多电平二级管钳位型逆变器电容电压平衡SVM算法及其应用研究

发布时间：2018-07-26 17:24

【摘要】：在大功率电力电子装置的研究和应用领域,多电平变换器因其比两电平变换器具有更好的输出电压波形和可获取更高的电压等级等突出优点而广受关注,已成为一种代表性的解决方案。然而目前并没有一种完美的多电平变换器拓扑,现有的拓扑在实际应用中均存在一定的局限性。二极管钳位型逆变器既不需要多个独立的直流电源,也不需要笨重的电容及其预充电系统,同时还具有开关控制易于实现、保护电路简单等优点。对这一拓扑的应用造成阻碍的是其直流侧分压电容在逆变器运行时会发生电压偏移,使得逆变器性能下降甚至不能正常工作。本文的工作就致力于从调制方案这一硬件结构最简单同时也最节省系统成本的途径来研究解决多电平二极管钳位逆变器的直流电容电压偏移问题。论文分析了传统SVM算法在多电平二极管钳位逆变器中实现的原理和步骤,指出了传统SVM算法需要进行大量三角函数运算或查表操作、因而运算效率低的不足。通过引入基于Kohonen竞争性神经网络的分类算法,提出了一种改进的多电平二极管钳位逆变器SVM算法。改进的算法并不需要对神经网络进行训练,在整个实现过程中不再需要任何三角函数计算或查表操作,而只需要进行简单的四则运算,所节省的运算时间可用于完成其他耗时的工作如实现电容电压均压等。对三电平和五电平的二极管钳位逆变器进行时域仿真,结果验证了文中数学分析的正确性和所提算法的可行性,同时表明改进的算法是一种对多电平二极管钳位逆变器通用的算法,将它应用于不同电平数的逆变器时不需要进行任何修改。为了能够利用调制算法消除多电平二极管钳位逆变器直流电容电压的偏移,论文从理论上对这一现象产生的原因进行了深入的分析。首先研究了三电平二极管钳位逆变器的调制矢量的特点,通过具体的图例分析了四类调制矢量对中点电位的影响,明确了三电平二极管钳位逆变器能够利用小矢量调节电容的电压。其次以五电平二极管钳位逆变器为例分析了电平数超过3时的多电平逆变器的电容电压偏移问题,分析表明这类拓扑无法在全调制比范围内利用小矢量控制电容电压。由于逆变器直流侧电容电流平均值为零时,直流电容电压才能保持平衡,因此推导出在SPWM调制算法下五电平二极管钳位逆变器直流侧电容电流平均值与调制比和交流侧功率因数之间的函数关系。分析结果说明在不采用辅助电路的情况下,传统的SPWM技术不能维持直流电容电压平衡；但SVM方法却有可能利用冗余开关状态、在不设置辅助硬件的前提下实现直流侧电容均压。在分析电容电压偏移的本质原因的基础上,说明了无论采用哪种PWM算法来维持多电平二极管钳位逆变器的电容电压,都存在一个由调制比和交流侧功率因数所限定的稳定运行范围,因而电容电压平衡SVM算法主要适合应用于二极管钳位逆变器进行无功功率交换的场合。根据多电平二极管钳位逆变器的最小能量特性建立了五电平逆变器的能量函数,研究了通过能量函数来选择恰当的冗余开关状态从而实现电容均压的方法。为了计算在不同开关状态下的能量函数值,推导了逆变器中间支路电流平均值与各个扇区开关状态之间的关系,得到了五电平二极管钳位逆变器的综合电流数学模型。在此基础上建立了多电平二极管钳位逆变器的通用电流模型,从而得到了对任意电平通用的电容电压平衡SVM算法。仿真研究表明基于能量函数的电容电压平衡算法能够在对称负载、不对称负载和畸变负载等稳态条件下使得偏离的电容电压回归到标准值。基于能量函数的平衡算法计算量非常大,且未考虑对开关频率的优化,因而不可避免地造成逆变器运行时开关频率较大。针对该算法的不足,本文提出了种新的基于有功电流的SVM算法。新算法利用有功电流来判断二极管钳位逆变器交流侧的能量传输方向,通过分析能量传输方向与不同的开关序列对电容电压的影响,从一组预定义的开关序列中选取使得偏离标准值最多的电容电压尽可能回归标准值的开关序列。新算法使逆变器在运行中摆脱了大量的乘法运算,只需几次比较运算就能得出调制结果,大大降低了运算量、提高了调制速度；同时算法对开关器件工作频率进行了优化,使得在一个开关周期内的器件开关次数最少,能有效降低开关频率从而降低开关损耗。大量不同工况下的仿真研究证明了基于有功电流算法的有效性以及它的性能优势。为了进一步考察电容电压平衡算法的应用效果,本文研究了一个基于五电平二极管钳位逆变器的STATCOM系统。该STATCOM的直流侧电容没有辅助平衡电路,它是完全依赖空间矢量调制算法来控制电容电压的。建立了STATCOM的数学模型,在此模型的基础上分别设计STATCOM交流侧电流控制器、直流端电压控制器和公共连接点电压控制器,再将控制器与电容电压平衡算法相结合,构成了完整的STATCOM控制系统。对STATCOM系统的性能进行了仿真研究,结果表明STATCOM的控制器能有效的控制STATCOM输出的无功功率、直流端电压和公共连接点电压。对基于能量函数和基于有功电流的电容电流平衡算法的性能进行了仿真对比,结果证明两种算法都能有效的在稳态和动态条件下将五电平二极管钳位逆变器的直流电容电压维持在标准值,但本文提出的基于有功电流的算法具有更好的动态性能。
[Abstract]:In the field of research and application of high-power power electronic devices, multilevel converters have become a representative solution because of their outstanding advantages of better output voltage waveform and higher voltage level than two level converters. However, there is no perfect multilevel converter topology at present. The existing topology has some limitations in practical application. The diode clamped inverter does not need multiple independent DC power supply, and does not need heavy capacitance and precharging system. It also has the advantages of easy switching control and simple protection circuit. The application of this topology is hindering its DC side. The voltage offset of the voltage divider will occur during the operation of the inverter so that the performance of the inverter can not even work normally. The work of this paper is devoted to the study of the problem of DC capacitance voltage offset from the simplest hardware structure and the most systematic cost saving of the modulation scheme.
The paper analyzes the principle and steps of the traditional SVM algorithm in the multilevel diode clamped inverter, and points out that the traditional SVM algorithm needs to perform a large number of trigonometric functions or look-up operations, so the operation efficiency is low. By introducing a classification algorithm based on Kohonen competitive neural network, an improved multilevel two is proposed. SVM algorithm of pole position inverter. The improved algorithm does not need to train the neural network. It does not need any trigonometric function calculation or look-up operation in the whole implementation process, but only requires four simple operations. The saving time can be used to complete other time-consuming tasks such as the realization of capacitor voltage mean voltage and so on. Three The level and five level diode clamped inverter is simulated in time domain. The results verify the correctness of the mathematical analysis and the feasibility of the proposed algorithm. At the same time, it shows that the improved algorithm is a common algorithm for multilevel diode clamped inverters. It does not require any modification when the inverter is used for different level number inverters.
In order to use the modulation algorithm to eliminate the offset of the DC capacitance voltage of the multilevel diode clamped inverter, the paper theoretically analyzes the cause of this phenomenon. First, the modulation vector characteristics of the three level diode clamped inverter are studied, and the four kinds of modulation vector pairs are analyzed by specific illustrations. The effect of point potential is made clear that three level diode clamp inverters can use small vector to adjust the voltage of capacitance. Secondly, five level diode clamp inverter is used as an example to analyze the capacitance voltage offset of multilevel inverters when the number of level exceeds 3. The analysis shows that this type of extension can not use small vector in the range of total modulation ratio. The capacitor voltage is controlled. The DC capacitor voltage can be balanced because the average current of the capacitor current of the DC side of the inverter is zero. Therefore, the function relationship between the average value of the capacitance current of the DC side of the five level diode clamped inverter under the SPWM modulation algorithm and the modulation ratio and the power factor of the AC side is deduced. In the case of the auxiliary circuit, the traditional SPWM technology can not maintain the DC capacitor voltage balance, but the SVM method may make use of the state of redundant switch to realize the DC side capacitor voltage without setting the auxiliary hardware. Based on the analysis of the essential reason of the capacitance voltage offset, it shows that no matter which PWM algorithm is used to maintain more, the capacitor voltage is maintained. The capacitor voltage of the level diode clamp inverter has a stable operating range limited by the modulation ratio and the AC side power factor, so the capacitor voltage balance SVM algorithm is mainly suitable for the occasion of the reactive power switching of the diode clamper inverter.
According to the minimum energy characteristic of the multilevel diode clamper inverter, the energy function of the five level inverter is established. The method of capacitive voltage sharing is realized by selecting the appropriate redundant switch state through the energy function. In order to calculate the energy function value in different switching states, the average current average of the middle branch of the inverter is derived. The relationship between the value and the switching state of each sector is obtained, and a comprehensive current mathematical model of the five level diode clamp inverter is obtained. On this basis, a general current model of the multilevel diode clamped inverter is established, and the SVM algorithm for the general capacitance voltage balance of any level is obtained. The simulation study shows that the energy function is based on the energy function. The capacitor voltage balancing algorithm can make the deviated capacitor voltage regress to the standard value under symmetrical conditions such as symmetrical load, asymmetrical load and distorted load.
The calculation of the balance algorithm based on the energy function is very large and does not consider the optimization of the switching frequency. Therefore, the switching frequency of the inverter is unavoidable. In this paper, a new SVM algorithm based on active current is proposed in this paper. The new algorithm uses active current to judge the communication of diode clamped inverter. The direction of the energy transmission in the side, by analyzing the influence of the energy transmission direction and the different switch sequence on the capacitance voltage, from a set of predefined switch sequences, the switching sequence which makes the capacitor voltage that deviates the most standard value is as much as possible to the standard value is selected. The modulation results can be obtained by several comparison operations, which greatly reduces the amount of operation and improves the modulation speed. At the same time, the algorithm optimizes the working frequency of the switch device, making the switching frequency of the device in a switching period least, reducing the switching frequency and reducing the switching loss effectively. A large number of simulation studies under different working conditions have proved that The effectiveness of active current algorithm and its performance advantages are also discussed.
In order to further investigate the application effect of the capacitor voltage balance algorithm, a STATCOM system based on five level diode clamped inverter is studied. The DC side capacitance of the STATCOM has no auxiliary balance circuit. It is completely dependent on the space vector modulation algorithm to control the electrical voltage of the capacitor. The mathematical model of the STATCOM is established, and the model is established in this model. On the basis of the model, the STATCOM AC side current controller, the DC voltage controller and the common connection point voltage controller are separately designed. The integrated STATCOM control system is formed by combining the controller with the capacitor voltage balance algorithm. The performance of the STATCOM system is simulated. The results show that the STATCOM controller can be effective. The reactive power, DC end voltage and common connection point voltage of the STATCOM output are controlled. The performance of the capacitive current balancing algorithm based on the energy function and the active current is simulated and compared. The results show that the two algorithms are effective for the DC capacitor voltage of the five flat diode clamped inverter under steady and dynamic conditions. It is maintained at the standard value, but the algorithm based on active current in this paper has better dynamic performance.
【学位授予单位】：湖南大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM464

【参考文献】