Detection of Island Networks with Voltage Source Inverters and Inverters

Abstract

The invention relates to a method for detecting an island network using a voltage-applied converter (14), comprising the steps of determining the frequency and phase of a voltage curve of a sub-network (10) connected to a network connection point of the converter (14), generating a voltage curve using a bridge circuit of the converter (14) at a specific frequency, the phase of the generated voltage curve having a first time variation with respect to the determined phase, determining a correlation between the first time variation and a second time variation of the output exchanged by the converter (14) and the sub-network (10), and detecting an island network if the determined correlation is below a specified correlation measure. The invention also relates to a corresponding voltage-applied converter (14).

Description

[Technical field]

[0001] The present invention relates to a method for detecting an island network with a voltage source inverter and to an inverter of this kind. [Background technology]

[0002] Distributed energy production from renewable energy sources, in particular photovoltaics, is becoming increasingly important for the energy transition. In this respect, the component of decentrally generated power in relation to the power generated by power plants continues to grow. The concern here is that this increase in distributed generation increases the risk of blackouts. The use of voltage-source inverters or network-forming inverters (the latter also called current-source inverters or current converters for short) instead of linearly commutated inverters can make a significant contribution to the stabilization of the network. Voltage-source inverters are configured to supply power by providing at their bridge output a voltage curve that is continuously synchronized with the voltage curve at the network connection point of the inverter. Since a decoupling impedance is arranged between the bridge output of the inverter and the network connection point, the defined active and reactive power exchanged with the network can be adjusted by a predefined, continuous, i.e. continuously or repeatedly updated phase and amplitude offset between the voltage at the bridge output and the voltage at the network connection point. For example, EP 2 003 759 A1 discloses a method for island network detection of voltage-source inverters, in which a test voltage is superimposed on the network frequency, the test frequency of said test voltage being lower than the network frequency. If the reactive power component assigned to the test voltage is below a predefined minimum value, an islanded network is detected.

[0003] For networks with a network impedance dominated by inductive components, the exchanged real power is regulated or determined by the continuously updated phase difference between the voltage at the bridge output and the voltage at the network connection point, and the reactive power is regulated or determined by the continuously updated amplitude difference between the two voltages. For networks with a network impedance dominated by resistive components, the roles of reactive and real power are usually reversed. Hybrid forms are also conceivable. As a result, the voltage-source inverter can independently establish an AC voltage network and maintain it in a stable manner.

[0004] However, this characteristic entails the risk that the voltage of the sub-network will also be maintained if the sub-network to which the voltage source inverter is connected is unexpectedly or forcibly disconnected from the upper network, which is undesirable in this case. In this case, the voltage source inverter can easily form an island network as long as it is capable of delivering the required power. This can lead to dangers if the sub-network is disconnected from the upper network, for example for maintenance purposes. Therefore, the voltage source inverter may need to be able to independently detect, within a predefined period of time, that it is connected to an island network and respond appropriately to this.

[0005] Such requirements for island network detection are typical for linearly commutated inverters and are described, for example, in the technical application standard VDE_AR-N 4105 for power plants in low-voltage networks. Corresponding methods are known. However, these methods cannot be easily transferred to voltage-source inverters. Summary of the Invention [Problem to be solved by the invention]

[0006] It is therefore an object of the present invention to provide a method that can reliably detect whether a voltage source inverter is connected to an island network. [Means for solving the problem]

[0007] This object is achieved by a method having the features of independent claim 1 and by a voltage source inverter having the features of independent claim 17. Preferred embodiments of the invention are set out in the dependent claims.

[0008] The method according to the present invention for detecting an island network by a voltage source inverter comprises the steps of: determining the frequency and phase of the voltage curves of the sub-networks connected to the network connection points of the inverter; generating a voltage curve having the determined frequency by a bridge circuit of an inverter, the phase of the generated voltage curve including a first time variation with respect to the determined phase; determining a correlation between a first temporal variation and a second temporal variation of the power exchanged between the inverter and the sub-network; detecting an island network if the determined correlation does not reach a predefined degree of correlation; Includes.

[0009] In this case, the power exchanged between the inverter and the sub-network can be the active power, especially in the case of networks with a network impedance dominated by inductive components. However, in certain applications, the apparent power, which is an easier variable to measure, or in general terms the weighted root mean square of the active and reactive powers can also be used here, in which case the weight of the averaging is determined based on the resistive and inductive components of the network impedance. This allows to improve the detection accuracy of the method. In the following, in some cases, the invention will be described with reference to the active power as the variable used for correlation for simplicity, without consequently excluding the other two possibilities.

[0010] The time variation of the phase is the intentionally induced deviation of the phase difference between the voltage curve of the bridge output and the voltage curve of the network connection point voltage from the time-constant phase difference Δφ, extrapolated from the previously determined frequency and phase of the voltage curve of the network connection point. If the voltage curve of the bridge output is TIFF2025505486000002.tif8170, The result is TIFF2025505486000003.tif7170.

[0011] Phase variation Δφ AID (t) can be generated as a periodic or aperiodic signal, so that it has a power component that is time-correlated to the phase fluctuations and proportional to the phase fluctuations, and if the voltage curve at the network connection point exactly matches the extrapolated characteristic, the curve of the network voltage is not influenced by the power supplied by the inverter. In this case, it is called a fixed network, since the phase of the network is constant, for example. The power components of the fixed network can be calculated in a known manner from the phase fluctuations, using the known inductance of a network filter placed between the bridge output and the network connection point.

[0012] In the presence of an island network, i.e. in the absence of a connection to a fixed network, the power supplied by the inverter influences the phase curve of the voltage at the network connection point, which to some extent follows the time variations of the phase generated by the inverter, so that the phase difference between the voltage characteristic at the network connection point and the bridge output is smaller than the generated variations and therefore smaller than the phase difference in the case of a fixed network. Here, it is conceivable that the phase of the network connection point perfectly follows the generated phase variations, so that the phase difference between the network connection point and the bridge output is constant in time, despite the occurrence of phase variations, especially if the inverter forms the only voltage source component of the sub-network. In this case, the power supplied by the inverter does not have a power component that is correlated in time with the generated phase variations. As a result, in this case, the island network is easier to detect.

[0013] However, situations may also occur where there is a certain phase constancy of the sub-network, i.e. the phase of the network connection points only partially follows the phase variations. In this case, the phase difference only reflects the generated phase variations to a lesser extent and the power components correlated in time with the generated phase variations are smaller than those that arise in the case of the fixed network described above. Island networks can be detected by a suitably selected degree of correlation, for example by the ratio of the power components determined from the correlation with the phase variations to the power components of the fixed network. The smaller this ratio, the more likely an island network is present.

[0014] One possibility for determining the correlation degree K is the following relationship: The method is given by TIFF2025505486000004.tif17170 and is particularly suitable for periodic phase variations. TIFF2025505486000005.tif9170 is assumed. In practice, the degree of correlation is often ascertained by the corresponding sum using a time series of the measured power values ​​and the phase fluctuations generated in association with them. Other ways of forming the correlation are of course conceivable, for example, the frequency content of the power fluctuations at the frequency of the phase fluctuations can be determined using a lock-in amplifier and substituted into the ratio with the amplitude of the phase fluctuations to determine the appropriate degree of correlation.

[0015] The selection of the limit value of the correlation degree can be adapted to the installation situation of the inverter, since if the limit value is not reached, it indicates that an island network has been detected. For example, if it is known that a further voltage-source inverter is installed in a network section, the sub-network itself, when disconnected from the upper network, will have a certain phase constancy caused by the further voltage-source inverter in response to the phase fluctuations caused by the method according to the invention. During normal operation, these inverters are synchronized with the phase of the network connection point continuously or at certain time intervals (hereinafter also called synchronization frequency). It is therefore preferable to generate phase fluctuations with a fluctuation frequency lower than the synchronization frequency or lower than the typical value range of the synchronization frequency of the further inverter, particularly preferably a part (for example 1 / 2, 1 / 3 or 1 / 4) of the synchronization frequency, since the synchronization is performed during normal operation, and therefore for such slow phase fluctuations the phase constancy is reduced.

[0016] Likewise, it is advantageous to generate the phase variation as a periodic variation having a first and a second frequency, the second frequency being greater than the first frequency, the first variation frequency being preferably less than the synchronous frequency, and the second variation frequency being preferably greater than the synchronous frequency. In particular, by comparing the ratio of the power components at the first variation frequency to the power components at the second variation frequency, it is possible to identify whether the phase constancy is caused by a higher-level network or another voltage-source inverter. If the phase constancy is caused by another voltage-source inverter, the power components will be lower at the second variation frequency than at the first, smaller variation frequency, and therefore the aforementioned ratio of the power components will be greater than if the higher-level network is still present. This response can be used additionally or alternatively to detect island networks.

[0017] In an advantageous embodiment of the invention, the method is repeated at time intervals. As soon as an island network is detected, the inverter preferably disconnects itself from the sub-network. Now, it is conceivable that after detecting the island network but before disconnecting from the sub-network, the inverter operates in such a way that the amplitude and / or frequency of the AC network is outside the range of permitted network parameters, which causes the disconnection of further network components connected to the sub-network. This increases the certainty that the detected island network is de-energized.

[0018] In addition to or instead of disconnecting from the sub-network, the inverter can provide a detection signal if an island network is detected. For this purpose, a dedicated signal output is provided or the inverter provides the detection signal using a digital communication interface. The detection signal can also be provided and transmitted as a signal modulated onto the generated AC voltage in the form of a high-speed power line communication (PLC) signal.

[0019] The detection signal thus provided can be transmitted to a network operator and / or can be used to cause further devices connected to the sub-network (in particular synchronous generators) to shift the amplitude and / or frequency of the AC network to values ​​outside the permitted network parameters and thus to disconnect further network components connected to the network.

[0020] In one embodiment of the present invention, the time variation of the phase comprises a periodic variation having a first component at a first variation frequency. Preferably, an island network is detected when the ratio of the component of the active power at the first variation frequency as a degree of correlation to the first component of the periodic variation falls below a predefined limit value (first detection criterion).

[0021] In a further embodiment of the invention, the temporal phase variation has, in addition to a first component at a first variation frequency, a second component with a second variation frequency higher than the first variation frequency. Here, for example, an island network is detected if the ratio of the component of the active power at the first variation frequency to the component of the active power at the second variation frequency falls below a predefined limit value. This second detection criterion can be used as an independent detection criterion or in logical combination with the first detection criterion.

[0022] In order to minimize the negative impact of the detection method on the network quality, the generation of the voltage curve may include gradually increasing the amplitude of the phase fluctuations, and if the fluctuations of the active power exchanged between the inverter and the AC network exceed a predefined end value, the method is terminated without detecting an island network. The end value may be predefined as a fixed value or may be determined as a function of a voltage component determined before executing the method, at the network connection point at the used fluctuation frequency. The latter allows taking into account potential interactions with other voltage-source inverters that also execute the method according to the invention for detecting island networks at the same or slightly shifted fluctuation frequency, so that the detection method is more reliable since they have less influence on each other. It is also conceivable to use the value of the amplitude of the phase fluctuations at the end of the method in order to define the starting amplitude of the phase fluctuations for the subsequent island network detection method. As a result, the negative impact of the method according to the invention on the network quality can be reduced.

[0023] To further take such interactions into account, determining the frequency and phase of the voltage curve of the AC network connected to the network connection point of the inverter may include determining a third time variation of the phase, and within the range of the island network detection, a voltage curve having a time phase variation that is synchronized with the previously determined third time variation of the phase is generated. In this way, multiple voltage-source inverters can constructively interact in the island network detection, increasing the detection sensitivity.

[0024] The method according to the present invention can be performed both with inverters connected to a single phase and with inverters connected to multiple phases having first and second phase conductors. In the latter case, the method can be performed simultaneously or alternately on the first and second phase conductors, and an island network is detected if a phase island network is detected on at least one of the phase conductors. Alternatively, an island network can be detected only if a phase island network is detected on both the first and second phase conductors.

[0025] According to these options, the method can also be extended to further phase conductors via which the inverter is connected to the network.

[0026] In a further aspect of the present invention, the voltage source inverter comprises: a measuring unit for determining the frequency and phase of the voltage curves of the sub-networks connected to the network connection points of the inverter and for determining the values ​​of the active and reactive power exchanged with the sub-network; a control unit for generating switching commands for a bridge circuit of the inverter, such that a voltage curve predefined by the control unit and having a determined frequency is generated at an output point of the bridge circuit, the control unit being configured to provide a predefined voltage curve having a time variation in phase relative to a phase of the determined voltage characteristic; a correlator for determining a correlation between the time variation of the phase and the time variation of the active power exchanged between the inverter and the sub-network, the correlator being configured to detect an islanded network if the degree of correlation of the determined correlation does not reach a predefined limit value; Equipped with. [Brief description of the drawings]

[0027] The invention is explained below with the aid of the following figures:

[0028] [Figure 1] FIG. 1 shows an example of an installation of a sub-network having multiple inverters. [Diagram 2] FIG. 2 shows the controller structure of the inverter according to the present invention. [Diagram 3] FIG. 3 shows a schematic diagram of an inverter according to the present invention. [Figure 4] FIG. 4 shows the time characteristics of the phase and switched power when carrying out the method of the invention in the case of connection to a higher-level network. [Diagram 5] FIG. 5 shows the time characteristics of phase and switched power when carrying out the method according to the invention in a first embodiment for the case of an island network. [Figure 6] FIG. 6 shows the time characteristics of phase and switched power when carrying out the method according to the invention in a second embodiment for the case of an island network. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0029] FIG. 1 shows an example design of a sub-network 10 connected to a higher-level network 11 via a circuit breaker 13 and a medium-voltage transformer 12. The sub-network 10 comprises a voltage-source inverter 14, a further inverter 15 and a number of loads 16 which are at least partially supplied with power by the inverters 14, 15 (possibly from the network 11). The further inverter 15 can likewise be designed to be voltage-sourced or externally excited. The sub-network 10 can have further components which supply power. The voltage-source inverter 14 can be connected on the DC side to a battery and / or a generator (e.g. a photovoltaic generator). It is also conceivable to connect loads rather than feeders via the power supply inverter. The inverters 14, 15 can each be designed as purely feed-in inverters or as bidirectional inverters.

[0030] The controller structure of the voltage source inverter according to the present invention is shown in FIG. NAPThe measured value of may be the current I flowing through the network connection point. NAP and the voltage at the network connection point U NAP Amplitude of TIFF2025505486000006.tif8170, are evaluated in a phase evaluation unit 20 to determine the frequency ω and the phase φ0. The phase evaluation unit can be designed, for example, as a phase-locked loop (PLL). The determined values ​​are fed as input variables to a pulse width modulation (PWM) generator to generate a pulse pattern with a duty cycle d, which operates a bridge circuit 22 to generate a voltage U at its bridge output. Br The bridge circuit 22 here can be a single-phase or multi-phase bridge circuit, and a dedicated pulse pattern can be generated for each of the phase conductors. Different topologies of the bridge circuit 22 with various numbers of bridge switches are contemplated here, in particular multi-level topologies, and the pulse pattern can have individual switching commands for each bridge switch.

[0031] Voltage U at the network connection point NAP Measured values ​​of and current I at the network connection point NAP The measured value of is the effective power P flowing through the network connection point. ist and reactive power Q ist The phase estimation unit 20 and the power determination unit 23 are coupled together to form a measurement unit 28.

[0032] The active power P flowing through the network connection point ist and reactive power Q ist The value of is supplied as an input variable to the characteristic curve unit 24, which determines the amplitude Based on the additional input variables TIFF2025505486000007.tif6158 and frequency ω, characteristic variables are determined which represent the setpoints of the reactive and active powers. During the determination of the setpoints, further variables can of course also be taken into account, such as the link circuit voltage or the state of charge of the accumulator connected on the DC side. In the case shown in FIG. 2, these characteristic variables are generated as the amplitude difference ΔU and the phase difference Δφ and are related to the voltage curve U generated by the bridge circuit 22. Br is the voltage curve U at the network connection point. NAP ω is supplied as a further input variable to the PWM generator 21 in order to generate a pulse pattern with an amplitude or phase difference corresponding to ω. In this case, the amplitude difference ΔU and the phase difference Δφ are determined as a function of the inductance of the decoupling inductor of a voltage-source inverter, where the decoupling inductor is arranged between the bridge output and the network connection point. To determine the setpoints of the active and / or reactive power, the characteristic curve unit 24 includes, for example, a function for determining the setpoint of the active power from the frequency ω and a function for determining the setpoint of the amplitude A characteristic curve for determining the setpoint of the reactive power is stored from TIFF2025505486000008.tif8170. However, the characteristic curve can also include further parameters, which can be implemented not only as fixed predefined parameters, but also as variable parameters, such as the above-mentioned link circuit voltage or the charge state of the accumulator.

[0033] During the period when island network detection is performed by the voltage source inverter, the additive phase difference Δφ AID is added to the phase difference Δφ determined by the characteristic curve unit 24, said added phase difference being generated by a generator 26 as a time-variable quantity. In this case, the time-variable quantity may be a periodic signal having one or more frequency components. However, it is also conceivable to use a non-periodic signal, such as a step function. The non-periodic signal can preferably be parameterized, in particular with respect to shape or time, or in another way, based on the communicated information.

[0034] During the period when island network detection is not performed, the additive phase difference Δφ AID is zero or is not added to the phase difference Δφ.

[0035] The correlator 25 receives the effective power P ist The amount of the curve and the phase difference Δφ generated by the generator AID and from there, a signal S, provided by the voltage source inverter as an internal or external signal and indicating whether the inverter detects a current island network, is determined. AID To generate the signal S, a correlation between these quantities is calculated, which is compared with a predefined limit value. Thus, if the calculated correlation does not reach the predefined limit value, a detected island network is indicated. AID Other devices receiving the signal can then respond accordingly.

[0036] In this case, the PWM generator 21, the generator 26 and the characteristic curve unit 24 are part of a control unit 27 of a voltage source inverter.

[0037] 3 shows a schematic design of an inverter according to the present invention. The DC side connections DC+, DC- are connected to a bridge circuit 22 via a link circuit 29, and the desired bridge output voltage U is obtained by appropriately controlling the bridge switches using a duty cycle d. Bris generated at the bridge output of said bridge circuit. The bridge output is connected to the network connection point 32 of the inverter via a decoupling inductor 31, which is part of the network filter. Further components (e.g. isolating relays) that may be arranged in the connection between the bridge output and the network connection point are not shown for the sake of clarity. Suitable measurement sensors are used to continuously or periodically measure the current and voltage at the network connection point 32, the measured values ​​being fed to a measurement unit 28, where they are evaluated to form network parameters 30 such as frequency, phase, amplitude, reactive power and active power. These network parameters 30 are transmitted to a control unit 27, which derives therefrom the desired bridge output voltage U Br Determine the duty cycle d for controlling the bridge switches required to generate

[0038] Here, the control unit 27 controls the periodic or aperiodic phase difference Δφ generated for carrying out the island network detection method according to the present invention. AID is sent to the correlator 25, which receives the phase difference Δφ AID and the power variable ascertained by the measuring unit 28 (here the active power P ist ) and calculate the corresponding correlation between the two variables. The island network detection signal S AID is generated based on the thus calculated degree of correlation, for example by comparing the latter with a predefined limit value, and provided for transmission within the inverter or at an output or communication interface.

[0039] 4 shows the time characteristics of variables relevant for understanding the present invention when a voltage source inverter implementing the method according to the present invention is connected to a higher-level network. The upper part of the graph shows the phase φ at the network connection point. NAP and the corresponding phase φ at the bridge output. Br In the example shown in FIG. 4, the phase φNAP is strictly linear with a slope that is a function of the network frequency ω. To this end, the phase φ of the bridge output Br is initially the phase of the network connection point φ NAP However, it periodically deviates from this offset characteristic, and here, the amplitude is Δφ AID The phase at the bridge output is φ Br The periodic deviation of NAP Due to the fact that the phase φ at the network connection point is not transferred to the curve of NAP The curve of is thus constant, exemplarily speaking, and the phase φ at the bridge output Br and the phase φ at the network connection point NAP The difference between AID The curve has a periodic shape having the following characteristics:

[0040] The lower subgraph shows the time curve of the active power P at the network connection point, which corresponds to the phase curve of the upper subgraph. The active power is calculated by the phase φ Br and the phase of the network connection point φ NAP and the amplitude ΔP around the average value P0 AID The mean value P0 corresponds to a constant phase difference Δφ. The curves of the active power P at the network connection point and the phase φ at the bridge output Br The coupling between the curves K and ΔP is caused by the inverter decoupling inductors placed between the bridge output and the network connection point. The relationship K=ΔP AID / Δφ AID corresponds to the degree of correlation that can be used to detect an island network. In the present case of the connection of a voltage source inverter to a higher-level network, this relationship is greater than the predefined limit value for detecting an island network, so this connection to the higher-level network is detected. However, the amplitude ΔP AID and amplitude Δφ AID It is contemplated that other relationships as a function of may be used to determine the degree of correlation.

[0041] 5 shows the time characteristics of the variables relevant for understanding the invention when the voltage source inverter implementing the method according to the invention has a faulty or very weak connection to the upper network, where the upper subgraph shows the phase φ at the network connection point. NAP and the corresponding phase φ at the bridge output. Br The time curve of the phase φ at the bridge output is shown. Br Here, the amplitude Δφ AID In contrast to the case shown in FIG. 4, this deviation is due to the phase φ NAP from a linear curve corresponding to the network frequency ω. For example, the phase φ NAP is not constant along a strongly linear curve, but the phase φ at the bridge output Br It follows to some extent.

[0042] This tracking can be perfect, i.e. the phase φ at the network connection point from a purely linear curve. NAP The magnitude of the deviation of the phase φ at the bridge output from its linear profile. Br The amplitude of the deviation Δφ AID and the phase of the bridge output φ Br and the phase φ at the network connection point NAP The difference between V and V becomes constant over time, which may be the case, for example, if there are no further voltage source inverters in an island network that is completely disconnected from the higher-level network.

[0043] However, the phase φ at the network connection point NAP may only partially track the phase φ at the network node. NAP The amplitude of the deviation of the curve from a purely linear one is the phase φ at the bridge output from its linear profile. Br The amplitude of the deviation ΔφAID and the phase of the bridge output φ Br and the phase φ at the network connection point NAP The difference between Br The amplitude of the deviation Δφ AID The voltage fluctuates periodically with an amplitude smaller than . This case occurs, for example, if a weak connection to the upper network still exists or if another voltage-source inverter is operating in the island network. Figure 5 is based on this case.

[0044] As a result of this situation, the curve of the active power P shown in the lower subgraph of Figure 5 has an amplitude ΔP around the average value P0. AID , which has a smaller amplitude than the amplitude of the active power P shown in FIG. 4 with an existing low impedance connection in the upstream network. Thus, for example, the relationship K=ΔP AID / Δφ AID 4. The degree of correlation, as determined by α=α+ ...

[0045] FIG. 6 shows an exemplary time profile of variables relevant for understanding the invention when implementing the method according to the invention for the present island network, and the phase φ at the bridge output. Br A non-periodic variation of φ at the network connection point, where a step function is used. NAP does not immediately follow this step, but only gradually re-approaches the phase difference Δφ that prevailed before the time of the step. The curve of the active power P therefore changes with the height of the step Δφ AID Then the curve again approaches the average value P0 of the active power before the jump. In this case, the degree of correlation that allows the presence of an island network to be detected is, for example, the deviation ΔP from the average value P0 of the active power AIDcan be determined as the length of time during which the jump height ΔP falls again below a predefined percentage of the jump height ΔP. Of course, other methods for determining the degree of correlation are also conceivable.

[0046] If the voltage source inverter is connected to the upper network with an existing low impedance connection, the phase φ at the network connection point NAP remains constant due to its linear characteristic, resulting in a phase φ at the bridge output. Br The jump in active power ΔP caused by the step does not decrease, or decreases only very slowly.

[0047] To detect the island network, the voltage source inverter, of course, detects the phase φ Br Other periodic or aperiodic variations in the correlation coefficient and correspondingly adapted methods for determining the degree of correlation or correspondingly adapted limit values ​​for distinguishing between island networks and sub-networks connected with a sufficiently low impedance to a higher-level network can also be used without departing from the scope of protection of the claims. [Explanation of symbols]

[0048] 10 Subnetworks 11 Network 12. Transformers 13. Circuit Breaker 14 Inverter 15 Inverter 16 Load 20 Phase Evaluation Unit 21 PWM Generator 22 Bridge Circuit 23 Power determining device 24 Characteristic Curve Unit 25 Correlator 26 Generator 27 Control Unit 28 Measurement Units 29 Link Circuit 30 Network parameters 31 Decoupling inductor

Claims

1. A method for detecting an island network by a voltage source inverter (14), comprising the steps of: determining the frequency and phase of the voltage curve of a sub-network (10) connected to the network node of said inverter; generating a voltage curve having the determined frequency by a bridge circuit of the inverter (14), the phase of the generated voltage curve including a first time variation with respect to the determined phase; determining a correlation between the first and second temporal variations of the power exchanged between the inverter (14) and the sub-network (10); detecting an island network if the determined correlation does not reach a predefined degree of correlation; A method comprising:

2. The method of claim 1 , wherein the method is repeated at time intervals.

3. 3. The method according to claim 1 or 2, wherein if an island network is detected, the inverter (14) is disconnected from the sub-network (10).

4. 4. The method of claim 3, wherein if an island network is detected, before the inverter is disconnected from the sub-network, the inverter is operative to drive the amplitude and / or frequency of the sub-network outside of a range of acceptable network parameters in order to cause disconnection of further network components connected to the sub-network.

5. The method according to any one of claims 1 to 4, wherein if an island network is detected a detection signal is provided at a signal output of the inverter (14).

6. The method according to any one of claims 1 to 5, wherein if an island network is detected, the inverter (14) modulates a detection signal as a PLC signal onto the voltage of the network node.

7. The method according to any one of claims 1 to 6, wherein the first temporal variation comprises a periodic variation having a first component at a first variation frequency.

8. 8. The method of claim 7, wherein an islanded network is detected when a ratio between a component of active power at the first variation frequency and the first component of the first time variation as the correlation measure does not reach a predefined limit value.

9. 8. The method of claim 7, wherein the first temporal variation includes a second component having a second variation frequency higher than the first variation frequency, and an islanded network is detected when a ratio of a component of power at the first variation frequency to a component of power at the second variation frequency does not reach a predefined limit value.

10. 10. The method according to claim 1, wherein the step of generating a voltage curve comprises a step of increasing an amplitude of the first temporal variation, and when the variation of the power exchanged between the inverter and the sub-network (10) exceeds a predefined value, the method is terminated without detecting an island network.

11. 11. The method of claim 10, wherein if the method terminates without detecting an island network, the amplitude of the first temporal variation is used to determine a starting amplitude of the first temporal variation the next time the island network detection method is performed.

12. The method according to any one of claims 1 to 11, wherein the step of determining the frequency and phase of a voltage curve of the sub-network (10) connected to the network connection point of the inverter (14) comprises the step of determining a third time variation of the phase, and the voltage curve is generated having a time variation synchronous with the determined third time variation of the phase.

13. 13. The method of claim 1, wherein the sub-network (10) is a network having first and second phase conductors, the method being performed simultaneously or alternately for the first and second phase conductors, and an island network being detected if a phase island network is detected for at least one of the phase conductors.

14. 14. The method according to claim 1, wherein the sub-network (10) is a network having first and second phase conductors, the method being performed simultaneously or alternately for the first and second phase conductors, and an island network is detected only if a phase island network is detected for both the first and second phase conductors.

15. The method according to any one of claims 1 to 14, wherein active power is used to determine the correlation between the first and second temporal variations of the power exchanged between the inverter (14) and the sub-network (10).

16. 14. The method according to claim 1, wherein a weighted root mean square of apparent power or active and reactive power is used to determine the correlation between the first and second temporal variations of the power exchanged between the inverter and the sub-network.

17. A voltage source inverter (14), a measuring unit (20) for determining the frequency and phase of the voltage curve of a sub-network (10) connected to a network connection point of said inverter (14) and for determining the values ​​of active and reactive power exchanged with said sub-network (10); a control unit (21) for generating switching commands for a bridge circuit (22) of the inverter (14), such that a voltage curve predefined by the control unit (21) and having the determined frequency is generated at a bridge output of the bridge circuit (22), the control unit (21) being configured to provide a predefined voltage curve including a first time variation of the phase relative to the phase of a determined voltage characteristic; a correlator (25) for determining a degree of correlation between a first temporal variation of the phase and a second temporal variation of the power exchanged between the inverter (14) and the sub-network (10), the correlator (25) being configured to detect an island network if the determined degree of correlation does not reach a predefined limit value; A voltage source inverter (14).

Images