Method and apparatus for detecting low-frequency oscillations

Abstract

Provided is a method for detecting low-frequency oscillations, in particular subsynchronous resonance, in an electrical supply grid, wherein the electrical supply grid has a grid voltage at a grid nominal frequency, comprising the steps of capturing at least one electrical signal from the electrical supply grid, and evaluating the electrical signal by means of wavelet analysis, during which a time-dependent frequency pattern is created by analyzing a correlation of the captured signal with a predetermined wavelet mother function, wherein the presence of a low-frequency oscillation is assumed if at least one further low-frequency frequency component is present in the time-dependent frequency pattern in addition to a fundamental component.

Claims

1. A method for detecting low-frequency oscillations in an electrical supply grid having a grid voltage associated with a nominal frequency, comprising: capturing at least one electrical signal from the electrical supply grid, the capturing including detecting a three-phase voltage of the grid voltage; transforming the three-phase voltage into a d/q representation; evaluating the transformed three-phase voltage using wavelet analysis, including: generating a time-dependent frequency pattern based on analyzing a correlation between the transformed three-phase voltage and a wavelet mother function; detecting a presence of a low-frequency component and a fundamental component in the time-dependent frequency pattern; and in response to detecting the presence of the low-frequency component and the fundamental component in the time-dependent frequency pattern, determining that a low-frequency oscillation is present, wherein the method is performed at a grid connection point where a wind power installation or wind farm supplies electrical current to the electrical supply grid, and wherein the at least one electrical signal is a voltage at the grid connection point or a voltage proportional to the voltage at the grid connection point.

2. The method as claimed in claim 1, comprising: determining whether the low-frequency component fluctuates; or determining whether the low-frequency component fluctuates periodically.

3. The method as claimed in claim 1, comprising: capturing a temporal profile of an amplitude of the low-frequency component.

4. The method as claimed in claim 1, comprising: evaluating the at least one electrical signal based on a test frequency range, wherein: the test frequency range is from a frequency lower limit to a frequency upper limit, the frequency lower limit is between 0.1 Hz and 2 Hz, and the frequency upper limit is between the nominal frequency and five times the nominal frequency or the frequency upper limit is the nominal frequency.

5. The method as claimed in claim 1, wherein the wavelet mother function has at least one property from a list of properties including: the wavelet mother function is scalable in time or frequency; the wavelet mother function is shiftable in time; the wavelet mother function has locality both in a frequency domain and in a time domain; the wavelet mother function has an integral of 0 in the time domain; and the wavelet mother function is a Morlet wavelet, a Daubechies D20 wavelet or a Mexican hat.

6. The method as claimed in claim 1, comprising: selecting the wavelet mother function based on a selection criterion; and selecting the wavelet mother function from a plurality of predetermined wavelet functions.

7. The method as claimed in claim 1, wherein: the low-frequency oscillation is a grid oscillation, the wind power installation includes a rotor with rotor blades, a generator, and an inverter that supplies electrical power to the electrical supply grid, and the method comprises: determining whether the wind power installation causes the grid oscillation by at least: identifying a low-frequency oscillation of the wind power installation as an installation oscillation using wavelet analysis of a DC link voltage of a DC link of the inverter; determining whether the grid oscillation and the installation oscillation are correlated; and determining the wind power installation causes the grid oscillation in response to determining that the grid oscillation and the installation oscillation are correlated.

8. The method as claimed in claim 7, comprising: in response to determining that the wind power installation causes the grid oscillation, determining whether the installation oscillation exceeds a predefined oscillation amplitude, and in response to determining that the installation oscillation exceeds the predefined oscillation amplitude, initiating an attenuation measure including at least one measure from a list of measures including: increasing an attenuation component of a controller of the generator; increasing a stator current of the generator; adjusting, using pitch control, a blade angle of the rotor blades to adjust the attenuation component; increasing an attenuation component of a supply controller to control the inverter to supply the electrical supply grid; shifting an operating point of power supply; reducing the power supply; and activating current control of the inverter to control a supply current and adjust the attenuation component.

9. The method as claimed in claim 1, wherein: capturing the at least one electrical signal includes sampling, at a sampling frequency, a measurement to respectively obtain a sampling signal of the at least one electrical signal, wherein the sampling frequency is higher than a frequency of the fundamental component, the sampling frequency is a multiple of the frequency of the fundamental component or the sampling frequency is at least ten times the frequency of the fundamental component, and evaluating the at least one electrical signal includes evaluating the sampling signal of respective samples using the wavelet analysis.

10. The method as claimed in claim 1, wherein the low-frequency oscillations are a subsynchronous resonance.

11. The method as claimed in claim 1, wherein the wavelet mother function is parameterized.

12. The method as claimed in claim 6, wherein the low-frequency component and the low-frequency oscillation have a frequency between 1 Hz and five times the nominal frequency.

13. The method as claimed in claim 6, wherein the selection criterion has at least one expected value of the at least one electrical signal.

14. The method as claimed in claim 9, wherein evaluating the sampling signal of the respective samples using the wavelet analysis is performed without filtering the respective samples.

15. A detection device for detecting low-frequency oscillations in an electrical supply grid having a grid voltage and an associated a grid nominal frequency, comprising: a sensor configured to capture at least one electrical signal from the electrical supply grid, the at least one electrical signal comprising a three-phase voltage of the grid voltage; and a controller configured to: transform the three-phase voltage into a d/q representation; evaluate the transformed three-phase voltage using wavelet analysis, including: generating a time-dependent frequency pattern based on analyzing a correlation between the transformed three-phase voltage and a predetermined wavelet mother function; detecting a presence of at least one low-frequency component and a fundamental component in the time-dependent frequency pattern; and in response to detecting the presence of the low-frequency component and the fundamental component in the time-dependent frequency pattern, determining that a low-frequency oscillation is present, wherein the detection device is positioned at a grid connection point where a wind power installation or wind farm supplies electrical current to the electrical supply grid, and wherein the at least one electrical signal is a voltage at the grid connection point or a voltage proportional to the voltage at the grid connection point.

16. A wind power installation or wind farm configured to supply electrical power to an electrical supply grid and to be connected to the electrical supply grid at a grid connection point, the wind power installation or wind farm comprising: a controller configured to: detect low-frequency oscillations in the electrical supply grid, wherein the electrical supply grid has a grid voltage and is associated with a grid nominal frequency, and the low-frequency oscillations to be detected have a lower frequency than the grid nominal frequency; and a sensor configured to capture at least one electrical signal from the electrical supply grid, the at least one electrical signal comprising a three-phase voltage of the grid voltage; wherein the controller is configured to: transform the three-phase voltage into a d/q representation; evaluate the transformed three-phase voltage using wavelet analysis, including: generating a time-dependent frequency pattern based on analyzing a correlation between the transformed three-phase voltage and a predetermined wavelet mother function; detecting a presence of a low-frequency component and a fundamental component in the time-dependent frequency pattern; and in response to detecting the presence of the low-frequency component and the fundamental component in the time-dependent frequency pattern, determining that a low-frequency oscillation is present, wherein the sensor is positioned at the grid connection point where the wind power installation or wind farm supplies electrical current to the electrical supply grid, and wherein the at least one electrical signal is a voltage at the grid connection point or a voltage proportional to the voltage at the grid connection point.

Description

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

(1) The invention is explained in more detail, by way of example, below on the basis of embodiments and with reference to the accompanying figures.

(2) FIG. 1 shows a perspective illustration of a wind power installation.

(3) FIG. 2 shows a schematic illustration of a wind farm.

(4) FIG. 3 shows a schematic illustration of a simplified structure for the capturing and further processing of low-frequency oscillations.

(5) FIG. 4 illustrates a section of the structure from FIG. 3.

(6) FIG. 5 illustrates possible signal evaluation of an element from FIG. 4.

(7) FIG. 6 illustrates a possible result of the analysis illustrated in FIG. 4.

DETAILED DESCRIPTION

(8) FIG. 1 shows a wind power installation 100 having a tower 102 and a nacelle 104. A rotor 106 having three rotor blades 108 and a spinner 110 is arranged on the nacelle 104. The rotor 106 is caused to rotate during operation by the wind and thereby drives a generator in the nacelle 104.

(9) FIG. 2 shows a wind farm 112 having, by way of example, three wind power installations 100 which may be identical or different. The three wind power installations 100 are therefore representative of fundamentally any desired number of wind power installations in a wind farm 112. The wind power installations 100 provide their power, specifically in particular the produced current, via an electrical farm grid 114. In this case, the respectively produced currents or powers from the individual wind power installations 100 are added and a transformer 116 is usually provided, which transformer steps up the voltage in the farm in order to then supply it to the supply grid 120 at the supply point 118 which is also generally referred to as a PCC. FIG. 2 is only a simplified illustration of a wind farm 112, which does not show a controller, for example, even though a controller is naturally present. The farm grid 114 may also have a different design, for example, by virtue of a transformer also being present at the output of each wind power installation 100, for example, to name just one other exemplary embodiment.

(10) FIG. 3 schematically shows a simplified structure as an overview of a proposed analysis of low-frequency oscillations for detecting such low-frequency oscillations. According to this overview structure 300, a wind power installation 301 is schematically illustrated and supplies an electrical supply grid 306 in a three-phase manner via a transformer 304, for example by means of a schematically illustrated frequency inverter 302. The currents i.sub.1, i.sub.2, i.sub.3 supplied in this case and the voltages v.sub.1, v.sub.2 and v.sub.3 of each phase which are present in this case are captured here at the measurement point 308 by means of a schematically indicated measurement device (voltmeter, ammeter or multimeter) 310 and are supplied to the evaluation system (controller) 312. The measurement point 308 may also form a grid connection point of the electrical supply grid 306. The voltages v.sub.1, v.sub.2 and v.sub.3 captured at the measurement point 308 therefore simultaneously also form the voltages of the electrical supply grid 306 at this grid connection point at the measurement point 308. In any case, however, these voltages captured at the measurement point 308 can provide information relating to the corresponding voltages in the electrical supply grid 306.

(11) The signals captured in this manner, specifically voltages and currents, can then be evaluated in the evaluation system 312. In this case, the evaluation can also use an external system which is illustrated here in the external block 314 as SCADA. This SCADA or SCADA system in the external block 314 can at least carry out or support the analysis for detecting low-frequency oscillations. It also comes into consideration that this SCADA system according to the external block 314 is additionally used to carry out additional analyses, for example, in order to improve the actual analysis for detecting the low-frequency oscillations, for example. The evaluation result could be remotely transmitted by means of a communication connection to the SCADA system according to block 314.

(12) According to one embodiment, the analysis for detecting the low-frequency oscillations is carried out in the evaluation system 312 and this is illustrated in FIG. 4.

(13) FIG. 4 therefore likewise schematically shows the evaluation system 312 but with further details. The possible communication to the external block 314 is not illustrated in FIG. 4 for the sake of simplicity.

(14) For the sake of simplicity, the evaluation system 312 is indicated as an evaluation system 412 in FIG. 4. It is shown there that the six measurement variables v.sub.1, v.sub.2, v.sub.3 and i.sub.1, i.sub.2, i.sub.3 form input variables in the filter block (filter, bandpass filter or low-pass filter) 416. Therefore, filtering which filters out, for example, high-frequency components from the captured signals can be carried out in the filter block 416. However, the filter block 416 also extracts components from these measurement signals in the process or subsequently for filtering. In particular, provision is made for a representative voltage v*, a representative current i*, a total active power p* and a total reactive power q* to be determined. In this case, these representative variables or total variables each together represent a corresponding signal for all three phases. These four variables which form the output of the filter block 416 are accordingly each input as a time signal in the determination block (controller) 418. The determination, that is to say the ascertainment, of the low-frequency oscillations is then carried out in the determination block 418 and at least a first important step thereof is carried out. The respective electrical signal, that is to say v*, i*, p* or q*, is specifically evaluated by means of wavelet analysis in the determination block 418. This is carried out in such a manner that a time-dependent frequency pattern is created by analyzing a correlation of the respective captured signal with a predetermined wavelet mother function. Said signals which are output by the filter block 416 and are input to the determination block 418, that is to say the voltage v*, the current i*, the power p* and the reactive power q*, therefore each form a captured signal.

(15) Such wavelet analysis can then be carried out with each of the four signals which are transferred from the filter block 416 to the determination block 418. The resulting four results can then be processed further and, in the simplest case, only the most meaningful result is used further.

(16) In any case, the result of the analysis in the determination block 418 is a time-dependent frequency pattern 420 which is only schematically indicated in FIG. 4. Only a three-dimensional coordinate cross is specifically used there to indicate that the result shows a time-dependent frequency, in which case the intensity can be additionally considered. This is intended to be indicated by the three axis labels t, f and int. A further illustration of this result, which indicates this time-dependent frequency pattern 420 here, is also explained in connection with FIG. 6.

(17) FIG. 5 schematically shows a possible evaluation which can be carried out in the determination block 418 in FIG. 4. In this example in FIG. 5, wavelet analysis is carried out by means of a correlation analysis, during which the captured signal 501 is correlated with a wavelet mother function 502. In this case, the wavelet mother function 502 is symmetrical over time and its maximum value in this case forms its temporal center and this is placed at the temporal zero point during the analysis. At this temporal zero point, on account of the high amplitude of the wavelet mother function, a high degree of correlation also results in a high intensity at that time. At times considerably before or considerably after this temporal zero point which is taken as a basis, the wavelet mother function 502 is substantially zero, with the result that a correlation can no longer occur there and accordingly no intensity either.

(18) Such an analysis illustrated in FIG. 5 is constantly repeated continuously or at least virtually continuously, thus resulting in a time-dependent frequency pattern with intensities.

(19) The captured signal 501 can therefore be the sinusoidal voltage signal which is representative of the three phase voltages v.sub.1, v.sub.2 and v.sub.3. This sinusoidal signal is fundamentally naturally known and the wavelet mother function 502 can be accordingly adapted thereto. The low-frequency oscillations to be captured are, in this respect, superimpositions on the ideally sinusoidal function, and these superimpositions are intended to be extracted by means of this wavelet analysis, that is to say also by the choice of the wavelet mother function 502, or to emerge at least to a greater extent.

(20) FIG. 6 illustrates a time-dependent frequency pattern which therefore corresponds to the result from the determination block 418, that is to say the time-dependent frequency pattern indicated in FIG. 4, and this is also the result of a continuously repeated analysis explained in connection with FIG. 5.

(21) In FIG. 6, the graph shows a frequency range, which is plotted logarithmically, over a time range from 0 to 40 000 s. The frequency is plotted here approximately from 30 Hz to 1 Hz. In order to improve the illustration, frequency lines for 1 Hz, 2 Hz and 8 Hz are depicted using dashed lines.

(22) The result of the wavelet analysis is to show intensities which have respectively occurred at the respective times and the respective frequencies. In this case, the representation usually uses a color scale in which, for example, the color scale extends from dark blue, via green and orange, to yellow, in which case dark blue can represent a low intensity and yellow can represent a high intensity. In FIG. 6, an attempt was made to illustrate such an intensity by means of accordingly pronounced regions in order to manage without a color scale. Therefore, there is a high intensity approximately in the range from 3 to 6 Hz. However, it can also be seen that this intensity is particularly pronounced in the range between 2 and 3, that is to say between 20 000 and 30 000 s. It can therefore be seen that a particularly strong oscillation has occurred in the range from 3 to 6 Hz in the three hours taken as a basis there. No significant intensity occurs in this frequency range in the range from 0 to 1, that is to say 0 to 10 000 s. Accordingly, a conclusion with regard to this low-frequency oscillation or low-frequency oscillations can be derived therefrom.

(23) In order to carry out control which counteracts a low-frequency oscillation, the practice of considering a considerably shorter period also comes into consideration, however. For this purpose, the described analysis can be carried out for an accordingly shorter period. The wavelet mother function can be adapted for this but can possibly also be used further without change. In any case, in so far as the wavelet mother function is tuned to the expected frequency range in which the low-frequency oscillations are expected, adaptation to a shorter capture period is not absolutely necessary.

(24) Provided is a unit (device) which can observe and assess energy system stability. In this case, it was recognized that oscillations in the energy system, in particular in the electrical supply grid, which are also referred to as power system oscillations (PSO) or energy system oscillations, may be a good indicator of the instability of energy systems. Furthermore, the observation and identification of power system oscillations may become an important part of future grid services which are aimed at attenuating such oscillations, for example.

(25) The present disclosure is therefore aimed at identifying power system oscillations. The observation of power system oscillations may be helpful not only for a warning system for operating wind farms, but rather this information can also be used to generate and supply suitable attenuation signals for attenuating the power system oscillations by means of a wind farm.

(26) Unlike in other known approaches for identifying low-frequency oscillations, a wavelet-based method is used here. In this respect, it was recognized that the use of wavelets makes it possible to detect non-stationary phenomena with locality properties both in the time domain and in the frequency domain.

(27) The proposed solution is fundamentally suitable for units connected to integrated grids, such as wind power installations or wind farms, but is not restricted thereto. The proposed solution can also be applied to consumer units (devices).

(28) The proposed method therefore relates to online detection of power system oscillations. It is suitable for production and for consumer installations.

(29) The method is based on an online analysis of transient measurement data, for example from a grid connection point.

(30) Provided herein is detecting power system oscillations on the basis of a transient measurement of the measured grid connection voltage and powers. A sampling rate of preferably at least 500 Hz for low-frequency oscillations in the frequency range of 0<f<five times the nominal frequency is proposed for all available measurement channels.

(31) The proposed wavelet-based method can be used to detect both periodic and aperiodic frequencies in the stated frequency range, which may depend on the duration of the time window to be analyzed because the proposed wavelets should have the locality property both in the frequency domain and in the time domain as far as possible.

(32) The selection of the wavelet mother function, which can also be referred to as a mother wavelet, and of the time window to be analyzed may play a role.

(33) Such a system or method for detecting power system oscillations can be implemented both at the level of wind power installations and at the level of wind farms, which is hereby proposed.

(34) The proposed method is also suitable for being automated. A proposed criterion is that the integral of the mother wavelets, preferably of all mother wavelets, should be zero. This is proposed, in particular, and is not generally obvious for a wavelet method since mother wavelets can also be different in other applications.