afterwards, the model predictive controller was divided into three modules including disturbance model, object calculation and dynamic optimization. Then, based on the analysis of the features of those modules, their mathematical model was established. At last, with the exertion of the emulation mode, the advantage of model predictive control on power control of wind turbine was validated.In chapter 4, the linearization method of wind turbine was analyzed and the three states linearization model was selected as control objects according to the research focus of this chapter. Based on the analysis of single neuron, the mathematical model of single neuron was deduced and its algorithm was ameliorated to attain the improved the mathematical model of single neuron. Then, by adopting the improved control algorithm to adjust conventional PID controller parameter, the self-adaptive PID controller which was based on the improved single neuron was gained. Then, the gained controller was used to regulate pitch system and comparative analysis of the functions of the two controllers in system was made.In chapter 5, based on kinematic equation, the filters in fore-afterward direction and in sideward direction were designed to increase damping and reduce vibration. Furthermore, by increasing damping for train and shielding the crossover frequency of blade, the resonance between tower and rotor was avoided. Next, the multi-blade coordinate transformation theory was analyzed and the method of independent pitch control was combined and kinematic equation of rotation coordinate was converted to fixed coordinate in individual pitch control. Thus, the design process of individual pitch was simplified. Then, the results were converted to rotation coordinate via inverse transformantion. And finally, the physical quantity was integrated with output of power control loop to become the input of pitch control of wind turbine, which could effectively reduce uncertain loads.In chapter 6, the conclusion of the major research contents and results was achieved and the task that should be accomplished further in the future was also advanced.更多还原