Detection and analysis of partial discharges in electrical signals generated by pulse width modulation (PWM)

Abstract

The present invention relates to a method for detecting and analyzing partial discharge (DP) in an electrical signal (1) to be analyzed, the electrical signal (1) being generated by PWM and originating from an electrical device (10) in an aviation environment, the method comprising: a detection phase (S1) including a step (E10) of detecting the electrical signal (1) to be analyzed by a sensor (91), and a step (E1) of identifying possible partial discharges by comparing the electrical signal with at least one characteristic signal (2) representing a partial discharge (DP); and an analysis phase (S2) including a step (E4) of decomposing the electrical signal (1) into a series of patterns, and a step (E5) of assigning detected partial discharges (DP) to these patterns.

Description

Technical Field

[0001] This invention relates to the identification of partial discharges under voltages synthesized by pulse width modulation (PWM). This invention is particularly applicable to the identification of fault types in electrical links and aerospace environments. Background Technology

[0002] Partial discharge, referred to as PD below, is a localized discharge that only partially short-circuits the insulating gap separating the conductors or electrodes.

[0003] The presence of these discharges accelerates the degradation of insulating materials, whether through oxidation causing the insulation to become liquid or corrosion causing it to become solid. These discharges can lead to significant reliability issues. Furthermore, in addition to premature wear of airborne equipment, recurring partial discharges can cause malfunctions that may be critical when the aircraft is in flight.

[0004] Until recently, the voltage levels used in the aviation industry were not high enough to warrant significant attention regarding the presence of these partial discharges, which were unlikely to occur and therefore caused minimal damage to equipment. For example, the voltage levels typically used ranged from 230 volts to 400 volts.

[0005] Furthermore, countries have adopted, are adopting, or will adopt various restrictions on carbon emissions. In particular, an ambitious standard applies to both new and currently operating aircraft, requiring the implementation of technological solutions to ensure compliance with current regulations. For years, civil aviation has been actively contributing to addressing climate change.

[0006] Technological research has resulted in significant improvements in the environmental performance of aircraft. The applicant has considered factors affecting all design and development phases to obtain aircraft components and products that are less energy-intensive, more environmentally friendly, and whose integration and use in civil aviation produces moderate environmental consequences, with the aim of improving aircraft energy efficiency.

[0007] Therefore, the applicant continues to strive to reduce the climate impact of aircraft by using benign development and manufacturing methods and operating processes that limit greenhouse gas emissions to the minimum possible level, thereby reducing the environmental footprint of aircraft activities.

[0008] These ongoing research and development efforts have a particular focus on using electrical technologies to ensure progress.

[0009] This trend toward hybrid and / or electrified propulsion systems has led to increased demand for electrical energy, and consequently, increased voltage levels. As a result, the voltage levels used on aircraft electrical grids can now exceed 400 volts, and even reach 1 kilovolt. This voltage increase, combined with harsh pressure and temperature conditions, increases the risk of these partial discharges.

[0010] Furthermore, in some applications, voltages are generated via pulse width modulation (PWM); these voltages are characterized by steep rising and falling edges (high dV / dt derivatives), which further increases the probability of partial discharge.

[0011] Whenever possible, components are designed to prevent such partial discharges. However, in the long run, it is impossible to completely prevent this phenomenon because components will naturally wear down, and in the aerospace field, these components are subjected to severe stress, especially in terms of temperature and pressure.

[0012] Due to the increased risk of partial discharges in harsh environments and their significant impact on the reliability of airborne equipment, it is necessary to reliably detect and identify these partial discharges as soon as they occur, without requiring excessive computing power. This last point is particularly important in embedded systems and in aerospace environments where computing resources are limited.

[0013] Furthermore, while partial discharge detection is relatively simple under sinusoidal voltage, the noise generated under pulse stress or pulse width modulation (PWM) tends to overlap with the partial discharge signal, which actually complicates the detection of partial discharge.

[0014] Because the discharge has a very low charge value, complex and robust measurement equipment must be implemented.

[0015] Detection methods exist, but setting them up is not simple because they require suitable and sophisticated electronic components, such as high-frequency acquisition. Furthermore, expert verification is usually necessary to confirm the possibility of partial discharge.

[0016] Therefore, one object of the present invention is to improve the current solution of the prior art, particularly, especially when generating voltage signals by pulse width modulation (PWM), by allowing the detection and subsequent identification (i.e., characterization) of partial discharges in the voltage signals.

[0017] In particular, this characterization may include phase-amplitude-frequency analysis of the PRPD (phase-resolved partial discharge) type. Summary of the Invention

[0018] To address the shortcomings of existing technologies related to the detection of partial discharges, particularly those related to the detection of partial discharges associated with voltage signals generated by pulse width modulation (PWM), according to a first aspect, the present invention can be implemented by a method for detecting and analyzing partial discharges (PDs) in an electrical signal to be analyzed, the electrical signal to be analyzed originating from electrical equipment in an aviation environment. The method includes: a detection phase comprising the steps of detecting the electrical signal to be analyzed generated by pulse width modulation (PWM) using a sensor, and identifying possible partial discharges by comparing the electrical signal to be analyzed with at least one characteristic signal representing a partial discharge; and an analysis phase comprising the steps of decomposing the electrical signal to be analyzed into a series of patterns, and assigning detected partial discharges to the patterns.

[0019] According to a preferred embodiment, the present invention includes one or more of the following characteristics, which may be used individually, in partial combination with each other, or in complete combination with each other.

[0020] The method also includes the step of associating each of the patterns described above with information related to partial discharge previously assigned to the pattern.

[0021] The correlation step involves superimposing multiple partial discharges detected within multiple cycles of the electrical signal to be analyzed for the same pattern.

[0022] The decomposition step includes determining the number of patterns in the electrical signal to be analyzed based on parameters of the control signal used to generate the electrical signal via pulse width modulation.

[0023] The number of patterns is determined based on the frequency of the modulated signal and the frequency of the carrier.

[0024] The decomposition step includes identifying patterns by comparing them with characteristic patterns.

[0025] The amplitude-phase-frequency diagram of the PRPD type is associated with the pattern.

[0026] The identification step includes calculating at least one value of a distance parameter between the electrical signal to be analyzed and at least one feature signal, and comparing the value of the distance parameter with a detection threshold and detecting possible partial discharges based on the comparison result, wherein the distance parameter is a function of the deviation between the electrical signal to be analyzed and at least one feature signal.

[0027] Another object of the present invention relates to a computer program comprising instructions that, when executed by a processor of a detection and analysis device, implement the method as previously described.

[0028] Another object of the present invention relates to a detection and analysis apparatus adapted to perform the steps of the method as previously described.

[0029] Another object of the present invention relates to an aircraft that includes at least one such detection and analysis device.

[0030] Other features and advantages of the invention will become apparent from the following description of a preferred embodiment of the invention, given by way of example and with reference to the accompanying drawings. Attached Figure Description

[0031] Other aspects, objects, advantages, and features of the invention will become more apparent from the following detailed description of preferred embodiments of the invention, given by way of non-limiting example and with reference to the accompanying drawings, in which: Figure 1 A flowchart illustrating the steps of a method for detecting and identifying partial discharge according to an embodiment of the present invention. Figure 2 This illustrates the technical principles behind partial discharge identification according to embodiments of the present invention. Figure 3 A system for detecting partial discharge to obtain identifying features according to an embodiment of the present invention is shown. Figure 4 The possible construction of a PWM electrical signal is shown. Figure 5 An example of decomposing an electrical signal into two patterns is shown. Figures 6A to 6D A simplified example of a PRPD plot within the framework of a sinusoidal signal is shown. Figure 7 A simplified example of a PRPD type diagram for a PWM type electrical signal is shown. Detailed Implementation

[0032] Electrical insulators enable electrical systems to function properly by preventing the flow of current. In power electronics, these materials are either used in passive components, where their dielectric properties (and their storage properties) are utilized, or they serve as insulation. In electrical engineering, the primary focus is on the function of bringing insulated components to different potentials.

[0033] Insulating materials, whether solid, liquid, or gaseous, are often weak points in electrical systems; in particular, if a certain voltage is exceeded, partial discharge (PD) phenomena and their consequences (e.g., short circuits or fault arcs) may occur.

[0034] To date, given the relatively low voltage levels used, the presence of these partial discharges has rarely been considered in the design of aircraft equipment. However, particularly in the aerospace field, the hybridization and electrification of high-power propulsion systems have led to an increase in operating voltage; in reality, it is currently impossible for any system to claim to be free from or resistant to PD.

[0035] In the following text, partial discharge (PD) will be understood as a localized discharge generated in the insulating gaps separating conductors under the influence of high voltage or voltage variations. Furthermore, "electric grid" will be understood as any type of network that can be found in an aerospace environment, such as any type of network found in any aircraft: airplane or helicopter, etc.

[0036] Figure 1 This is a flowchart of the steps in a method for detecting and identifying partial discharge (PD). The main steps of this method are as follows: Figure 2 As shown in the image.

[0037] The proposed method includes a detection phase S1 and an analysis phase S2. In this paper, the method is implemented by a computer.

[0038] In the detection phase S1, the method includes step E10 of detecting the electrical signal 1 to be analyzed.

[0039] Pulse width modulation (or PWM) is a common technique for synthesizing pseudo-analog signals using digital circuitry (all-or-nothing, 1 or 0), or more generally, a common technique for synthesizing pseudo-analog signals using digital circuitry (all-or-nothing, 1 or 0) in discrete states. The general principle is that by applying a rapidly occurring series of discrete states with carefully chosen duration ratios, any intermediate value can be obtained by simply looking at the average value of the signal.

[0040] There are different techniques for generating electrical signals via PWM.

[0041] One traditional technique is the method called the intersection method.

[0042] refer to Figure 4 The method includes comparing the value of carrier 41 and the value of modulation signal 42 at each time step, and assigning discrete values ​​to modulation signal 1 based on the result of the comparison.

[0043] The carrier wave can be a triangular signal, such as... Figure 4 As shown, but it could also be other waveforms (such as sine waves).

[0044] Figure 4 A possible construction of a PWM electrical signal from a triangular carrier 41 of a modulation signal 42 is shown. In this paper, the modulation signal 42 is a sinusoidal modulation signal.

[0045] When the modulating signal 42 is greater than the carrier signal 41, such as in region 43, the output electrical signal 1 takes a first discrete value, which corresponds, for example, to logic "1". Conversely, such as in region 44, the electrical signal 1 takes a second discrete value, which corresponds, for example, to logic "0". In this way, an output signal with alternating discrete values ​​(usually two) is obtained based on the comparison between the modulating signal 42 and the carrier signal 41.

[0046] The carrier 41 can also be a sinusoidal carrier or a sawtooth carrier, etc.

[0047] Another method for generating PWM signals is the third harmonic injection method. This method was first described in Buja's 1975 article "An Improvement of Pulse Width Modulation Techniques", Archives of Electrical Engineering, Vol. 57, No. 5, pp. 281-289.

[0048] This method can be seen as a variant of the intersection method, but it includes the following preparatory steps: namely, by... Adding a factor modulates the modulation signal; that is, modifying the modulation signal by inserting the third harmonic into the modulation signal 42. These are the parameters of the method. Indicates time.

[0049] Another method for generating PWM signals is the space vector modulation (SVM) method. This method was first described in "Design and Experimental Results of Brushless AC Servo Drivers" by G. Pfaff et al., July / August 1984, IEEE Transactions on Industrial Applications, Vol. IA-20, No. 4, pp. 814-821, and is also widely described in related scientific literature, such as in MP. Kazmierkowski; R. Krishnan and F. Blaabjerg (2002), "Control in Power Electronics: Selected Problems," San Diego: Academic Press, ISBN 978-0-12-402772-5.

[0050] Another technique is the pre-computation method, also known as offline PWM or "Optimal Pulse Pattern" (OPP). The pattern of the output signal 1 is determined (offline) in advance and stored in a table, which is then read in real time. This method is described, for example, in "Synchronous Optimal Pulse Width Modulation with Different Modulation Waveform Symmetry Characteristics for Feeding High Magnetic Anisotropy to Synchronous Motors," published by ADBirda, J. Reuss, and C. Hackl in 2017, EPE'17 ECCE Europe, pp. 1-10, September 2017, doi:10.23919 / EPE17ECCEEurope.2017.8098963.

[0051] Another technique is full-wave control. In this type of operation, the switch used to generate the PWM signal operates at the frequency of the output electrical quantity. The on-time of the switch is T / 2, where T is the electrical cycle. The generated signal is a periodic square wave signal with a period of T.

[0052] Other methods for generating electrical signals via pulse width modulation (PWM) have been proposed and described in technical literature. As will be seen later, the described methods are applicable to PWM signals generated by different methods.

[0053] Once the electrical signal to be analyzed 1 is detected, it is advantageously preprocessed. The detection step is accompanied by filtering the electrical signal to be analyzed 1 through a filter E11. Preferably, the filter is a high-pass filter to minimize the noise level present in the electrical signal to be analyzed 1. The cutoff frequency of the high-pass filter is on the order of several hundred MHz and higher, for example, about 300 MHz.

[0054] The method then includes step E12 of acquiring the electrical signal 1 to be analyzed. Preferably, the electrical signal 1 to be analyzed is acquired by an acquisition device (described below).

[0055] Once the electrical signal 1 to be analyzed has been processed, step E1, which identifies any partial discharge, is performed. This step may be based on a comparison between the electrical signal to be analyzed and one or more characteristic signals 2 representing partial discharges (PDs).

[0056] Different embodiments of the identification step E1 can be implemented within the framework of the proposed method.

[0057] According to one embodiment, the identification step E1 includes a step E2 that calculates the value of distance parameter 3. Distance parameter 3 is a function of the deviation between two signals or a portion of signals.

[0058] The value of distance parameter 3, E2, is calculated between the electrical signal to be analyzed, 1, and the characteristic signal, 2. Characteristic signal 2 is a signal representing partial discharge (PD). Therefore, the value of distance parameter 3 is calculated between the electrical signal to be analyzed, 1, and the signal representing partial discharge (PD).

[0059] According to one implementation, the value of distance parameter 3 is calculated as a normalized Euclidean distance. That is, for each point where the amplitude of the electrical signal 1 being analyzed is centered and decreases, the deviation between that point and the corresponding point where the amplitude of the characteristic signal 2 is centered and decreases is determined. In other words, the value of distance parameter 3 is the result of the following equation:

[0060] in It is the number of points in feature signal 2. It is the amplitude of a point in characteristic signal 2. This refers to the amplitude of a point in the electrical signal 1 to be analyzed. Advantageously, the point... and The amplitude is voltage. Preferably, the amplitude is centered and reduced to obtain an amplitude that is independent of the selected unit or scale and has the same average value and the same dispersion.

[0061] According to another implementation, the value of distance parameter 3 is calculated using a dynamic time warping function. Such a function allows for a more robust calculation of the value of distance parameter 3 between the electrical signal 1 being analyzed and the characteristic signal 2, relative to time and amplitude expansion / contraction.

[0062] According to another implementation, the value of distance parameter 3 can be calculated using any function suitable for evaluating the deviation between two signals.

[0063] Furthermore, advantageously, calculation step E2 includes calculating a series of values ​​8 for the distance parameter 3. A series of values ​​8 are obtained by performing the following steps.

[0064] In the analysis window 6, select part 7 of the electrical signal 1 to be analyzed, E21. The analysis window 6 has a predetermined width 61. Calculate the value of distance parameter 3 between the portion 7 selected through analysis window 6 and characteristic signal 2 of electrical signal 1 to be analyzed (E22); Move the E23 analysis window 6 through predetermined step 62; and Repeat steps E24, including selecting E21, calculating E22, and moving E23, to obtain a series of values ​​for distance parameter 3.

[0065] In other words, the analysis window 6 of the selected portion 7 of the electrical signal 1 to be analyzed is shifted in time on the electrical signal 1 to be analyzed, and for each position of the analysis window 6, the distance parameter 3 is calculated as E22. In this way, a series of values ​​8 of the distance parameter 3 represent the deviation between the electrical signal to be analyzed and the characteristic signal of each portion 7 of the electrical signal 1 to be analyzed.

[0066] Advantageously, the width 61 of the analysis window 6 represents the time interval corresponding to the duration of the partial discharge PD represented on the feature signal 2.

[0067] Advantageously, step 62, which moves the analysis window 6 along the time scale of the electrical signal 1 to be analyzed, is equal to one acquisition period of the electrical signal 1 to be analyzed. However, step 62 may be different from this acquisition period, for example, step 62 may be a multiple of the acquisition period or any other time period.

[0068] Furthermore, advantageously, the value of the distance parameter 3 between part 7 of the electrical signal 1 to be analyzed (E22) and multiple characteristic signals 2 representing different types of partial discharges (PDs) can be calculated.

[0069] According to one implementation, multiple characteristic signals 2 are derived from a database. The detection device C described below enables the generation of a database, comprising multiple characteristic signals 2 representing partial discharge PD, in optional step E0. Indeed, the characteristic signals can be identified and recorded in the database before other steps of the detection and identification method are implemented. The signal characteristics 2 can evolve according to the conditions of the power grid 10 during its use. Therefore, the signal characteristics 2 representing partial discharge PD can be generated in a laboratory setting, just as they would under the same conditions. To monitor the insulation condition of the power grid 10, it is advantageous to implement a detection and identification method that includes using a database 10 comprising multiple characteristic signals 2 representing partial discharge PD.

[0070] Furthermore, different characteristic signals 2 can be collected for the same type of partial discharge, especially depending on the location where the measurement is performed. Therefore, the same partial discharge phenomenon will generate different signals depending on whether the signal is measured by probe 91 or probe 92. Figure 3 In this configuration, probes 91 and 92 are located on the same line 10, but are a certain distance apart from each other (e.g., 1 or 2 meters), as described below.

[0071] Following calculation step E2, the detection and identification method includes a comparison step E3. During this step, the value of the distance parameter 3 derived from calculation step E2 is compared with a detection threshold 4. Comparison E3 is dedicated to detecting potential partial discharge (PD) based on the obtained results, and therefore based on the distance between the electrical signal to be analyzed 1 and the characteristic signal 2. In practice, the smaller the value of the distance parameter 3, the smaller the deviation between the electrical signal to be analyzed 1 and the characteristic signal 2. A small deviation between the electrical signal to be analyzed 1 and the characteristic signal 2 implies substantial similarity, and thus, a partial discharge PD may be detected.

[0072] In an embodiment where a series of values ​​8 for distance parameter 3 are calculated, comparison step E3 compares the series 8 with a detection threshold 4 to detect one or more possible partial discharge PDs and the location of these one or more possible partial discharge PDs in the electrical signal 1 to be analyzed.

[0073] Advantageously, the detection threshold 4 is determined as a function of the average of a series of values ​​8 of the distance parameter 3, calculated by subtracting the standard deviation of the series of values ​​8 of the distance parameter 3 n times from this value. The number n is a real number, preferably an integer greater than or equal to 1.

[0074] According to one embodiment, the detection and identification method presented above can be coupled with other methods to ensure even more accurate identification. For example, this method can be coupled with a partial discharge PD identification method that implements wavelet transform of the electrical signal 1 to be analyzed.

[0075] Figure 3 Device C is schematically shown. Device C is used to pre-detect feature signal 2 in a laboratory, thereby enabling the generation of a database of E0 feature signals 2 and / or obtaining the electrical signal 1 to be analyzed in order to implement the detection and identification method as described above.

[0076] In addition, the source S of the electrical signal to be analyzed and the motor M are shown. The motor M can be replaced by any power-consuming device.

[0077] This source is suitable for generating electrical signals via pulse width modulation. To do this, the source can receive commands from a control device (not shown) to form a modulated signal. The modulated signal can then modulate the "carrier" or carrier signal source (typically forming a pure sine wave) to form a PWM electrical signal.

[0078] For example, device C is a test bench that includes a data acquisition device 9 and a sensor 91. For example, sensor 9 is a capacitively coupled sensor 91. Advantageously, sensor 91 includes a metal tip that contacts a component 11 of the power grid 10. Advantageously, component 11 of the power grid 10 is covered with a copper layer 12 to amplify the detected electrical signal.

[0079] However, other sensors besides capacitive sensors can also be used to perform step E10 of detecting electrical signals.

[0080] Device C enables the pre-generation of an E0 database and the acquisition of the electrical signal 1 to be analyzed. For this purpose, sensor 9 acquires electrical signals, for example, in millivolts, representing electrical physical quantities, such as current, that vary in the power grid 10.

[0081] Filtering E11, for example, using a high-pass filter, is applied to the electrical signal to minimize the noise level. Advantageously, the cutoff frequency of the filter used is approximately 100 MHz or higher, for example, approximately 300 MHz. Finally, step E12 is required to obtain the electrical signal 1 to be analyzed or the characteristic signal 2.

[0082] Once a signal is detected, filtered, and acquired, if the signal is intended to enhance measurements in the E0 laboratory database, it is recorded in the database, or the signal is used as electrical signal 1 to be analyzed, to achieve the detection and identification method described above.

[0083] The analysis phase S2 of the proposed method includes step E4, which decomposes the electrical signal 1 to be analyzed into a series of patterns.

[0084] In practice, two periodic signals—a modulation signal and a carrier—are used to generate electrical signals via pulse width modulation (PWM). If necessary, the resulting signal 1 is also a periodic signal.

[0085] However, the period of signal 1 can be longer than the period of the input signal, and it may not correspond to the period of the sine wave corresponding to the PWM signal. Therefore, different PWM patterns can exist for each period of the sine wave.

[0086] Figure 5 The diagram illustrates how electrical signal 1 is decomposed into two patterns, M1 and M2. These two patterns follow each other, thus forming signal 1.

[0087] Several methods can be used to determine the patterns that an electrical signal 1 can be decomposed into. These methods may depend on the method used to generate the PWM signal.

[0088] According to one embodiment, a frequency method is used. This method is based on prior knowledge of the carrier frequency. and the frequency of the modulation signal Then, the number of patterns that can be generated can be deduced. This information is sufficient to fully characterize the output signal because the output signal must necessarily be a series of these different patterns. Therefore, the number of known patterns... From this, it can be deduced that the output signal has a modulus of 1. A series of patterns , … .

[0089] In the case of electrical signal 1 generated by the intersection method, the number of patterns can be determined by the following expression:

[0090] The notation LCM() indicates the least common multiple.

[0091] In the case of electrical signals generated by other methods, the number of patterns can be determined in different ways.

[0092] In the case of electrical signals generated via harmonic injection or SVM, the frequency method can also be used. In fact, these methods are derivatives of the intersection method, but with different modulation shapes. Then, based on prior knowledge of the carrier frequency... and the frequency of the modulation signal The number of patterns that can be generated can be derived.

[0093] Regarding the pre-calculation method, it is preferable to determine the waveform a priori by calculating the switching angles to optimize the spectrum of the generated signal. Knowing the values ​​of these angles allows the determination of the number of patterns in the input electrical signal.

[0094] Finally, under full-wave control, the switch operates at the frequency of the output electrical physical quantity. Therefore, the generated PWM signal pattern is unique and is a square wave signal with a period of T, where T is the electrical period of the signal.

[0095] According to another embodiment, the pattern constituting the electrical signal 1 can be determined by identifying these patterns within the signal.

[0096] In particular, a method similar to that described previously for detecting and identifying partial discharges in signals can be used. In fact, in the same manner as before, a dictionary of characteristic patterns can be established, and then the value of the distance parameter between the electrical signal 1 and the characteristic pattern can be calculated, compared with a threshold, and then the pattern can be identified based on this comparison.

[0097] Then, the analysis phase S2 of the proposed method includes the step E5 of assigning the detected partial discharge PD to the previously identified pattern.

[0098] In fact, since electrical signal 1 is decomposed into a series of patterns, each detected partial discharge can be located in time relative to the pattern derived from the decomposition of the signal. Therefore, this PD / pattern assignment can be easily performed.

[0099] Subsequently, or in parallel, information related to these partial discharges can be correlated for each pattern: when electrical signal 1 is analyzed and a partial discharge is detected, the partial discharge can be assigned to the corresponding pattern. Therefore, data related to the detected PD is used to progressively enhance the information associated with that pattern.

[0100] Therefore, the method includes step E6 of associating each pattern with information related to the partial discharge (PD) assigned to that pattern.

[0101] This information can be structured in various ways. According to a preferred embodiment, the information may include an amplitude-phase-frequency diagram of the PRPD (phase-resolved partial discharge) type.

[0102] In this type of PRPD plot, PDs are represented by point clouds, which are characterized by phase (which allows the point cloud to be positioned facing the signal being analyzed), amplitude (which allows the point cloud to be located on the vertical axis), and frequency (corresponding to the number of PDs with that phase and amplitude).

[0103] More typically, the information generated and associated with the pattern enables the visual PRPD diagram to be produced on a human-machine interface, but the information can also be used in other ways, particularly by digitally processing the information without going through a display stage.

[0104] In particular, digital processing enables the automation of diagnostic steps, which allow information to be assigned to one or more possible faults at the local discharge power source.

[0105] In the case of PRPD, a sine wave can be used ( Figures 6A to 6D The identification of patterns associated with each fault type is performed in the form of a dictionary of patterns / faults generated by PWM.

[0106] In particular, by using supervised machine learning tools, these different patterns can be classified and associated with each type of fault at the local discharge point.

[0107] Figures 6A to 6D A simplified example of a PRPD plot within the framework of a sinusoidal signal is shown.

[0108] Each graph represents one cycle (360°) of a sinusoidal signal, plotting the (most) distribution of detected partial discharges over this cycle.

[0109] Figure 6A This demonstrates the presence of partial discharge within cavities created by the wear of semiconductor coatings.

[0110] Figure 6BPartial discharge is shown in the cavity originating from the delamination phenomenon of the insulating tape layer.

[0111] Figure 6C The study shows surface discharge at the coil head of the motor due to surface contamination.

[0112] at last, Figure 6D This demonstrates the presence of internal partial discharge within a microcavity within an insulator.

[0113] If possible Figures 6A to 6D As seen in the various examples, the PRPD diagram forms the characteristic features of the fault type.

[0114] The proposed method enables the generation of such a PRPD diagram within the framework of an electrical signal generated by PWM, that is, by associating the PD with the constituent patterns of the signal, rather than associating the PD with the entire signal.

[0115] This contribution is significant because the waveforms that make up the pattern itself enable fault identification. Therefore, it is of interest to be able to correlate the PD with different constituent patterns of electrical signals.

[0116] Figure 7 A simplified example of a PRPD type graph of a PWM type electrical signal is shown, which can be derived from the implementation of the detection and analysis method described above.

[0117] The diagram illustrates two component patterns, M1 and M2, of an electrical signal. Therefore, the information generated in step E6 and associated with each pattern enables the generation of a PRPD diagram for each of these two patterns.

[0118] The pattern can be obtained by superimposing the detected partial discharge PDs based on their temporal phase and amplitude in a reference frame identical to the pattern, and facing the pattern corresponding to the detection of the partial discharge PDs. Preferably, the superposition is performed in the form of a point cloud.

[0119] The distribution of points corresponding to PD within patterns M1 and M2 can form fault characteristics at the PD source.

[0120] Therefore, in summary, the proposed method enables PRPD to be performed under PWM voltage by indicating the presence of partial discharges in different phases of the analyzed electrical signal. Thus, this paper proposes a solution to the problem of PD analysis under PWM voltage.

[0121] Therefore, this method enables the use of PRPD-type graphs for signals generated by pulse width modulation, both now and in the future.

[0122] Overall, the described method enables: reliable and easy tracking of PRPDs under PWM voltages; identification of partial discharge occurrences (if any); identification of faults and analysis of critical flight phases, where the PD frequency may be higher; elimination of the need for more complex processing; and differentiation of all PDs from other high-frequency noise (e.g., one of the PWM converters).

[0123] The described method is compatible with all types of electrical systems operating under PWM voltage, regardless of the power and purpose of such electrical systems, as long as they have: a sensor capable of providing a signal containing the characteristics of the phenomenon, and a feature library of the features to be searched.

[0124] The proposed method is also compatible with all the different techniques used to generate PWM signals (such as generating PWM by intersection, generating PWM by harmonic injection, etc.).

[0125] Of course, the present invention is not limited to the examples and embodiments described and illustrated. The invention is particularly capable of having many variations available to those skilled in the art.

Claims

1. A method for detecting and analyzing partial discharge (PD) in an electrical signal (1) to be analyzed, the electrical signal (1) to be analyzed being generated by pulse width modulation (PWM) and originating from an electrical device (10) in an aviation environment, the method comprising: The detection phase (S1) includes the step (E10) of detecting the electrical signal (1) to be analyzed by a sensor (91), and the step (E1) of identifying possible partial discharges by comparing the electrical signal (1) to be analyzed with at least one characteristic signal (2) representing partial discharge (PD). The analysis phase (S2) includes the steps of decomposing the electrical signal (1) to be analyzed into a series of patterns (E4) and assigning the detected partial discharges (PDs) to the patterns (E5).

2. The method according to claim 1, wherein, The method further includes the step (E6) of associating each pattern in the pattern with information related to partial discharge (PD) previously assigned to the pattern.

3. The method according to the preceding claim, wherein, The information includes an amplitude-phase-frequency diagram of type PRPD.

4. The method according to claim 3, wherein, The amplitude-phase-frequency diagram is obtained by superimposing multiple partial discharges detected in multiple cycles of the electrical signal (1) to be analyzed for the same pattern.

5. The method according to any one of the preceding claims, wherein, The decomposition step (E4) includes determining the number of patterns in the electrical signal to be analyzed (1) based on parameters of the control signal used to generate the electrical signal by pulse width modulation. Preferably, the number of patterns is determined based on the frequency of the modulation signal and the frequency of the carrier.

6. The method according to any one of claims 1 to 4, wherein, The decomposition step (E4) includes identifying the pattern by comparing it with a feature pattern.

7. The method according to any one of the preceding claims, wherein, The identification step (E1) includes a step (E2) of calculating at least one value of a distance parameter (3) between the electrical signal to be analyzed (1) and the at least one feature signal (2), and a step (E3) of comparing the value of the distance parameter (3) with a detection threshold (4) and detecting the possible partial discharge (PD) based on the result of the comparison, wherein the distance parameter is a function of the deviation between the electrical signal to be analyzed (1) and the at least one feature signal (2).

8. A computer program comprising instructions that, when executed by a processor of a detection and analysis device (C), implement the method according to any one of the preceding claims.

9. A detection and analysis device (C) comprising at least one sensor (91) and a data acquisition device (9) adapted to be combined with a database to implement the steps of the method according to any one of claims 1 to 7.

10. An aircraft comprising at least one device according to the preceding claim.

Images