SYSTEM AND METHOD FOR DETECTING FAULTS BETWEEN TURNS IN WINDINGS OF ALTERNATING CURRENT ELECTRICAL MACHINES

ES3078634A1Undetermined Publication Date: 2026-09-15UNIVERSIDAD DEL PAÍS VASCO EUSKAL HERRIKO UNIBERTSITATEA (75 00) +1
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
ES2025030121
Authority / Receiving Office
ES · ES
Patent Type
Applications
Current Assignee / Owner
Filing Date
2025-02-17
Publication Date
2026-09-15

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

A system and method for detecting turns in the windings of AC electrical machines utilizes at least the current measurements of the machine being diagnosed and uses them to obtain instantaneous components that serve as diagnostic characteristics. These instantaneous components are subjected to frequency domain analysis, allowing the generation of harmonics characteristic of the turns-to-turn fault, which then triggers an alarm to external devices. This invention is capable of detecting turns-to-turn faults in the pre-short-circuit stage, relying solely on the fault current flowing between the turns, thus enabling early diagnosis of the operating electrical machine.
Need to check novelty before this filing date? Find Prior Art

Description

SYSTEM AND METHOD FOR DETECTING FAULTS BETWEEN TURNS IN AC ELECTRICAL MACHINES WINDINGS OBJECT OF THE INVENTION The present invention relates to a system and method for detecting faults between turns for windings of alternating current electrical machines, generally but not limited to rotating electrical machines, which may also be linear electrical machines or transformers, which detects insulation defects between nearby conductors with the machine in operation. The method and system of the present invention are of particular relevance in electrical systems that use electrical machines in their drives, such as, but not limited to, wind turbines with permanent magnet electrical machines, where early diagnosis of faults between turns of the electrical machine can be critical to avoid major damage to the machines. BACKGROUND OF THE INVENTION With the measures being implemented to combat climate change, electric machines are gaining prominence over conventional thermal machinery. This implies an increase in the number of these machines to decarbonize sectors such as electricity generation and transportation. However, with the inclusion of power converters to control these machines, and due to their modulation capabilities, the insulation of these devices is being compromised more rapidly. When the number of units in operation is considered, the risk of power outages is increasingly high. Regarding electrical faults, there are two fundamental types that can occur within an electrical machine and are studied because they are the most common. First, there are ground faults, where a conductor with defective insulation makes contact with the iron and chassis of the electrical machine. Second, there are turn-to-turn faults, where the insulation between at least two conductors, whether adjacent or not, is compromised. This second phenomenon involves the creation of a current that flows between the short-circuited turns, chaining together only within the machine and causing very slight effects externally, making its detection very difficult.Although the voltage between the fault points between both turns is very low, the circuit impedance can also be very low, so a direct fault between turns can result in very high currents that can burn out the machine's windings and render it unusable. Currently, there are many techniques for detecting faults between turns, the most representative of which are presented below: Turn-to-turn faults can be detected using techniques that require electrical machines to be disconnected from their loads or completely switched off. This is the case with Russian patent RU2642521C2, which, by using two voltage transformers and galvanometers connected to the secondary winding of a transformer, is able to detect turn-to-turn faults in the transformer winding. This requires disconnecting the load from the secondary side and supplying power from the mains to the primary side. These techniques can detect turn-to-turn faults even in early stages; however, they cannot detect them during operation. Another technique is frequency response analysis (FRA), which injects a voltage signal and measures the impedance of the dipole between the input and output. By comparing the sound footprint with the one obtained during maintenance, faults between turns can be detected if the difference between the two measurements is considerable, generally in the range of 10 kHz to 1 MHz. However, if such faults occur during operation, the machine will suffer irreparable damage before a diagnosis can be performed. It should be noted that phase-by-phase analysis of winding resistance using direct and alternating current with the machine disconnected from the grid is another known technique for resolving this type of diagnostic. However, it suffers from the same problem as the two previously mentioned methods. Other techniques, such as the one described in Chinese patent publication CN110988751A, propose using resonant circuits connected to the winding under study and linking them via a voltage and current transient. By analyzing this transient, the total inductance of the circuit can be measured. If this deviates from the inductance obtained under sound conditions, where C = L and therefore the imaginary part of the equivalent impedance is zero, then there is a fault between turns that makes L less than C. - In Chinese patent publication CN117538746A, a device detects turn-to-turn faults in electrical machines connected to the grid via a transformer, with both neutrals (machine and transformer secondary) grounded. This is achieved by measuring the zero-sequence component of the machine's voltage, the three phase voltages, the circulating ground current between the transformer neutral and the connected machine, and a predefined model. By detecting changes in the zero-sequence voltage and phase voltages, the invention is able to detect turn-to-turn faults. However, a predefined model or simultaneous simulation of the machine is necessary to perform the diagnosis. - In the US patent publication US2022393625A1, the three phase currents of a synchronous generator and the rotor position are collected with an "encoder," which allows for greater resolution when obtaining the harmonic frequency components since the fundamental frequency of the machine is available by measuring the rotation of its shaft. Therefore, new techniques are needed to overcome several limitations identified in the current state of the art for detecting turns-to-turn faults in AC electrical machines. To achieve this, diagnostic and protective systems and methods must be available that are easy to implement, use few sensors, are fast and efficient to prevent severe machine damage, and are capable of detecting incipient faults (i.e., turns with partially deteriorated insulation between them, or in other words, minor faults) in electrical machines before they develop into serious faults that cause severe damage. All of this must be done while the machines being protected are in operation. DESCRIPTION OF THE INVENTION The present invention solves the problems presented by the prior art discussed above, by means of the system and method defined in the accompanying independent claims. To this end, a first aspect of the invention relates to a system for detecting faults between turns in alternating current electrical machines that allows the detection of at least one fault between turns while the machine is operating, and which comprises at least the following subsystems: - a current recording subsystem that collects at least the currents from all input or output terminals of the winding of the electrical machine to be protected; - a subsystem for estimating instantaneous sequence currents from the previously measured currents. These instantaneous sequence components are preferably arranged as follows (for a three-phase example): where: ia, ib and ic are the phase currents of the electrical machine to be protected; i0r, i1r and i2r are the real instantaneous sequence components; i0i, i1i and i2i are the instantaneous sequence components transverse to the previous ones; i0, i1 and i2 are the absolute instantaneous sequence components; - a turn-to-turn fault detection subsystem in the time domain using a current that records the ripple measurement of component i1 and / or i2 and detects the turn-to-turn fault if said component exceeds a threshold value T1 previously established by the installation operator; - a turn-interference fault detection subsystem in the frequency domain using a current that transforms the time-domain measurement of i1 and / or i2 into a frequency domain, and that detects the fault if the frequency components exceed a certain threshold T2 previously established by the installation operator; and - an alarm emission subsystem that sends an alarm signal with information of at least the detection of the fault between turns to devices external to the invention. In a preferred embodiment of the invention, the time-domain turn-fault detection subsystem calculates the current ripple as: where i is the ripple of the current "i" which refers to a generic instantaneous current which can be i1 or i2; imax and imin refer to the maximum and minimum values ​​of the previous current for a cycle of ia, ib or ic; and im refers to the average value of the current "i". In another preferred embodiment of the invention, the turn-fault detection subsystem in the frequency domain compares the second harmonic component of the current i1 or i2, obtained from a conversion process, such as a fast Fourier transform, in successive non-overlapping windows and uses the difference converted to decibels such that: where: Ai is the difference in decibels between the second harmonic (k=2) normalized with respect to its first harmonic of the current is, which is the current of the successive window, and the second harmonic (k=2) normalized with respect to the first harmonic of the current ip, which is the current of the previous window, for i1 or i2. In another preferred embodiment of the invention, the recording, instantaneous sequence component estimation, and time-domain and frequency-domain fault detection subsystems are for voltages instead of currents. In this case, the voltages to be considered can be the phase voltages or line voltages at the terminals of the electrical machine to be protected. In another preferred embodiment of the invention, both the voltage-related subsystems and the current-related subsystems can be used in parallel to detect the fault between turns. In another preferred embodiment of the invention, the system for detecting turns between turns in AC electrical machines uses the current and voltage recording subsystems to estimate the machine load. This allows for adaptive monitoring of current and / or voltage ripple and relative comparison in decibels of the second harmonic of current and / or voltage, preventing unwanted tripping due to load fluctuations. The system will then detect a fault when a fluctuation occurs in (or for voltages) or when there is a difference greater than T2 in (or for voltages) when there is no fluctuation in the machine load. Another aspect of the invention relates to a method for detecting faults between turns in alternating current electrical machines that allows the fault between turns to be detected while the machine is in operation, where the method comprises at least the following steps: - a current recording stage that collects at least the currents from all input or output terminals of the winding of the electrical machine to be protected; - an instantaneous sequence current estimation stage using the previous currents. Preferably, these instantaneous sequence components are arranged as follows (for a three-phase example): where: ia, ib and ic are the phase currents of the electrical machine to be protected; i0r, i1r and i2r are the real instantaneous sequence components; i0i, i1i and i2i are the instantaneous sequence components transverse to the previous ones; i0, i1 and i2 are the absolute instantaneous sequence components; - a time-domain fault detection stage between turns using current that records the ripple measurement of component i1 and / or i2 and detects the fault between turns if said component exceeds a threshold value T1 previously established by the installation operator; - a frequency-domain fault detection stage between turns using current that transforms the time-domain measurement of i1 and / or i2 into a frequency-domain form, and detects the fault if the frequency components exceed a certain threshold T2 previously established by the installation operator; and - an alarm emission stage that sends an alarm signal with information of at least the detection of the fault between turns. In the method of detecting faults between turns in alternating current electrical machines, the stages can be applied to the measurement of phase or line voltage instead of currents, or both voltage and current measurements can be used in a complementary way to perform the detection of the fault between turns of the electrical machine to be protected. By using voltage and current measurements together, the machine's load can be estimated to prevent unwanted tripping. Trips exceeding thresholds T1 and T2 are therefore valid if no load variation has been detected during the process. Therefore, the invention provides a diagnostic and protection system and method that is easy to implement, requires few sensors, and is fast and efficient in preventing severe machine damage. Furthermore, the invention is capable of detecting incipient faults—that is, faults in turns with partially deteriorated insulation between them, or in other words, minor faults—in electrical machines before they develop into serious faults that cause severe damage. All of this occurs while the machines to be protected are in operation. BRIEF DESCRIPTION OF THE FIGURES A series of figures, non-limiting examples, are briefly described to help to better understand the invention: Figure 1 shows an electrical schematic of a non-limiting example of a possible embodiment of the system of the invention for an electrical machine directly connected to the grid. Figure 2 shows an electrical schematic of a non-limiting example of a possible embodiment of the system of the invention for an electric machine used in a variable speed drive with converters. Figure 3 shows a block diagram of a possible configuration of the subsystems of a possible embodiment of the invention. Figure 4 shows a block diagram of a possible configuration of the stage flowchart of a possible embodiment of the method of the invention. Figures 5.ay and 5.b show graphs of example estimation of instantaneous sequence currents in the time and frequency domains, respectively, for turn-to-turn fault detection for an electrical machine. NUMERICAL REFERENCES FOR THE FIGURES (1) Alternating current network side; (2) Transformer; (3) Secondary side of the alternating current transformer; (4) Alternating current electric machine; (5) Load / turbine; (6) Current meter; (7) Voltage meter; (8) Voltage signal; (9) Current signal; (10) System for detecting faults between turns; (11) Alarm signal; (12) Network-side electronic converter; (13) Positive pole of direct current; (14) Negative pole of direct current; (15) Capacitor; (16) Electric machine side converter; (17) Alternating current side of the electric machine; (18) Current recording subsystem; (19) Voltage recording subsystem; (20) Load estimation subsystem; (21) Subsystem for estimating instantaneous sequence currents; (22) Subsystem for estimating instantaneous sequence voltages; (23) Turn-to-turn fault detection subsystem in the time domain using current; (24) Turn-to-turn fault detection subsystem in the time domain using voltage; (25) Turn-to-turn fault detection subsystem in the frequency domain using current; (26) Turn-to-turn fault detection subsystem in the frequency domain using voltage; (27) Alarm emission subsystem; (28) Voltage and / or current recording stage; (29) Load estimation stage; (30) Stage of estimation of instantaneous sequence components; (31) Turn-to-turn fault detection stage in the time domain; (32) Turn fault detection stage in the frequency domain; (33) Alarm emission stage. DETAILED DESCRIPTION OF PREFERRED FIGURES AND MANUFACTURES Next, the figures introduced previously are described in more detail, which are preferred embodiments that do not limit the invention. Figure 1 shows a rotating AC electrical machine (4) connected directly to the electrical grid (1) via a transformer (2). This transformer separates the AC grid side from the secondary side of the AC transformer (3), which typically operates at a lower or equal voltage. The machine (4) is connected to a load / turbine (5), allowing it to function as a motor or generator, respectively. Current and line and / or phase voltage are measured at the stator terminals of this machine, which could be, but is not limited to, a squirrel-cage induction machine or a synchronous machine, using current meters (6) and voltage meters (7), respectively. The current (9) and voltage (8) signals are sent to the turn-fault detection system (10), which performs diagnostics and sends an alarm signal (11) to external devices in case of a fault. Figure 2 shows an electrical machine connected via a variable speed drive or controlled drive. In addition to the elements described in Figure 1, an electrical drive has been subsequently incorporated into the secondary side of the AC transformer (3). This drive consists of a grid-side electronic converter (12) that rectifies or inverts the current (depending on whether the electrical machine is operating as a motor or generator, respectively). When functioning as a rectifier, the AC current from the transformer (2) is rectified to DC between the positive (13) and negative (14) DC terminals. A capacitor (15) has been provided to reduce voltage ripple in the DC stage.Subsequently, a converter on the electrical machine side (16) inverts this current under varying voltage and frequency conditions to control the machine at the desired operating point for the installation. The voltage (7) and current (6) meters remain connected to the stator terminals of the electrical machine, which can be either synchronous or asynchronous. These meters allow both feedback to the converter on the electrical machine side (16) and the input of these signals into the turn-fault detection system (10) for machine diagnostics. Figure 3 shows a possible distribution of the subsystems of a possible embodiment of the invention for the turn-fault detection system (10). In this system, the voltage (8) and current (9) signals are first introduced into the voltage recording (19) and current recording (18) subsystems, respectively. From these subsystems, digital signals are generated and sent to different subsystems that enable various functions. Specifically, these signals are sent to a load estimation subsystem that estimates the machine's load status using U (voltage), I (current), and the machine's equivalent impedance, which is previously determined through tests performed on the electrical machine to be protected.On the other hand, the current recording subsystem (18) introduces its digital current signals into the instantaneous sequence current estimation subsystem (21) which allows the matrix transformation of the currents ia, ib and ic (in the case of figures 1 and 2) into i0, i1 and i2, such that:. where: ia, ib and ic are the phase currents of the electrical machine to be protected; i0r, i1r and i2r are the real instantaneous sequence components; i0i, i1i and i2i are the instantaneous sequence components transverse to the previous ones; i0, i1 and i2 are the absolute instantaneous sequence components and where: And on the other hand, the voltage recording subsystem (19) introduces its digital voltage signals into the instantaneous sequence voltage estimation subsystem (22) which allows the matrix transformation of the voltages ua, ub and uc (in the case of figures 1 and 2) into u0, u1 and u2, such that: where: ua, ub and uc are the phase voltages of the electrical machine to be protected; u0r, u1r and u2r are the real instantaneous sequence components; u0i, u1i and u2i are the instantaneous sequence components transverse to the previous ones; u0, u1 and u2 are the absolute instantaneous sequence components and where: It is worth noting that instantaneous sequence components provide more information than traditional phasor sequence components. These can be subjected to harmonic decomposition in the frequency domain, since they are not solely based on the first harmonic measurement, as is the case with direct, inverse, and zero-sequence phasor components. Subsequently, in a current-based time-domain turn-fault detection subsystem (23) or a voltage-based time-domain turn-fault detection subsystem (24), a turn-fault is detected if the peak-to-peak or ripple measurement of the current or voltage exceeds a certain threshold T1, previously defined by the installation operator, such that: or where i is the ripple of the current "i" which refers to a generic instantaneous current which can be i1 or i2; imax and imin refer to the maximum and minimum values ​​of the previous current for a cycle of ia, ib or ic; and im refers to the average value of the current "i"; or such that: or where u is the ripple of the voltage "u", which refers to a generic instantaneous voltage that can be u1 or u2; umax and umin refer to the maximum and minimum values ​​of the voltage mentioned above for a cycle of ua, ub, or uc; and um refers to the average value of the voltage "u". These peak-to-peak or ripple changes are subject to the load estimation subsystem (20) not detecting changes in the load. In this way, it is possible to distinguish whether these ripple or peak-to-peak changes are due to faults or changes in the load.In parallel, a slower tripping mechanism is developed for detecting turns-to-turn faults than in the time-domain detection subsystems (23) / (24). This occurs in a frequency-domain turn-to-turn fault detection subsystem using current (25) and in a frequency-domain turn-to-turn fault detection subsystem using voltage (26), which perform this transformation using techniques such as the Fast Fourier Transform and compare harmonics before and after the turn-to-turn fault. As a non-limiting indicator of the invention, the comparison using the second harmonic of the fundamental frequency (k = 2) is provided. Detection occurs when the difference between this harmonic in the time window of the current / voltage before and after the turn-to-turn fault, transformed to the frequency domain, exceeds a threshold value T2 previously defined by the operator of the electrical installation to be protected, such that: where: Ai is the difference in decibels between the second harmonic (k=2) normalized with respect to its first harmonic (k=1) of the current is, which is the current of the successive window, and the second harmonic (k=2) normalized with respect to the first harmonic (k=1) of the current ip, which is the current of the previous window, for i1 or i2; or such that: where: Au is the difference in decibels between the second harmonic (k=2) of normalized with respect to its first harmonic (k=1) of us, which is the voltage of the successive window, and the second harmonic (k=2) normalized with respect to the first harmonic (k=1) of the voltage up, which is the voltage of the previous window, for u1 or u2. Finally, if any of the indicators exceeds T1 or T2, the subsystem emits an alarm signal (11) to devices external to the invention by means of an alarm emission subsystem (27). Figure 4 shows a possible embodiment of the method of the invention. In this figure, the method comprises several stages. This method begins by collecting the currents, ia, ib, and ic, and the voltages ua, ub, and uc (in the case of a three-phase alternating current system) in a voltage and / or current recording stage (28). In this stage, the recorded currents are the winding currents of the electrical machine to be protected, and the voltages can be phase voltages or line voltages of the electrical machine to be protected. These are then entered into a load estimation stage (29), which determines the machine's load based on the parameters of its equivalent circuit, previously entered by the electrical installation operator. Next, an instantaneous sequence component estimation stage (30) is performed, which can be applied to both currents (30.a) and voltages (30.b).b) creating two sufficient and complementary detection branches. This instantaneous sequence component estimation stage (30) allows the estimation of currents i0, i1, and i2 in the time domain and / or of u0, u1, and u2 in the time domain, yielding more information than from the conventionally used first-harmonic phasor sequence components. With the instantaneous sequence components, fault detection between turns is carried out in a time-domain fault detection stage between turns (31), which allows for virtually instantaneous tripping if the current ripple (31.a) / voltage (31.b) or its peak-to-peak value exceeds a first tripping threshold, T1, previously set by the electrical installation operator.If the ripple or peak-to-peak value is greater than the trigger threshold and this does not coincide with a variation with the machine load estimated in the load estimation stage (29), then proceed to an alarm emission stage (33) with at least information on the detection of a fault between turns. If, on the other hand, the peak-to-peak or ripple value remains within the limits of the T1 threshold, then a fault detection stage between turns in the frequency domain is performed, where i1 and / or i2 (32.a) or u1 and / or u2 (32.b) are transformed to the frequency domain, for example, using a Fast Fourier Transform (FFT). These currents or voltages in the frequency domain are compared using measurement windows of a length determined by the installation operator. A longer measurement length will provide more detail about the harmonics but will be slower, and vice versa.Subsequently, the "previous" and "next" current and / or voltage windows will be compared to obtain Ai and / or Au respectively, for a characteristic harmonic, this being (without limitation to this stage) the second harmonic of the fundamental frequency of the electrical machine, and if either of the two indicators (Ai or Au) is greater than a threshold T2, then it implies a fault which implies sending an alarm signal through an alarm transmission stage (33). Otherwise, the system will be operating under healthy conditions and the voltage and / or current readings will be taken again in the corresponding recording stage (28). The coil fault detection system (10) can be implemented using a programmable electronic device, programmed to perform the diagnosis described above. Figure 5 shows a numerical example of a fault in winding W1W2 of a 3.5 kW permanent magnet electric machine operating at 350 V and a fundamental frequency of 42 Hz. In this machine, a short circuit is introduced between turns of this winding with a fault current of 1.5 A, which is considered a minor fault (0.25 pu). The load applied to the electric machine is 60% of its three-phase balanced rated power, operating in "motor" mode. Furthermore, T1 is configured for a 10% ripple and a peak-to-peak difference of 0.5 A, and T2 is configured for a maximum difference of 5 dB. The three currents ia, ib, and ic applied to the different stages and subsystems of the possible embodiment of the invention described above are recorded. Figure 5.a shows a 20 s time window of the instantaneous sequence current, i1, calculated according to the previously described values ​​of the currents ia, ib, and ic. It can be observed that at 10.1 seconds, a fault occurs between turns because, with no change in the load, a disturbance is produced that increases the ripple from 7.16% and a peak-to-peak magnitude of 0.2 A to 11.0% and a peak-to-peak magnitude of 0.3 A. In this case, the system could emit an alarm signal since T1 has been exceeded due to current ripple. - In Figure 5.b, the current i1 from the previous figure is analyzed in two windows. On one hand, the current prior to the fault (up to t = 10 s) is observed, and on the other hand, the current during the fault (from t = 10 s to 20 s). Both spectra have been normalized for the first harmonic current of i1 (at 42 Hz). Comparing both results, the frequency component of the second harmonic of i1 (at 84 Hz) is -16.74 dB in the healthy case and 5.96 dB in the faulted case, so Ai = 22.7, which far exceeds the value of T2.

Claims

1. - System for detecting turns-to-turn faults in alternating current electrical machines that allows the detection of at least one turns-to-turn fault while the machine is in operation, characterized in that it comprises at least the following subsystems: - a current and / or voltage recording subsystem that collects at least the currents and / or voltages of all the input or output terminals of the winding of the electrical machine to be protected; - a subsystem for estimating instantaneous sequence currents from the above recorded currents; a time-domain turns-to-turn fault detection subsystem using current that records the ripple measurement of component i1 and / or i2 and detects the turns-to-turn fault if said component exceeds a threshold value T1 previously established by the operator of the installation,where i1 and i2 are the absolute instantaneous sequence components; - a frequency-domain turn-fault detection subsystem using current that transforms the time-domain measurement of i1 and / or i2 to the frequency domain, and detects the fault if the frequency components exceed a certain threshold T2 previously established by the installation operator; and - an alarm transmission subsystem that sends an alarm signal with information on at least the detection of the turn-fault to devices external to the invention. 2.- Turn-fault detection system in alternating current electrical machines according to claim 1, characterized in that, in the sequence current estimation subsystem, said instantaneous sequence components are arranged as follows, for a three-phase example: where: ia, ib and ic are the phase currents of the electrical machine to be protected; i0r,i1r and i2r are the actual instantaneous sequence components; i0i, i1i, and i2i are the instantaneous sequence components transverse to the former; i0, i1, and i2 are the absolute instantaneous sequence components.

3. A system for detecting faults between turns in alternating current electrical machines according to claim 1, characterized in that the time-domain fault detection subsystem between turns calculates the current ripple as: where i is the ripple of the current "i" which refers to a generic instantaneous current, which may be i1 or i2; imax and imin refer to the maximum and minimum values ​​of the current for a cycle of ia, ib, or ic; and im refers to the average value of the current "i".

4. A system for detecting faults between turns in alternating current electrical machines according to claim 1,characterized in that the turn-fault detection subsystem in the frequency domain compares the second harmonic component of current i1 or i2, obtained from a conversion process in successive non-overlapping windows, and uses the difference converted to decibels such that: where: Ai is the difference in decibels between the second harmonic (k=2) normalized with respect to its first harmonic of current is, which is the current of the successive window, and the second harmonic (k=2) normalized with respect to the first harmonic of current ip, which is the current of the previous window, for i1 or i2. 5.- Turn-fault detection system in alternating current electrical machines according to claim 4,characterized in that the current compared in Ai is the second harmonic current with respect to the fundamental supply frequency of the electrical machine to be protected from the direct or reverse instantaneous sequence component (i1 or i2).

6. Turn-to-turn fault detection system in alternating current electrical machines according to claim 1, characterized in that the recording, instantaneous sequence component estimation, and time-domain and frequency-domain fault detection subsystems are based on recorded voltages instead of currents, the voltages being: line voltages or phase voltages at the terminals of the windings to be diagnosed.

7. Turn-to-turn fault detection system in alternating current electrical machines according to claims 1 and 6,characterized in that the time-domain turn-fault detection subsystem calculates the voltage ripple as: where u is the voltage ripple "u" which refers to a generic instantaneous voltage, which can be u1 or u2; umax and umin refer to the maximum and minimum values ​​of the above voltage for a cycle of ua, ub or uc; and um refers to the average value of the voltage "u". 8.- Turn-fault detection system in alternating current electrical machines according to claims 1 and 6, characterized in that the turn-fault detection subsystem in the frequency domain compares the second harmonic component of the voltage u1 or u2, obtained in a conversion process, in successive non-overlapping windows and uses the difference converted to decibels such that: where: Au is the difference in decibels between the normalized second harmonic (k=2) with respect to its first harmonic of the voltage us,which is the voltage of the successive window, and the second harmonic (k=2) normalized with respect to the first harmonic of the voltage up, which is the voltage of the previous window, for u1 or u2. 9.- System for detecting faults between turns in alternating current electrical machines according to claims 1 and 6, characterized in that the system uses voltage and current measurements simultaneously to use them in parallel in the diagnosis of faults between turns, each of the branches being either diagnosed by currents, or diagnosed by complementary voltages between them and a sufficient condition for individual fault diagnosis. 10.- System for detecting faults between turns in alternating current electrical machines according to claims 1, 6 and 9,characterized in that the system comprises at least one electrical machine load estimation subsystem that uses current and voltage recording subsystems to estimate the machine load, and that is communicated with the time-domain, current and / or voltage, turn-to-turn fault detection subsystem, and that prevents tripping if the preset threshold T1 is exceeded when a load variation occurs, and that subsequently, if T1 remains exceeded, the fault is detected. 11.- Turn-to-turn fault detection system in alternating current electrical machines according to any of the preceding claims, characterized in that the electrical machine is: a squirrel-cage induction machine,a synchronous machine or a transformer.

12. Method for detecting turn-to-turn faults in alternating current electrical machines that allows the detection of turn-to-turn faults while the machine is in operation, characterized in that it comprises at least the following stages: - a current and / or voltage recording stage that collects at least the currents and / or voltages of all the input or output terminals of the winding of the electrical machine to be protected; - an instantaneous sequence current estimation stage from the previously recorded currents; a time-domain turn-to-turn fault detection stage using current that records the ripple measurement of component i1 and / or i2 and detects the turn-to-turn fault if said component exceeds a threshold value T1 previously established by the installation operator,where i1 and i2 are the absolute instantaneous sequence components; - a frequency-domain turn-fault detection stage using current that transforms the time-domain measurement of i1 and / or i2 to the frequency domain, and detects the fault if the frequency components exceed a certain threshold T2 previously established by the installation operator; and - an alarm transmission stage that sends an alarm signal with information of at least the detection of the turn-fault to devices external to the invention.

13. A method for detecting turn-faults in alternating current electrical machines that allows the turn-fault to be detected while the machine is in operation, according to claim 12, characterized in that said instantaneous sequence components are arranged as follows, for a three-phase example: where: ia, ib and ic are the phase currents of the electrical machine to be protected; i0r,i1r and i2r are the actual instantaneous sequence components; i0i, i1i, and i2i are the instantaneous sequence components transverse to the former; i0, i1, and i2 are the absolute instantaneous sequence components.

14. A method for detecting turn-to-turn faults in alternating current electrical machines according to claim 12, characterized in that the steps are applied to the measurement of phase or line voltage instead of currents, or both voltage and current measurements can be used in a complementary manner to perform the detection of the turn-to-turn fault in the electrical machine to be protected.

15. A method for detecting turn-to-turn faults in alternating current electrical machines according to claim 12,characterized in that the time-domain turn-fault detection stage calculates the current ripple as: where i is the ripple of the current "i" which refers to a generic instantaneous current, which can be i1 or i2; imax and imin refer to the maximum and minimum values ​​of the previous current for a cycle of ia, ib or ic; and im refers to the average value of the current "i". 16.- Method for detecting turn-faults in alternating current electrical machines according to claim 12, characterized in that the turn-fault detection stage in the frequency domain compares the second harmonic component of the current i1 or i2, obtained from a conversion process such as a fast Fourier transform,in successive windows that do not overlap and uses the difference converted to decibels such that: where: Ai is the difference in decibels between the second harmonic (k=2) normalized with respect to its first harmonic of the current is, which is the current of the successive window, and the second harmonic (k=2) normalized with respect to the first harmonic of the current ip, which is the current of the previous window, for i1 or i2. 17.- Method for detecting turns faults in alternating current electrical machines according to claim 12, characterized in that the turn fault detection stage in the time domain calculates the voltage ripple as: where u is the voltage ripple "u" which refers to a generic instantaneous voltage which can be u1 or u2; umax and umin refer to the maximum and minimum values ​​of the previous voltage for a cycle of ua,ub or uc; e um refers to the average value of the voltage "u".

18. Method for detecting turns-to-turn faults in alternating current electrical machines according to claim 12, characterized in that the turn-to-turn fault detection stage in the frequency domain compares the second harmonic component of the voltage u1 or u2, obtained from a conversion process such as a fast Fourier transform, in successive non-overlapping windows and uses the difference converted to decibels such that: where: Au is the difference in decibels between the second harmonic (k=2) normalized with respect to its first harmonic of the voltage us, which is the voltage of the successive window, and the second harmonic (k=2) normalized with respect to the first harmonic of the voltage up, which is the voltage of the previous window, for u1 or u2.

19. Method for detecting turns-to-turn faults in alternating current electrical machines according to any of claims 12 to 18,characterized in that it uses voltage and current measurements together to estimate the machine load in order to avoid unwanted tripping, considering trips that exceed thresholds T1 and T2 to be valid if no load variation has been detected during the process.

Citation Information

Patent Citations

  • Induction motor stator winding turn-to-turn short circuit early detection method based on voltage signal

    CN117074937A

  • Motor winding fault diagnosis

    EP4199346A1