Method for producing an output voltage and assembly for performing the method

Abstract

In a method for generating with a modular multilevel power converter (M2C) having a plurality of sub-modules a frequency-variable output voltage, the sub-modules are switched on and off to generate from an input voltage discrete voltage steps approximating an approximately sinusoidal alternating output voltage having a first angular frequency located between a zero frequency and a second angular frequency, and the input voltage is controlled or regulated as a function of the first angular frequency so as to be located between a lower angular frequency, which is equal to or greater than the zero frequency, and a third angular frequency, such that the input voltage increases with increasing first angular frequency, thereby reducing capacitor complexity in the power converter.

Claims

1. A method for generating a frequency variable AC output voltage from a controllable DC input voltage utilizing a multilevel power converter having a plurality of sub modules, the method comprising: switching the sub-modules on and off to generate from the DC input voltage discrete voltage steps approximating a sinusoidal alternating AC output voltage at a M2C controlled output frequency, and controlling or regulating the DC input voltage in correlation with the output frequency of the AC output voltage such that the DC input voltage increases with increasing output frequency.

2. The method of claim 1, wherein the sub-modules are switched on and off in stages by switching at least one the sub-modules on and off.

3. The method of claim 2, wherein the at least one sub-module is switched on and off by pulse-width-modulation (PWM).

4. The method of claims 1, wherein the input voltage increases monotonously as a function of the first angular frequency.

5. The method of claim 1, wherein the input voltage increases linearly at least in one interval as a function of the first angular frequency.

6. The method of claim 1, wherein the input voltage increases non-linearly at least in one interval as a function of the first angular frequency.

7. The method of claim 1, further comprising additionally controlling or regulating the input voltage as a function of at least one of a load current, a terminal voltage or load voltage of the M2C, and a torque of a machine controlled with the output voltage.

8. The method of claim 1, further comprising additionally controlling or regulating the input voltage as a function of a modulation of the common-mode voltage of a load.

9. The method of claim 1, further comprising additionally controlling or regulating the input voltage as a function of current resulting from an energy shift between energy storage devices of differing sub-modules within at least one branch of the power converter.

10. An assembly for generating a frequency variable AC output voltage from a controllable DC input voltage, the assembly comprising: a modular multilevel power converter (M2C) comprising a DC input side and a plurality of sub-modules connected in series and each forming a branch, with each branch generating a corresponding AC output voltage having a controlled frequency from the DC input voltage, and a device producing a controllably variable DC output voltage and being coupled to the DC input side of the power converter, wherein the variable DC output voltage is varied commensurate with the output frequency such that the DC input voltage to the converter increases with increasing output frequency.

11. The assembly of claim 10, wherein the device producing the variable DC output voltage comprises an adjustable transformer and a rectifier.

12. The assembly of claim 10, wherein the device producing the variable DC output voltage comprises a power converter with an intermediate current circuit.

13. The assembly of claim 10, wherein the device producing the variable DC output voltage comprises a mains-side power converter and a DC/DC converter.

14. The assembly of claim 10, wherein the device producing the variable DC output voltage comprises a power converter with an intermediate voltage circuit.

15. The assembly of claim 10, wherein the device producing the variable DC output voltage comprises a converter selected from the group of 2-level converter, 3-level NPC, 3-level flying capacitor voltage source converter, multilevel flying capacitor voltage source converter, and active neutral point clamped multi-level converter (ANPC).

Description

BRIEF DESCRIPTION OF THE DRAWING

(1) The invention will now be described in greater detail in the following on the basis of an exemplary embodiment. The related drawings show the following:

(2) FIG. 1 a circuitry implementation of a sub-module of an M2C based on the prior art.

(3) FIG. 2 a possible circuit construction of the M2C with N sub-modules per branch, based on the prior art.

(4) FIG. 3a a representation of the dependency in principle of the voltage U.sub.d on .sub.0

(5) FIG. 3b a representation of the dependency in principle of the voltage .Math..sub.LN on .sub.0

(6) FIG. 4 a circuitry example of the variation of the voltage U.sub.d. (Load: Motor/Generator (Asynchronous machine, synchronous machine, . . . )/3-phase supply)Circuitry combination of (poss. stepped) adjustable transformer, diode rectifier and M2C,

(7) FIG. 5 a further circuitry example of the variation of the voltage U.sub.d with a circuit combination of CSR (current-source-rectifier) and M2C,

(8) FIG. 6 a third circuitry example of the variation of the voltage U.sub.d with a circuit combination comprising a mains-side power converter NSR, a DC/DC converter and an MC2 and

(9) FIG. 7 a schematic representation of the output voltage of the converter (line-to-neutral voltage) as a function of the voltage U.sub.d; no PWM modulation of the sub-modules takes place.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

(10) For the M2C, various regulation and control methods are known, such as that described, for example, in M. Hagiwara, H. Akagi, PWM control and experiment of modular multilevel converters, Power Electronics Specialists Conference (PESC) 2008, PESC 2008, IEEE, pages 154-161, which have an effect on the energy fluctuations in the sub-module capacitors. In addition to the implemented control and regulation methods, the operating mode of the power converter has also a decisive impact on the energy fluctuation. For this reason the essential dependencies of the energy fluctuations are described by way of example for a method, and the influence of the voltage U.sub.d is also shown.

(11) The structure of a cell arranged in series connection or of a so-called sub-module of an M2C is shown in FIG. 1, wherein other circuitry variants are also possible.

(12) FIG. 2 shows the circuit construction of an M2C, for example for the creation of a 3-phase voltage system on the terminals U, V, W. However, the statements also apply for an arbitrary number of phases m, m{1, 2, 3, . . . }. The arm reactors Lz are important for the mode of operation, in order to avoid power compensation of the sub-modules of the converter at the moment of the switching of a cell. By this means the arm reactors can be magnetically coupled in a phase (U, V, W). Any resistors involved are not indicated in FIG. 2. The voltage U.sub.d is provided, for example, by a mains-side power converter (passive-front-end (PFE) or active-front-end (AFE)) and can be smoothed by means of capacitors in the direct voltage intermediate circuit.

(13) Known combinations of the M2C are: M2C with diode bridge (PFE) M2C with M2C (AFE) M2C with 3L-NPC-VSC as AFE.

(14) Other variations are also possible. The following characteristics are particularly advantageous for the switching of the M2C: High availability through redundancy Strong modular construction Low harmonic output voltage; quality of the output voltage can be increased by the number of sub-modules Simple adaptation to various power and voltage levels Containment of short-term mains supply failures Sure start from zero-voltage state (black start) Network applications: Island mode operation possible Network applications: Minimal or zero filter requirement Machine applications: Arbitrary machine types (asynchronous/synchronous, . . . ) possible Operation with standard motors, without filter DC bus configurations possible Standard transformer or transformerless operation possible.

(15) The essential disadvantages of the circuit are summarized as follows: High expense in terms of capacitive energy storage High expense in terms of control and/or regulation Dependency of the energy fluctuation in the energy storage devices of load current and fundamental frequency f0 of the load, and also the operating mode of the power converter With low output frequencies Increased expense on capacitors Possible increased neutral point voltage Possible additional current components in the branch current (increased losses) Considerable distortion of the output voltage.

(16) A known disadvantage from the prior art with so-called circulating-current-free operation exists, in that with the M2C the energy fluctuation in the capacitors of the sub-modules is dependent on the amplitude .sub.L and the fundamental frequency f.sub.0 of the load current.

(17) Thus without suitable additional measures, with .sub.L=constant, U.sub.d=constant in the case of a reduction in the fundamental frequency f.sub.0, an increase in the energy fluctuation in the sub-module capacitors is to be expected.

(18) This technical problem is solved by the invention.

(19) In the following very simplified considerations are examined on the M2C. These are inadequate to describe the real operating behavior, but they are adequate for the core of the disclosure of the invention. The aim of the embodiments is to show the influence of the variable voltage U.sub.d taking the example of an operating mode.

(20) At the outset a symmetrical structure of the converter is considered. Related components such as arm reactors L.sub.z and sub-modules (SM) have identical parameters. The structure of the converter with N sub-modules per branch is shown in FIG. 2. In the model used the arm reactors in a phase are not magnetically coupled. The lines of argument relating to the energy fluctuations on an M2C with magnetically coupled arm reactors are similar and are therefore not separately specified.

(21) In the literature, various methods are known for the regulation of an M2C. For this description no reference is made to these. Simple physical considerations are sufficient for the stationary operation. These relate merely to the fundamental oscillation. An influencing of the common mode voltage through modulation of the sub-modules does not take place. Further, no additional branch current components are noted. Furthermore, the voltage drops on the arm reactors L.sub.z are neglected.

(22) The stationary symmetrical operation is considered as the operating mode. This is characterized in that all 6 branches have to achieve similar sum voltages of the modules, which are, however, offset in time. During symmetrical operation of the converter the upper and lower modules of a phase must respectively provide the time-averaged voltage

(23) $\begin{array}{cc}{u}_{\mathrm{dSM}}=\frac{u_d}{2N}& 1\end{array}$
at the terminals. For the generation of a rotating field at the terminals U, V, W a sinusoidal modulation of the form

(24) $\begin{array}{cc}{u}_{\mathrm{SinSM}}=\frac{0_{LN}}{N}\sin \left({}_{0}t+x\right)& 2\end{array}$
is also required. Thus it follows that the voltages to be achieved at the terminals of the sub-modules are as follows:

(25) $\begin{array}{cc}{u}_{K1,1}=\frac{1}{N}\left(\frac{u_d}{2}-{\hat{U}}_{LN}\sin \left({}_{0}t+{}_u\right)\right)u_{K1,2}=\frac{1}{N}\left(\frac{u_d}{2}+{\hat{U}}_{LN}\sin \left({}_{0}t+{}_u\right)\right)u_{K1,3}=\frac{1}{N}\left(\frac{u_d}{2}-{\hat{U}}_{LN}\sin \left({}_{0}t+{}_u-\frac{2}{3}\right)\right)u_{K1,4}=\frac{1}{N}\left(\frac{u_d}{2}+{\hat{U}}_{LN}\sin \left({}_{0}t+{}_u-\frac{2}{3}\right)\right)u_{K1,5}=\frac{1}{N}\left(\frac{u_d}{2}-{\hat{U}}_{LN}\sin \left({}_{0}t+{}_u+\frac{2}{3}\right)\right)u_{K1,6}=\frac{1}{N}\left(\frac{u_d}{2}+{\hat{U}}_{LN}\sin \left({}_{0}t+{}_u+\frac{2}{3}\right)\right)& 3\mathrm{to}8\end{array}$
wherein voltage drops at the arm reactors and any resistances present are neglected.

(26) A load at the terminals U, V, W is taken into account, which draws the following sinusoidal currents:

(27) $\begin{array}{cc}{i}_{L1}={\hat{I}}_L\sin \left({}_{0}t+{}_{\mathrm{iL}}\right)i_{L2}={\hat{I}}_L\sin \left({}_{0}t+{}_{\mathrm{iL}}-\frac{2}{3}\right)i_{L3}={\hat{I}}_L\sin \left({}_{0}t+{}_{\mathrm{iL}}+\frac{2}{3}\right)& 9\end{array}$

(28) The apportionment of the load currents to the branches 1 to 6 is assumed to be uniform. Thus for the currents in the branches it follows that:

(29) $\begin{array}{cc}{i}_{z1}=\frac{1}{3}{i}_d+\frac{1}{2}{i}_{L1}{i}_{z2}=\frac{1}{3}{i}_d-\frac{1}{2}{i}_{L1}{i}_{z3}=\frac{1}{3}{i}_d+\frac{1}{2}{i}_{L2}{i}_{z4}=\frac{1}{3}{i}_d-\frac{1}{2}{i}_{L2}{i}_{z5}=\frac{1}{3}{i}_d+\frac{1}{2}{i}_{L3}{i}_{z6}=\frac{1}{3}{i}_d-\frac{1}{2}{i}_{L3}& 10\mathrm{to}15\end{array}$
wherein i.sub.d is the current at the DC terminal. On the basis of the stationary operation,

(30) $\frac{{\mathrm{di}}_d}{\mathrm{dt}}=0,$
thus for the voltage drop across the reactors L.sub.d it follows that u.sub.Ld=0.

(31) Also in the case of practical applications the voltage drop at the resistances R.sub.d can frequently be neglected, so that in the following it can be taken that u.sub.d=U.sub.d.

(32) In the time-averaged mean at each of the j sub-modules of a branch, the i, (i{1, 2, . . . , 6}, j{1, 2, . . . , N}) power
p.sub.SM,ij =u.sub.Kl,ijl.sub.SM,i =u.sub.Kl,ijl.sub.zi16
is converted. The following considerations relate to modules of the branch 1; however they similarly apply to the modules of the remaining branches 2 and 3. The following power values are converted on the modules of the branch 1:

(33) $\begin{array}{cc}{p}_{\mathrm{SM},1j}=\frac{1}{N}\left(\begin{array}{c}\begin{array}{c}\frac{1}{6}{U}_d{i}_d-\frac{1}{4}{\hat{U}}_{LN}{\hat{I}}_L\cos \left({}_U-{}_{\mathrm{iL}}\right)-\\ \begin{array}{c}\frac{1}{3}{\hat{U}}_{LN}{i}_d\sin \left({}_{0}t+{}_U\right)+\\ \frac{1}{4}{U}_d{\hat{I}}_L\sin \left({}_{0}t+{}_{\mathrm{iL}}\right)+\end{array}\end{array}\\ \frac{1}{4}{\hat{U}}_{LN}{\hat{I}}_L\cos \left(2{}_{0}t+{}_U+{}_{\mathrm{iL}}\right)\end{array}\right),\mathrm{.Math.})& 17\end{array}$

(34) In order for the energy to remain constant over a period in the energy store of the sub-module, the following must apply:

(35) $\begin{array}{cc}{i}_d=\frac{3}{2}\frac{{\hat{U}}_{LN}}{U_d}{\hat{I}}_L\cos \left({}_U-{}_{\mathrm{iL}}\right)& 18\end{array}$

(36) With this condition equation 17 becomes:

(37) 0$\begin{array}{cc}{p}_{\mathrm{SM},1j}=\frac{{\hat{I}}_L}{4N}\left(\begin{array}{c}-2\frac{{\hat{U}}_{LN}^2}{U_d}\cos \left({}_U-{}_{\mathrm{iL}}\right)\sin \left({}_{0}t+{}_U\right)+\\ U_d\sin \left({}_{0}t+{}_{\mathrm{iL}}\right)+{\hat{U}}_{LN}\cos \left(2{}_{0}t+{}_U+{}_{\mathrm{iL}}\right)\end{array}\right)& 19\end{array}$

(38) In order to further simplify the discussion, on the assumption of an inductive load, i.e.

(39) ${}_U={}_{\mathrm{iL}}+\frac{}{2},$
the following conclusions can be drawn and also relate to general load cases.

(40) The current direction .sub.u is fixed as .sub.u=0. It then follows from equation 19, that:

(41) $\begin{array}{cc}{p}_{\mathrm{SM},1j}=\frac{{\hat{I}}_L}{4N}\left(U_d\sin \left({}_{0}t\right)-{\hat{U}}_{LN}\sin \left(2{}_{0}t\right)\right)& 20\end{array}$

(42) The relationship describes the periodic energy input into a sub-module. This is directly proportional to the load current .sub.L and leads to a periodic change in the energy content in the energy storage device of the sub-module. The energy content can be determined by integration of equation 20,

(43) $\begin{array}{cc}{W}_{\mathrm{SM},1j}=-\frac{U_d{\hat{I}}_L}{4N{}_0}\left(\cos \left({}_{0}t\right)-\frac{1}{2}\frac{{\hat{U}}_{LN}}{U_d}\cos \left(2{}_{0}t\right)\right)+W_{\mathrm{ij}0}& 21\end{array}$

(44) with the, on average, stored energy W.sub.ij0 on the stationary work regime. The time-dependent portion of the energy content is directly proportional to the load current and indirectly proportional to the angular frequency .sub.0.

(45) In the special case

(46) ${\hat{U}}_{LN}\frac{U_d}{2}$
equation 21 can be simplified to

(47) $\begin{array}{cc}{W}_{\mathrm{SM},1j}-\frac{U_d{\hat{I}}_L}{4N{}_0}\cos \left({}_{0}t\right)+W_{\mathrm{ij}0}& 22\end{array}$

(48) In this case the amplitude of the energy change is directly proportional to the voltage U.sub.d; this provides a strong impetus for the voltage U.sub.d not to be constant, but rather variable. In this way U.sub.d cannot be varied arbitrarily.

(49) Due to the structure of the MC2 the maximum achievable line-to-neutral voltage without influence of the common mode voltage is restricted by modulation to:

(50) $\begin{array}{cc}0.Math.{\hat{U}}_{LN}.Math.\frac{1}{2}{U}_d& 23\end{array}$

(51) This equation can also be otherwise interpreted. With the required voltage .Math..sub.LN the following must apply:
U.sub.dz .Math..sub.LN24

(52) It remains to be established that, through variation of the voltage U.sub.d the energy fluctuation in the energy stores can be suitably influenced. In addition to the reduction in the current .sub.L this also represents a further possibility to limit the energy fluctuation in the energy stores with variable angular frequency .sub.0. However, in particular in the case of machine applications the reduction of the current .sub.L on the basis of the thus reduced torque is mostly only possible with restrictions.

(53) According to the invention, the voltage U.sub.d is variable in accordance with U.sub.d=U.sub.d (.sub.0). By suitable selection of U.sub.d (.sub.0) the energy fluctuation in the modules in the application can be appropriately influenced.

(54) As an approach, a linear dependency of the DC voltage on the angular frequency .sub.0

(55) $\begin{array}{cc}{U}_d\left({}_0\right)=U_{d,n}\frac{{}_0}{{}_N}& 25\end{array}$

(56) is considered, wherein other approaches are also possible. This linear relationship cannot be achieved in practice for arbitrary angular frequencies .sub.0, so that as a range of validity, for
.sub.2=.sub.N .sub.1<.sub.0<.sub.226
must be considered. The dependencies are sketched in FIG. 3a. A motor is considered as a load on the terminals U, V and W. In order to simplify the considerations, it is assumed that the machine is U/f controlled. The dependency of the line-to-neutral voltage can be approximately described with

(57) $\begin{array}{cc}{\hat{U}}_{LN}\left({}_0\right)={\hat{U}}_{LN,n}\frac{{}_0}{{}_N}& 27\end{array}$

(58) The curve is sketched in FIG. 3b. In order to determine the energy fluctuation in the energy storage devices of the sub-modules of the branch 1 integration of equation 19 is required.

(59) $\begin{array}{cc}{w}_{\mathrm{SM},1j}=\frac{U_d{\hat{I}}_L}{4N{}_0}\left(\begin{array}{c}\begin{array}{c}-\cos \left({}_{0}t+{}_{\mathrm{iL}}\right)+\\ \frac{1}{4}{\hat{u}}_{LN}\sin \left(2{}_{0}t+{}_U+{}_{\mathrm{iL}}\right)+\end{array}\\ \frac{1}{2}{\hat{u}}_{LN}^2\cos \left({}_U-{}_{\mathrm{iL}}\right)\cos \left({}_{0}t+{}_U\right)\end{array}\right)+W_{\mathrm{ij}0},& 28\end{array}$
wherein the amplitude .Math..sub.LN is expressed by the reference value .sub.LN

(60) 0$\begin{array}{cc}{\hat{u}}_{LN}=2\frac{{\hat{U}}_{LN}}{U_d}& 29\end{array}$

(61) Insertion of equations 25 and 27 in equation 28 yields:

(62) $\begin{array}{cc}{w}_{\mathrm{SM},1j}=\frac{U_{\mathrm{dn}}{\hat{I}}_L}{4N{}_N}\left(\begin{array}{c}\begin{array}{c}-\cos \left({}_{0}t+{}_{\mathrm{iL}}\right)+\\ \frac{1}{4}{\hat{u}}_{LN,n}\sin \left(2{}_{0}t+{}_U+{}_{\mathrm{iL}}\right)+\end{array}\\ \frac{1}{2}{\hat{u}}_{LN,n}^2\cos \left({}_U-{}_{\mathrm{iL}}\right)\cos \left({}_{0}t+{}_U\right)\end{array}\right)+W_{\mathrm{ij}0},\mathrm{with}{\hat{u}}_{LN,n}=2\frac{{\hat{U}}_{LN,n}}{U_{d,n}}.& 30\end{array}$

(63) From this, it follows that the amplitude of the energy fluctuation in the sub-modules for angular frequencies .sub.0
.sub.1<.sub.0<.sub.2
with .sub.L, .sub.u, .sub.iL being equally constant irrespective of the angular frequency .sub.0.

(64) In the following, circuitry examples for the variation of the voltage U.sub.d are presented. In the foregoing considerations the effect of a variable voltage U.sub.d has been described. In the FIGS. 4, 5 and 6 examples of several circuitry principles are sketched, which facilitate a variable voltage U.sub.d. No claim is made to completeness here. A discussion of the advantages/disadvantages of the options is not undertaken here.

(65) If the voltage U.sub.d is provided via a diode rectifier, then a variable voltage U.sub.d can be obtained by changing the input voltage of the rectifier. This is represented in FIG. 4.

(66) A circuitry example for the variation of the voltage U.sub.d, which also can be operated without a transformer, is shown in FIG. 5. The variable voltage U.sub.d is in this case provided via a current-source-rectifier (CSR) with semiconductors which can be switched on, or on and off.

(67) Switching combinations as shown in FIG. 6 are also possible. Here the direct voltage of the mains-side power converter (NSR) is again adapted via a DC/DC converter.

(68) In the case of multilevel converters it is known that the quality of the output voltage reduces as the phase control factor decreases. This effect is illustrated in FIG. 7 (solid line, dashed and dotted lines).

(69) An improvement in the voltage quality can be achieved through different voltage levels of the individual stages, the sub-modules. These different voltage levels of the stages can be achieved through different voltages in the sub-module capacitors. However, this approach is generally unfavorable from a modularity point of view.

(70) A further possibility for the increase of the voltage quality is the matching of the mean sub-module voltage to the voltage U.sub.d by changing the energy stored in the energy storage devices of the sub-modules.

(71) The matching of the mean voltage corresponds to a changed step height of the output voltage. In FIG. 7 the resultant output voltage curve for the matched sub-module capacitor voltage is indicated by the dashed line

(72) The reduction in the energy fluctuation leads to a minimization of the sub-module capacitor requirement, a simplification of the control and regulation of the M2C, a reduction in overall losses, the avoidance of high modulation of the common mode voltage and/or additional branch current components (circulating currents) and also to an extension of the range of application of the M2C, for example for continuous operation at low frequencies.

(73) An additional advantage is the matching of the capacitor voltage to the variable voltage U.sub.d. This enables a higher quality of output voltage with a lower output voltage amplitude with the already available sub-modules. Additional redundancies are possible, but are not necessary.