【作者】 王鹿军；

【导师】 吕征宇；

【作者基本信息】 浙江大学 ， 电气工程， 2013， 博士

【摘要】 分布式发电以其发电方式灵活、能源利用率高、环境污染小等优点成为近年来发展较快的一种供电模式。并网逆变器作为分布式发电连接电网的接口,在风力发电、光伏发电和燃料电池发电等分布式发电中具有重要的作用,甚至在一定程度上决定了分布式发电所能提供的电能质量。分布式发电并网逆变器所接入的电网通常都是相对较弱的配电网,容易受负载的影响而发生畸变,也可能因短路等故障而产生三相不平衡现象。多电平逆变器不通过器件的直接串联提高了变流器的电压,随着分布式发电功率等级的提高,多电平逆变器有望在分布式发电中得到广泛应用。在分布式发电的并网技术标准中,孤岛检测、低电压穿越、有功功率和无功功率的控制都涉及并网逆变器的两项关键技术：锁相技术和并网逆变器的控制技术。所以本文主要选择电网谐波和不平衡条件下的锁相方法、中点钳位型三电平并网逆变器的控制方法作为重点内容开展研究。针对小功率分布式发电通常接入的单相电网,本文比较了几种典型的单相电网锁相方法,深入分析了其在理论上或实用中的优点和不足,评估了在电网谐波情况下的性能,然后在此基础上提出了一种改进的单相畸变电网锁相方法,改进方法的核心部分是一个由主模块和多个谐波模块组成的交错耦合反馈结构,其中主模块是依据梯度下降法来估测基波幅值频率的锁相环,谐波模块是简单的二阶陷波滤波器。主模块作为主要的功能模块来估测输入信号中的基波分量,谐波模块配置在谐波频率处将得到的谐波分量反馈回输入信号。改进的方法易于实现,不但在电压信号严重畸变时具有较高的锁相精度,而且在电压信号发生突变时还具有较快的响应速度。针对三相不平衡电网,本文介绍了三相不平衡电网的一些常用锁相算法,分析了这些算法的基本原理和优缺点,在此基础上提出了一种基于静止坐标系下直接梯度下降法的锁相环。该锁相环结构简单,不需要将电网电压变换到同步旋转坐标系,在静止坐标系下通过数值算法求解根据梯度下降原理得到的非线性方程组,直接估测出正负序分量的幅值、相位和频率。通过实验评估了该锁相方法的性能,实验结果表明,所提方法对对输入电压的噪声不敏感,在三相电压不平衡时稳态精度高,在电网发生不平衡故障时,响应速度快,即使电网电压在故障后相位、幅值和频率均有大幅跳变,也基本能在两个工频周期以内完成锁相。针对三电平并网逆变器的控制策略,本文提出了一种有限集最优预测(Finite Set Optimal Predictive, FSOP)控制方法,该方法将27个开关矢量看成一个有限集,然后在此有限集中搜索使罚函数最小的开关矢量,其中罚函数是由输出电流误差、中点不平衡电压和器件开关次数组成的加权和。本文详细推导了中点钳位型三电平并网逆变器的状态空间模型,并给出了FSOP控制方法的设计步骤和程序流程图,最后通过仿真和实验评估了FSOP控制方法的性能。实验结果表明,与传统方法相比,FSOP控制方法跟踪参考电流和平衡中点电压的速度快,平均开关频率低,对电网谐波扰动不敏感,对模型参数具有一定的鲁棒性,通过调整罚函数中的加权系数,该方法还可以非常灵活方便的优化并网逆变器的综合性能。针对三电平并网逆变器的工程实践问题,本文提出了一种适用于多电平并网逆变器的高效仿真方法。多电平逆变器的控制方法复杂,拓扑结构还在不断发展,在对多电平逆变器的拓扑结构及控制算法没有准确掌握的基础上,直接进行实物实验需要花费较多的人力、物力,还存在很多安全隐患。所以,非常需要一种高性能的多电平逆变器的仿真方法,仿真方法与实物的差别越小越好。本文设计并实现了一种基于Saber软件的高性能仿真方法,该方法对影响控制系统性能的死区时间、量化误差、数字延迟等因素都进行了有效的模拟,仿真速度快,仿真结果可靠性高,并可方便移植到实物系统中。该仿真方法对控制算法还具有离线调试功能,能细致入微地分析算法的具体实现细节,可以作为工具来探索一些高级的控制方法。本文还介绍了比例谐振(PR)控制器在定点型DSP中实现时遇到的问题,介绍了实验样机的硬件和软件保护方法,给出了实验现场的部分照片。更多还原

【Abstract】 Distributed generations develop rapidly in recently years with substantial benefits, including flexible electricity production, high energy utilization and low environment pollution. Grid-connected inverters which are the interface of grid and wind power generation, photovoltaic power generation and fuel cell power generation play an important role in distributed generations. The grid connected by distributed generations is usually weak and may be distorted due to nonlinear loads and unbalanced due to short-circuit faults. With the increase of distributed generation power ratings, multi-level inverters are expected to be widely used in distributed generations since the voltage rating is increased without direct parallel switching devices. In the standard of grid-connected inverters in distributed generations, islanding detection, low voltage ride through, active and reactive power control involve two important technologies, which are the phase-locked loop (PLL) technology and the inverter control strategy. So in this paper, the research focuses on the PLL technology in distorted and unbalanced grid and the control strategy of neutral point clamped three-level inverter.Based on the fact that the low-power distributed generation is usually connected to single-phase grid, this paper compared several typical single-phase PLL methods, analyzed the advantages and disadvantages in theory and practice, and evaluated the performance in distorted grid. On the basics of above work, this paper proposed a modified PLL method especially for distorted grid. The proposed PLL method has a coupling feedback structure composed of a main module aiming to estimate fundamental component and multi notch filters aiming to get harmonic components. The main module is an amplitude-frequency-phase-locked loop (AFPLL) using gradient descent method and the harmonic modules are simple second order notch filters (NF). The proposed PLL method can be easily implemented in DSP and the experimental results show that the propose PLL not only can eliminate steady state error in harmonic distorted grid voltage, but also has a quick response when grid voltage suddenly changes.For unbalanced three-phase grid voltage, this paper firstly analyzed the basic principle of commonly used PLL methods, and then proposed a PLL method to estimate positive and negative components with gradient descent algorithm in stationary reference frame. The proposed method has a very simple structure and does not require transformation of grid voltage form stationary reference frame to synchronous reference frame. It can estimate the amplitude, phase angle and frequency of positive and negative components by solving a nonlinear equation set with gradient descent algorithm. The performance of the PLL method is verified by experiments, and the experimental results show that the proposed PLL has a quick response when grid voltage suddenly becomes unbalanced. The response time of the PLL is within two frequency cycles even if all the amplitude, phase angle and frequency have large changes.As for the control strategy of neutral point clamped three-level inverter, this paper proposed a finite set optimal predictive (FSOP) control method. In the proposed method, the optimal switching vector is selected from a finite set according to a cost function. The finite set is composed of27switching vectors, and the cost function is composed of the error of output current, the neutral point unbalanced voltage and the switching times of device. The state space model of neutral point clamped three-level grid-connected inverter is detailed derived and the flowchart of FSOP control is given. Finally, the performance of FSOP control is verified though simulations and experiments. Experimental results show that FSOP control has a quick response in tracking output reference current and balancing neutral point voltage, insensitive to harmonic distorted voltage and model parameters. FSOP control has a flexible control purposes and we can optimize the synthesize performance just by adjusting weighting coefficients of cost function.As for engineering practice, this paper proposed a high performance simulation methodology for multi-level inverters. In comparison with two-level inverters, the control algorithm of multilevel inverters is very sophisticated and many new control algorithms have been proposed recently. Without a full understanding of the control algorithm, the experiment will take both money and time even with safety issues. Therefore, an effective simulation methodology for multilevel inverters is needed. The simulation methodology can simulate the factors that have great impact on control performance such as dead band, digital delay, and quantization error. The simulation results are more convincing and can be easily ported to an actual system. The simulation methodology also provides an off-line debugging function, which will be helpful for exploring advanced control algorithms. This paper also described the implementation problem of PR controller in fixed point DSP. The software and hardware protection methods and some experiment photos are also demonstrated.更多还原