METHOD FOR FEEDING ELECTRICAL POWER INTO AN ELECTRICAL SUPPLY NETWORK

Description

[0001] The present invention relates to a method for feeding electrical power into an electrical supply grid. The present invention also relates to a device, in particular a wind turbine, for feeding electrical power into an electrical supply grid.

[0002] It is well known that wind turbines can feed power into a three-phase electrical grid. It is also known that, in addition to pure power supply, wind turbines can also perform grid support functions. In particular, to support the electrical grid, the feed-in can be adjusted depending on grid conditions such as grid voltage or grid frequency.

[0003] When evaluating or considering such grid conditions, it is often assumed that large power plants with directly coupled synchronous generators determine the behavior of the grid. Wind turbines or other decentralized feeders assume similar behavior and react accordingly. It is particularly important to note that systems that feed in power via converters, especially wind turbines that feed in power via a full-scale converter or full-scale converter concept, can react very dynamically to changes in grid conditions. Their dynamics are essentially freely definable. In particular, unlike directly coupled synchronous generators, they exhibit hardly any physically determined dynamics, such as a certain inertial behavior.

[0004] With the increasing dominance of decentralized feeders, fully or partially controlled by converters, such physical dynamics as the dynamics of directly coupled synchronous generators are losing influence and dominance in the electrical grid. In particular, the grid-stabilizing or at least calming inertia of directly coupled synchronous generators can be suppressed and would have to be accounted for elsewhere. One option is to emulate a synchronous generator or the behavior of a synchronous generator using converter-controlled feeders.

[0005] For this purpose, a physical model of a synchronous generator can be stored in the control system of a converter for feeding in the electrical power and the converter can be controlled in such a way that it behaves essentially like a synchronous generator.

[0006] However, this can have the disadvantage that such synchronous generator models are complex and can therefore sometimes be difficult to handle during programming and parameterization. An overly complicated emulation can also lead to excessively slow control, or expensive control hardware may be required to ensure sufficiently fast control. Filtering can also cause unwanted signal delays.

[0007] Furthermore, emulating a synchronous generator can not only achieve the desired effects, such as the aforementioned stabilization or calming of the high moment of inertia, but can also lead to other potentially undesirable behaviors of the synchronous generator. In addition to the potential risk of a synchronous generator falling out of step, it should also be noted that suboptimal, particularly suboptimal sinusoidal, currents can also occur. These can also be the result of an inadequacy in the synchronous generator model used.

[0008] The increasing penetration of converter-controlled feeders in some electrical networks may also mean that such feeders will have to take on tasks in the area of ​​a black start or at least a grid restoration in the future.

[0009] WO 2011 / 092193 A2 concerns a method for emulating synchronous machines. Published patent application EP 2 963 759 A1 relates to the control of an inverter for feeding electrical power into an electrical grid, in particular a microgrid.

[0010] The present invention is therefore based on the object of addressing at least one of the above-mentioned problems. In particular, a solution is to be proposed which works in such a way that the feed-in of electrical power by means of a generator controlled by a converter is adapted as well as possible to an electrical supply grid with a small proportion of conventional large-scale power plants with directly coupled synchronous generators. In particular, a feed-in with the highest possible quality and / or good control stability is to be proposed for such a grid or even an isolated grid. At least an alternative solution to previously known solutions is to be proposed. According to the invention, a method according to claim 1 is proposed. This method relates to the feed-in of electrical power into a three-phase electrical supply grid at a grid connection point by means of a wind turbine and in this case using an inverter or converter.In principle, it is also possible to use another decentralized generator, as long as it uses an inverter or converter. Any descriptions and explanations regarding an inverter also apply to converters, and vice versa.

[0011] To this end, it is initially proposed that the electrical grid voltage be measured at the grid connection point. Measuring the electrical grid voltage at the grid connection point is particularly advantageous because the feed-in is also planned there, and therefore the electrical grid voltage should be taken into account there. However, it is also possible to perform the measurement at another point in the electrical supply grid or at another point upstream of the grid connection point, i.e., between the feeding-in wind turbine and the grid connection point, if a correspondingly representative voltage can be measured there.

[0012] A virtual generator voltage is then determined using a machine model. The machine model emulates the behavior of a synchronous machine. The virtual generator voltage thus determined is therefore a voltage that corresponds to the voltage of a generator whose behavior is emulated by the machine model.

[0013] Furthermore, the recorded grid voltage is prepared for comparison with the virtual generator voltage. This means that the determined virtual generator voltage, which is initially essentially a calculated value, can be compared with the recorded grid voltage that was actually measured.

[0014] In addition, a target current is specified as a feed-in current based on the virtual generator voltage and the grid voltage prepared for comparison. This can mean, in particular, that a deviation between these two voltages leads to a corresponding target current, e.g., a higher voltage deviation leads to a higher target current.

[0015] A feed-in current is then generated according to the specified target current and fed into the electrical grid at the grid connection point. A comparison between the virtual generator voltage and the measured voltage at the grid connection point thus leads to a dependent target current and thus also to a correspondingly adjusted feed-in current.

[0016] It is now proposed that preparing the measured grid voltage for comparison with the virtual generator voltage involves transforming the measured grid voltage into a space vector representation. The measured grid voltage, namely that of the three-phase electrical supply network, and thus in particular the three-phase grid voltage thus recorded, is thus transformed into a space vector representation. It has been recognized that preparing the measured grid voltage in a space vector representation particularly facilitates the consideration of the sinusoidal fundamental signal. The virtual generator voltage is also the result of a calculation and can therefore also be assumed to be sinusoidal with low distortion and little noise.

[0017] The transformation of the recorded mains voltage into a space vector representation is preferably a d / q transformation, which can also be referred to as a qd transformation or a Park transformation. Such a d / q transformation assumes a three-phase system and transforms this into a two-axis, rotating coordinate system with the axes d and q or the values ​​d and q. Ideally, the values ​​d and q are constant because they rotate with the coordinate system. The transformation basically assumes a system with sinusoidal quantities, even if quantities deviating from sinusoidal quantities can also be taken into account due to changes over time. Nevertheless, such sinusoidal quantities form the basis and the d / q transformation orThe calculation in the space vector representation can therefore also have a non-linear filter function, which can filter out disturbances or other deviations from a sinusoidal signal to a certain straight line or can keep them out of the target current to be ultimately generated.

[0018] It is particularly advantageous to use such a sinusoidal voltage for the comparison between the recorded mains voltage and the virtual generator voltage.

[0019] According to one embodiment, it is proposed that the grid voltage be filtered in the space vector representation into which it was transformed and then re-transformed, so that the target current is specified as a function of the virtual generator voltage and as a function of the re-transformed grid voltage. The detected grid voltage is thus filtered in the space vector representation. This prevents distortion or delay of the sinusoidal base signal. In particular, after re-transformation, a filtered sinusoidal signal is obtained, which is also assumed to be sinusoidal due to the underlying transformation rule or re-transformation rule.

[0020] It was recognized that simple filtering, particularly filtering with a PT1 element, can be implemented in the space vector representation, resulting in a beneficial filtering effect in the back-transformed sinusoidal signal. In particular, filtering in the time domain with the same filtering result would have required a much more complex filter. However, such a complex filter would be complex to parameterize and would not be very robust. If the type of interference to be filtered out changes in such a complex filter, the quality of the filter may deteriorate or even its effectiveness may be called into question. The proposed filtering in the space vector representation, on the other hand, is robust and essentially specifically adapted to the fact that the basic signal to be filtered is sinusoidal. It is particularly important that the values ​​d and q in the space vector representation are ideally constant.To put it simply, the filter can essentially filter to a constant value. Despite a nonlinear input variable or an input variable with a nonlinear response, a simple linear filter can be used.

[0021] This achieves a sinusoidal waveform of the measured mains voltage that is as interference-free and delay-free as possible, which can then be easily compared with the virtual generator voltage, which is also sinusoidal. The target current can thus be specified with high quality based on this inversely transformed mains voltage and the virtual generator voltage.

[0022] Preferably, the machine model is based on a virtual synchronous machine with a stator and rotor, which are thus also virtual, even without this being mentioned. To determine the generator voltage, the machine model uses one, several, or all of the variables from the following list: a virtual angle of rotation ϑ of the rotor, a virtual speed ω of the rotor, a virtual excitation voltage, a virtual stator current, a virtual moment of inertia J of the rotor, a virtual torque T e of the rotor and a virtual friction Dp of the rotor.

[0023] For simplicity, these and other virtual variables can also be referred to below without the suffix "virtual," because as far as they refer to the virtual generator, they are also virtual themselves. The position of the rotor is recorded via the virtual rotation angle (ϑ) of the rotor, i.e., the virtual rotor, which influences the phase angle of the generator voltage.

[0024] By taking into account the virtual speed (ω) of the rotor, the frequency of the generator voltage can be taken into account and influenced.

[0025] The virtual excitation voltage can be used to influence and take into account the amplitude of the generator voltage.

[0026] The virtual stator current can also be used to take into account the behavior of the virtual synchronous machine with respect to a connected impedance.

[0027] The dynamics of the virtual synchronous machine can also be influenced via a virtual moment of inertia (J) of the rotor.

[0028] The dynamic behavior of the virtual synchronous machine can also be taken into account and influenced via a virtual torque (T e ) of the rotor.

[0029] Virtual rotor friction can be implemented particularly as a reinforcement of a frequency-dependent power control. This power control is preferably part of the machine model and takes into account deviations between the rotational frequency of the machine model and a reference frequency. The reference frequency can be a measured grid frequency or a specified frequency. This allows the frequency-dependent behavior of the virtual synchronous machine to be taken into account in the machine model.

[0030] Preferably, the generated feed-in current is used as a virtual stator current. This establishes a connection between the virtual synchronous machine and the actual feed-in. The generated and then actually fed-in feed-in current also depends on the conditions of the inverter output wiring, particularly on the conditions in the electrical supply grid. Thus, these influences can be taken into account in the virtual synchronous machine, i.e., in the machine model, via the feed-in current, which then forms the virtual stator current.

[0031] Additionally, or alternatively, it is proposed that the virtual moment of inertia be adjustable. This can particularly influence the dynamics of the virtual synchronous machine. Depending on the specific situation or requirement, a larger or smaller virtual moment of inertia can be selected. A small virtual moment of inertia can be particularly useful for rapid synchronization of the virtual synchronous machine or machine model with the electrical grid when power is not yet being fed into the grid but is only being prepared. A higher virtual moment of inertia can be particularly useful when feeding into the electrical grid to support and stabilize the electrical grid.A particularly high moment of inertia is considered the preferred setting, particularly in cases where the electrical supply grid is an isolated grid and must be stabilized by the inverter, or where the inverter is even intended to act as a grid former. However, even in an electrical supply grid that is not an isolated grid, different grid situations can arise, to which a correspondingly adjusted virtual moment of inertia can be responded. It is also possible that the grid operator of such an electrical supply grid sets specifications for the dynamics of feed-in, to which a correspondingly adjusted virtual moment of inertia can be responded.

[0032] Preferably, the virtual moment of inertia is set depending on a grid state or a grid property. A grid state is, for example, an overfrequency or an underfrequency, or even the current value of the frequency. The current grid voltage is also a grid state, and this also includes an overvoltage or undervoltage, to name another example. A grid property that can be taken into account, for example, is grid sensitivity, which can be specified as the ratio of a voltage change at a grid connection point in question to a change in the power fed into the grid at the grid connection point in question. In this respect, grid sensitivity is also a property of the electrical supply grid related to the grid connection point. It can indicate how sensitively the grid reacts to changes.Especially in the case of high grid sensitivity, it can be advantageous to choose a correspondingly high virtual moment of inertia for stabilization.

[0033] According to one embodiment, it is proposed that a virtual grid impedance be considered for specifying the target current. This impedance is considered as an impedance between an output of the machine model or the virtual synchronous machine and the grid connection point. For this purpose, it is proposed that the virtual impedance be variable in magnitude. The virtual synchronous machine is thus connected to a virtual impedance, so that this virtual impedance also influences a current depending on the generator voltage.

[0034] The virtual impedance allows different properties of the electrical supply network to be taken into account. The behavior of the virtual synchronous machine can thus be easily modified to account for changing properties of the electrical supply network. This may make dedicated consideration unnecessary.

[0035] Preferably, the virtual impedance is selected depending on whether the feed-in is in a normal state of the electrical supply grid or whether the feed-in is in a restoration mode after the electrical supply grid has been interrupted or failed and in which the electrical supply grid must be ramped up to a normal operating point. For the grid restoration mode, it is specifically provided that the virtual impedance is selected to be greater than in the normal state of the electrical supply grid in order to facilitate ramping up to the normal operating point. In particular, during ramping up to the normal operating point, a generator voltage can be gradually increased and at the same time the virtual impedance can be adjusted so that a low current flows as long as not much power is to be fed in in this grid restoration mode.

[0036] This allows for a simple implementation of grid restoration or a black start. In particular, the requirement that a black start should initially focus on generating and maintaining a voltage while feeding in only a small amount of power can be met. At the same time, a normal mode can also be implemented with the same configuration. By adjusting the virtual impedance, the feeder can be adapted to feed in in normal mode.

[0037] According to one embodiment, it is proposed that in the machine model a speed difference is formed between the virtual speed and a reference speed, a filtered value of the virtual speed or a predetermined frequency is used as the reference speed, the difference speed is calculated to an auxiliary torque via a difference speed gain, the auxiliary torque acts on the virtual moment of inertia of the machine model via a summing point in order to thereby regulate the virtual speed to the reference speed, wherein the difference speed is preferably set to zero for synchronizing the machine model with the electrical supply network.

[0038] By taking this differential speed into account and feeding it back, the virtual speed can be adjusted to the reference speed, although steady-state accuracy with a deviation of 0 does not necessarily have to be achieved. In particular, this differential speed feedback via the differential speed gain can be referred to as so-called droop control. This takes into account the behavior of a synchronous machine, which initially slows down when higher power is called up. Such a higher power output, which is particularly characterized by higher output currents, i.e. feed-in currents in this case, increases the torque of the synchronous generator and accordingly the virtual torque of the synchronous generator. This virtual torque of the synchronous generator acts at the same summing point at which the auxiliary torque acts or is added.

[0039] Preferably, the differential speed gain is a positive factor, and the auxiliary torque is subtracted at the summing point. If the rotation of the virtual synchronous machine slows down, i.e., the virtual speed decreases, while the reference speed remains the same or changes more slowly, the differential speed is negative and the auxiliary torque is also negative. Subtracting this negative auxiliary torque at the summing point results in an increase in the total torque at the summing point by the amount of the auxiliary torque. For this exemplary case, the virtual synchronous machine is therefore accelerated or counteracts an increased virtual generator torque.

[0040] To synchronize the machine model with the electrical supply grid, however, it is suggested to override this droop control. This can be done by directly setting the differential speed to 0. However, this can also be done indirectly by calculating the differential speed as the difference between two equal speeds, which then also results in the differential speed being 0. To do this, the filter that filters the virtual speed can be set to one.

[0041] A particularly suitable option is to use the filtered virtual speed as the reference speed. The filtered virtual speed is then subtracted from the virtual speed. For a filter function that has a total gain of 1 but a dynamic response, this differential speed gradually becomes 0 at steady-state speed.

[0042] In any case, for synchronization, it is proposed to set the differential speed and thus the auxiliary torque to 0 and to disable the droop control, thereby suppressing such speed regulation. Especially when no or no significant currents flow during synchronization, and in particular when the target power is 0, the virtual synchronous machine is essentially in an idle state, and this state is not disturbed by this droop control. The virtual synchronous machine can thus be brought into or maintained in a state in which it runs synchronously with the electrical supply grid into which it is ultimately to be fed. If this is the case, and if the voltage and frequency of the electrical supply grid do not change, the virtual synchronous machine is essentially in a stable, idle-like operating state.

[0043] Preferably, it is proposed that to synchronise the machine model with the electrical supply network, a target power has the value zero, a calculation model is used to calculate internal virtual generator voltages and / or the virtual torque T e , where the calculation model uses one, several or all of the variables from the list comprising the virtual angle of rotation ϑ of the rotor, the virtual speed ω of the rotor, the virtual excitation voltage, and the feed-in current or the target current, whereby in particular no frequency of the electrical supply network is recorded.

[0044] Thus, to synchronize the machine model with the electrical supply grid, the target power is first set to 0. In the machine model, the target power acts as a drive torque on the virtual moment of inertia via a corresponding conversion factor. By setting this power to 0, the drive torque of the virtual machine is also set to 0. The virtual synchronous machine then receives no drive. Internal virtual generator voltages and the virtual torque are then calculated using a calculation model. In principle, only one of the variables is calculated, but it is advantageous to calculate all of these variables. The calculation model preferably uses the machine equations of a synchronous machine, which are parameterized according to the assumed virtual synchronous machine.

[0045] For this purpose, the calculation model receives the virtual rotor angle of rotation, the virtual rotor speed, the virtual excitation voltage, and the feed current as input variables. These are required in the machine equations to calculate the aforementioned variables. However, simplified calculations using a simplified model are also theoretically feasible. The virtual speed can also be calculated from the virtual angle of rotation, eliminating the need to input both variables into the calculation model.

[0046] The feed-in current can be any current actually generated according to the specified target current. Furthermore, here and in the rest of the description, a target current, a feed-in current, and any resulting currents are each understood to be a three-phase current. The feed-in current, or the target current, therefore constitutes three individual currents, each of one phase. In any case, a small feed-in current can also be generated during synchronization, which can flow into smaller consumers located upstream of the grid connection point. Particularly in the case of a wind farm with multiple wind turbines, small consumers can be used in the corresponding farm grid. These could be, for example, line inductors or the wind turbine transformer. A choke can also be considered.Furthermore, the virtual impedance can be adjusted to the corresponding values ​​or behavior of these small consumers.

[0047] Alternatively, if no current is flowing at all, the specified target current can be included in the calculation model as the actual current.

[0048] According to one embodiment, it is proposed that the virtual excitation voltage is determined at least as a function of a specified reactive power and optionally a specified electrical grid voltage at the grid connection point.

[0049] The virtual excitation voltage can be determined using a specified reactive power, which is compared with a measured reactive power, so that such a reactive power deviation is used. This reactive power difference or reactive power deviation is applied via a reactive power gain and integrated into the virtual excitation voltage. The reactive power gain is thus a gain factor of the integrator for reactive power control.

[0050] Optionally, a specified electrical grid voltage is determined at the grid connection point, which also influences the virtual excitation voltage. For this purpose, a difference is calculated between this measured grid voltage and a specified voltage. This voltage difference can form a reactive power value via a voltage gain and then, like the described difference between the specified and measured reactive power, influence the virtual excitation voltage. The sum of these reactive powers can be integrated together to form the virtual excitation voltage, with the integrator having the described reactive power gain as the integration constant.

[0051] Preferably, it is proposed that electrical quantities of the machine model be calculated in space vector representation, in particular according to a d / q transformation. This particularly applies to a three-phase stator current and a three-phase stator voltage. The virtual synchronous machine can thus be calculated efficiently. In particular, a symmetrical three-phase system can be easily calculated and, in particular, also guaranteed.

[0052] According to one embodiment, it is proposed that the feed-in current be generated using a tolerance band method. In particular, it is thus controlled according to the tolerance band method. The tolerance band method basically works in such a way that a tolerance band is placed around the desired signal shape of the current to be generated, i.e. in particular around the sinusoidal signal shape of the current to be generated. The current generated in each case is measured and then controlled via corresponding switch positions depending on the measured value, namely such that the current is reduced when it reaches an upper tolerance band limit and such that it is increased when it reaches a lower tolerance band limit. The use of a tolerance band method for feeding in or generating an electrical sinusoidal current is basically known.However, a method is proposed here that can achieve voltage modulation despite using the tolerance band method. This is particularly possible using virtual impedance, which ultimately allows an output voltage to be not only achieved but also specifically specified despite a specified current. This voltage can be specifically and controlledly increased in the event of grid restoration or even a black start.

[0053] In particular, island grid operation is also possible and it is particularly suggested that in island operation, when the inverter sets the grid frequency, the virtual speed ω depends on a predetermined mains frequency, in particular such that an auxiliary torque acting in the machine model is regulated as a function of a difference between the virtual speed and the predetermined mains frequency, and / or the virtual excitation voltage depends on a predetermined voltage of the electrical supply network and a detected voltage of the electrical supply network, in particular such that an auxiliary reactive power value, on which the virtual excitation voltage depends, is regulated as a function of a difference between the predetermined voltage and the detected voltage.

[0054] The proposed method, at least according to this embodiment, is thus particularly suitable for island operation or island grid operation. Island grid operation is considered here to be an operation in which there is a closed electrical supply grid in which the wind turbine, or other inverter- or converter-controlled feed-in device, has such dominance that it essentially controls the grid, particularly with regard to frequency and voltage. For this purpose, it is proposed that the virtual speed depend on a predetermined grid frequency. Here, the grid frequency is therefore predetermined, particularly as a fixed reference frequency. Depending on a difference between the virtual speed, i.e. a speed in the machine model, and the predetermined grid frequency, an auxiliary torque can be determined, which is then regulated, controlled, or adjusted accordingly.This auxiliary torque counteracts the virtual electrical torque in the model. The difference between the auxiliary torque and the virtual electrical torque, possibly after taking into account other torques, affects the virtual speed, which is then calculated by integration with the corresponding integration time constant. The virtual speed can then serve as an input variable for the calculation model and thus specify the frequency of the generator voltage and ultimately also the frequency of the reference currents. This method can therefore be used to implement a fixed frequency using the virtual synchronous machine in such a way that the island electrical grid is controlled in terms of frequency.

[0055] The auxiliary torque control preferably uses a PI controller to regulate the virtual speed with steady-state accuracy to the specified mains frequency or a corresponding speed.

[0056] Additionally or alternatively, it is proposed here that the virtual excitation voltage depends on a predetermined voltage of the electrical supply network and a detected voltage of the electrical supply network. This is specifically proposed by determining an auxiliary reactive power value based on a difference between the predetermined voltage and the detected voltage. If necessary, a detected reactive power value is subtracted from this. This reactive power value thus obtained can be integrated into the virtual excitation voltage, particularly via reactive power amplification, as already explained.

[0057] By taking the voltage into account, voltage regulation can be achieved, which can also be referred to as droop voltage control. This voltage regulation can be activated or deactivated. By taking into account the optional use of the specified voltage and the detected voltage, i.e. activating it, this voltage regulation is activated. The virtual synchronous machine is thus controlled in such a way that it can react to voltage changes in the electrical supply network, particularly at the grid connection point, with voltage regulation. It can therefore act to stabilize or track the voltage. This variant is particularly recommended for island grid operation, as it enables voltage regulation and thus voltage control. This regulation then takes effect via the electrical excitation voltage. As a result, the method can not only specify the frequency in an island grid, but also...but also the voltage, i.e. the voltage amplitude.

[0058] For island grid operation, it is therefore particularly suggested that the auxiliary torque is controlled via a PI controller and the virtual speed is obtained in particular by integrating a differential torque, as the difference between the auxiliary torque and the virtual electrical torque, via an integration time constant and / or the auxiliary reactive power value is controlled via a PI controller, and the virtual excitation voltage is obtained in particular by integrating a differential reactive power, as the difference between the auxiliary reactive power and a detected reactive power, via an integration time constant.

[0059] The machine model is thus subjected to a torque resulting from the difference between the controlled auxiliary torque and the virtual torque. This torque is integrated into the machine model using the inverse virtual moment of inertia as the integration time constant, thus determining the virtual speed. If no additional torque is added, this integration results in the virtual speed. This is how the speed of the virtual synchronous machine is controlled.

[0060] The output voltage is regulated or controlled via the reactive power. For this purpose, the voltage deviation from a reference value or setpoint is converted into a reactive power value as a control deviation via a PI control, and the voltage is thus regulated by the virtual synchronous machine via the virtual excitation voltage.

[0061] According to the invention, a wind turbine is also proposed that is designed to feed electrical power into a three-phase electrical supply network at a grid connection point. For this purpose, it has at least one inverter for generating a feed-in current. Such a feed-in current is then generated with the inverter, for example, using a tolerance band method. Furthermore, a detection means for detecting an electrical grid voltage is provided, and this electrical grid voltage is detected in particular at the grid connection point. The detection means can accordingly be designed as a measuring device or measuring sensor and measure the voltage at the grid connection point. However, the actual sensor can also be present as a separate element that is not part of the wind turbine.The wind turbine then has an interface through which the detected grid voltage can be received and this interface can then serve as a detection device for detecting the electrical grid voltage.

[0062] In addition, a control device is provided to control the feed-in. This control device specifically controls the inverter. For this purpose, a special microprocessor can be installed in the inverter, which performs switching operations within the inverter and is controlled by the control device, particularly receiving setpoints for the feed-in current. However, the control device can also act directly on the inverter.

[0063] In any case, it is proposed that the control device be configured to execute a method that determines a virtual generator voltage using a machine model, wherein the machine model emulates the behavior of a synchronous machine. Thus, at least this machine model is implemented in the control device. For this purpose, the control device may comprise a corresponding microprocessor or other computing devices.

[0064] The implemented method also includes the step of comparing the recorded grid voltage with the virtual generator voltage. Specifically, the grid voltage is filtered so that, if possible, only a fundamental wave of the grid voltage is compared with the virtual generator voltage. This is based on the idea that the virtual generator voltage is generated in the control unit, particularly with the aid of the machine model, and therefore exhibits no, or no significant, measurement interference. Since the comparison is not about detecting differences in measurement interference, the recorded grid voltage should be prepared so that it exhibits no, or at least as few, measurement interference as possible.

[0065] Furthermore, the implemented method includes the step of specifying a target current as a specification for the feed-in current, depending on the virtual generator voltage and the grid voltage prepared for comparison. The target current is thus specified depending on the virtual generator voltage and the grid voltage.

[0066] Furthermore, the inverter is designed to generate feed-in current based on the target current and feed the generated feed-in current into the electrical grid at the grid connection point. Appropriate wiring at the inverter output is particularly advantageous for this purpose. This may include a suitable choke. Furthermore, the inverter may be connected to the grid connection point via at least one additional transformer.

[0067] It is further proposed that preparing the detected grid voltage for comparison with the virtual generator voltage includes transforming the detected grid voltage into a space vector representation. The control device is thus also prepared to provide such a transformation. For this purpose, appropriate transformation algorithms can be implemented in the control device. Furthermore, the control device is equipped with appropriate computing capacity, in particular with a corresponding microprocessor or other computer device.

[0068] Preferably, such a wind turbine, in particular the control device, is prepared to carry out a method for feeding electrical power into an electrical supply grid according to at least one embodiment described above. For this purpose, it is proposed to implement the explained steps in the control unit.

[0069] According to the invention, a wind farm is also proposed which has a plurality of wind turbines, wherein the wind turbines are designed according to at least one embodiment described above. In addition or alternatively, it is proposed that the wind farm have a farm feed-in device which is connected to a grid connection point and is prepared to carry out a method according to an embodiment described above. In particular, a farm inverter is proposed for this purpose. Such a farm feed-in device or the farm inverter can, particularly in the case of a wind farm with a DC grid, feed power from the wind turbines of the wind farm as a whole into the electrical supply grid. If necessary, this can also be done at a high voltage. In particular, a central farm inverter is proposed here for this feed-in.Such a central park inverter can also operate based on the proposed emulated synchronous machine and apply the proposed methods, or one of them.

[0070] The invention is explained in more detail below by way of example with reference to the accompanying figures. Fig. 1 shows a wind turbine in a perspective view. Fig. 2 shows a wind farm in a schematic representation. Fig. 3 shows an equivalent circuit diagram for power transmission between two AC voltage sources. Fig. 4 shows a schematic structural diagram of a feed-in method according to an embodiment for grid operation. Fig. 5 shows a structural diagram of a method according to an embodiment for island grid operation. Fig. 6 shows a structural diagram illustrating an adaptation of a virtual impedance according to an embodiment. Fig. 7 shows a structural diagram illustrating a proposed voltage filtering process according to an embodiment. Fig. 8 shows voltage curves to explain a filtering effect.

[0071] Figure 1shows a wind turbine 100 with a tower 102 and a nacelle 104. A rotor 106 with three rotor blades 108 and a spinner 110 is arranged on the nacelle 104. During operation, the rotor 106 is set into rotation by the wind and thereby drives a generator in the nacelle 104.

[0072] Figure 2shows a wind farm 112 with, for example, three wind turbines 100, which may be identical or different. The three wind turbines 100 are thus representative of essentially any number of wind turbines in a wind farm 112. The wind turbines 100 provide their power, namely in particular the generated electricity, via an electrical farm grid 114. The currents or power generated by the individual wind turbines 100 are added together, and a transformer 116 is usually provided, which steps up the voltage in the farm and then feeds it into the supply grid 120 at the feed-in point 118, which is also generally referred to as a PCC. Fig. 2is only a simplified representation of a wind farm 112, which, for example, does not show a control system, although a control system is of course present. The farm network 114 can also be designed differently, for example, by also providing a transformer at the output of each wind turbine 100, to name just one other embodiment.

[0073] The proposed method is a current control method for a full-scale converter in island and grid operation based on the equation of motion of a synchronous machine. The invention thus relates to a current control method for a full-scale converter whose core control is based on the equation of motion of a synchronous machine. This control enables, among other things, the moment of inertia of a synchronous machine to be emulated as freely as possible during grid state changes, particularly changes in the grid frequency and the grid voltage. This property also depends on the energy stored in the DC link of the full-scale converter. Energy fed into the DC link by a primary energy source via an DC link voltage control is also considered. This can then also be available in the DC link.

[0074] In the following, the procedure is explained in more detail using several examples.

[0075] In an alternating current system, the active and reactive power transmitted between two nodes with voltages V and E are determined by the following equations: P = V ∗ E ∗ sinδ X Q = E 2 − E ∗ V ∗ cosδ X where X is the line reactance between the two nodes and δ is the phase shift between the two voltages. This is shown in the equivalent circuit of the Figure 3 shown.

[0076] Assuming that E is the output voltage of a full-scale converter connected to the grid with the voltage V, it is clear from equations (1) and (2) that the transmitted active and reactive powers can be influenced by the amplitudes and phase position of the converter's output voltage.

[0077] The control system consists of an outer and inner control loop. In the outer control loop, the amplitude and angle of the rotor voltage of a virtual synchronous machine are dynamically changed so that the active and reactive power fed into the grid by a full-scale converter is regulated to specified setpoints. Active power control is based on the equation of motion of a synchronous machine, while reactive power or voltage control is based on a P controller or PI controller. dω r dt = T m − T e − K D Δω r J

[0078] Equation (3) describes the equation of motion of a synchronous machine, where T m and T e are the mechanical and electromagnetic torque, respectively, ω r the rotor speed, KD is the damping factor and J is the inertia of the synchronous machine.

[0079] T m is determined by an active power setpoint, as described in Figure 4is explained, while K d and J are adjustable parameters. The variable KD can also be referred to as gain DP.

[0080] The electrical quantities of the virtual synchronous machine are transformed into d / q components with respect to the rotor coordinate system and used for the calculation of Te. The d / q transformation can also be referred to as qd transformation or transformation into qd coordinates, and for further explanations on the transformation, please refer to the reference [Lit1] mentioned below. For the transformation, the angle θ, which results from the solution of (3) and integration of ω r results, is used.

[0081] The abc / qd transformation matrices are defined as follows: T 1 = 2 3 cosθ cos θ − 2 π 3 cos θ + 2 π 3 sinθ sin θ − 2 π 3 sin θ + 2 π 3 1 2 1 2 1 2 T 1 − 1 = cosθ sinθ 1 2 cos θ − 2 π 3 sin θ − 2 π 3 1 2 cos θ + 2 π 3 sin θ + 2 π 3 1 2

[0082] For a full-pole machine, T e is calculated as follows: T e = 3 2 pL m i f i q

[0083] Here p represents the number of pole pairs, L m the mutual inductance, ifthe excitation current and IQ represents the q component of the stator current, where the proportion L mif results from reactive power and voltage control.

[0084] The amplitude of the virtual pole wheel voltage in qd components is calculated from: E q = ω r L m i f E d = 0

[0085] Using θ and (5), (7) and (8) are transformed back into abc. It should be noted that the voltage angle θ is not separately obtained from the grid voltage using a method such as a phase-locked loop, but rather results from the solution of the system of equations described above, in particular the equation of motion (3). This is a decisive advantage of the class of control methods based on synchronous machines, as delays of the order of several tens of milliseconds, which are common in phase-locked loops and similar methods, are avoided.

[0086] The internal control loop consists of a current hysteresis controller, which can also be referred to as a tolerance band method, whose setpoints are derived from the instantaneous values ​​of the virtual pole wheel voltages e abc , measured voltages at the inverter terminals v abc , and a virtual stator impedance Z s , result.

[0087] The virtual stator impedance can be adjusted to ensure control stability and consists of the following equation: Z s = R s + s L s

[0088] Where R s and L s represent the stator resistance and stator inductance of the virtual synchronous machine.

[0089] The current setpoints are thus calculated from: i sabc = e abc − v abc Z s

[0090] The measured terminal voltages of the converter are not necessarily sinusoidal and, depending on the operating mode or grid condition, can contain many harmonics, which can have a negative impact on the current setpoints according to equation (10) and the entire control system. This will be the case, for example, in no-load operation, i.e., when a voltage is applied without load, or with highly non-linear loads. In principle, the always symmetrical rotor voltage of the synchronous generator equation system counteracts this asymmetry and harmonic content, but it cannot always compensate for it completely. To counteract this behavior, the converter terminal voltages are transformed into qd coordinates, their components are filtered by a first-order low-pass filter, and then transformed back into abc components. Thus, no amplitude and phase compensation is required. This is shown in the equivalent circuit of the Figure 3 shown.

[0091] The following explains grid operation in which power is fed into an electrical supply network, such as the European interconnected grid. This may differ from feeding into an isolated grid, as described below.

[0092] Because the control system emulates the characteristics of a synchronous machine, the converter is capable of synchronizing itself with the electrical supply network, which can also be referred to simply as the grid. A synchronization phase takes place before the converter is connected to the grid.

[0093] The active and reactive power setpoints are set to zero and the control ensures that the conditions according to (11) are met. This means that the amplitude and phase position of the virtual rotor voltages e abc equal to those of the mains voltages v abcat the node point (VKP). Thus, the active and reactive powers in the steady state according to equations (12) and (13) are zero at the VKP. E = V g θ = θ g P = V g ∗ E ∗ sin θ − θ g X s = 0 Q = V g E ∗ cos θ − θ g − V g X s = 0

[0094] Since the converter is not yet connected to the grid during the synchronization phase, the currents according to equation (10) are virtual. At the end of the synchronization phase, when equation (11) is satisfied, the following results: e abc − v abc = 0 → i sabc = 0 → i q = 0 → T e = 0

[0095] Once the synchronization phase is complete, the inverter can be connected to the grid and P set and Q set can be set to a new setpoint. The structure proposed for grid operation is shown in the Figure 4 shown.

[0096] The procedure is intended to specify the target current i sabc. This target current i sabc is 3-phase and results from a differential voltage U DV that is applied to the virtual impedance ZV. In the structural diagram of the Figure 4This is shown in such a way that the differential voltage U DV is input into this virtual impedance.

[0097] The differential voltage U DV results from the difference between the measured voltage V PCC , which is subtracted from the virtual generator voltage E gen . For this purpose, an output summing element 10 is provided. The voltage V PCC is the voltage measured at the grid connection point, which is, however, filtered, preferably in the manner described above using the Figure 7 is explained.

[0098] The generator voltage E gen is calculated by the calculation model 12 and output by it. The calculation model 12 uses the machine equations of the synchronous generator, which were described above. The difference between the generator voltage E gen and the measured voltage V PCC at the grid connection point thus affects the virtual impedance ZV, from which the target current I sabc is calculated. This essentially means that an impedance, represented by the virtual impedance ZV, lies between the generator voltage and the voltage at the grid connection point.

[0099] Calculation model 12 uses the virtual rotor rotation angle δ and the virtual rotor speed ω as input variables. These two variables, which are mechanical variables, ultimately depend on an active power, specifically a specified active power.

[0100] The generation of the target current I sabc as a function of the virtual impedance ZV and the differential voltage U DV is based on the equation (10) explained above. Therefore, the generator voltage E gen corresponds to the generator voltage e abc in equation (10), and the measured and filtered voltage V PCC corresponds to the voltage v abc in equation (10). The virtual impedance ZV corresponds to the impedance Z s in equation (10).

[0101] In addition, the virtual excitation voltage U e and the output current I abc are included in the calculation model 12. The output current I abc can be a measured current, namely, in particular, the three-phase output current generated by the inverter according to the target current I sabc.

[0102] Furthermore, it is pointed out that the Figure 4 and the Figure 5contain a virtual machine model and any quantities that refer to a synchronous machine therein are therefore quantities of this virtual synchronous machine and are therefore generally to be regarded as virtual quantities, even if this is not explained in detail.

[0103] In addition, the calculation model 12 outputs a virtual electrical torque T e . During operation, this virtual electrical torque T e counteracts a mechanical torque T m at the torque summing element 14. An effective torque T w results from the torque summing element 14. The mechanical torque T m results from a power that can be specified here as the target power P s . The mechanical torque T m is derived from the entered target power P s via the torque conversion 16, which essentially takes the speed into account.

[0104] The effective torque T w is divided by the virtual moment of inertia J, which implements the inertia gain 18, and then leads to the speed ω via the first mechanical integrator 21. The speed ω leads to the rotor's angle of rotation δ via the second mechanical integrator 22. These two mechanical integrators 21 and 22 thus essentially reproduce the mechanical behavior of the rotor, which is acted upon by the effective torque T w.

[0105] In addition, a torque controller with a torque controller gain 24 is provided. The torque controller gain 24 results in an auxiliary torque T h , which is taken into account in the torque summing element 14 with a negative sign and thus reduces the effective torque T w as long as the auxiliary torque itself has a positive value.

[0106] For this purpose, the result of the speed summing element 26 acts on the torque controller gain 24. In the speed summing element, according to the Figure 4 In the configuration shown, a filtered speed ω* is subtracted from the virtual speed ω. The result is the differential speed Δω.

[0107] With this torque control, the speed ω can be influenced, especially for synchronization. As long as the speed ω changes, a difference can arise between the speed ω and the filtered speed ω*. For this purpose, it is particularly provided that the speed filter 28 has a gain of 1 and thus the filtered speed ω* corresponds to the speed ω in the steady state. In this steady state, the difference speed Δω will then be 0 and thus also the auxiliary torque T h will be 0. If the specified power P s is 0 during synchronization and the virtual electrical machine as a whole is running in such steady-state idling operation, the virtual electrical torque T e also becomes 0 and thus the effective torque T w also becomes 0. The speed ω then no longer changes. Synchronization is then completed and, for example,a connection for feeding into the electrical supply network via the grid connection point.

[0108] Especially for isolated grid operation, a frequency setting can be made via the frequency setting block 30, which is especially intended for isolated grid operation. Island grid operation will be discussed below in connection with the Figure 5 described.

[0109] The virtual excitation voltage U e results from the integration of an effective reactive power QW, taking into account a reactive power gain G, which can also be referred to as the integration time constant. The reactive power gain block 32 and the reactive power integrator 34 are provided for this purpose.

[0110] The effective reactive power Q w is the result of the difference between the specified reactive power Q s and the measured reactive power Q i . The difference is calculated at the reactive power summing element 36. The measured reactive power Q i is the reactive power fed in by the inverter at that moment. The virtual excitation voltage U e is thus adjusted or influenced by the reactive power.

[0111] A voltage control switch 38 can be used to activate or switch on the voltage control. This voltage control can be used to control the voltage to a predetermined voltage V PCCS. At least a voltage control can be carried out depending on this. For this purpose, a voltage V PCCI measured at the grid connection point is subtracted from this target voltage V PCCS. The voltage summing element 40 is provided for this purpose. For the voltage control, the differential voltage ΔV thus obtained is passed via a voltage gain D q in the voltage amplification block 42. This results in a control reactive power QR which influences the effective reactive power Q w via the reactive power summing element 36 and thus the virtual excitation voltage U e is then influenced via the reactive power amplification block 32 and the reactive power integrator 34. This control can be activated by closing the voltage control switch 38.Preferably, the voltage gain D q of the voltage gain block 42 is also variable in order to influence in particular the dynamics of this voltage regulation.

[0112] In island grid operation, the frequency and voltage of the grid are regulated to the desired setpoints. The frequency control is still based on equation (3) with the difference that T m is determined by a PI controller that reacts to a frequency deviation and is Figure 5 referred to as TR. Similar to frequency control, voltage control is also carried out by a PI controller, whose output provides the reactive power setpoint.

[0113] The corresponding control enables the converter to be black-start capable. This feature is achieved by connecting a small resistive or inductive load in parallel with the converter terminals if the no-load currents flowing through the converter's line filter or other equipment are too low for a satisfactory use of a hysteresis current controller. This allows a minimal converter output current to develop, ensuring the functionality of the hysteresis control. This results in a voltage at the converter's terminals, whose frequency and amplitude are regulated to the desired setpoints.

[0114] Island grid operation is in Figure 5 which shows a structure like in Figure 4As far as the structure or the elements in both figures are the same or have the same function, the same designations and reference symbols are used. For an explanation of the functions that have not changed for the island grid operation, please refer to the explanation of Figure 4 referred to.

[0115] In island grid operation, an additional PI power controller component 50 is provided for power control, which essentially affects the mechanical movement of the rotor. Based on a differential speed Δω, this generates a controller torque TR, which influences the effective torque T w via the torque summing element 14.

[0116] The differential speed Δω results from the difference between the speed ω of the machine model and a target speed ω s , which is specified by the frequency setting block 30. The frequency setting block 30 specifies the frequency for the island network and, in this case, specifies a target frequency, converted into a target speed ω s . The difference is formed in the speed summing element 26. A filtered speed ω* according to the structure of the Figure 4 is not used here. However, the structure can also be Figure 4 be used insofar as the speed filter 28 is switched to the frequency setting block 30 for island operation, as is done by the frequency switch 29 in Figure 4This also allows for a temporary island grid situation to be taken into account, i.e., when the system is not only deployed in an island grid, such as on an actual small geographical island, but also when, due to switching operations in a larger supply grid, a section of this supply grid forms an island grid, i.e., is temporarily disconnected.

[0117] By applying the control torque TR, the rotor movement or virtual rotor movement of the virtual synchronous machine can be controlled in such a way that the specified frequency or the specified target speed ω s is established. The consideration of a target power P s via the torque conversion 16 is not initially provided here. It can be selected via the target power switch 52 if necessary. The virtual electrical torque T e , however, continues to act on the effective torque T w via the torque summing element 14. In this respect, the machine model has remained unchanged, but a different control system has been implemented. For voltage control, the island grid operation according to the structure of the Figure 5 also provides for the use of a Pl component, namely as Pl voltage control component 54. The Pl voltage control component 54 outputs a control reactive power QR, which for the sake of simplicity has the same designation as in the Figure 4receives. The control reactive power QR is now the output of the PI voltage control component 54. This means that an integral component is included in this control reactive power QR. The input of this PI voltage component 54 is again a difference between the specified voltage V PCCS at the grid connection point and a voltage V PCCI measured there. The effective reactive power Q w is now the result of the difference between the control reactive power QR and the measured reactive power Q i . A specified reactive power QS is not effective because the reactive power switch 56 is connected to the output of the PI voltage control component 54.

[0118] The PI voltage control component 44 essentially sets a reactive power that depends on the voltage deviation between the specified and measured voltage at the grid connection point. The integral component of the PI voltage component 54 is intended to achieve steady-state accuracy of the specified voltage at the grid connection point.

[0119] For switching from black start operation to load operation, i.e., in an island grid, and vice versa, an adaptation of the virtual impedance is performed. This is proposed because during operation without significant loads, a stronger low-pass filtering of the difference between the virtual rotor voltage and the measured terminal voltage would be necessary to ensure stability.

[0120] The adaptation of the virtual impedance is in the Figure 6The adaptation when switching from black start operation to load operation is based on the measured output power of the converter, which can be active and / or reactive power, or alternatively on the measured power gradient dP / dt and / or dQ / dt and is shown in the Figure 6 referred to as Condition 1, or Condition C1. Here, the inductive component Ls of the virtual impedance Zs is changed, and this change is limited by a maximum gradient, which can be referred to as a rate limiter.

[0121] During the transition from load operation to black start operation, the load impedance increases significantly. Since the setpoint currents do not change during the transition, this leads to a significant transient increase in the terminal voltage. If the difference between the amplitudes of the virtual rotor voltage and the measured terminal voltage exceeds a certain limit, which is referred to as Condition 2 or Condition C2 in Figure 6is shown, the virtual impedance of the control is changed.

[0122] The Figure 6 shows in this respect an implementation of the virtual impedance ZV of the structures of the Figures 4 and 5 , in which the value of this virtual impedance ZV is inversely multiplied by the differential voltage U DV at the multiplier 60, resulting in the target current I abc. Due to its three-phase nature, the three-phase target current I abc can also be referred to as target currents, namely one current per phase. The same applies to other three-phase quantities.

[0123] The virtual impedance ZV , which can also be referred to as Zs, is composed of an inductive part Ls and an ohmic part Rs and these quantities thus form an input for the impedance block 62, in which these two parts are combined and the inverse of the impedance is formed and output for further calculation.

[0124] According to the embodiment shown, the inductive component LS is changed, while the resistive component RS remains unchanged. However, it can also be changed in principle.

[0125] The change or setting of the inductive component LS depends particularly on whether black start operation or load operation is present. Accordingly, the operating mode switch 64 can be used to switch between the two inductive components depending on these operating modes, namely the inductive component L S1 for black start operation and L S2 for load operation. For this purpose, the two conditions C1 and C2 are entered into the operating mode switch 64. Since this switchover can be carried out during operation, but the change in the virtual inductance ZV or ZS should not occur abruptly, a gradient block 66 is provided which ensures that the change is passed on as a ramp with a maximum gradient or that the change is limited to such an edge with a maximum gradient. Such a maximum gradient is therefore limited in terms of amount. This means that both a rise and a fall are limited.Thus, if the operating mode switch 64 switches between two values ​​of the inductive components and the output of the operating switch 64 is a step, the output of the gradient block 66 is a ramp.

[0126] As a result, the virtual impedance ZV or ZS changes gradually and this change can thus be made during operation and acts directly on the target current I abc via the multiplier 60.

[0127] However, it is also possible, especially in black start mode, to not only switch between the two inductive components L S1 and L S2, but also to change the inductive component that is active depending on the position of the operating mode switch 64. This change can also be limited to a maximum gradient, i.e., a maximum edge steepness, via the gradient block 66.

[0128] Figure 7then illustrates the filtering of the voltage V PCC measured at the grid connection point. The measured voltage is referred to here as V PCC ', whereas the filtered result is then the voltage V PCC. The measured voltage V' PCC is entered into the transformation block 70, which transforms this 3-phase voltage into a basically familiar representation with a q-component and a d-component. The transformation is also known as a d / q transformation and concerns the same transformation that is also known, for example, as a Park transformation. It converts the 3-phase quantities, here namely the voltage of the three phases, into a 2-axis coordinate system with the axes d and q. For illustration, a q-block 72 contains the q-component and a d-block 74 contains the d-component. To clarify that these two components still belong to the unfiltered voltage signal V' PCC, they are also shown as primed quantities, namely as q' and d'.They then each enter a filter block, namely the q-filter block 76 and the d-filter block 78. Both filter blocks 76 and 78 can be identical or different. The values ​​in . Figure 7 The proposed structure uses two identically parameterized first-order linear filters, i.e., a PT1 element and a PT1 filter, respectively. Both components q' and d' are thus passed through a first-order low-pass filter, and the result is the filtered component q and d, respectively. These filtered components q and d are then passed to the inverse transformation block 80 and transformed back into the 3-phase system. The result is the filtered 3-phase voltage V PCC , which is particularly useful in the Figures 4 and 5 to the output summation element 10.

[0129] Figure 8shows two voltage diagrams, each showing a voltage amplitude U in volts over time in seconds. The voltages shown thus essentially represent a sinusoidal voltage with a period of 20 ms and thus a voltage of a 50 Hz signal.

[0130] In the upper illustration of the Figure 8 The voltage V' 1,PCC of one phase recorded at the grid connection point is shown. This corresponds to one phase of the three-phase voltage V' PCC at the input of the transformation block 70 of the Figure 7 . It should be noted that the results of the Figure 8 Simulation results are. It should also be noted that the letter U and the letter V are sometimes used to represent electrical voltage, without any technical difference being intended.

[0131] The upper diagram of the Figure 8thus shows an unfiltered voltage, which is also noticeably noisy.

[0132] In the diagram below the Figure 8 , which has the same time axis as the upper diagram, two filtered voltages are shown. The voltage V 1,PCC shows the filtered curve to the unfiltered voltage V' 1,PCC of the upper diagram, which results from the filtering according to Figure 7 The voltage V 1, PCC is therefore the voltage corresponding to the upper diagram of the Figure 8 corresponding phase of the three-phase voltage signal V PCC at the output of the inverse transformation block 80 of the Figure 7 . If a three-phase voltage signal is fed into the transformation block 70 of the Figure 7 entered, with all three phases following a course similar to the upper diagram of the Figure 8 shown, the output of the inverse transformation block 80 is the Figure 7a three-phase voltage signal, of which all three phases have approximately the curve V 1, PCC according to the lower illustration of the Figure 8 The unfiltered voltage V' 1,PCC and the filtered voltage V 1,PCC relate to the same phase.

[0133] It can be seen that there is practically no phase shift between the unfiltered voltage V' 1,PCC and the corresponding filtered voltage V 1,PCC.

[0134] For filtering, according to Figure 7 For the two filter blocks 76 and 78, a PT1 element with the same gain, namely 1, and the same time constant, namely 10 milliseconds, is used.

[0135] For comparison, the same noisy signal V' 1,PCC from the upper diagram was filtered directly through a PT1 filter, which was parameterized in the same way as the two filter blocks 76 and 78. The result of this direct filtering using a PT1 filter is shown in dashed lines in the lower diagram and labeled V PT1R . In addition to the reduced amplitude, which could certainly be adjusted by adjusting the gain of the PT1 filter, a phase shift of almost 90 degrees is also clearly visible. This actually reflects the phase behavior known from a PT1 filter.

[0136] It can therefore be seen that the proposed filtering, which is not only specifically adapted to the three-phase signal but also takes into account the fact that a sinusoidal signal is expected, delivers a very good filtering result. The phase fidelity is particularly noteworthy. With the PT1 filter used internally, direct filtering with similar filter quality, i.e. similar noise suppression, would only be possible with a large phase shift or phase lag. Alternatively, a much more complex filter, particularly of a higher order, could be used to reduce the problem of phase delay. However, this filter would then have to be much more complex, adapted very precisely to the expected noise behavior, and would therefore be unlikely to be robust against changes in the signal to be filtered, especially against superimposed interference.

[0137] Thus, a power control of a converter in grid operation, both for strong and weak grids, and a black-start capable converter in off-grid operation are proposed. Voltage and frequency control for off-grid operation were also presented.

[0138] In principle, the control method is suitable for use with various forms of primary energy, such as wind turbines, battery storage systems, flywheels, and others. Furthermore, the self-controlled construction of a system supply network, such as an uninterruptible power supply (UPS), is also possible.

[0139] This has created a way to emulate a synchronous machine using a converter. This is also based on the following findings. Synchronous machine emulation is particularly important with regard to increasing the proportion of loads and generator units controlled by power electronics, which can lead to a power supply system based on power electronics. Such a system can cause grid instabilities because, for example, when the grid frequency changes, the moment of inertia or the stored rotational energy of conventional power plant generators is no longer present but can be emulated using targeted control strategies. With suitable parameterization of the virtual synchronous machine, this has a positive effect on various frequency-related phenomena, such as a frequency drop following the failure of a larger power plant unit or an HVDC line. Frequency oscillations can also be relevant.

[0140] After the loss of the electrical supply grid, especially after the loss of an interconnected grid, the formation of an island grid system can be achieved through black start capability, in which the grid voltage and frequency can be specified by the converter.

[0141] The use of the core equations of a synchronous machine using a current-controlled converter to easily implement current limitation is made possible.

[0142] The use of the method for forming an island grid during a black start and switching from no-load to island grid operation and back is an advantage.

[0143] It is also proposed that instantaneous voltage values ​​be used to generate the sinusoidal setpoint currents, as shown in equation (10). The measured terminal voltages of the converter are not sinusoidal and may contain numerous harmonics, which could cause problems in calculating the setpoint currents and, in the worst case, lead to instability. Filtering the terminal voltages in qd components, compared to the abc coordinate system, results in better voltage quality in the steady state with a small filter time constant, without the need to compensate for amplitude and phase shift.

[0144] The virtual impedance according to equation (9) is adaptively adjusted depending on the operating mode, so that the control stability is ensured, at least according to one embodiment.

[0145] In particular, stable operation of a current-controlled converter can be achieved with the following properties: Black start by applying a voltage to the converter terminals without any load connected.

[0146] Switching to island operation with load and regulating voltage and frequency to selectable setpoints.

[0147] Synchronization with an existing grid and fast control of active and reactive power depending on grid frequency and voltage.

[0148] Mains operation even with very small short circuit conditions.

[0149] The following literature is particularly noted: [Lit1] PC Krause, O. Wasynczuk and SD Sudhoff, "Analysis of Electric Machinery and Drive Systems", 2nd Edition, New York, 2002, John Wiley & Sons.

Claims

1. A method for feeding electrical power into a three-phase electrical power supply network (120) at a network connection point, in particular by means of a wind power installation (100), using an inverter, comprising the following steps: - detecting an electrical network voltage (V, E), in particular at the network connection point, - determining a virtual generator voltage (Egen) using a machine model that emulates a behavior of a synchronous machine, - preparing the detected network voltage (V, E) for comparison with the virtual generator voltage (Egen), - predefining a setpoint current (Isabc) as predefinition for an infeed current as a function of the virtual generator voltage (Egen) and as a function of the network voltage prepared for comparison, and - generating the infeed current depending on the setpoint current (Isabc) and feeding the generated infeed current at the network connection point into the electrical power supply network (120), wherein preparing the detected network voltage (V, E) for comparison with the virtual generator voltage (Egen) comprises transforming the detected network voltage (V, E) into a space vector representation, characterized in that a virtual impedance (Zv) is taken into account for predefining the setpoint current (Isabc), which virtual impedance (Zv) is taken into account as an impedance between an output of the machine model or the virtual synchronous machine and the network connection point, and the magnitude of the virtual impedance (Zv) is variable, and the virtual impedance (Zv) is selected depending on whether - infeed is effected in a normal state of the electrical power supply network (120), or - infeed is effected in a recovery mode after interruption or failure of the electrical power supply network (120), in which recovery mode the electrical power supply network (120) must be run up to a normal operating point.

2. The method as claimed in claim 1, characterized in that transforming the detected network voltage (V, E) into a space vector representation is a d / q transformation.

3. The method as claimed in claim 1 or 2, characterized in that the network voltage (V, E) in the space vector representation into which it was transformed is filtered and is subsequently subjected to inverse transformation, such that the setpoint current (Isabc) is predefined as a function of the virtual generator voltage (Egen) and as a function of the inverse-transformed network voltage (V, E).

4. The method as claimed in any of the preceding claims, characterized in that the machine model takes as a basis a virtual synchronous machine having a stator and a rotor and for determining the generator voltage (Egen) uses one, a plurality or all of the variables in the list comprising the following: - a virtual angle of rotation (ϑ) of the rotor, - a virtual rotational speed (ω) of the rotor, - a virtual excitation voltage (Ue), - a virtual stator current, - a virtual moment of inertia (J) of the rotor, - a virtual torque (Te) of the rotor, and - a virtual friction (Dp) of the rotor.

5. The method as claimed in claim 4, characterized in that - the generated infeed current is used as virtual stator current and additionally or alternatively - the virtual moment of inertia (J) is adjustable.

6. The method as claimed in claim 4 or 5, characterized in that the virtual moment of inertia (J) is set as a function of a network state or a network property.

7. The method as claimed in any of the preceding claims, characterized in that in the machine model - a rotational speed difference between a virtual rotational speed (ω) and a reference rotational speed is formed, - a filtered value of the virtual rotational speed (ω) or a predefined frequency is used as the reference rotational speed, - the difference rotational speed (Δω) is calculated by way of a difference rotational speed gain with respect to an auxiliary torque, - the auxiliary torque acts on the virtual moment of inertia (J) of the machine model by way of a summing point in order thereby to control the virtual rotational speed (ω) to the reference rotational speed, wherein preferably - the difference rotational speed (Δω) is set to zero for the purpose of synchronizing the machine model with the electrical power supply network (120).

8. The method as claimed in any of the preceding claims, characterized in that - for the purpose of synchronizing the machine model with the electrical power supply network (120), - a setpoint power has the value zero, - a calculation model is used for calculating - internal virtual generator voltages and / or - a or the virtual torque (Te), wherein the calculation model uses for calculating one, a plurality or all of the variables in the list comprising - a or the virtual angle of rotation (ϑ) of the rotor, - a or the virtual rotational speed (ω) of the rotor, - a or the virtual excitation voltage (Ue), and - the infeed current or the setpoint current (Isabc), wherein in particular no frequency of the electrical power supply network (120) is detected.

9. The method as claimed in any of the preceding claims, characterized in that - a or the virtual excitation voltage (Ue) is determined at least as a function of - a predefined reactive power and optionally - a predefined electrical network voltage (V, E) at the network connection point.

10. The method as claimed in any of the preceding claims, characterized in that electrical variables of the machine model are calculated in space vector representation, in particular in accordance with a d / q transformation.

11. The method as claimed in any of the preceding claims, characterized in that the infeed current is generated by means of a tolerance band method.

12. The method as claimed in any of the preceding claims, characterized in that in island network operation, if the inverter predefines the network frequency, - the virtual rotational speed (ω) is dependent on a predefined network frequency, in particular such that a or the auxiliary torque acting in the machine model is controlled as a function of a difference between virtual rotational speed (ω) and predefined network frequency, and / or - the virtual excitation voltage (Ue) - is dependent on a predefined voltage of the electrical power supply network (120) and a detected voltage of the electrical power supply network (120), in particular such that an auxiliary reactive power value is controlled as a function of a difference between the predefined voltage and the detected voltage, the virtual excitation voltage (Ue) being dependent on said auxiliary reactive power value.

13. The method as claimed in claim 12, characterized in that in island network operation - the auxiliary torque is controlled by way of a PI controller and - the virtual rotational speed (ω) results in particular from an integration of a difference torque, as difference between the auxiliary torque and the virtual electrical torque, by way of an integration time constant (1 / J), and / or - the auxiliary reactive power value is controlled by way of a PI controller, and - the virtual excitation voltage (Ue) results in particular from an integration of a difference reactive power, as difference between the auxiliary reactive power and a detected reactive power, by way of an integration time constant (G).

14. A wind power installation (100) for feeding electrical power into a three-phase electrical power supply network (120) at a network connection point, comprising - an inverter for generating an infeed current, - a detection means for detecting an electrical network voltage (V, E), in particular at the network connection point, - a control device for controlling the infeed, and the control device is prepared for carrying out a method, comprising the following steps: - determining a virtual generator voltage using a machine model that emulates a behavior of a synchronous machine, - preparing the detected network voltage (V, E) for comparison with the virtual generator voltage, and - predefining a setpoint current (Isabc) as predefinition for the infeed current as a function of the virtual generator voltage and as a function of the network voltage (V, E) prepared for comparison, wherein - the inverter is prepared to generate the infeed current depending on the setpoint current (Isabc) and to feed the generated infeed current at the network connection point into the electrical power supply network (120), wherein preparing the detected network voltage (V, E) for comparison with the virtual generator voltagecomprises transforming the detected network voltage (V, E) into a space vector representation, and wherein characterized in that a virtual impedance (Zv) is taken into account for predefining the setpoint current (Isabc), which virtual impedance (Zv) is taken into account as an impedance between an output of the machine model or the virtual synchronous machine and the network connection point, and the magnitude of the virtual impedance (Zv) is variable, and the virtual impedance (Zv) is selected depending on whether - infeed is effected in a normal state of the electrical power supply network (120), or - infeed is effected in a recovery mode after interruption or failure of the electrical power supply network (120), in which recovery mode the electrical power supply network (120) must be run up to a normal operating point.

15. The wind power installation (100) as claimed in claim 14, characterized in that the wind power installation (100), in particular the control device, is prepared to carry out a method as claimed in any of claims 1 to 15.

16. A wind farm (112) comprising a plurality of wind power installations (100), wherein - wind power installations (100) as claimed in claim 14 or 15 are used and / or - a farm infeed device, in particular a farm inverter, is provided, which is connected to a network connection point and is prepared to carry out a method as claimed in any of claims 1 to 13.