DE102013110805A1 - Method for controlling converters - Google Patents

Abstract

The invention, which relates to a method for controlling converters, the object is to provide a method for controlling converters, with which a better decoupling of the switch positions with respect to the input variables of the approximated Gleitregimeregler is achieved. This object is achieved by making the selection of the further subsystem, the selection of the further reference curves and the choice of the functions that act as input of the approximated sliding regime controller from a selection of all system variables such that the manipulated variables are approximated with respect to the input variables of the approximated ones Gleitregimeregler have an adjustable coupling.

Description

Method for controlling power converters, in which a control system in the form of a mathematical model describing a power converter and the load connected thereto as a mathematical relationship between system variables, consisting of subsystems, is used, whereby approximate sliding regime controllers are used to calculate the manipulated variables used for the control Reference curves are selected for some system variables, whereby the control system is extended by a further subsystem for which reference curves are also selected and in that the inputs of the approximated slip regime controller are selected from a selection of all system variables by means of which the power converter is controlled ,

When regulating converters, it may happen that fewer manipulated variables are required to meet the requirements of the application than are available in the system to be controlled. This problem can be avoided by using modulation techniques. However, if approximated sliding-sequence controllers, for example hysteresis controllers or switching tables, are used, the system to be controlled can be extended by a further subsystem for which a reference profile is specified in order to adapt the number of sliding functions to the number of manipulated variables. For example, in [ UTKIN, V .: Sliding mode control design principles and applications for electric drives, IEEE Trans. Ind. Electron. vol. 40, no. 1, pp. 23-36, 1993 ] this is applied, but the chosen approach only solves the problem mentioned and does not allow any further optimizations.

Since the derivation of a sliding function, such as a string current can also be dependent on the other switch positions, this has the consequence that switching actions that are triggered due to reaching the switching threshold of a string current affect the course of the other string currents. This phenomenon is referred to herein as "coupling of the switch positions".

A major disadvantage of the previous approach is that the adjusting switching frequencies can not be determined in advance in a simple manner, which complicates in particular the interpretation of power electronic components.

The problem of determining the occurring switching frequencies is usually solved simulatively by considering the solution of the closed-loop system over a sufficiently long period of time. This procedure has the disadvantage that, depending on the initial state, only one solution is considered at a time, and therefore the largest possible time interval must be taken into account in order to obtain a meaningful result. This method is therefore relatively time consuming. Moreover, it is very difficult to determine the switching frequency as a function of time-varying system parameters, such as the mains voltages. Another disadvantage of the previous solution is that regardless of the implementation of the approximated Gleitregimeregler injury to the tolerance bands occur.

• a) eine zeitdiskrete Implementierung,
• b) (gewollte oder parasitäre) Totzeiteffekte und
• c) im Entwurf der Regler vernachlässigte Kopplungseffekte.

Causes of the breaches of the tolerance band are in particular:

• a) a discrete-time implementation,
• b) (intentional or parasitic) dead time effects and
• c) Neglected coupling effects in the controller design.

• • Das Toleranzband wird nicht immer voll ausgenutzt [ JENNI, FELIX und DIETER WÜEST: Steuerverfahren für selbstgeführte Stromrichter. B. G. Teubner Stuttgart, 1995 .]
• • Es können Toleranzbandverletzungen um das Zweifache auftreten. D. h., bei einer Hysterese mit unterer Schaltschwelle – Δ / 2 und oberer Schaltschwelle Δ / 2 gilt für den Fehlerbetrag |e| ≤ Δ [JENNI, FELIX und DIETER WÜEST: Steuerverfahren für selbstgeführte Stromrichter. B. G. Teubner Stuttgart, 1995].
• • Ermittlung der Schalthäufigkeit ist zeitaufwendig, da im Allgemeinen eine Simulation notwendig ist.

Disadvantages of the previous procedure:

• • The tolerance band is not always fully exploited [ JENNI, FELIX and DIETER WÜEST: Control procedure for self-commutated converters. BG Teubner Stuttgart, 1995 .]
• • Tolerance band violations can occur twice. That is, at a hysteresis with lower threshold - Δ / 2 and upper switching threshold Δ / 2 applies to the error amount | e | ≤ Δ [JENNI, FELIX and DIETER WÜEST: Control method for self-commutated converters. BG Teubner Stuttgart, 1995].
• • Determining the switching frequency is time-consuming, since a simulation is generally necessary.

The invention is therefore based on the object of specifying a method for controlling converters, with which a better or complete decoupling of the switch positions of a power converter and thus a lower or no influence of the switches are achieved with each other.

In addition, optional complete decoupling is to be achieved, so that in [ FEHR, H. and GENSIOR, A. and RUDOLPH, J .: Flatness-based switching frequency estimation of sliding mode controllers for single-input systems. IECON 09 Conference Proceedings, pp. 1555-1560, 2009 .] described method for estimating the switching frequency can be applied.

In addition, a better utilization of the available intermediate circuit voltage is to be achieved. By this is meant that to demonstrate the stability compared to the conventional method, a lower DC link voltage is required.

Another object of the invention is to avoid violations of the tolerance band by the coupling effects described above.

According to the invention, the object is achieved in a method for controlling power converters of the type mentioned in that the choice of the further subsystem, the choice of other reference curves and the choice of functions that act as input of the approximated Gleitregimeregler, from a selection of all system variables are made such that the manipulated variables with respect to the input variables of the approximated Gleitregimeregler have an adjustable coupling.

The gist of the invention is the choice of a further subsystem as well as the choice of new system variables, the planning of suitable reference trajectories for these quantities, and the choice of sliding functions used as inputs to the approximated slip regroup controllers. An example of this is the introduction of size z _ν , which can be understood as a virtual stream. The peculiarity is that this subsystem does not directly describe a physical situation, so that its solution can be determined numerically by simulation in real time, but from a control point of view is considered as part of the overall system to be controlled. A central point for the invention is that the choice of the new subsystem, parameterized in the exemplary embodiment by L _ν , the new system variables and the new sliding functions on the coupling of the switch positions can be influenced.

Here, the sliding functions are chosen such that the system can achieve an approximated sliding regime.

In one refinement of the method, it is provided that the selection of the further subsystem, the selection of the further reference curves and the choice of the functions which act as input of the approximated sliding regime controllers are made from a selection of all system variables in such a way that the manipulated variables are related to the input variables the approximated Gleitregimeregler are decoupled.

This is particularly relevant if the above-mentioned method for predetermining the switching frequency is to be used.

In one embodiment of the invention it is provided that a time-dependent coupling is achieved.

In another embodiment, it is provided that the system to be controlled is a mains converter with or without a filter upstream of the mains converter.

This configuration is considered in the embodiment in the case of a two-point converter, which is connected via an L-filter to a network. In principle, however, many other technical configurations are possible for this application, for example by instead of the two-point rectifier, a three- or multi-point converter, z. Eg "3L (active) neutral point clamped converter", "3L flying capacitor converter", "series connected H-bridge converter", "modular multilevel converter" or others, or an LCL filter instead of the filter , a transformer or another filter is used or is completely dispensed with a filter.

In an embodiment variant of the invention it is provided that the system to be controlled is an electromechanical energy converter with or without a filter upstream of the energy converter.

As electromechanical energy converters are mainly electrical machines in question, for example, asynchronous machines, synchronous machines, double-fed asynchronous machines, reluctance machines and others, but also linear actuators and magnetic bearings and combinations of magnetic bearings and drives.

The invention will be explained in more detail with reference to embodiments. In the accompanying drawing shows

1 a schematic of a two-step inverter from the prior art.

The 1 shows a schematic representation of a two-point AC converter on the network.

To explain the method, a two-point converter is considered, which is schematically in the 1 is shown.

The quantities V _i with i = 1, 2, 3 denote the mains voltages, z _i the mains currents and s _i ε [-1, 1], i = 1, 2, 3 the switch positions of the power converter. The DC link _{voltage is} described by V _c . The mathematical model is in rotating coordinates L _{d d} = V _d - V _c s _d / 2 + Lωz _q (1a) Lz _q = V _q - V _c s _q / 2 - Lωz _d (1b) the streams, and similarly all other string-related quantities, were transformed as follows:

Where θ is the network angle and ω is the network frequency and θ is true. = ω. The choke is parameterized by the inductance L. In addition, apply z ₀ = 0 (3) z _e = V ₀ -s ₀ V _c / 2 -V _c / 2 (4) which, however, will not be considered further below.

The model (1) is differentially flat with the flat output y = (z _d , z _q ). Thus reference trajectories t ↦ z _{d, d} (t) and t ↦ z _{q, d} (t) can be planned for which a stable follow-up behavior of the system is to be achieved. For this purpose, an approximated Gleitregimeregler should be used.

Since the flat output of the system (1) has two components, but three switch positions, which are understood here as input variables, are available in stationary coordinates, an additional condition is required in order to be able to construct three independent sliding functions.

erweitert, das auch als Differentialgleichung für einen „virtuellen” Strom z_ν interpretiert werden kann, wobei L_ν dann die korrespondierende „virtuelle” Induktivität repräsentiert.Therefore, the system (1) becomes the subsystem

which can also be interpreted as a differential equation for a "virtual" current z _ν , where L _ν then represents the corresponding "virtual" inductance.

The system consisting of (1) and (5) is differentially flat with the flat output y = (z _d , z _q , z _ν ) and now it is possible to have three independent sliding functions for the controlled variables z _d , z _q and z _ν to construct. A choice of sliding functions is e = (e _d , e _q , e _ν ) ^T = (z _d -y _{1, d} , z _q -y _{2, d} , _{zν -y 3, d} ) ^T (6) where t ↦ y _{i, d} (t), i = 1, 2, 3 are reference trajectories for the components of the flat output.

Transform e as follows: e ₁₂₃ = (e ₁ , e ₂ , e ₃ ) ^T = (BS) ^T e (7) This gives new sliding functions, where B = diag (L ^-1 , L ^-1 , L _v ^-1 ).

mit f = (f_d, f_q, f_ν)^T und

It should be noted that the achievement of a slip regime for e implies that a slip regime is achieved for e ₁₂₃ , and vice versa. For the derivatives of the new sliding functions applies

with f = (f _d , f _q , f _ν ) ^T and

You can show that s _i = -sign (e _i ), i = 1, 2, 3 (12) is a law of rules, which under certain conditions ensures that e reaches zero in finite time.

One implements e ₁ , e ₂ , e ₃ as inputs of e.g. As hysteresis, wherein the performance is determined by the effect of each other switch on the error e ₁ , e ₂ , e ₃ . This influence can be examined by the derivation of (7), thus (8). Of particular interest is therefore the term K = (BS) ^T BS, because it determines which switching position s _i has an effect on the change of the errors e _i , with i = 1, 2, 3.

K can be represented as follows:

If one chooses _Lν = L / √2 , we get the diagonal matrix K = 1 / 3L² diag (2, 2, 2). In this case, therefore, a decoupling of the switch positions with respect to the sliding functions e _{123 is achieved} since the switching positions s _i only influence the corresponding error e _i , with i = 1, 2, 3.

The method in the application for mains converter can be generalized, so that a decoupling of the switch positions is also achieved for other sliding functions. This succeeds by also choosing the virtual subsystem accordingly. A method for choosing an alternative virtual subsystem can be given as follows: The starting point is the representation e ₁₂₃ = (b ₁ (e _d , e _q , e _ν ), b ₂ (e _d , e _q , e _ν ), b ₃ (e _d , e _q , e _ν )) ^T (14) for alternative sliding functions, where b _i , i = 1, 2, 3 functions of the errors e _d , e _q and e _ν are with the property that the image e ₁₂₃ assumes zero only if e is zero, and vice versa.

In principle, it is possible, and possibly also useful for other systems, if the functions b _i , i = 1, 2, 3 also depend on other system variables. For the sake of simplicity, however, this description will be omitted.

For the derivatives of the errors e _i , i = 1, 2, 3 then applies

Taking into account (1), one obtains using the approach

wobei g_i, i = 1, 2, 3 Funktionen der Systemvariablen sind. Ersetzt man s_d und s_q unter Verwendung von S, so erhält man mit γ = 2π/3

Further

where g _i , i = 1, 2, 3 are functions of the system variables. Substituting S _d and s _q using S yields γ = 2π / 3

gelten.If one wants to reach that e _i is independent of s and s _{j k} with i, j, k ε {1, 2, 3} and i ≠ j, j ≠ k and i ≠ k, so must

be valid.

If one chooses ∂b _i / ∂e _ν = 1, then one can choose ∂b _i / ∂e _d and ∂b _i / ∂e _q according to the conditions mentioned above and the result is

The functions b _i , i = 1, 2, 3 can finally be determined by integration according to e _d and e _q and a coefficient comparison.

Finally, it should be noted that the described method basically also comes into question for other applications in which more input variables than controlled variables occur.

This is often the case for power electronic applications, such as the operation of electrical machines on such a power converter or the use of other line filters as examples, the method also being applicable in connection with other converter topologies.

In the following, the invention will be described in another generalized example to illustrate the basic approach when the method is to be applied to other systems.

beschreiben lassen, wobei f_i und g_i,k Funktionen von x sind.Starting point are systems that are characterized by a model of form

Let f _i and g _{i, k be} functions of x.

erweitert (wobei dieses beispielsweise in Echtzeit im Regler integriert werden kann), wobei x ~ neue Systemvariablen und f ~_i und g ~_i,k Funktionen der Größen x und x ~ sind.This model will be around the subsystem

(for example, this can be integrated in the controller in real time), where x ~ are new system variables and f ~ _i and g ~ _{i, k are} functions of the quantities x and x ~.

It is then possible to schedule reference curves x _d for x and x ~ _d for x ~.

i = 1, 2, ..., m betrachtet. Der Einfachheit halber soll davon ausgegangen werden, dass Ṡ_i nur von u_i, i = 1, 2, ..., m abhängen soll, im Allgemeinen können die Abhängigkeiten aber jedoch auch vertauscht werden. Eine Entkopplung der Schalterstellungen bezüglich der Funktionen S_i erreicht man, wenn

gelten, für alle i = 1, 2, ..., m, k ≠ i. Um dies zu erreichen, sind die Funktionen g ~_i,k, i = 1, 2, ..., n ~, k = 1, 2, ..., m und S_i, i = 1, 2, ..., m geeignet zu wählen. An die Wahl von f ~_i, i = 1, 2, ..., n ~ resultieren aus diesen Betrachtungen keine Anforderungen, sodass diese auch zu Null gewählt werden können.The goal is now, functions S _i : (x, x ~) ↦S _i (x, x ~), S _i (x _d , x ~ _d ) = 0 i = 1, 2, ..., m whose derivation depends only on one manipulated variable and each manipulated variable appears only in one of the derivations. These are the derivatives

i = 1, 2, ..., m considered. For the sake of simplicity, it should be assumed that Ṡ _i should depend only on u _i , i = 1, 2, ..., m, but in general the dependencies can also be interchanged. A decoupling of the switch positions with respect to the functions S _i can be achieved if

apply, for all i = 1, 2, ..., m, k ≠ i. To achieve this, the functions g ~ _{i, k} , i = 1, 2, ..., n ~, k = 1, 2, ..., m and S _i , i = 1, 2, .. ., m suitable to choose. From the considerations of f ~ _i , i = 1, 2, ..., n ~, no requirements result from these considerations, so that these can also be chosen to be zero.

If one can prove with the aid of the prior art that S _i , i = 1, 2,..., M are sliding functions and if a stable slip regime can be established when using slip control controllers, then these functions are used as inputs to approximated slip regimens. By selecting x _d and in particular x ~ _d , for example, the switching frequency can be influenced.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited non-patent literature

• UTKIN, V .: Sliding mode control design principles and applications for electric drives, IEEE Trans. Ind. Electron. vol. 40, no. 1, pp. 23-36, 1993 [0002]
• JENNI, FELIX and DIETER WÜEST: Control procedure for self-commutated converters. BG Teubner Stuttgart, 1995 [0007]
• FEHR, H. and GENSIOR, A. and RUDOLPH, J .: Flatness-based switching frequency estimation of sliding mode controllers for single-input systems. IECON 09 Conference Proceedings, pp. 1555-1560, 2009 [0009]

Claims (5)

Method for controlling power converters, in which a control system in the form of a mathematical model, which describes a power converter and the load connected thereto as a mathematical relationship between system variables, consisting of subsystems is provided for control, wherein approximated Gleitregimeregler used to calculate the control variables used for the control Reference curves are selected for some system variables, whereby the control system is extended by a further subsystem for which reference curves are also selected and in that the inputs of the approximated slip regime controller are selected from a selection of all system variables by means of which the power converter is controlled , characterized in that the selection of the further subsystem, the selection of the further reference curves and the choice of the functions which function as input of the approximated sliding regime controllers, are selected from a selection of a System variables are made such that the manipulated variables with respect to the input variables of the approximated Gleitregimeregler have an adjustable coupling. A method according to claim 1, characterized in that the choice of the further subsystem, the choice of further reference curves and the choice of functions that act as input of the approximated Gleitregimeregler, is made of a selection of all system sizes such that the manipulated variables with respect to the input variables the approximated Gleitregimeregler are decoupled. A method according to claim 1 or 2, characterized in that a time-dependent coupling is achieved. Method according to one of claims 1 to 3, characterized in that the system to be controlled is a mains converter with or without a mains converter upstream filter. Method according to one of claims 1 to 3, characterized in that the system to be controlled is an electromechanical energy converter with or without an energy converter upstream filter.

Images