双流环控制的单相逆变器并网 外文文献翻译 毕业设计（论文）

双流环控制的单相逆变器并网 外文文献翻译 毕业（） 附录 英文翻译 1. 英文文献原文 Control of inverter-based micro-grids T.C. Green, M. Prodanovic Imperial College London, Department of Electrical and Electronic Engineering, Control and Power Group, London SW7 2AZ, United Kingdom Available online 28 September 2006 Abstract The predicted growth of small-scale non-50/60 Hz power sources and the desire to be able to support loads independently of the public electricity grid requires the development inverter-based micro-grids. Power electronic interfaces have very different characteristics to conventional electrical machines and, therefore, different operation, control and protection schemes are required. Attention also needs to be given to the dominance of single-phase harmonically distorting loads in some networks and control schemes put in place that maintain voltage quality. A control scheme that exploits the controllability of inverters to operate a micro-grid and provide good power quality is examined and compared with both traditional power systems and with control of dc/dc power converters. The limitations of communication and control bandwidth are discussed. Experimental results are used to illustrate the performance that can be achieved with various combinations of linear and non-linear, three-phase and single-phase loads. ? 2006 Elsevier B.V. All rights reserved. Keywords: Distributed generation; Micro grids; Inverters; Power quality 1. Introduction Micro-grids are thought to be a likely direction for evolution of power systems that incorporate distributed generation. The term micro-grid is not strictly defined and covers a wide range of possible systems [1–3]. Here, the term is taken to mean a sub-set of a power distribution system that contains sufficient energy sources to supply most or all of the local load. The micro-grid is thus able to operate with only a small power exchange with the rest of the system and at times can separate from the public system and run as an independent island. A micro-grid differs from existing island power systems (both physical islands and electrical islands, such as offshore oil/gas platforms, ships and aircraft) in that connection to and disconnection from a public grid is a regular event. It is anticipated that the loads will be nor-mal domestic and commercial consumer loads. As such, large numbers of single-phase diode-rectifier loads are anticipated and the supply current is expected to be unbalanced and distorted. In some island systems, for instance an oil platform, the initiation of major loads (hoists, pumps, etc .) can be anticipated and the generation resources can be put in readiness for the transient. This may be possible in some sorts of micro-grid but, if not, load forecasting is likely to be important and demand-side participation in control may be needed. The generating technology found in a micro-grid may take many forms and an individual micro-grid is likely to include a mixture of these. Sources may be small, such as domestic CHP systems or roof-top photovoltaic panels or relatively large, such as office-block-scale CHP using gas-turbines or reciprocating engines. Sources may be intermittent (photovoltaic and heat-led CHP) or fully controllable (electric-led CHP or simple diesel/gas fuelled generators). To make best use of intermittent sources and/or to provide security, a micro-grid may include a storage element, such as a flywheel, battery or hydrogen electrolysis/fuel-cell system. Relatively large generators, especially existing peak-lopping diesel generators that are turned to micro-grid duty, are likely to be conventional synchronous machines generating at 50/60 Hz three-phase. Newer technology will be different. Some sources are natural dc sources, such as photovoltaic’s, fuel-cells and batteries. Others are variable frequency or high frequency ac, such as micro-turbines and fly-wheels. The non-50/60 Hz sources will need to be interfaced to the grid via inverters. The form the micro-grid takes and the type of loads sup-plied will have a large impact on the operating and control regime of the system. As yet, there are no established standard operating schemes and it may never be possible or desirable to have a universally applied method. This paper will describe the operation and control of a micro-grid for a particular situation, namely, a micro-grid that is formed of a relatively small number of generation sources that are interfaced to the grid through inverters and expected to supply unbalanced and distorting load current. 2. The power park The example micro-grid system will be a power park. A power park consists of several power sources in relatively close proximity used to improve the electricity service to a group of customers. The customers may be a retail park, shopping mall or group of office buildings. If these are new consumers the power park may be motivated by network congestion or power shortages. Other motivations might be the wish to install cogeneration or to provide premium power. The premium aspect would be the ability to continue supplying high value customers in the event of outage of the public grid or voltage sag. It might also mean avoidance of other power quality problems, such as waveform distortion, flicker or swell. The improved power quality might come about from a lowered supply impedance, active management (such as active filtering) and/or disconnection from the public grid when the grid is deemed to be the source of the problem. Because some system operators presently do not allow distributed generators to control the voltage at the point of common connection, some of the power quality benefits may only be available when the micro-grid is running as an island. For increased reliability, flexibility and efficiency (when running part-load) several generators are required. The power park might be owned by a landlord, a building ser-vices company, the consumers themselves or the existing utility company. Power export to the public grid might or might not be used whatever the ownership of the generation. Micro-grids in general may have a variety of owners of generation but the power park variant is likely to see a single owner or a single operator. Further, the generators will be in groups that are physically and electrically close together. The generation technology may be a mix of traditional electrical machines and inverters. There are several possible operating modes of the power park: (1) Grid-connected operation with fixed local generation: The grid connection will see a fluctuating power flow as loads change. The power flow could be in either direction. (2) Grid-connected operation with load following local generation: The grid will see a fixed power flow which might be import (to cover short fall of local generation), export (to sell excess capacity) or zero (floating of the micro-grid but with a connection maintained to cover contingencies). (3) Island operation : The local generation will load follow, possibly using storage or demand-side participation to increase security. The operation and control issues to be faced are: (1) How to share steady-state load between generating sets? (2) How to share transient load between sets? (3) How to switch seamlessly from grid-connected to island operation? (4) How to incorporate additional features, such as active management of waveform quality? 3. Characteristics of electrical machines and inverters The history of power system generation is the history of the synchronous machine and the characteristics of this machine have influenced the development of other aspects of the system. For instance, protection schemes rely on the short-term over current rating of an electrical machine (resulting from its large thermal mass) to provide the fault current necessary to discriminate faults. In turn, the required clearance times for faults are dictated by the over-swing and loss of synchronism characteristics of machines. On the other hand, electronic power converters are well established in the fields of OEM power supplies and industrial drives but have not influenced the development of traditional power systems. Electronic power converters have very different characteristics to electrical machines and would have directed engineers toward a very different transmission and distribution system. The relevant characteristics of electrical machines are: (1) Operate as voltage sources whose amplitude can be adjusted. The adjustment is normally part of a closed-loop excitation control scheme with a relatively low bandwidth. (2) Sine-wave voltage is a feature designed into the construction of the machine. The total harmonic distortion (THD) of the voltage is low. Distortion in the load current will increase the distortion in the voltage to an extent dependent on the source impedance. (3) The short-circuit current is high because the source impedance is low and no current limiting is employed. (4) The current rating is set by temperature rise of the winding insulation. The thermal time-constant of the winding and surrounding steel is relatively large and a useful short-term rating is available. Fault currents of as much as 10 times the steady-state maximum can be sustained for several mains cycles. (5) Real power exchange is dictated by the torque applied to shaft. Steady-state load sharing can be applied by a closed-loop governor setting that makes power output a function of the (common) system frequency. This is a designed in feature of the governor system but it is similar to the natural tendency for the speed of the prime mover to droop when electrical load is drawn. The corresponding characteristics of inverters are: (1) Operate as voltage sources (although current source versions are known) with near instantaneous and independent control of the magnitude of each phase. (2) Sine-wave voltage can be achieved through use of a suitable reference waveform and modulator but any shape can be used at will. Alternatively, closed-loop current regulation can be applied to achieve various current waveforms. The low-frequency spectrum of an inverter is well controlled but the switching action of the inverter produces high frequency distortion that can only be addressed through filtering. (3) Potential short-circuit current is high but protection against it must be provided in the form of current limiting action. (4) The current rating is set by the temperature rise of the semiconductors. The thermal time-constant of the semiconductor is very short and large over currents cause device failure in considerably less than 1 ms. The cooling system also has a relatively short thermal time-constant and so even moderate-over currents can only be tolerated for short periods unless the inverter has been over-rated to accommodate them. (2)Real power exchange is dictated by the references applied to the control system (subject to the dc-link being able to source or sink this power). There are several low-power application areas where electronic power converters are used for power distribution, such as in telecommunication systems and large computers [4]. These power systems often consist of several dc power supplies running in parallel. The systems that have developed here are quite different from public ac power grids. The operating schemes are hierarchical control-loops that actively share power between modules through distribution of current demand signals. The current demand signals are arranged to ensure that modules are not subjected to fault currents beyond their ratings and the over-all system has a limited fault current that causes collapse of the system voltage. There are several low-power application areas where electronic power converters are used for power distribution, such as in telecommunication systems and large computers [4]. These power systems often consist of several dc power supplies running in parallel. The systems that have developed here are quite different from public ac power grids. The operating schemes are hierarchical control-loops that actively share power between modules through distribution of current demand signals. The current demand signals are arranged to ensure that modules are not subjected to fault currents beyond their ratings and the over-all system has a limited fault current that causes collapse of the system voltage. of the high order filters, is likely to lie near the upper limit of the controller bandwidth. This is problematic. There is barely enough control bandwidth to suppress the filter resonance. Placing the resonance safely above the controller frequency limit is not feasible because insufficient filter attenuation would be achieved. Reducing the resonant frequency to well within the control bandwidth will require physically large components and will make the system response sluggish above that frequency. 4. Control of an inverter for power export 4.1. Grid-connected mode In grid-connected mode, the objective is to export a controlled amount of power into the established voltage. This mode of operation has been extensively examined for power flow predominantly in the opposite direction under the guise of active rectifiers [5]. Most active rectifiers employ simple first-order filters (because power quality in the medium frequency range may not be particularly important) but second-order filters are also reported. Control of the exported power is through control of the in-phase component of current. Phase-locked loop (PLL) techniques are used to ensure synchronism [6]. Control in dq-axis (synchronous reference frame) form is usually preferred to magnitude-angle form. The current demands are generated from the power demands using the local voltage magnitude (available from the PLL). The power demands themselves come from the micro-grid supervisor (MGS) with the exception of photovoltaic and heat-led CHP sources which generate autonomously. The MGS will operate either to run the generators at fixed output (presumably optimum in some sense) or load following so as to regulate export from or import to the micro-grid. The overall control structure will be similar to Fig. 3. The current is controlled in closed-loop form. The current references (d- and-q axes) are set on the basis of the required real and reactive powers. Strictly speaking, the power is set open-loop but the closed-loop current control ensures proper output power levels and power sharing between parallel units. Fig. 3 shows some standard inverter control techniques in addition to the simple current-control-loop. These are the addition of decoupling terms to compensate the natural coupling between d- and q-axes (introduced by the filter reactance) and the feed-forward of the capacitor voltage into the current control-loop. For clarity, Fig. 3omits terms that might be added to compensate variation of the dc-link voltage of the inverter. The grid voltage may not be available for monitoring so alternative connections for the PLL are shown. Because each inverter controls its output current, current sharing and power sharing are assured according to the demands established by the MGS. The design of the controller and the design of the filter need to be considered together in order to establish good attenuation of both high and low-frequency interference [7]. Controllers implemented in discrete time will also require additional decoupling terms [8]. Public electricity grids (at distribution voltage levels) often have significant harmonic distortion of the voltage. Choosing to control the output current of the inverter in an explicit current control-loop has the advantage of making the inverter a high impedance path that prevents flow of harmonic currents due to grid voltage distortion. (The filter capacitor does provide a path for harmonic currents and this must be account for in selecting its value.) In contrast, a voltage source synchronous machine will present a low-impedance path to non-triplen harmonics. optional connections). Thought must be given how the voltage magnitude signal is used in the power calculation. The current calculated as necessary to export constant power will be harmonically distorted if the voltage is harmonically distorted. It may be preferred low-pass filter the calculated current reference to remove harmonic terms. The consequence of this is that the power export will contain ripple at frequencies related to the voltage harmonics. In other words, if the voltage is distorted there are three choices: (i) export undistorted current but distorted power, (ii) export undistorted power but distorted current or (iii) suffer some degree of distortion to both current and power. Fig. 3 shows the current demand fed through a low-pass filter to ensure undistorted current references (option (i)).It would be possible to set all inverters to export balanced fundamental-frequency-only current. However, the single-phase diode-rectifier loads will still required unbalanced and harmonic current components and these will have to be provided by any machines on the micro-grid or by the public grid. Drawing these components through significant line impedances will worsen the voltage distortion. Drawing a high proportion of distortion current from the public grid may be disallowed by the utility operator or by regulations. To achieve good voltage quality for local consumers extra control functions can be added. The voltage at the filter capacitor can be sensed and filtered to identify the harmonic distortion. This feedback is used to generate a current reference to add to the active and reactive current references in order to correct the distortion. (The terms reactive current and reactive power are here used to refer to the quadrature component of fundamental frequency current and its associated reactive power.) Fig. 4 shows a suitable control structure. It requires that there is some significant impedance (at harmonic frequencies) between the filter and the voltage source of the public grid in order that there can be some influence over the local voltage. This is one approach to incorporate active power filter functionality into a power export inverter. Such techniques generally require a greater degree of instrumentation that the simple inverter controllers and a cost and complexity penalty is paid for this. If control of the voltage magnitude at the point of common coupling is allowed (or if a local bus separated from the utility grid is arranged for supply of premium loads) then control of the whole of the voltage spectrum (up to the control bandwidth limit) is possible. One such scheme has been reported [9] using a technique known as repetitive control to suppress distortion. 4.2. Island mode Whereas the task in grid-connected mode is to control current into an established voltage, the task in island mode is to establish that voltage. Fig. 5 shows the obvious approach of closing a control-loop on the second order plant. There are short comings to this approach that have long been recognized in switch-mode power supplies and machine drives [10,11] : (1) Although the voltage is controlled, there is no explicit control of current and large transient currents (dangerous to the semiconductors) can occur. (2) The plant is not simply second order: it includes a distribution network and loads. It can be viewed as a second order plant with a disturbance current. In addition, the plant parameters may not be known accurately. (3) Feed-forward of load current disturbances can be readily incorporated. The standard alternative is to use nested control-loops, Fig. 6. The plant is considered as two first-order systems. They are reasonably well decoupled because of their different time-constants. An inner current-control-loop is formed around the inductor and is arranged to have a fast response. This subsystem behaves as a near perfect controlled current from the point of view of the outer voltage-control-loop. If the current demand generated by the outer loop is limited then the inverter will not be subjected to excessive current. The controller is again shown in a rotating reference frame. The angle used to convert between abc and dq coordinates can be obtained from an oscillator that sets the system frequency. Further improvement in transient response to load changes can be obtained by feed-forward of the load (disturbance) current. Another advantage of the nested control-loop approach with its explicit current demand is that it lends itself to current (and power) sharing between parallel modules[12] . This is illustrated in Fig. 7. The technique can be readily applied to power converters in very close proximity. However, when there is some distance between them then the current demand signals need to be transmitted over an explicit communication channel. In addition, a synchronizing signal will also need to be communicated to all inverters. The bandwidth of the communication links would need to be above 100 kb/s to avoid degrading the performance of the controllers. An alternative approach to providing sharing between parallel modules that is popular in modular dc power supplies is the master–slave approach[13,14] . Nested control-loops are used for the master power converter (or inverter). The output current of this converter (or, alternatively, the total load current) is measured by the slave units and used as their current reference. Thus, the slaves attempt to supply the same current as the master (or a scaled version of that current) and steady-state sharing is achieved. The master–slave approach can be readily applied to inverters and ac systems by forming the control-loops in a rotating reference frame, Fig. 8 or even a stationary reference frame [13] . The asymmetry of the system leads the master converter to take on most of the transient duty and most of the supply of unbalanced current. This is principally because of the limited bandwidth or noise filtering in the distribution path of the current sharing signal. (Negative-sequence unbalance currents appear as double frequency terms in a rotating reference frame, a potential disadvantage of applying dq-axis control to unbalanced systems. The zero-sequence terms have not been shown in these diagrams but would need to be controlled also if four-leg inverters and a four-wire generator bus were used.) This effect is heightened if the bandwidth of the communication channel for the slave current demand has been deliberately limited. Because the transient capacity of an inverter is limited (in comparison to electrical machines) it would be helpful to provide transient-state sharing between close inverters. This can be achieved without high bandwidth communication links by partitioning the frequency spectrum and applying different controllers to different partitions, Fig. 9. Steady-state sharing of power is achieved through control of the low-frequency portion of the voltage in a control system similar to Fig. 6. The current demand for distribution to the various modules is in dq-form and limited in bandwidth to around 500 Hz (perhaps 10 kb/s). Local control-loops (for each inverter) that sense the local voltage then act to reduce the harmonic distortion (at relatively high frequency) toward zero. This gives rise to the term distributed control [13] . There are some weaknesses in the control scheme that need to be guarded against in implementation. The control is distributed but relies on one control-loop for the most important feature of voltage control. This leaves the system open to a single-point failure mechanism despite its modularity. The voltage control loop could be implemented in each inverter such that any inverter is able to assume this duty should one inverter fail. A second weakness is that the feed-forward term assumes that the total load current can be measured. In a power park of several generators in close proximity feeding one or two master busses this may be possible; in other circumstances it may not. The feed- forward term is not essential but does improve transient voltage performance proving noise and accuracy issues are addressed. 4.3. Mixed networks The distributed control scheme was not designed explicitly to incorporate traditional electrical machines as peers of the inverters. There are two approaches that could be used to form a mixed network. If the inverters dominate the network then distributed control could be used to establish an isochronous network. A synchronous machine can then be run up and synchronized to it. The machine will run as a constant power device with the power set by the combination of its governor settings and the choice of reference frequency for the inverter controller [15] . Alternatively, if the electrical machines dominate then they can form a micro-grid and the inverter can be controlled in grid- connected mode. The inverter can be set to run in constant (dispatched) power mode or given a power demand derived from the system frequency in a manner that mimics a governor droop characteristic. 5. Experimental results The experimental rig was built with an inverter of 10 kV A and was connected to the 50 Hz voltage grid via a -Y isolating transformer (rated at 10- kVA and with the -winding connected to the grid and with a voltage ratio of 415 V/208 V). The inverter parameters were: switching and sampling frequencies, 8.192 kHz, current control-loop bandwidth 1.9 kHz; C =50; L = 1.35 mH. The isolating transformer parameters were: ; Rm,,20F,s ;; . Control systems and pulse-width modulators were LH,100,LmH,100Rk,,1smm implemented on a TMS320LF2407 in discrete time form. The processor is also used to monitor several signals within the control blocks and make these available for recording by an oscilloscope via digital to analogue converters. 5.1. Grid-connected mode Fig. 10 shows the response of the power exported to the grid when the demands are subject to step changes. Both the real and reactive powers are closely regulated through the action of the current control-loops. Because the loops are in dq-form and the natural coupling is cancelled, the real power is not perturbed when the reactive power is stepped (and vice versa). The public grid voltage available in the laboratory is slightly distorted and unbalanced so the exported power shows a small ripple. As noted in Section 4.1, it is possible to suppress this ripple but only by introducing distortion to the current. Fig. 11 shows the pre-existing distortion of the available grid voltage and in particular significant fifth and seventh harmonic distortion. Fig. 12 shows the current flowing into the grid. It has a similar degree of low order harmonic distortion despite the high impedance of the inverter in this frequency range. The harmonic current is being driven by the grid voltage harmonics through the capacitor of the inverter’s filter. A voltage source generator would present a much lower impedance to voltage harmonics and larger harmonic currents would flow. Fig. 12 reveals some resonance lift at around 2 kHz that results from the inability of the controller to suppress completely the resonance of the filter. Fig. 13 shows the current over a wider frequency range and demonstrates adequate attenuation of the switching frequency components to more than 60 dB below the fundamental component. 5.2. Island mode Three inverters (30 kVA system) running the distributed control algorithm were connected to a variety of linear and non -linear loads in both steady-state and transient conditions. Fig.4 demonstrates that the voltages are reasonably free of both switching frequency and low-order harmonic distortion. The total harmonic distortion was calculated as 1.32%. This is largely due to the limited ability to suppress distortion in the region of the filter resonance because of the limited control band-width. A micro-grid is likely to supply several single-phase loads and load imbalance must be anticipated. Fig. 15 shows that the voltage controller maintains reasonable balance in the presence of a pure single-phase load. A disadvantage of control in a synchronously rotating reference frame (compared with control in a stationary frame) is that imbalance is more difficult to control. Notwithstanding this the resulting negative- and zero-sequence imbalance in voltage were only 2.9% and 1.3%. The voltage THD remained at 1.3%. The closed-loop voltage control will give good voltage regulation properties in steady-state but the transient response is a concern. Fig. 16 shows that the transient response of the voltage is well-damped and lasts less than 2 ms. Loads in domestic and commercial environments are likely to contain many uncontrolled diode-rectifiers (although IEC 1000 will put a halt to the growth of the worst of these devices).Fig. 17 shows the phase current drawn by a typical single-phase rectifier (e.g. a group of PCs). It is clear that the waveform has a very high crest factor, high harmonic distortion and low-power factor. Although the real power load is only 0.25 p.u. the current is peak is 0.7 p.u. The phase-voltage shows the common distorted profile experienced in many distribution systems. Fig. 18 confirms the presence of considerable harmonic distortion in the phase-voltage, principally third and fifth harmonics as a result of the current harmonic demand. 6. Conclusions Island mode operation of an inverter has been demonstrated in a control scheme specifically designed to accommodate parallel operation and to exploit the controllability of an inverter to deal with distorting loads. The parallel operation scheme has been designed to use low-bandwidth communication channels. Although communication is not absolutely necessary, its provision does allow for better sharing of duty in transient conditions. The control bandwidth of an inverter is limited by the switching frequency and the discrete nature of the control system. However, there is sufficient bandwidth to control actively the low-order harmonic disturbances (from non-linear loads). Acknowledgements The work described covers projects supported by Turbo- Genset PLC, Engineering and Physical Science Research Council (UK) (GR/N38190/1) and CEU Framework V (DispowerENK5-CT-2001-00522). References [1] R.H. Lasseter, Microgrids, in: IEEE Power Engineering Society Winter Meeting, vol. 1, 2002, pp. 305–308. [2] C. Marnay, F.J. Robio, A.S. Siddiqui, Shape of the microgrid, in: IEEE Power Engineering Society Winter Meeting, vol. 1, 2001, pp. 150–153. [3] G. Venkataramanan, M. Illindala, Microgrids and sensitive loads, in: IEEE Power Engineering Society Winter Meeting, vol. 1, 2002, pp. 315–322. [4] A.T. Wojciech, M.M. Jovanovi ?c, F.C. Lee, Present and future of distributed power systems, in: IEEE Applied Power Elec. Conference (APEC ’92), 1992, pp. 11–18. [5] T.C. Green, Impact of EMC directive on power converter designs, IEE Power Eng. J. 8 (1) (1994) 35–43. [6] N. Bruyant, M. Machmou, Simplified digital and analogical control of shunt active power filters under unbalanced conditions, in: IEE Power Electronics and Variable Speed Drives Conference (PEVD ‘98), September 1998, pp. 11–16. [7] M. Prodanovi?c, T.C. Green, Power quality improvement in grid connection of inverters, in: IEE PEMD Conference, Bath, Conf. Pub. No. 487, April 2002, pp. 24–29. [8] M. Prodanovi?c, T.C. Green, Control and filter design of three-phase inverters for high power quality grid connection, IEEE Trans on Power Electronics 18 (1) (2003) 373–380, Part 2. [9] J. Liang, T.C. Green, G. Weiss, Q.-C. Zhong, Evaluation of repetitive control for power quality improvement of distributed generation, in: IEEE Power Elect. Specialist Conference PESC ’02, Cairns, June 2002. [10] N. Mohan, T.M. Undeland, W.P. Robbins, Power Electronics-Converter, Applications, and Design, 3rd edition, John Wiley & Sons, 2003. [11] Murphy, Turnbull, Power Electronic Control of AC Machines, Pergamon Press, 1988. [12] T.C. Green, M.H. Taha, N.A. Rahim, B.W. Williams, Three-phase stepdown reversible AC–DC power converter, IEEE Trans. Power Elect. 12 (2) (1997) 319–324. [13] M. Prodanovi ?c, T.C. Green, Comparison of control strategies of parallel connection of inverters to a distribution grid, in: Power Electronics and Variable-Speed Drives Conference (PEVD 2000), IEE Conf. Publ. No. 475, London, U.K., September 2000, pp. 472–477. [14] K. Siri, Q. Lee, T.F. Wu, Current distribution control for parallel connected inverters, IEEE Trans. Aero. Electron. Sys. 28 (3) (1992) 829–851. [15] L.L.J. Mahon, Diesel Generator Handbook, Butterworth-Heinemann Ltd., Oxford (U.K.), 1992 (Chapter 8). 2.英文文献翻译 文摘 预测增长的小规模非50/60 Hz的电源以及希望能够支持独立于公共电力网格的负荷，需要开发变频微电网。电力电子接口不同于传统的电力设备，它有非常不同的特性,因此,不同的操作,控制和保护是必需的。 为了保持电压质量，我们还需要留意在一些网络和控制方案中用到的单相谐波控制负荷。利用逆变器的可控性来操作微型电网并提供良好的电能质量，对这样的控制方案进行检查,并与传统的电源系统和控制的直流/直流电源转换器进行比较。还讨论了沟通和控制带宽的局限性。实验结果是用来说明用不同的线性和非线性、三相和单相负载的组合所能达到的性能。 关键词:分布式发电;微电网;逆变器;电能质量 1、介绍 微电网被认为是一个整合了分布式发电的电力系统的发展方向。微电网这个术语并非严格定义，它涵盖了广泛范围内可能的系统[1 - 3]。在这里,这个术语的意思是指一个配电系统的子集，包含足够的能源供应大部分或所有的局部负载。因此，微电网能够仅通过与系统其余部分的微小的功率交流运行,有时可以独立于公共系统，作为一个独立体来运行。一个微型电网不同现有的岛状电力系统(包括物理岛屿和电气岛,如海上石油/天然气平台、船舶和飞机)在那些场合，连接和断开公共网格是一个常规的项目。有人预测说,这种负荷将会成为常态化的的民用和商业消费性负荷。这样的话,大量的单相二极管整流器负载被期望使用并且供应将会不平衡和扭曲。在一些岛状系统,例如石油平台,主要的负载(起重机、水泵等)的启动可以预测并且生成的资源可以瞬时准备就绪。这在某些类型的微型电网中是有可能的,但如果不是,负荷预测可能是重要的并且可能需要附加参与控制的要求。 微电网中使用的发电技术可能有多种形式，个人微型电网可能糅合这些。电源可能是很小的部位,如国内热电联合系统或屋顶光伏电池板或者相对大一些的,如使用燃气涡轮机或往复引擎的办公大楼规模的CHP。电源可能是间歇性的(光伏和热导CHP)或全控(电导CHP或简单柴油/天然气推动发电机)。为了充分利用间歇源和/或提供安全保护措施,微型电网可能包括一个存储元件,如飞轮、电池或氢电解/燃料电池系统。相对大的发电机,特别是现有具有微电网职能的削峰值式柴油发电机,很可能是传统的能发50/60赫兹电的三相同步发电机。新技术将是不同的。一些电源是自然直流源,如太阳能电池、燃料电池和电池组。其他的是变频或高频交流,如微型涡轮机和飞轮。这种非50/60赫兹的电源需要通过逆变器连接到微电网。这种微电网所需要的形式以及负载所支持的类型会对操作和系统的控制体制有很大的影响。然而，到目前为止,没有建立操作方案，它可能永远无法实现或提出一个普遍适用的方法。本文将描述某种特定情况下微电网的操作和控制,也就是说,微型电网由相对少量的发电电源形成,而这些电源通过逆变器连接到微电网，将供应不平衡和扭曲的负载电流。 2、电力综合园区 微电网系统就像是一个电力综合园区，它由几种电源组成，这些电源相当于提高电力服务的一群客户。客户可能是一个商业区，购物中心，或集团办公楼。如果有新的消费者，电力综合园区可能被电力短缺和网络拥堵激发。其他动力可能是希望安装热电联产或者提供优质电源。其优质的一面表现在，在发生公共电网供应中断或是电压下降时， 能够持续给重要负荷供应能量。这可能意味着避免了其他电能质量的问题。如波形失真，闪烁或膨胀。电能质量可能会因降低电源阻抗，积极的管理(如有源滤波)和/或当认为公共电网是问题的根源时断开公共电网。因为一些系统运营商目前不允许分布式发电机在常见的连接中控制电压。一些电力质量效益只有在微电网孤立运行时才会显现出来。为提高可靠性，灵活性和效率(部分负荷运行)，就需要用到几个发电机。 电力综合园区可能属于一个地主，一个建筑服务公司，消费者自己或是现有的公共设施公司。无论什么类型的发电源，发送到公共电网的电能都可能也可能不会用到。微电网一般可能有几类发电源，但电力综合园区变体可能是一个独立的主体或是一个独立的操作员。此外，发电机将成组存在，以使其物理和电紧密联系。发电技术可能是传统电气设备和逆变器的混合体。 电力综合园区有几类可能的运行模型: (1) 有固定局部发电的微电网并网运行:微电网并网将把波动的功率流认为是负载变 化。功率能够流向任何一个方向。 (2) 伴随着局部发电的带负载微电网并网运行:微电网将会有一个固定的功率流动， 它可能输入(弥补局部发电负荷的短暂下降)，可能输出(出售过剩的产能)或 保持不变(微电网波动但保持一个连接以避免意外) 3) 孤岛运行方式:局部发电将随机装载。可能使用所储存的电能或附加需求参与来( 提高安全性。 操作和控制中必须面对的问题是: (1) 如何分担发电机组之间的稳态负荷, (2) 如何分担机组之间的暂态负荷, (3) 如何做到从并网运行向独立运行的无差切换, (4) 如何兼容额外特性，例如波形质量的主动管理, 3.电机和转换器的特点 电力系统发电的历史是同步机的历史，同步机的特性影响到了整个系统的其他方面的发展。例如，保护方案依赖于电机的短期过电流(由大量热量导致)来提供必要的故障电流来判别故障电流类型。反之，故障所需的清除时间由电机的摇摆和失步特性决定。一方面电力电子转换器在OEM能源供应和工业驱动领域广泛存在，但并不影响传统的电力系统的发展，电力电子转换器具有与电机截然不同的特性，能够指导工程师制定出一套不同的输配电系统。 电机的相关特性有: (1) 作为振幅可调的电压源来操作。调节通常是有相对较低带宽的闭环励磁控制方案 的一个环节。 (2) 正弦波电压是电机功能设计的一个特点，电压的总体谐波失真很低。负载电流中 的失真将会导致一定程度上依赖于源阻抗的电压的失真。 (3) 短路电流很高，是因为源阻抗低，没有采用限流。 (4) 额定电流由绕组绝缘的温升设定。要求绕组的热时间常数以及周围的钢相对大， 还要有一个有益的短时评级。故障电流时稳态电流最大值的十倍能维持几个周 期。 (5) 有功功率交换是由转轴决定的，稳态负载共享可被应用到闭环调速器控制中，从 而使功率输出功能的(普通)系统频率。这是控制系统所设计的功能特点，但它 类似于原动机的转速在电力负荷下降时减小的自然趋势。 逆变器相应的特点是: (1) 作为电压源运作(虽然电流源类型是已知的)，这些电压源每相大小都有瞬时， 独立的的控制。 2) 正弦波电压能通过使用一个合适的参考波形和调制器来获得，但任何波形可能被( 随机用到。另外，闭环电流调节能用于回去不同的电流波形。逆变器的低频频谱 可以实现良好的控制，但逆变器的开关动作产生高频失真，只有通过过滤才能解 决。 (3) 潜在的短路电流高但对它的保护必须以一种电流的方式实现限流作用。 (4) 额定电流由半导体的温升决定。半导体的热时间常数很小并且过电流引起的设备 故障大大小于1毫秒，冷却系统也有 一个相对较小的热时间常数，所以即 使适度的过电流也只能容许短时存 在，除非逆变器已经过量设定来容纳 它们。 (5) 有功功率交换由应用于控制系统的 参量决定(受连接到电源的直流支 路的影响或使电能汇集起来) 图1.一阶和三阶衰减型滤波器的电压频率组件 在电力电子转换器被用来分配电能的场合有几类低功耗应用领域。例如，在电信系统和大型计算机中这些电力系统通常有几个并列运行的直流电源组成。这里已经开发的系统不同于公共交流电网，操作方案是分层控制回路，通过当前要求的信号分布主动的共享组块之间的电能，当前要求的信号被安排用以确保模块不受故障电流超出额定电流的影响，从而使整个系统有一个有限的可以引起系统电压崩溃的故障电流。用来补充逆变器所需要的控制和过滤功能对系统的性能有显著的影响。逆变器的开关频率受功率损耗(由开关产生)的限制。这些损耗可以再冷却系统中积累，也可能受到改性能的要求的限制。开关频率可能介于5kHz到20kHz之间，控制系统将成为一个离散时间控制器(在DSP上实现)，将有一个等同于开关频率或其他有利因素的采样率，这样的话，逆变器的控制带宽将在1 —5Hz之间，第二个问题是逆变器电压的开关频率分量将需要减小，一阶的感应滤波器是最优的选择，因为接口感应器在任何情况下都需要一个电压源逆变器连接到电压网格。然而，一阶的滤波器需要一个大的感应器来获得足够的衰减，因此滤波器可能是一个二阶的LC电路。接口感应器，耦合了变压器或线电感也可能出现这样的滤波器。因此这样的滤波器是三阶的显式或隐式的LCL电路。图 一显示了滤波器类型，图二是频谱，我 们能看到滤波器的过渡区域，包括高阶 滤波器的谐振频率，可能接近控制器带 宽的上限。这是有问题的，几乎没有足 够的控制带宽来抑制滤波器的谐振。将 共振频率设定的高于控制器频率极限， 这并不可行，因为不能实现足够的谐振 衰减。减小谐振频率到控制带宽的范围 内需要体积很大的元件，将使系统的响 应慢于上述频率。 图2.逆变电压和控制带宽的频4(功率输出逆变器的控制 谱 4.1 并网发电模式 在并网发电模式中，其目的是输出受控电量来形成电压，运行模式的已经经过广泛的检验，因为功率在有源整流器下向相反的方向流动。滤波器大多数有源整流器采用简单的一阶(因为中频范围内的电能质量可能不是特别的重要)但二阶过滤器在有应用，输出电能的控制是通过控制电流元件的同相分量实现的。锁相环(PLL)技术用来确保同步。Dqaxis(同步参考系)形式中的控制通常是首选幅角模式，当前的需求产生于使用局部电压幅值(可以来自于锁相环)时的电能需求，电能要求自身来自于由光伏和热导的能自动产生CPH源的微电网主管(MGS)。MGS要么使发电机以固定的输出运转(从某种角度来讲大概是最佳的)，要么随之装载从而使调整输入到微电网或从微电网输出的的负荷。总体控制结构类似于图3。电流以闭环的形式被控制，参考电力(d轴和q轴)基于所需的有功和无功而设定。严格的讲，电能设置于开环，但闭环电流控制确保适当的输出功率水平和并联单元之间的功率分配。 图3.并网发电模式下电力出口的控制结构(粗线代表电源连接，细线代表信号连接，虚线代表可选的连接) 图3展示了一些除了简单电流控制回路之外的标准的逆变器控制技术。这些都是补偿d轴和q轴(由滤波器电抗引入)自然耦合和电容器电压到电流控制回路的前馈反应的附加解耦。为清楚起见，图3省略了那些可能被添加到逆变器用来补偿直流母线电压的变量。微电网电压不能用来监控，因此显示了锁相环的可选连接。因为每一个逆变器控制自己的电流输出，根据MGS设定的要求需要保证共同分担电流和功率。控制器的设计和滤波器的设计需要一起考虑以使高频和低频干扰实现良好的衰减。在离散时间里使用的控制器要求额外的解耦条件。 公共电网(配电电压水平)经常有明显的电压谐波失真。在一个明确的电流控制回路中选择控制逆变器的电流输出有一大优势，使逆变器有一个高阻抗路径，从而阻止由于微电网电压畸变而导致的谐波电流的流动。(滤波电容为谐波电流提供了一个路径，并且我们必须考虑它的价值)相反，电压源同步电机将为非三相谐波提供一个低阻抗路径。必须考虑电压等级信号如何在功率计算中应用，如果电压谐波失真，电流需要必要的计算，输出恒功率也将谐波失真。计算参考电流可能首选低通滤波器，以消除谐波。这样的结果是电能输出将包含与谐波电压有关的频率波动。换句话说，如果电压发生畸变，有三种选择。(i)输出无失真电流，但功率失真。(ii)输出无失真功率但电流发生失真;(iii)或者电流和功率都遭受到某种程度上的失真。图3显示了通过低通滤波器得出的电流要求以保证参考电流不失真。(选项(i))将所有的逆变器设置的输出平衡的基频电流是由可能的。然而，单相二极管整流器负载将需要不平衡和谐波电流元件， 这些都必须由微电网中的电机提供或者由公共微电网提供。通过重要线路的阻抗来画这些元器件将是电压失真变得更加严重。画一个来自公共电网的高比例的电流失真图可能不被设备操作员或管理员接受。 为了使局部消费者获得良好的电压质量，可以用到附加功能。滤波器电容的电压可以被感知和过滤，以确定谐波失真。这种反馈用来产生一个参考电流添加到有功和无功电流来纠正失真。(无功电流和无功功率在这里指基频电流的正交分量和相关的无功功率)图4显示了一个合适的控制结构，它要求在过滤器和公共电网的电压源之间有几个重要的阻抗(以谐波频率)以便可以有部分影响局部的电压。这是一种可以将有源电力滤波器功能兼容到功率输出逆变器的方法。这样的技术一般要求更大的仪器，需要花费一个简单的逆变控制器，既昂贵又复杂。 图4.并网发电模式和附加失真抑制 如果常见耦合中电压幅值控制是被允许的(或者如果从实际微网中分离出来的局部总线被安排供应重要负荷)那么整个电压频谱的控制(达控制带宽的极限)就成为可能。同类型的方案已经有前例，使用一种被称为重复控制的技术来抑制失真。 岛状模式 鉴于并网连接模式中的任务是控制电流以产生电压，岛状模式中的任务是产生这种电压。图5显示了一种关闭二阶控制电路的方法。 这种方法有缺点，那就是长期被认为在开关模式的电能供应和电机驱动: (1) 虽然电压受控，电流没有明确的控制。而且，大的暂态电流(对半导体来说是危 险的)有可能发生。 (2) 微网不只是二阶:包括配电网和负荷。我们可以把二阶电网看作是一个干扰电流， 另外，电网参数可能无法精准的知道。 (3) 负载电流扰动的前馈行为可以很容易的结合在一起。 图5.岛状模式的单回路电压控制 可选标准是使用嵌套的控制回路，图6、电网被认为是两个一阶系统。这两个系统因为有不同的时间常数，从而理应得到很好的解耦。一个内部电流控制的回路存在于电感器周围，这样安排是为使电感器有快速的响应。这个子系统充当了一个从我外部电压回路看来近乎完美的控制电流。如果外电路产生的电流需求受限，逆变器将不会承受过电流。控制器又一次显示在旋转参考结构中。角度在abc轴和dq轴之间转换，能够从设定系统频率的振荡器获得。通过负载(扰动)电流前馈可以提高对负荷变化的暂态响应。 图6.用来控制栅极电压的嵌套型电压和电流控制环 有固定的电流需求的嵌套式控制回路方法的另一个优点是它可以给自身的各并联组块之间供应电流(和功率)，这在图7得以说明。这项技术可以很容易的应用到功率转换器中，并且两者之间的差异很小。然而，当两者之间有一定的距离时，电流要求信号通过一个显式通信信道传输。另外，所有的逆变器也需要接同步信号。通信带宽要求高于100kb/s，以防止控制器性能的退化。 图7.多个逆变器和单个电流控制回路 在并联组块之间实现共享的另一种方法是主从法，在直流电源模块中用的很多。嵌套式控制回路用于主要的功率转换器(或逆变器)，转换器的输出电流(或者，可选择用，总的负荷电流)被从属器件测量，并作为参考电流使用。这样的话，从属器件试图作为主器件提供相同的电流(或该电流按比例缩小的版本)，从而实现了稳态共享。主从法能通过在旋转参考结构中形成控制电路很容易的应用于逆变器和交流系统并且中。图8,或者甚至一个固定的参考框架。系统的不对称导致了主转换器承担了大多数的暂态影响，供应了绝大多数的不平衡电流。这主要是因为带宽有限和电流共享信号的分布路径中的噪声过滤。(负序不平衡电流作为旋转参考结构中的倍频分量出现，这是将dq轴控制应用到不平衡系统中的一个潜在的缺点，零序分量没有出现在图中，但如果用到四出脚逆变器和四线发电机总线，则零序分量也将需要控制。)倘若从属电流所要求的 信道宽度有意限制，则其带来的影响将会被强化。 图8.多个逆变器的主从安排结构 因为逆变器的暂态电容有限(与电机比较)，这将有助于在密切联系的逆变器之间实现暂态共享。即使没有高带宽的信道连接，我们也可以通过分割频谱，在不同的频带用不同的控制器来获得。图9，功率的稳态共享通过在与图6相似的控制系统中控制电压的低频分量。不同模块的电流要求以dq形式分配，并且将带宽限制在500Hz左右。(可能是10kb/s)。能测量局部电压的局部控制回路(每个逆变器)然后采取措施将谐波失真(在相对较高的频率)减小到零。这大的增加了分布控制。在控制方案中有些缺点需要防范，控制分布式的，但依赖于一个控制回路，因为具有电压控制的最为重要的特征。这使得系统适用于单点故障机制而不用在意它的模块化。任何一个逆变器中都存在电压控制回路，任何一个逆变器都能够承担这份职责。第二个缺点是前馈项假定总的负载电流是可以衡量的。在相近的几个发电机的电力网格区间铺设一到两条主控总线是可能的。但在其他情况下，也许就行不通。前馈行为并非必不可少，但确实提高了暂态电压性能，意味着噪声和精确度的问题得以解决。 4.3混合电网 图9.使用频率分区的分布式控制 分布式控制方案并没有明确的设计将传统电机作为逆变器的同类型器械进而兼并。有两种方法可以用来形成一个混合网络。如果逆变器控制网络，那么分布式控制能够用来形成同步网。同步电机可以同步运行起来，电机将作为一个衡功率设备运行，功 率由管理器设定和逆变器控制器参考频率选择联合制定。另外，如果电机主导，它们能形成一个微电网，并且逆变器能在并网连接模式下得到控制。逆变器可以设置为运行在恒(离散)功率模式，或给定一个来源于 系统频率的功率需求，用一种模拟管理器 下行特性的方式。 实验结果 5. 实验台是由10kVA逆变器建成，并通 过Y-?隔离变压器连接到50Hz电压网络 中。(额定参量为10kVA和?绕组连接到微 电网且有一个415V/208V的电压比)。逆变 器参数为:开关和采样频率，8.192kHz， 电流控制回路带宽1.9 kHz，C = 50μf、 L = 1.35 mH。隔离变压器参量为RS = 20m Ω; LS = 100μH;RM =1kΩ; LM = 100 mH. 控制系统和脉冲宽度调制器以离散的时间形图10.所要求的渐变的有功无功响应 式在TMS320LF2407上实现。处理器被用来在 控制块中的一些信号，并使示波器通过数模转换 器记录。 5.1.并网连接模式 图10显示了当电能需求受到阶跃变化的影响，输 出到电网的功率响应。有功和无功都通过电流控 制回路得到密切的控制和管理。因为回路在dq 形式，且自然解耦被取消，当无功发生阶跃时， 有功并未受到扰动(反之亦然)。实验室可达到的 公共电网电压发生轻微的失真和不平衡，因此输 出功率表现出一个较小的波动，正如4.1节中所图11.显示现有网络变化的电网电压 提到的，可以抑制这种脉动但只有引入失真电流。图11显示了电网电压已经存在的畸变，尤其是五次和七次谐波尤为明显。图12显示了 图12.输出电网电流示出对应的电压畸变的低次谐波失真。 图13.所显示的包括切换频率在内的频率范围的输出电流 电网中的电流流动。它有一个类似的低阶谐波失真，除了频谱中逆变器的高阻抗。谐波电流由电网电压谐波分量通过逆变滤波器的电容驱动所得到。电压源发电机将给电压谐波提供一个更低的阻抗。使得更大的谐波 电流流通，图12揭示了控制器无法完全抑 的共制滤波器共振而导致产生2kHz左右 振。图13显示了更宽的频率范围内的电 流，并且演示了开关频率分量得到充分的 衰减，达到低于基频60dB的频率状况。 5.2.岛状模式 运行在分布式控制算法的三个逆变器 (30kVA系统)在稳态和暂态条件下连接 各种线性和非线性的负载。 图14演示了电压应当不受开关频率和低阶谐波失真影响，总的谐波失真计算为1.32%。这很大程度上是因为由控制带宽有限导致的滤波器共振区域有限的抑制畸变的能力。微电网很可能供应一些单相负载，负载的不平 衡性必须已知。图15显示了电压控制器维图14.相电压为标幺值1.0的阻性负载，总谐波失真为1.32, 持纯单相负载合理的平衡。同步旋转的参考 结构(与稳态结构中的控制作比较)控制的一 个缺点是失衡更难控制。尽管负序和零序电压 的不平衡只有2.9%和1.3%。电压总的谐波失真 维持在1.3%。闭环控制将使稳态电压得到良好 的管理，但暂态响应一直是一个问题。图16 显示了电压的暂态响应得到很好的抑制，并且 持续时间低于2ms。国内的商业条件下负荷可 能包含许多不可控二极管整流器(尽管IEC 1000将终止这些设备最糟 糕的一部分继续恶化)图17显示了由典型 的单相整流器产生的相电流，我们可以很清图15.标幺值为1.0的相电压单相负载，电压的负序和零序不平 衡量为2.9,和1.3,, 总谐波失真为1.32,。 楚的看到波形具有很高的峰值系数，高的谐 波畸变和低的功率因数。虽然有功负载只有 0.25(标幺值)电流峰值只有0.7。相电压 显示了在许多配电系统中常见的失真。图 18确认相电压中存在相当大的谐波失真， 主要6(结论 逆变器的岛状模式运行已经在控制系统中 得到证实，尤其是用来说明并行操作和利用 逆变器的可控性处理畸变负荷。并行操作方 图16.标幺值为1.0，带有阶梯应用的相电压和相电流 案在设计中用到了低带宽的信道。虽然通信 的单相负载。 并非绝对必要，其条件允许暂态环境中更好 的责任分担。逆变器的控制带宽受开关频率 和控制系统的离散特性限制。然而，有足够的带宽来控制低阶谐波干扰的活动。(来自非线性负载) 感谢 本文提到的这项工作得到了包括工程物理科学研究委员会(英国)(GR / N38190/1)和CEU框架V(Dispower ENK5 - CT - 2001 - 00522)的支持。 引文 [ 1 ] R.H.拉塞特，微电网，在IEEE电力工程学会:冬季会议，2002卷1，305-308页。 [ 2 ] C.马尔奈，F.J.Robio，A.S. Siddiqui，在微网形状:IEEE 电力工程协会冬季会议，2001卷1，150-153页。 [ 3 ] G.温卡特摩南，M. illindala，微电网和敏感的负载，在IEEE电力工程协会冬季会议，2002卷1，315-322页。 [ 4 ] a.t.wojciech，M.M. jovanovi?F.C，李，现在和未来的分布 电力系统，在IEEE应用电力电子:会议(APEC 92)，1992，11–18页。 [ 5 ] T.C.格林，在功率转换器设计EMC指令的影响，评价电力工程J 8(1)(1994)35–43。 [ 6 ] n. bruyant，M. machmou，简化数字和模拟控制并联，有源电力滤波器不平衡条件下，在电力电子学会和变速驱动器会议(PEVD’98)，九月1998，11–16页。 [ 7 ] M. prodanovi?C，T.C.格林，并网连接的逆变器中电力品质的改善，IEE会议:光电磁探测器，Bath，Conf.Pub。487号，四月 2002，24–29页。 [ 8 ] M. prodanovi?C，T.C.格林，三相逆变器的控制和滤波器设计并网连接的高电能质量，IEEE反式电力电子18(1)(2003)373–380，部分2。 [ 9 ] J.梁，T.C.格林，G.斯，Q-C.Zhong，评价重复分布式发电的电力品质改善控制，在IEEE电力。专家会议和02，凯恩斯，六月 2002。 [ 10 ]mohan，T.M.昂德兰，W.P.罗宾斯，电力电子变换器，应用，与设计，第三版，约翰威利& Sons，2003。 [ 11 ]特恩布尔，墨菲，交流电机的电力电子控制，帕加马出版社，1988。 [ 12 ] T.C.格林，M.H.塔哈，N.A.拉希姆，B.W.威廉姆斯，三相降压可逆交流–直流功率转换器，IEEE反式。电力。12(2)。(1997)319–324。 [ 13 ] M. prodanovi?C，TC格林，并行控制策略的比较逆变器对配电网，在连接:电力电子与变速驱动器会议(PEVD 2000)，IEE会议服务。475号，伦敦，英国，九月2000，472–477页。 [ 14 ] k. Siri，Q:李，t.f.wu，并联电流分布控制逆变器，IEEE反式。航空。电子。系统。28(3)(1992)829–851。 [ 15 ] l.l.j. Mahon，柴油发电机手册，巴特沃斯海涅曼公司，牛津(英国)，1992(8章)。