【作者】 王国强；

【导师】 王志新；

【作者基本信息】 上海交通大学 ， 电机与电器， 2012， 博士

【摘要】 开发和规模化利用以风能为代表的可再生能源，对促进国民经济发展、保护环境、应对全球能源危机等方面具有重要的社会意义和战略价值，现已成为世界各国实现可持续发展的重要战略手段之一。近年来，我国风电事业得以蓬勃发展，风能的开发利用已由陆地逐步转向海上。开发海上风能具有风资源蕴藏量丰富、风能利用率高和大型风电机组不占用土地资源等特点，因此越来越受到业界的关注。随着大型海上风电机组技术的日趋成熟，海上风电场建设规模的不断扩大，且与陆地电网的距离越来越远，急需解决海上风电场的输电问题，使得研究与应用适合海上风电场实施远距离传输的柔性直流输电(VSC-HVDC)技术已成为当前的研究热点；同时，VSC-HVDC系统对海上输电变流器的电压等级、系统动态性能、并网及电能质量等方面都提出了更高的要求。论文针对海上风电场并网VSC-HVDC系统，侧重于研究海上风电场柔性直流输电变流器主电路拓扑结构及其控制算法，主要解决基于模块化多电平结构的大功率变流器的调制方法和均压控制技术难题，以及各种先进控制算法在海上风电场VSC-HVDC系统中的应用等问题。论文的具体研究内容简述如下：1、根据VSC-HVDC系统具有电压高、功率大的特点，建立模块化多电平变流器(MMC)拓扑结构的数学模型，同时，针对对直流侧电容电压不平衡问题，提出分别控制桥臂内电容电压和桥臂间电容电压实现直流侧电容的平衡，将电容电压排序法与参考信号叠加环流抑制分量的方法有机结合。并采用多载波PWM调制技术，分别针对带直流输出电容和直接输出的MMC结构进行仿真，结果表明采用论文所设计的调制和电容均压控制方法，可使输出电平数提高近一倍，同时，多电平波形质量明显优于常规均压控制方法。2、分析了MMC主电路参数选择方案，采用传统双闭环矢量控制策略，设计了20MW大容量VSC-HVDC系统，并结合论文所提出的电容均压控制方法搭建了系统仿真模型。同时，还针对传统矢量控制中存在的PI参数较多、整定困难等不足，提出将粒子群(PSO)搜索算法与PID神经网络(PIDNN)相结合，基于系统线性模型，以PIDNN控制器代替传统的PI调节器，以PSO算法搜索PIDNN最优权值，避免PI参数的试凑过程，使系统动态跟踪误差达到最小，优化了系统控制性能。3、研究非线性控制策略在VSC-HVDC系统中的应用，包括：虚拟磁链直接功率控制(VF-DPC)、反馈线性化，无源性、反步法、无差拍等方法。其中对于VF-DPC，给出了该控制策略应用于MMC结构时磁链和功率估计的方法，将均压控制策略和多载波调制方法相结合代替传统查表法，使得VF-DPC能够适用于MMC拓扑结构。仿真研究结果表明，相对于传统矢量控制方法，将VF-DPC应用于MMC，能够有效提高大容量VSC-HVDC系统动态性能；对于反馈线性化方法，将变流器非线性数学模型转化为等效的线性模型，并基于系统精确线性化模型设计其控制器；无源性方法利用系统能量耗散特性，设计其控制器，同时，配合适当的阻尼注入使系统快速收敛于平衡点；反步法针对变流器的高阶、非线性等特点，给出了一种寻找系统李雅普诺夫函数的方法，从而推导出全局渐近稳定的控制律；无差拍控制主要解决实际系统中存在的采样和计算延迟造成的系统不稳定问题，通过数学模型递推的方法，可预测下一拍的控制量，从而弥补延时造成的影响。通过对比各种非线性控制方法在VSC-HVDC中的应用，表明VF-DPC方法既能提高系统性能又易于工程实现。4、VSC-HVDC实验系统设计，包括实验系统的硬件和软件设计，采用工程上常用的双闭环矢量控制方法，实现了双端系统的并网运行，能够跟踪功率、电压等控制目标，且稳态精度较高，验证了所设计实验平台硬件系统的正确性和基本控制算法的有效性。同时，实验结果为进一步设计满足实际工程应用的大容量海上风电场柔性直流输电变流器(如20MW)奠定了基础。论文的创新点如下：（1）针对MMC直流侧电容电压平衡问题，提出通过将桥臂电容电压排序法和参考信号叠加环流抑制分量方法相结合，并利用多载波调制策略控制变流器工作，提高了MMC的多电平输出波形质量和电平数；（2）提出利用改进的PID神经网络控制器代替传统PI调节器，并通过与粒子群搜索算法相结合，来优化神经网络权值参数，使系统动态跟踪误差达到极小；（3）将直接功率控制方法应用于MMC结构，提出了适用于MMC结构的功率及磁链估计方法，提高了大容量系统的动态响应性能。更多还原

【Abstract】 Development and large-scale utilization of new energy represented bywind power is of great significance and strategic value in the aspects ofpromoting national economic development, environmental protection, solvingglobal energy crisis and so on, which has become one of the importantstrategic goals of the world to achieve sustainable development. In recentyears, Chinese wind power industry is to flourish and has gradually turned tooffshore from land-based wind farm. There are many advantages such asabundant wind energy reserves, high utilization of the wind generator, nottaking up land resources and so on. Therefore, it is one of the focuses ofindustry. As the technology of wind turbine control constantly improve,continues to mature, offshore wind farms have been expanding and distancewith the onshore grid is farther. So, offshore wind farm transmission problemsneed to be resolved. The technology of high voltage direct currenttransmission based on voltage source (VSC-HVDC), which is suitable forlong-distance transmission of offshore wind energy, has become one of thecurrent research focus. This has put forward higher requirements on theaspects of offshore converter voltage level, system dynamic performance, andnetwork power quality.This paper researches the related technologies of VSC-HVDC system foroffshore wind farm grid integration, mainly including two aspects of the maincircuit topology and control algorithm. The following problems are solved,including voltage balance and triggering method of high-power converterbased on the MMC structure, as well as a variety of advanced control algorithms application on the offshore wind farm VSC-HVDC system. Theresearches are outlined as follows.1. High voltage and high-power are the characteristics of theVSC-HVDC system applications. According to it, a new topology of modularmulti-level converter (MMC) was introduced, and its mathematical modelwas established. Based on these, for the control difficulty of unbalancedcapacitor voltage, the method was proved, that controlling the balancing ofcapacitor voltages between arms and within one, respectively. Two methodsof capacitor voltage sequencing and component added to reference signal forcirculating current suppression. Simulation studies were carried out for theMMC with DC output capacitance and the direct output one, using themulti-carrier PWM triggering technology. The simulation results show thatthe MMC modulation and voltage balancing control strategy designed in thispaper can significantly improve the quality of the output waveform andincrease the number of output level by nearly double compared toconventional methods.2. The method of MMC main circuit parameter selection is analyzed.The20MW VSC-HVDC system was researched based on the strategy ofdouble closed-loop vector control and the simulation model of whole systemwas established, including the capacitor voltage balancing control methodproposed. For the problems of too many PI parameters and difficult to adjust,the method of combination of particle swarm (PSO) searching algorithm andPID neural network was proposed. The traditional PI regulator was instead byPIDNN controller. The optimal weight of PIDNN is searched by PSOarithmetic, which made the dynamic tracking error to minimum, optimizingthe system control performance and avoiding the repeated adjustment of thePI parameters.3. Nonlinear control strategy for VSC-HVDC system was researched,including the virtual flux direct power control (VF-DPC), feedbacklinearization, passivity control, back-stepping method and deadbeat controland so on. For the VF-DPC, estimation method of flux was given when it was used in MMC. The capacitor voltage balancing control method andmulti-carrier trigger strategy was used to instead of the traditional look-uptable method, which enabled the VF-DPC could be applied to the MMCtopology. The VF-DPC strategy was compared with the traditional vectorcontrol method by simulation, the results show that it used in MMC caneffectively improve the dynamic performance of the large capacityVSC-HVDC system. For the feedback linearization control method, thenonlinear mathematical model of the converter was transformed intoequivalent linear model, and controller was designed based on exactlinearization model. For passivity control method, the controller was designedbased on system energy dissipation characteristics, and combined with theappropriate damping injection to make the system quickly converges to theequilibrium point. For the high-order and nonlinear characteristics of VSC,the back-stepping method is used for searching the Lyapunov function, andthen the control strategy of global asymptotic stability can be derived.Deadbeat control was used mainly to solve the system instability caused bythe delay of the sampling and calculating in actual system. It can predict thecontrol quantity of next-beat and compensate the impacts because of delaybased on recursive derivation of mathematical model. Then, all the nonlinearcontrol methods above mentioned were compared in the application ofVSC-HVDC, and the conclusion that the VF-DPC can improve systemperformance and ease for actual implementation.4. The experimental system was built, and then the system hardware andsoftware were designed. The double closed-loop vector control strategycommonly used in engineering was realized for the double-ended systemconnecting to grid running. The system power and voltage could track thegiven values, and the accuracy of steady-state is high. It verified thecorrectness of the control algorithm and effectiveness of the designedexperimental platform hardware. The experimental results lay a foundation ofhigh-power (20MW) offshore wind farm VSC-HVDC converter for practicalengineering application. The innovations are as follows:(1) For the problem of DC capacitor voltage balance of MMC, The bothmethods of capacitor voltage sorting and superimposing circulationrestraining components into reference signals were combinated. Andmulti-carrier modulation strategy was used to control the converter. This canimprove the quality and level number of MMC output wave.(2) Improved PID neural network controller was used for instead of thetraditional PI regulator, and combination of particle swarm search algorithmto optimize the neural network weights parameters. It can make systemdynamic tracking error minimum.(3) Direct power control method was applied to the MMC structure.Power and flux estimation methods for the MMC structure were approved.This can improve the dynamic response performance of the high-capacitysystem.更多还原