Computer-implemented method for monitoring generator of wind turbine, wind turbine, computer program and electronically readable storage medium

Abstract

A computer-implemented method for monitoring a generator (13) of a wind turbine (1) to detect a turn-to-turn short circuit fault in at least one stator winding set (15, 23) of the generator (13), in which the intensity of the second harmonic of the power generated from the stator winding set (15, 23) and the DC value of the power are determined, -evaluating the DC value of the power at a first analysis criterion describing that the DC value of the power is constant,-evaluating the intensity of the second harmonic at a second analysis criterion describing that a transient of the intensity of the second harmonic occurs, and-generating a turn-to-turn short-circuit fault signal if both the first analysis criterion and the second analysis criterion are true for a given number of time steps.

Description

Field of the Invention

[0001] The present invention relates to a computer-implemented method for monitoring a generator of a wind turbine to determine an inter-turn short circuit fault in at least one stator winding group of the generator, wherein the intensity of the second harmonic of the power generated from the stator winding group and the DC value of the power are determined. The present invention further relates to a wind turbine, a computer program, and an electronically readable storage medium. Background Art

[0002] As is known in the prior art, wind turbines typically use generators to convert mechanical energy (specifically, wind energy) from the blade assembly of the wind turbine into electrical energy (specifically, electric power) to drive electrical equipment and / or be fed into the power grid. A generator generally includes a rotor that rotates by the wind hitting the wind turbine blades of the wind turbine and a stator having stator windings, in which an electric current is induced due to the rotation of the rotor. In addition to the electric machine forming the generator, the generator device of a wind turbine generally further includes at least one channel, where each channel includes a circuit breaker, a converter unit, and a set of stator windings of the generator connected to the converter unit via the circuit breaker.

[0003] In particular, multi-stator electric machines have been proposed, the stator of which includes two or more sets of stator windings. Using multiple sets of stator windings allows, for example, a wind turbine to continue operating even if one set of stator windings or the corresponding generator fails.

[0004] Generally speaking, with regard to the generator device of a wind turbine, various fault conditions may occur. Such faults can lead to serious damage to the wind turbine, especially even to the destruction of components of the generator device, for example, by a fire that suddenly breaks out due to the fault. In order to detect such faults as early as possible to react with certain protective measures, it has been proposed, for example, to monitor the operation of the wind turbine, especially the operation of the generator device, by providing a monitoring device in the wind turbine. However, known monitoring methods may not be able to detect all kinds of fault conditions in time before damage occurs.

[0005] In WO 2019 / 037 908 A1, methods and devices for monitoring a first winding group and at least one second winding group of a stator of a generator during operation have been proposed. In particular, high-resistance connections are addressed here, which may be caused by loose connections (attributed to poor workmanship, thermal cycling, and corrosion or contamination). Such high-resistance connections may evolve into open-circuit faults. In addition, secondary damage is expected as a result of overheating and excessive vibration. To detect such faults, a fault diagnosis method is proposed, which is configured for online condition monitoring and protection of generators employed in wind turbines, wherein high-resistance connections in multi-stator permanent magnet generators are detected such that the converter controller can stop operation or apply other measures, such as shutting down the faulty winding group. Here, a first intensity of the second harmonic of the power generated from the first winding group and a second intensity of the second harmonic of the second power generated from the second winding group are obtained, and the first winding group and / or the second winding group are diagnosed based on the second harmonic power difference between the first intensity and the second intensity. Both the first power and the second power may include active power and / or reactive power.

[0006] The second harmonic is, for example, an oscillation of the output voltage and / or output current of the corresponding stator winding group, which oscillates at twice the fundamental electrical frequency of the generator. In an ideal situation without degradation or faults, it is expected that any winding group only delivers a power flow with the fundamental electrical frequency, such that the intensity of the second harmonic is substantially zero. In the case where the second harmonic of the corresponding power is not zero or not less than a threshold, it may already indicate that the corresponding winding group is at least partially faulty. It has been found that if there is a deviation between the first intensity and the second intensity, this may indicate that one of the stator winding groups (in particular the one with the higher second harmonic intensity) is at least partially faulty, in particular exhibiting a high-resistance connection.

[0007] WO 2020 / 128 058 A1 discloses a method for determining the location of a short-circuit fault in a generator device. Here, the short-circuit fault is detected by a large current that exceeds the current during normal operation of the generator device. To detect whether a short-circuit fault has occurred in the generator or the converter unit, a measurement of the torque pulsation of the machine is performed after the breaker is opened. However, if a large current that causes the opening of the breaker is detected, there may already be damage to the generator device or elsewhere in the wind turbine.

[0008] Such a large short - circuit current can be the result of a less obvious fault condition, for example, an inter - turn short - circuit fault. Here, the physical quantities by which an inter - turn short - circuit fault can be detected may be known. However, these effects are usually very small, such that a sufficiently low detection threshold may result in a large number of false alarms, which is disadvantageous. In particular, in known methods, it is not possible to reduce the threshold to the required level due to production tolerances (manufacturing differences between generators of the same type) and the dependence on operating conditions (such as stator and rotor temperature, generator speed, and current). As a result, inter - turn short - circuit faults are often detected too late and cause extensive damage. Experience shows that wind turbines often stop minutes or even hours after an inter - turn short - circuit fault occurs, triggered by a smoke or temperature alarm. Summary of the Invention

[0009] The object of the current invention is to provide a monitoring method that allows for early, in particular reliable and / or robust detection of inter - turn short - circuit faults.

[0010] This object is achieved by providing a computer - implemented method, a wind turbine, a computer program, and an electronically readable storage medium according to the independent claims. Advantageous embodiments are described by the dependent claims.

[0011] According to the invention, in the method as initially described, in each time step, - a first analysis criterion that describes a DC value of power evaluates the DC value of the power for constancy, - a second analysis criterion that describes the intensity of a transient in which a second harmonic occurs evaluates the intensity of the second harmonic, and - if both the first analysis criterion and the second analysis criterion are true for a given number of time steps, an inter - turn short - circuit fault signal is generated.

[0012] Here, the given number of time steps can be one. However, for increased robustness and reliability, the given number can be at least two. Usually, in each time step, measurement results for the power and thus its DC value and the intensity of the second harmonic are obtained, and the criteria are evaluated accordingly. By requiring both results to be true for more than one time step to detect an inter - turn short - circuit fault, a certain degree of "fault maturing" is achieved, improving robustness and reliability by, in particular, reducing the number of false alarms. Here, the analysis can refer to consecutive time steps, and the given number can correspond to a time interval or a time threshold. For example, the given number can be chosen to correspond to a time interval of 2 to 10 seconds, which has been shown to be advantageous for the filtering measures applied in the embodiments, as discussed further below.

[0013] Generally, it has been found that, especially by analyzing on-site data of a wind turbine, an inter-turn short circuit fault can be detected from the transient of the intensity (especially the amplitude) of the second harmonic of the power, especially regarding the difference between the relative intensity of the second harmonic and the expected value of the relative intensity. However, it has also been found that the amplitude variations in these intensity-related signals are usually very low, especially significantly lower than the fault detection threshold employed. It has also been observed that due to the low signal-to-noise ratio, fault detection is difficult under varying operating conditions (such as varying power). Therefore, in the present disclosure, it is proposed to detect the transient of the intensity of the second harmonic only in a time window in which the power described by the DC value of the power is approximately constant. Thus, two analysis criteria are employed, namely a first analysis criterion that generates a first result signal, wherein according to the first analysis criterion, if the DC value is at least substantially (i.e., approximately) constant, the first result signal indicates true. In other words, the first analysis criterion describes the required degree of constancy (or stability, since the requirement is to be in steady-state operation). On the other hand, the second analysis criterion generates a second result signal that indicates true if a transient of the intensity (especially the amplitude) of the second harmonic is detected, i.e., the second analysis criterion is satisfied. Only when both result signals indicate true, i.e., the corresponding analysis criteria are satisfied, can the presence of a fault be detected.

[0014] In other words, the present invention is based on the recognition that, except for an inter-turn short circuit fault, generally (i.e., if there is no inter-turn short circuit fault), the second analysis criterion is satisfied only in the presence of a relevant change in the power (especially its DC value). Therefore, as an additional condition for detecting an inter-turn short circuit fault, stability (constancy) of the DC value of the power is required, such that the number of false alarms can be significantly reduced or even completely prevented.

[0015] In summary, the present invention provides a solution for on-line inter-turn short circuit fault detection and protection of a generator employed in a wind turbine. In such a permanent magnet generator, an inter-turn short circuit fault can be reliably and robustly detected while the number of false alarms can be kept at a sufficiently low level.

[0016] Here, it should be noted that the intensity of the second harmonic can be determined as disclosed, for example, in the already mentioned WO 2019 / 037 908 A1. In particular, the second harmonic is, for example, an oscillation of the output voltage and / or output current of a stator winding group, which oscillates at twice the fundamental electrical frequency of the generator. The fundamental electrical frequency is proportional to the rotational speed of the rotor relative to the stator. In particular, the fundamental electrical frequency can be calculated as the product of the rotational speed of the rotor and the number of pole pairs of the permanent magnets. Under normal operating conditions, the second harmonic (of voltage, current, and / or power) should be substantially zero.

[0017] In principle, embodiments of the present invention allow for considering only the active power or only the reactive power or a combination of reactive and active power for diagnosis. However, in a preferred embodiment, the power is the active power generated by at least one stator winding group. In particular, if the deviation from the desired intensity of the second harmonic as further discussed below is analyzed, it has been shown that by selecting the active power as the power (in particular using the amplitude of the DC component as the DC value and the amplitude of the second harmonic as the intensity of the second harmonic), the value of this deviation is significantly reduced over the entire operating range of the generator.

[0018] As a general note, in a generator having a stator with multiple stator winding groups, the methods described here can of course be applied to each stator winding group so that inter-turn short circuit faults can be specifically detected for each stator winding group.

[0019] In a specific embodiment, the DC value of the power can be determined by low-pass filtering the measurement signal of the power, and / or the intensity of the second harmonic can be determined by band-pass filtering the measurement signal using a band-pass filter centered at the second harmonic frequency. As already mentioned, the second harmonic frequency is twice the fundamental electrical frequency, which can be determined by the rotational speed of the rotor and the number of pole pairs of the permanent magnets of the rotor. The measured signal can be obtained, for example, by measuring the current in the stator winding group using at least one current sensor. To determine the second harmonic, an adaptive band-pass filter is used, to which the second harmonic frequency or directly the fundamental electrical frequency is supplied. Then, the intensity of the second harmonic can be determined as the amplitude of the band-pass filtered measured signal. Note that time averaging can also be applied as described in WO 2019 / 037 908 A1.

[0020] In a preferred embodiment, the false result of the first evaluation criterion can be maintained for a predetermined number of time steps. In particular, if another false result is determined during these time steps, the holding time interval can be reset. In other words, if the DC value is determined to be unstable by the first analysis criterion, the result remains false for a predetermined number of time steps to increase robustness, especially with respect to short-term constancy at high values, etc.

[0021] In a specific preferred embodiment, the first analysis criterion may compare the absolute value of the high-pass filtered DC value with a first threshold value, where as long as the absolute value of the high-pass filtered DC value is below the first threshold value, the power is assumed to be constant. By applying a high-pass filter, if the DC value is approximately constant, the high-pass filtered DC value is assumed to be approximately equal to zero. Then the high-pass filtered DC value is compared with the first threshold value, which indicates that the DC value and thus the power are stable, i.e., approximately constant. The first threshold value can be understood as defining a boundary for the constancy of the DC value and can be generally applied, that is, applied under all operating conditions of the generator, because the deviation from zero is checked due to high-pass filtering.

[0022] Regarding transient detection, in an advantageous embodiment, the second analysis criterion may compare a particularly high-pass filtered deviation value derived from the intensity of the second harmonic with a second threshold value, where the deviation value is determined as follows: - By dividing the absolute value of the intensity (especially the amplitude) of the second harmonic by the absolute value of the DC value of the power to determine the relative intensity of the second harmonic, - Determining the expected value of the relative intensity, and - Determining the deviation value as the absolute value of the difference between the relative intensity and the expected value, where if the deviation value exceeds the second threshold value, the second analysis criterion indicates a transient.

[0023] Therefore, analyzing the deviation from the expected value is proposed, which has the advantage that under normal operating conditions (healthy operation), the deviation value can again be at least approximately zero in some cases. However, it is preferred to apply a high-pass filter to the deviation value signal because, for example, due to tolerances, etc., the expected value may deviate from the actually measured relative intensity. The expected value (reference value) can be constant or preferably depends on the load, where the load can be described by the DC value of the power. Such an expected value and / or the correlation between the load and the expected value can preferably be determined empirically and / or by analysis, for example, by statistically evaluating the DC value and intensity of the second harmonic measurement during normal, healthy operation, especially in a time window that satisfies the first analysis criterion. The correlation can be provided, for example, as a look-up table that associates the current load with the expected value.

[0024] In a specific embodiment, as described in WO 2019 / 037 908 A1 for the determination of the diagnostic signal for each channel / stator winding group, the relative intensity of the second harmonic can be determined, for example, as: , (1) where X represents the active power (preferably) or reactive power of the stator winding group, especially X DC represents the DC value (amplitude of the DC component), and X2f represents the intensity of the second harmonic, especially the amplitude of the second harmonic. d is the relative intensity. If now the expected value d* is provided, the deviation value that describes the deviation between the expected value and the currently measured value can be obtained by the following formula: (2).

[0025] As already mentioned, the evaluation of experimental and statistical data from the field shows that when an inter-turn short circuit fault occurs, the deviation value changes suddenly. Therefore, the deviation value or the intensity of the second harmonic in general is significantly different from other signals that are usually measured, such as the sound pressure level in the rotor hub or the generator temperature, which usually remain unchanged when an inter-turn short circuit fault occurs.

[0026] Since the transient of the deviation value (the transient for indicating the intensity of the second harmonic) is to be determined, the deviation value can also be high-pass filtered before the comparison with the second threshold. Here, it should be noted that since the additional conditions of the first analysis criterion are checked, the second threshold can be selected to be very low, thus resulting in high sensitivity. However, since the additional requirements of the first analysis criterion are met, false alarms are avoided.

[0027] It should be noted that in principle, it may be conceivable to compare the intensities of the second harmonics between a pair of stator winding groups, which requires that the two DC values are at least substantially constant. However, such a deviation value will be affected by two kinds of variations and will require additional criteria for power constancy, resulting in such an embodiment being less preferred.

[0028] Regarding both the first threshold and the second threshold, they can be determined empirically, especially by evaluating data from healthy operating wind turbines and wind turbines in which an inter-turn short circuit fault has been detected, and selecting two thresholds such that the number of false detections is minimized while the maximum value of correctly detected inter-turn short circuit faults is achieved. Such analyses and experiments have been carried out, and the applicability and advantages of the method of the present invention have been shown. In other words, the verification has been successful.

[0029] In this context, in an embodiment, the deviation value and / or the DC value can be pre-filtered before high-pass filtering. For example, a suitable low-pass filter can be applied. Here, in an embodiment, pre-filtering of the DC value may not be necessary, especially if it has already been low-pass filtered. However, regarding the deviation value, pre-filtering may be beneficial for shaping the resulting deviation value signal to allow for optimal detection performance. In particular, by applying a low-pass filter and a high-pass filter, a band-pass filter can be created and parameterized for optimal signal shaping.

[0030] If a turn - to - turn short - circuit fault is detected, corresponding measures can be taken to protect the generator and thus the wind turbine / the entire generator installation. For example, if a turn - to - turn short - circuit fault signal indicates that a turn - to - turn short - circuit fault has been detected, the generator (especially at least the corresponding stator winding group) is shut down and / or a warning is output. For example, the affected stator winding group and the corresponding channel can be deactivated by opening the circuit breaker in the corresponding channel. A warning can be output at the wind turbine and / or preferably at a remote monitoring center.

[0031] It should be noted that in a preferred embodiment, the short - circuit fault detection implemented by the method according to the present invention can be combined with the monitoring / fault - detection process described in WO2019 / 037908 A1, because the intensity of the second harmonic is also determined and used, especially the relative intensity is also mentioned in the embodiment. For example, the intensity of the second harmonic can be further evaluated with respect to at least one other fault condition, especially by diagnosing two different stator winding groups based on the second - harmonic power difference between the corresponding intensities of the second harmonic. In this way, in particular, high - resistance connections can be additionally detected.

[0032] The present invention further relates to a wind turbine, which includes: a generator having at least one stator, the stator including at least one stator winding group; measuring means for measuring the power measurement signal of the at least one stator winding group; and a monitoring device configured to perform the steps of the method according to the present invention. All features and remarks regarding the method according to the present invention can be correspondingly applied to the wind turbine according to the present invention, so that the same advantages can be achieved. In one example, the measuring means may include a current sensor for the winding.

[0033] In a preferred embodiment, the monitoring device can be provided as part of a control device of the wind turbine (e.g., a controller of a frequency converter, a controller of the wind turbine), or be provided in a dedicated wind - turbine condition monitoring and protection system. The monitoring device can include at least one processor and / or at least one storage device. Functional units can be implemented in hardware and / or software to perform the steps of the method according to the present invention. For example, the monitoring device can include a pre - processing unit for determining the intensity and DC value from the measured signal, an optional determination unit for determining the deviation value, an analysis unit for evaluating the analysis criterion and detecting a turn - to - turn short - circuit fault from the result, and / or an optional control unit for performing measures (e.g., opening the circuit breaker and / or outputting a warning).

[0034] A computer program according to the present invention comprises program means configured to perform the steps of the method according to the present invention if the computer program is executed on a monitoring device of a wind turbine. The computer program may be stored on an electronically readable storage medium according to the present invention, which electronically readable storage medium thus comprises control information stored thereon, the control information at least comprising the computer program according to the present invention, such that when the electronically readable storage medium is used in a monitoring device of a wind turbine, the monitoring device is configured to perform the steps of the method according to the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Other objects and features of the present invention will become apparent from the following detailed description considered in conjunction with the accompanying drawings. However, the drawings are only schematic sketches for illustrative purposes and do not limit the present invention. In the drawings: Figure 1 A wind turbine according to an embodiment of the present invention is schematically illustrated; Figure 2 A graph explaining the monitoring process of the present invention is shown; Figure 3 A first functional diagram explaining an embodiment of the method according to the present invention is shown; Figure 4 A second functional diagram explaining an embodiment of the method according to the present invention is shown, and Figure 5 shows Figure 1 a full view of the wind turbine. DETAILED DESCRIPTION

[0036] Figure 1Schematically illustrates relevant components of a wind turbine 1 according to an embodiment of the present invention. The wind turbine 1 includes a rotor hub 3 to which a plurality (e.g., three) wind turbine blades 5 are connected. The rotor hub 3 is connected to a rotating shaft 7, and the rotating shaft 7 is coupled to an optional gearbox 9. A second shaft 11 coupled to the other end of the gearbox 9 is coupled to a generator 13. In this case, the stator of the generator 13 includes two stator winding groups 15, 23. At the first stator winding group 15 (particularly a three-winding group), the generator 13 outputs a first power as three phase currents 17, 19, 21. The second winding group 23 outputs a second power as three phase currents 25, 27, 29. The first power generated by the first stator winding group 15 is supplied to a first AC-DC-AC converter 31 of a first channel, while the second power is supplied to a second AC-DC-AC converter 33 of a second channel. Both converters 31, 33 include an AC-DC converter section 35, a DC link 37, and a DC-AC converter section 39. The converters 31, 33 are respectively configured to convert a variable-frequency AC power flow (currents 17, 19, 21 or 25, 27, 29) into a fixed-frequency power flow, particularly a three-phase power flow 41, 43, 45 or 47, 49, 51. Two power flows from the first stator winding group 15 and the second stator winding group 23 are transformed into a higher voltage in a common transformer 53 so as to be provided to a power grid 55 as a three-phase power flow.

[0037] The wind turbine 1 further includes a monitoring device 2 for monitoring the first winding group 15 and the second winding group 23 of the generator 13 during operation with respect to turn-to-turn short circuit faults. The monitoring device 2 is configured to perform a method of monitoring the generator 13 according to the present invention. Here, the monitoring device 2 includes a processor 4 and a storage device, and can be implemented as an independent device or integrated into another controller, such as a controller for the converters 31, 33 or generally for the wind turbine 1.

[0038] The monitoring device may include a preprocessing unit configured to determine a first intensity of a second harmonic of a first power generated from a first stator winding group 15, wherein the first power is supplied to the monitoring device 2 using a first measurement signal 6, which may be obtained by measuring a current in the first winding group 15 using a current sensor 57. A second intensity of a second harmonic of a second power generated from a second winding group 23 is correspondingly determined from a second measurement signal 8 indicative of the second power output by the second winding group 23. It should be noted that in the wind turbine 1, with respect to the permanent magnet synchronous generator 13, the measured phase current and reference voltage are readily available and are used for monitoring purposes in embodiments of the present invention. Herein, the currents 17, 19, 21 output by the first stator winding group 15 measured using the current sensor 57 and a first reference voltage 59 supplied to the first converter 31 by the controller 61 are used to derive a first power supplied to the monitoring device 2 as the first measurement signal 6. In a similar manner, currents 25, 27, 29 and a reference voltage 65 (supplied by the controller 63) of a second power flow generated by the second winding group 23 are obtained, and the corresponding second power is supplied to the monitoring device 2 as the second measurement signal 8.

[0039] In an embodiment, the first intensity and the second intensity may be used to diagnose a high resistance connection with respect to the first winding group 15 and / or the second winding group 23 based on a second harmonic power difference between the first intensity and the second intensity, as described in WO 2019 / 037908 A1. However, in the present invention, it has been found that the intensity of the second harmonic of each stator winding group 15, 23 also indicates an inter-turn short circuit fault caused by a sudden transient.

[0040] Figure 2 In the upper figure, a graph 10 of the power (measurement signals 6, 8) output by the stator winding groups 15, 23 measured in the generator 13 is shown, while in the lower figure, a graph 12 of the expected value of the relative intensity of the second harmonic and a graph 14 of the measured relative intensity of the second harmonic (according to equation (1)) are shown. Between time points 16 and 18 and between time points 20 and 22, clearly distinguishable transients appear in graph 14, while graph 12 remains at the same level. These transients are not caused by a change in the DC value, which remains substantially constant at least, see graph 10. Thus, if a transient occurs in the intensity of the second harmonic of the power during at least approximately constant (stable) power, the monitoring device 2 is also configured to detect an inter-turn short circuit fault.

[0041] Hereinafter, the active power is analyzed. However, the method may also be applied to reactive power and / or combinations. In addition, for the two stator winding groups 15, 23 (i.e., for each channel), monitoring for an inter-turn short circuit fault is performed independently, and hereinafter, it is described for one of them.

[0042] Figure 3 Illustrated is how the preprocessing unit of the monitoring device 2 determines the DC value of the power and the intensity (in this case the amplitude) of the second harmonic of the power from the measurement signals 6, 8 of the first channel and the second channel respectively. To determine the DC value of the power (in this case the active power), the low-pass filter 67 is applied to the measurement signals 6, 8. To determine the intensity, the band-pass filter 69 is applied to the measurement signals 6, 8, which are centered at the second harmonic frequency (i.e., twice the fundamental electrical frequency of the generator 13). As discussed above, the fundamental electrical frequency can be calculated as the product of the rotational speed of the rotor of the generator 13 and the number of pole pairs of the permanent magnets of the generator 13. Here, the rotor speed 71 is provided in revolutions per minute. Based on the filtering results, in step 73, the amplitude of the second harmonic can be calculated as the intensity of the second harmonic. To determine the relative intensity of the second harmonic, as discussed in equation (1), both the DC value and the amplitude of the second harmonic are supplied to step 75.

[0043] In the current embodiment, the deviation value defined in the above equation (2) is used to detect the transient of the intensity of the second harmonic. Thus, in step 77, the expected value of the relative intensity d, i.e., d* in equation (2), is determined. The expected value can be constant or, as indicated by the arrow 79, can be load-dependent, in which case it depends on the DC value determined by applying the filter 67. In step 81, the absolute value of the difference between the relative intensity determined in step 75 and the expected value determined in step 77 is calculated and output as the deviation value according to the arrow 83. The filters 67 and 69 and step 73 are applied in the preprocessing unit, while steps 75, 77, and 81 can be performed in the determination unit of the monitoring device 2.

[0044] Then the determined DC value and the determined deviation value are provided to the analysis unit of the monitoring device 2, where the processing steps of the analysis unit are illustrated in Figure 4 As illustrated by the arrows 85 and 87, the DC value of the power and the deviation value are respectively supplied to the analysis unit. The block 89 indicates an optional pre-filtering step, in particular by applying the low-pass filters 91 and / or 93. Although the low-pass filter 91 can be omitted if the low-pass filter 67 has achieved the desired signal shape, the low-pass filter 93 can improve the turn-to-turn short circuit fault detection by providing additional signal shaping for the deviation value.

[0045] The block 95 indicates high-pass filtering, as indicated by the application of low-pass filters 97, 99 and subsequent subtraction. For the DC value, high-pass filtering allows reducing the signal to the actual variations, thus allowing analyzing the steady power by considering the deviation from zero, which should be the high-pass filtered DC value if the power is constant. Thus, since the expected value d* and the relative intensity d do not necessarily have to match (as also shown in Figure 2 ), by high-pass filtering, as long as no transient to be detected occurs, the deviation value signal can also be reduced to at least substantially zero. In particular, the combination of the low-pass filter 93 and high-pass filtering provides a band-pass filter, which can be parameterized empirically to improve the detection performance.

[0046] In steps 101 and 103, a first analysis criterion for the high-pass filtered DC value and a second analysis criterion for the filtered deviation value are evaluated respectively. The first analysis criterion in step 101 is used for the detection of steady power, i.e., whether the power is at least approximately constant. Here, the high-pass filtered DC value is compared with a first threshold. As long as the high-pass filtered DC value is below the first threshold, the power is considered to be at least substantially constant. Thus, the first analysis criterion is satisfied and "true" is output.

[0047] In the optional step 105, for increasing robustness, a "false" output (i.e., not satisfying the first analysis criterion) can be held for a predetermined number of time steps, e.g., time steps corresponding to a time interval of several seconds. If an additional "false" result of the first analysis criterion occurs during this holding time interval, the time interval is extended accordingly.

[0048] In the second analysis criterion, the high-pass filtered deviation value is compared with a second threshold. If the high-pass filtered deviation value exceeds the second threshold, the second analysis criterion is satisfied and "true" is output, thereby meaning that a transient has been detected.

[0049] The first threshold and the second threshold can be determined correspondingly, e.g., empirically and / or analytically from the evaluation field data from a wind turbine.

[0050] In step 107, a fault detection logic is applied. A turn-to-turn short circuit fault is detected only when both analysis criteria result in true, that is, only when the power is stable and a transient has been detected in the intensity of the second harmonic. In this case, step 107 outputs "true". In step 109, fault maturing is performed. Here, a turn-to-turn short circuit fault signal is generated only when the first analysis criterion and the second analysis criterion are true for a given number of consecutive time steps. For example, a given number of consecutive time steps can be selected such that a transient must be detected within a detection time interval of 2 to 10 seconds.

[0051] Note that the monitoring device 2 may of course also include a control unit for performing measures to protect the generator 13 and the wind turbine 1 typically when an inter-turn short circuit fault has been detected. For example, if the inter-turn short circuit fault signal indicates that an inter-turn short circuit fault has been detected, in particular by indicating "true", the generator 13, at least the channels with the corresponding stator winding groups 15, 23 can be shut down. Preferably additionally, a warning can be output.

[0052] Figure 5 A full view of the wind turbine 1 is shown. The wind turbine 1 includes a tower 111, and a nacelle 113 is arranged on the tower 111. A rotor hub 3 with its blades 5 is mounted to the nacelle 113. The hub 3 can rotate about the axial direction 115, where the corresponding rotation is driven by the wind interacting with the blades 5. As already explained with respect to Figure 1 as explained, the rotation of the rotor hub 3 is transmitted to the generator 13 via the rotating shaft 7 (and optionally a gearbox 9 and a second shaft 11, which are not shown in Figure 5 for simplicity). The axial direction 115 is arranged horizontally, but can also be inclined with respect to the horizontal direction. Although the total height of the wind turbine 1 is on the order of dozens or hundreds of meters, the output power of the wind turbine 1 is in the range of several megawatts, in particular between 1 megawatt and 40 megawatts.

[0053] The generator 13 includes a housing 117, a stator 119, and a rotor 121, where the stator 119 and the rotor 121 are arranged within the housing 117. The stator 119 does not rotate. As described above, the rotor 121 is coupled to the rotor hub 3, and the rotor 121 can also rotate about the axial direction 115.

[0054] Although the present invention has been described in detail with reference to preferred embodiments, the present invention is not limited to the disclosed examples, and those skilled in the art can derive other variations from these disclosed examples without departing from the scope of the present invention.

Claims

1. A computer-implemented method for monitoring a generator (13) of a wind turbine (1) to detect an inter-turn short circuit fault in at least one stator winding group (15, 23) of the generator (13), wherein the intensity of the second harmonic of the power generated from the stator winding group (15, 23) and the DC value of the power are determined, It is characterized in that, At respective time steps, - A first analysis criterion describing that the DC value of the power is constant evaluates the DC value of the power, - A second analysis criterion describing a transient in the intensity of the second harmonic evaluates the intensity of the second harmonic, and - If both the first analysis criterion and the second analysis criterion are true for a given number of time steps, an inter-turn short circuit fault signal is generated.

2. The method according to claim 1, wherein The power is the active power generated from the at least one stator winding group (15, 23).

3. The method according to claim 1 or 2, characterized in that, The DC value of the power is determined by low-pass filtering the measurement signal (6, 8) of the power, and / or the intensity of the second harmonic is determined by band-pass filtering the measurement signal through a band-pass filter (69) centered at the second harmonic frequency.

4. The method according to one of the preceding claims, characterized in that, The false result of the first evaluation criterion is maintained for a predetermined number of time steps.

5. The method according to one of the preceding claims, characterized in that, The first analysis criterion compares the absolute value of the high-pass filtered DC value with a first threshold, wherein as long as the absolute value of the high-pass filtered DC value is below the first threshold, the power is considered to be constant.

6. The method according to one of the preceding claims, characterized in that, The second analysis criterion compares a deviation value, especially a high-pass filtered deviation value, derived from the intensity of the second harmonic with a second threshold, wherein the deviation value is determined by: - Determining the relative intensity of the second harmonic by dividing the absolute value of the intensity, especially the amplitude, of the second harmonic by the absolute value of the DC value of the power, - Determining an expected value of the relative intensity, and - Determining the deviation value as the absolute value of the difference between the relative intensity and the expected value, wherein if the deviation value exceeds the second threshold, the second analysis criterion indicates a transient.

7. The method according to claim 5 or 6, characterized in that, Before performing high-pass filtering, the deviation value and / or the DC value are pre-filtered.

8. The method according to one of the preceding claims, characterized in that, The given number of the time steps is at least 2.

9. The method according to one of the preceding claims, characterized in that, If the inter-turn short circuit fault signal indicates that an inter-turn short circuit fault is detected, the generator (13), especially at least the corresponding stator winding group (15, 23), is shut down and / or a warning is output.

10. The method according to one of the preceding claims, characterized in that, The intensity of the second harmonic is further evaluated with respect to at least one additional fault condition, especially by diagnosing two different stator winding groups (15, 23) based on the second harmonic power difference between the respective intensities of the second harmonic.

11. Wind turbine (1), comprising: A generator (13) having at least one stator (119), the at least one stator (119) including at least one stator winding group (15, 23); measuring means for measuring a measurement signal of the power of the at least one stator winding group (15, 23); And a monitoring device (2) configured to perform the steps of a method according to one of the preceding claims.

12. A computer program which, when executed on a monitoring device (2) of a wind turbine (1), performs the steps of a method according to one of claims 1 to 10.

13. An electronically readable storage medium, on which is stored a computer program according to claim 12.

Images