Flying capacitor converter fault identification method and device based on auxiliary winding, and computer readable storage medium

Abstract

The invention relates to a flying capacitor converter fault identification method and device based on an auxiliary winding, and a computer readable storage medium. The method comprises the steps: obtaining an inductance voltage signal through the induction of the auxiliary winding, and carrying out the synchronous sampling through taking a carrier period as a window; synchronously extracting a direct-current component of the inductive voltage and a harmonic complex component at a carrier frequency in each window; constructing a two-dimensional judgment point by taking the phase angle of the harmonic complex component as a fault unit positioning feature and taking the direct current component as a fault type judgment feature; and matching the judgment point with a fault feature mapping relation pre-established based on a harmonic vector symmetric offset principle under carrier phase-shifting modulation to realize accurate output of a fault unit and type. According to the method, non-intrusive sensing is combined with double-feature fusion analysis, so that the fault positioning precision and the type identification capability are remarkably improved, and the anti-interference robustness and the real-time diagnosis reliability are enhanced by utilizing a continuous period confirmation strategy.

Description

Technical Field

[0001] This application relates to the field of power electronic converter fault diagnosis technology, and in particular to a method, device and computer-readable storage medium for fault identification of a flying capacitor converter based on an auxiliary winding. Background Technology

[0002] Multilevel flying capacitor DC-DC converters contain multiple switching units and flying capacitors. Short circuit or open circuit faults in any of these switching devices can cause voltage stress imbalance and degrade system reliability. Existing diagnostic solutions often rely on additional voltage / current sensing and complex online calculations, which suffer from high hardware costs, deployment difficulties, insufficient anti-interference capabilities, and unstable judgment in the early stages of fault occurrence.

[0003] Existing technologies typically diagnose faults by monitoring inductor current or the voltage of each flying capacitor. A common approach is to equip each switching unit with a dedicated voltage and current sensor. While this can pinpoint the fault, it significantly increases hardware cost and size. Furthermore, in high-frequency switching environments, the sensor's sampling bandwidth and anti-interference capability limit diagnostic accuracy. Another approach utilizes the frequency domain characteristics of inductor voltage for analysis. However, in practical applications, conventional frequency domain analysis often relies on strict carrier synchronization. Moreover, for latent faults such as drive delays or weak parameter drifts caused by early aging of switching devices, their characteristic signals are easily overwhelmed by strong carrier interference, resulting in insufficient diagnostic sensitivity and difficulty in providing early warnings of faults.

[0004] Chinese patent CN107306083B discloses a voltage balancing control device and method for a flying capacitor. The method uses a current direction prediction unit to obtain the voltage change of any selected flying capacitor from the multi-level conversion circuit of the flying capacitor, and receives the feedback signals of the two adjacent switches of the selected flying capacitor. The method then averages or accumulates the feedback signals generated within the adjustment period and outputs the calculation result. The sign of the multiplication and / or division of the voltage change and the calculation result is taken as the prediction of the current direction.

[0005] Chinese patent CN111983524B discloses a method for assessing transformer winding faults based on time-frequency transformation of oscillating waves. The method obtains the time-frequency characteristics of the denoised oscillating wave signal by Hilbert transformation, calculates the standard offset variance coefficients of the peaks and troughs of the time-frequency characteristic curves to determine the type of transformer winding fault, and then performs binarization processing on the graph enclosed by the time-frequency characteristic curves under normal and fault conditions and the horizontal axis to obtain the centroids of the peaks and troughs of the selected frequency band. Finally, the offset index of the centroid coordinates is proposed to assess the degree of transformer winding fault.

[0006] Of the above solutions, the former mainly relies on feedback signals to achieve balanced control and fault prevention of capacitor voltage. Although it helps maintain the normal operating point of the flying capacitor, it does not extract and diagnose the early fault characteristics of the switching device. The operational feedback signal it relies on is easily interfered with under strong switching noise and cannot effectively distinguish the fault type and degree. Although the latter uses Hilbert transform for signal processing, its application object and fault mechanism are different from the high-frequency switching faults in multi-level switching converters. When this method is directly transplanted to high-frequency, multi-level flying capacitor converters, it will face problems such as high switching frequency, complex signal modulation, and dense interference components.

[0007] Furthermore, combining the two approaches described above, while it's possible to combine logic-based state monitoring (such as capacitor voltage balance and switching timing consistency) with feature-enhanced fault feature extraction to create a richer fault feature set, and then using pattern recognition or machine learning algorithms (such as support vector machines and lightweight neural networks) for online analysis and classification of these features—achieving a leap from "simple voltage imbalance alarms" to "precise fault type (such as specific switch short circuit, open circuit, aging) and severity identification"—still presents challenges for high-frequency, multi-level flying capacitor converters. Their high switching frequency and complex modulation result in signals mixed with a large amount of high-frequency noise, switching glitches, and dense sideband components generated by modulation. Signals directly obtained from main circuit sensors (such as current or capacitor voltage sensors) have a low signal-to-noise ratio, making it easy for key fault features to be obscured, thus compromising the accuracy and robustness of feature extraction.

[0008] Furthermore, the establishment of the aforementioned fault feature set relies heavily on computationally intensive pattern recognition or machine learning algorithms for online analysis, which cannot meet the stringent requirements of fault diagnosis for real-time performance and speed. It cannot provide an effective response in the extremely short time it takes for a fault to occur and spread. Moreover, such data-driven methods usually rely on a large amount of historical fault data or accurate simulation models for training, while obtaining comprehensive and realistic fault samples in practical applications is extremely difficult. The generalization ability of the model is uncertain, and the reliability of diagnosis may drop significantly when faced with unseen operating conditions or noise interference.

[0009] Therefore, we hereby propose a method, device and computer-readable storage medium for fault identification of flying capacitor converters based on auxiliary windings. Summary of the Invention

[0010] The main objective of this application is to provide a method, device, and computer-readable storage medium for fault identification of flying capacitor converters based on auxiliary windings, aiming to solve the problems of insufficient fault location accuracy, incomplete fault type identification, poor robustness to noise and operating condition changes, and high real-time requirements in existing flying capacitor converter fault diagnosis.

[0011] To achieve the above objectives, this application provides a fault identification method for a flying capacitor converter based on an auxiliary winding, comprising the following steps; S1. The inductor voltage signal of the converter is synchronously sampled, and its carrier period is used as the calculation window; S2. Within each calculation window, a DC component value and a harmonic complex component located at the carrier frequency are synchronously extracted from the sampled signal. S3. Using the phase angle of the harmonic complex component as the fault unit location feature and the DC component value as the fault type discrimination feature, construct a decision point P(k). S4. Match the decision point with the pre-stored fault feature mapping relationship, and output the fault switch unit number and fault type according to the matching result; wherein, the establishment of the fault feature mapping relationship is based on the principle of symmetrical cancellation of harmonic vectors of each switch unit under carrier phase shift modulation; when any switch unit fails, the symmetry is destroyed, causing the phase angle of the harmonic complex component to exhibit characteristics related to the modulation phase of the unit.

[0012] Preferably, the inductor voltage signal is obtained by induction through an auxiliary winding disposed on the magnetic core of the converter's magnetic element.

[0013] Preferably, in step S2, the extraction of the DC component and the harmonic complex component is specifically achieved by performing a discrete Fourier transform on the sampling sequence with the carrier period as the window.

[0014] Preferably, step S2 further includes; Envelope analysis is performed on the sampled signal to extract the envelope spectrum, and the early aging state of the switching device is identified based on the changes in the energy of specific sidebands in the envelope spectrum.

[0015] Preferably, the fault types include at least short-circuit faults, open-circuit faults, and early aging faults; The short-circuit fault and the open-circuit fault are distinguished by the positive and negative offset of the DC component, and the early aging fault is identified by the low-frequency sideband features of the envelope spectrum. Furthermore, when it is determined that the fault is in any of the following states: short circuit fault, open circuit fault, or early aging fault, the phase angle of the harmonic complex component at the carrier frequency is linearly mapped to the modulation phase of each switching unit; and a target angle partition corresponding to the fault unit is constructed based on the linear mapping relationship for matching the horizontal coordinate of the discrimination point.

[0016] Preferably, it also includes a fault confirmation step; If the decision point of the first calculation window does not clearly fall into the fault area, the decision points of at least one subsequent cycle will be analyzed, and the final fault confirmation will be based on the fact that consecutive decision points stably fall into the same fault area.

[0017] Preferably, the flying capacitor converter is a single-phase N-level flying capacitor buck converter, which includes at least N-1 switching units and N-2 flying capacitors.

[0018] Preferably, when the converter level N is greater than 3, the higher-order harmonic characteristic ratio is extracted as the third-dimensional coordinate, and the decision point P(k) is expanded into a three-dimensional feature vector for cluster analysis to distinguish the fault attribution of different switching units.

[0019] To achieve the above objectives, this application provides an apparatus for implementing the above-described method for fault identification of a flying capacitor converter based on an auxiliary winding, comprising: An auxiliary winding disposed on the magnetic core of the converter's magnetic element is used to acquire an induced voltage signal that reflects the high-frequency characteristics of the inductor voltage. A sampling circuit is used to convert the induced voltage signal into a digital sequence; The processor is used to execute Hilbert transform, discrete Fourier transform, feature spectral line extraction, discrimination point construction, lookup table matching, and safety margin and continuous period confirmation logic. And a memory for storing fault diagnosis lookup tables, discrimination rules, and higher harmonic characteristic ratio data when N>3.

[0020] To achieve the above objectives, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for fault identification of a flying capacitor converter based on an auxiliary winding.

[0021] The beneficial effects of the technical solution of this invention are as follows: By utilizing the principle of symmetrical cancellation of harmonic vectors under carrier phase-shift modulation, when any switching unit fails, the symmetry is disrupted, causing the phase angle of the harmonic complex components at the carrier frequency to exhibit characteristics related to the modulation phase of the faulty unit, thereby enabling accurate identification and location of the faulty switching unit.

[0022] By extracting the DC component of the inductor voltage signal, short-circuit and open-circuit faults can be effectively distinguished. Combined with envelope analysis, early aging conditions of switching devices can also be identified, achieving comprehensive identification of multiple fault types. Meanwhile, the inductor voltage signal induced by the auxiliary winding has high-frequency characteristics and is less affected by main circuit noise and load changes. Extracting the DC and harmonic components through discrete Fourier transform further enhances noise suppression capabilities.

[0023] Synchronous sampling and feature extraction using the carrier period as the calculation window enable rapid fault diagnosis, meeting the needs of online real-time monitoring. Only an auxiliary winding is required on the magnetic core of the magnetic component, eliminating the need for additional high-precision current / voltage sensors, thus reducing hardware costs and system complexity. Furthermore, a confirmation mechanism is introduced to ensure that consecutive decision points stably fall within the same fault region, avoiding false positives and improving diagnostic reliability. Attached Figure Description

[0024] Figure 1 A schematic diagram of a single-phase N-level flying capacitor buck converter; Figure 2 This is a schematic diagram of the carrier waveform under carrier phase-shift pulse width modulation and the corresponding switching function values; Figure 3 This is a vector diagram of the high-frequency harmonics of the voltage during normal operation of an N-level flying capacitor buck converter. Figure 4 To divide the fault region of the three-level flying capacitor buck converter, the coordinate system is... It is divided into five zones, containing four types of fault and normal zones; Figure 5 A timeline diagram for converter normal operation—fault occurrence—fault identification—fault location; Figure 6 relative time of failure For equivalent switching function Impact: When SW1 experiences a short circuit fault, right The influence of position in the coordinate system ( ); Figure 7 relative time of failure For equivalent switching function Impact: When SW1 experiences a short circuit fault, right The influence of position in the coordinate system ( ); Figure 8 relative time of failure For equivalent switching function Impact: When SW1 experiences an open circuit fault, right The influence of position in the coordinate system ( ); Figure 9 relative time of failure For equivalent switching function Impact: When SW1 experiences an open circuit fault, right The influence of position in the coordinate system ( ).

[0025] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0026] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0027] Furthermore, descriptions using terms such as "first" and "second" in this application are for descriptive purposes only (e.g., to distinguish identical or similar elements) and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one feature. Additionally, technical solutions from different embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If a combination of technical solutions is contradictory or impossible to implement, such a combination should be considered nonexistent and not within the scope of protection claimed in this application.

[0028] In existing technologies, the flying capacitor converter (FCC), as a multilevel converter, has been widely used in new energy power generation, energy storage systems, and electric vehicles due to its advantages such as low output voltage ripple, low switching losses, and high power density. However, because it contains a large number of switching devices and flying capacitors, it is prone to failure during long-term operation, such as short circuits and open circuits of the switching transistors, and aging and failure of the capacitors. This can affect the normal operation of the converter, potentially leading to system collapse or even safety accidents.

[0029] Currently, traditional fault diagnosis methods for flying capacitor converters mainly include: Voltage / current signal-based monitoring: Faults are identified by monitoring changes in the input / output voltage and current of the converter or the voltage and current waveforms across the switching transistors. However, these signals are often affected by factors such as load changes and noise interference, and the fault characteristics are not obvious, making it difficult to achieve accurate fault location and type identification.

[0030] Model-based methods: A mathematical model of the converter is established, and faults are diagnosed by comparing the deviation between actual operating data and model predictions. However, this method requires high model accuracy and is computationally complex, making it difficult to apply in real-time online.

[0031] Signal processing-based methods: These methods analyze the converter signals using Fourier transform and wavelet transform to extract fault features. While this approach can identify faults to some extent, it often requires complex algorithms and substantial computational resources, and there is still room for improvement in the accuracy of fault type and location.

[0032] Artificial intelligence-based methods utilize machine learning algorithms such as neural networks and support vector machines to train and classify fault data. However, this method requires a large amount of fault data for training, and the model's generalization ability is limited by the diversity of the training data.

[0033] See Figure 1 This invention proposes a fault identification method for flying capacitor converters based on auxiliary windings. By synchronously sampling and extracting features from the inductor voltage signal, combined with the carrier phase-shift modulation principle, it achieves accurate identification of faulty switching units and comprehensive identification of fault types. Specifically, it includes the following steps: S1. The inductor voltage signal of the converter is synchronously sampled (i.e., the sampling frequency is synchronized with the carrier frequency of the converter to ensure that a fixed number of sampling points are collected in each carrier cycle), and its carrier cycle is used as the calculation window.

[0034] In another preferred embodiment, Figure 1 This is the basic circuit structure of a single-phase N-level flying capacitor buck converter used in one embodiment of the present invention.

[0035] The converter circuit includes a DC input voltage source V. DC A multilevel switching network consisting of N-1 connected switching units, N-2 flying capacitors, a filter inductor L, and an output capacitor C. o And the DC output load Z. The input of the multilevel switching network is connected to the DC input voltage source V. DC The connection is as follows: its output terminal is connected to one end of the filter inductor L; the other end of the filter inductor L is connected to the DC output terminal; the output capacitor C... o Both the load Z and the DC output terminal are connected in parallel to the DC output terminal, and the voltage at the DC output terminal is denoted as the output voltage v. o The multilevel switching network consists of N-1 switching units connected in series, labeled sequentially as switching unit 1 to switching unit N-1. The k-th switching unit (k=1, 2, ..., N-2) includes a controlled switching device and an SW. k With freewheeling diode D kThe node voltage near the output side of a multilevel switching network is denoted as the switching node voltage V. SW The voltage across the filter inductor L is denoted as the inductor voltage v. L Its reference polarity is as follows Figure 1 As shown. The flying capacitors are configured in multiple groups, each connected across adjacent switching units.

[0036] The inductor voltage signal is obtained by induction through an auxiliary winding disposed on the magnetic core of the converter's magnetic element. The auxiliary winding can sense high-frequency magnetic flux changes in the magnetic core, thereby reflecting the high-frequency characteristics of the inductor voltage. Furthermore, its induced signal is less affected by main circuit noise and load changes, exhibiting a good signal-to-noise ratio.

[0037] S2. Within each calculation window, a DC component value and a harmonic complex component located at the carrier frequency are synchronously extracted from the sampled signal.

[0038] The extraction of the DC component and harmonic complex components is specifically achieved by performing a Discrete Fourier Transform (DFT) on the sampling sequence with the carrier period as the window. The Discrete Fourier Transform (DFT) can decompose a time-domain signal into components of different frequencies, thereby accurately extracting the DC component (Hz component) and the harmonic components at the carrier frequency.

[0039] The DC component reflects the average value of the inductor voltage, which exhibits different offsets under normal operation and different fault modes. The harmonic complex components (including amplitude and phase angle) reflect the characteristics of the signal at the carrier frequency. Under carrier phase-shift modulation, the harmonic vectors generated by each switching unit normally cancel each other out symmetrically. When any switching unit fails, this symmetry is disrupted, causing significant changes in the amplitude and phase angle of the harmonic complex components at the carrier frequency.

[0040] In another preferred embodiment, let the voltage drop across the flying capacitor be v. ck (k = 1, 2, ...,N-2), the th The output voltage of each switching unit is the voltage drop across the lower bridge arm diode, which is v. l (l = 1, 2, ..., N-1). From Kirchhoff's voltage law, we can obtain: ; Define a switching function S k ∈{0,1} (k=1, 2, ..., N-1). If S k If the value is 0, the top switch of switch unit k is closed; otherwise, it is open. k The value is determined by both the carrier wave and the reference waveform. Using, for example... Figure 2The carrier phase-shift pulse width modulation strategy shown controls the flying capacitor converter.

[0041] Wherein, the switching function S k It is a square wave signal with an amplitude of 1 and a duty cycle of α. For ease of analysis, S can be... k Performing Fourier decomposition, the spectrum of the DC component and each harmonic frequency is obtained as follows: ;

[0042] in, It is the carrier frequency. This is the carrier phase of the k-th switching unit. The carrier phase... It is related to the carrier phase sequence of each switching unit. For example... Figure 2 As shown, when the phase of the (k+1)th switching unit carrier leads the phase of the kth switching unit carrier, the phase shift of the kth switching unit carrier is... It can be represented as: ; Furthermore, the switching function can be extracted from equation (2). Harmonic components at carrier frequency It can be expressed as the product of the magnitude term and the phase term: ; Therefore, based on the above, taking into account the inductor voltage... There is a linear mapping relationship between the switching functions of each switching unit and the inductor voltage. The harmonic components of the switching functions at the carrier frequency will be synchronously reflected in the inductor voltage. To ensure the electrical stress balance of the devices in the multi-level flying capacitor structure, under carrier phase-shift modulation, the voltages of each flying capacitor satisfy a predetermined distribution relationship, and the output voltage of each switching unit can be expressed as: ; Among them, v ck V represents the voltage across the k-th flying capacitor. l Indicates the first The output voltage components of each switching unit. Therefore, the switching node voltage V SW It can be expressed as the sum of the output voltages of each switching unit, and the inductor voltage satisfies: ; Because of v o Under steady-state conditions, it mainly exhibits a DC component, therefore The harmonic components at the carrier frequency can be derived from V. SWThe result obtained by superimposing the harmonic components at the carrier frequency is: ; in Indicates the first The harmonic components at the carrier frequency in the output voltage of each switching unit.

[0043] S3. Using the phase angle of the harmonic complex component as the fault unit location feature and the DC component value as the fault type discrimination feature, construct a decision point P(k).

[0044] The phase angle of the harmonic complex component is used as the abscissa (or one dimension) of the decision point P(k), because when any switching unit fails, its modulation phase directly affects the phase angle of the harmonic complex component. The DC component value is used as the ordinate (or another dimension) of the decision point P(k) for fault type identification. Furthermore, the positive and negative offset of the DC component can distinguish between short-circuit and open-circuit faults.

[0045] In summary, to facilitate the explanation of the phase relationship between harmonic components at the carrier frequency, in another preferred embodiment, the phase relationship between the harmonic components at the carrier frequency is... It is represented in the complex plane as a vector.

[0046] like Figure 3 As shown, and They are respectively The magnitude and phase angle. In the complex plane, the harmonic vector... With angular velocity Rotate, and Figure 3 Not all vectors are given in the text.

[0047] Furthermore, under carrier phase-shift modulation conditions, the harmonic vectors corresponding to each switching unit exhibit a characteristic of being distributed at equal phase intervals in the complex plane, thereby dividing the complex plane into N-1 sectors. Based on this symmetry, when the voltages of each flying capacitor satisfy a predetermined balance relationship (as shown in Equation (5)), the harmonic vectors cancel each other out under the meaning of vector superposition, resulting in the inductor voltage... The harmonic components at the carrier frequency satisfy the following under ideal conditions. =0, meaning that there are no high-frequency harmonic components at the carrier frequency in the inductor voltage spectrum.

[0048] When any switching device fails, the flying capacitor voltage associated with that switching device will deviate from the predetermined balance relationship, thereby causing the aforementioned symmetrical cancellation condition to fail and resulting in an inductor voltage... Non-zero harmonic components appear at the carrier frequency.

[0049] Therefore, this embodiment can utilize The occurrence and changes of harmonic components at the carrier frequency serve as the basis for fault detection, and the fault location is further achieved by combining the harmonic phase characteristics.

[0050] At the beginning of the kth cycle, the inductor voltage The instantaneous value of its DC component is obtained online by sampling and using Discrete Fourier Transform (DFT). and phase angle Furthermore, a phase angle is established. The x-axis represents the instantaneous value of the DC component. Two-dimensional decision points with ordinate as the vertical axis : ; Wherein, phase angle It can be used to locate faulty units and measure the instantaneous value of the DC component. It can be used to determine the type of fault, including short-circuit faults and open-circuit faults. When any switching unit fails, the symmetry of the inductor voltage spectrum will be disrupted, and higher harmonics will appear in the voltage spectrum, thus affecting the decision point. Migrating from the normal area to the faulty area, therefore, the embodiments of the present invention are based on the above. The movement trajectory is used to identify and locate faults.

[0051] In this embodiment of the invention, based on the instantaneous value of the DC component of the inductor voltage... The variation characteristics under short-circuit and open-circuit faults, and the phase angle of the harmonic components of the inductor voltage at the carrier frequency. The correspondence between the fault switch unit location and the corresponding pattern is used to predetermine the decision point of the N-level flying capacitor buck converter when a short-circuit / open-circuit fault occurs in each switch unit. The target coordinates are determined, and a fault lookup table is established and stored (as shown in Table 1 below). During operation, a discrete Fourier transform is performed on the sampled inductor voltage signal to obtain... And the phase angle, construct the decision point and will The fault is matched against the fault lookup table to output the fault type and fault switch unit location result.

[0052] Table 1 - Faults in different switching units of the N-level flying capacitor buck converter coordinate values

[0053] Note: Duty cycle; the above Can be equivalent to angles in or The inner part indicates.

[0054] For ease of understanding, the basis for obtaining the vertical and horizontal coordinates in Table 1 is explained below; it should be understood that this explanation is for illustrative purposes only and does not constitute a limitation on the scope of protection of this invention.

[0055] In one embodiment of the present invention, using Representing the DC components in v1, v2, … vN-1 respectively, under normal operating conditions of the converter, we can obtain: ; According to equation (6), we can obtain: ; in It is the DC component of the inductor voltage. It is the DC component of the output voltage. When a short-circuit fault occurs, One of them will jump from αVDC / (N-1) to VDC / (N-1). This increment is reflected in the DC component of the inductor voltage: ; When a circuit breaker fault occurs One of them will transition from αVDC / (N-1) to 0. This decrease is reflected in the DC component of the inductor voltage: ; Therefore, the sign and magnitude of the vertical axis in Table 1 can be used to distinguish between short circuits and open circuits.

[0056] In summary, the horizontal axis The value of is determined by the phase sequence of the carrier phase-shift modulation. Specifically, based on the carrier phase relationship (3) and the harmonic phasor expression (4) of the switching function, and combined with the superposition relationship (7) of the harmonic components of the inductor voltage at the carrier frequency, the target phase angle value corresponding to different fault switching units can be calculated in advance and stored in the form of a lookup table; during operation, the calculated value will be used in real time. Matching the phase angle target value to achieve fault switch unit location.

[0057] In a preferred embodiment of the present invention, when the converter level N is greater than 3, in order to more accurately distinguish the fault attribution of different switching units, the higher-order harmonic characteristic ratio can be extracted as the third-dimensional coordinate, and the decision point P(k) can be expanded into a three-dimensional feature vector for cluster analysis.

[0058] In a more preferred embodiment of the present invention, the formation process of harmonic components at the carrier frequency is described below in conjunction with a three-level flying capacitor buck converter. However, it should be understood that this description is only for explaining the principle of the present invention and does not constitute a limitation on the scope of protection.

[0059] In a three-level flying capacitor buck converter, only one flying capacitor is used, and its voltage is denoted as... Under carrier phase-shift modulation conditions, the inductor voltage The harmonic components at the carrier frequency can be obtained by superimposing the harmonic components of the output voltages of each switching unit at the carrier frequency. Based on the harmonic expression of the switching function at the carrier frequency, the harmonic components of the inductor voltage at the carrier frequency can be expressed in a form related to the switching functions of the first switching unit, the second switching unit, and the flying capacitor voltage, i.e. ; in, The carrier frequency amplitude coefficient related to the duty cycle. , These are the carrier phases of the two switching units, respectively. From (13), it can be seen that... The harmonic components of the switching function at the carrier frequency and the voltage across the flying capacitor. Relatedly, when a fault occurs, the switching function of the switching unit will no longer change according to the modulation law, but may exhibit constant conduction or constant turn-off. Therefore, an equivalent description needs to be introduced to uniformly characterize the impact of the fault.

[0060] To describe the "failure" effect of the k-th switching unit on modulation after a fault, an equivalent switching function is introduced. To replace fault conditions Where tF represents the time point of the fault occurrence, and ε(·) is a step function with an amplitude of 1. For short-circuit and open-circuit faults, the equivalent switching functions can be expressed as follows: ; Due to the occurrence of the fault It can only take the value 0 or 1. Therefore, the exponential modulation term corresponding to the fault switch unit will no longer be a tunable wave component at the carrier frequency, but will mainly show a change in DC characteristics, which provides a basis for subsequent fault diagnosis.

[0061] Flying capacitor voltage The change is mainly determined by the current flowing through the flying capacitor. The flying capacitor current can be expressed as: ; Taking the moving average of (15), we can obtain the average current across the capacitor over one period: ; in, The average current flowing through the flying capacitor during the period. This is the average value of the inductor current (which can be approximately constant over one cycle). , These are the equivalent duty cycles of switching unit 1 and switching unit 2 under fault equivalent conditions, respectively. When the converter is operating normally, the average net current flowing into the flying capacitor in one cycle is 0, and the flying capacitor voltage remains stable. Conversely, when a short circuit or open circuit fault occurs, It may not be zero, which would cause the flying capacitor voltage to shift in subsequent cycles, thus leading to (13) and The change in the parameters manifests as different trends in the carrier frequency harmonic amplitude depending on the type of fault.

[0062] The following four typical fault scenarios illustrate the variation pattern of carrier frequency harmonics: SW1 short circuit fault: At this time, the equivalent duty cycle of switch unit 1 is... Substituting into (16), we get: ; The voltage across the flying capacitor will gradually decrease from 0.5VDC towards 0, while diode D1 remains off throughout the entire cycle. Correspondingly: ; Therefore, the harmonics of the inductor voltage carrier frequency are mainly composed of contribute: ; SW2 short-circuit fault: The flying capacitor voltage will increase from 0.5VDC to VDC, and diode D2 will be in the off state throughout the cycle. The inductor voltage carrier frequency harmonic can be expressed as: ; SW1 open circuit fault: At this time, the equivalent duty cycle The flying capacitor's net current is positive, and its voltage increases from 0.5VDC towards VDC, with D1 remaining on throughout the cycle. The inductor voltage carrier frequency harmonic is: ; SW2 open circuit fault: The flying capacitor voltage decreases from 0.5VDC to 0, and D2 remains open throughout the cycle. The inductor voltage carrier frequency harmonics are: ; As can be seen from (19) and (21), when switch unit 1 malfunctions (whether short circuit or open circuit), The phase angle is determined by Decision; as can be seen from (20) and (22), when switch unit 2 malfunctions, The phase angle is determined by The decision is made because: after a faulty unit loses its modulation function, the high-frequency modulation term of its corresponding carrier frequency point is no longer effective, and the phase information of the carrier frequency point harmonics can only be determined by the modulation phase of the healthy unit. Therefore, the carrier frequency point harmonic phase angle can be used as a faulty unit location characteristic.

[0063] Furthermore, it can also be seen from equations (19)–(22) that short circuits and open circuits will cause the capacitor voltage to jump. Shifting in different directions, thus causing The amplitude of the fault exhibits different trends under different fault types; however, relying solely on amplitude and phase may still be insufficient to reliably distinguish fault types. Therefore, this invention further introduces a DC component as a supplementary criterion.

[0064] Furthermore, the DC component of the inductor voltage is used. As an indicator for fault type discrimination. Let... , , They are respectively , , The DC component, during normal operation, has: ; in This represents the duty cycle. According to the inductor-voltage relationship, we can obtain: ; When a short circuit fault occurs or One of them will be by Leap to The increment will be reflected in the DC component of the inductor voltage, thus yielding the DC component characteristics under short-circuit fault conditions: ; When a circuit breaker fault occurs or One of them will be by Leap to The reduction will be reflected in the DC component of the inductor voltage, thus yielding the DC component characteristics under open-circuit fault conditions: ;

[0065] Therefore, in this embodiment, the fault unit is located using the harmonic phase angle of the carrier frequency, and the short-circuit / open-circuit fault type is distinguished using the DC component of the inductor voltage, thereby forming the decision point. The horizontal and vertical coordinates are used for subsequent coordinate system division and table lookup determination.

[0066] Under three-level conditions, substituting N = 3 into the aforementioned general lookup table for N-level (Table 1) yields the decision coordinate system partitioning rules for the three-level converter. Specifically, as follows... Figure 4 As shown, the instantaneous value of the DC component of the inductor voltage is... The vertical axis represents the harmonic component phase angle at the inductor voltage carrier frequency. Construct decision points for the x-axis In this embodiment, the decision coordinate system is divided longitudinally into a short-circuit fault region, a normal operation region, and an open-circuit fault region; when the duty cycle is taken as... At that time, the region is symmetrically distributed about the horizontal axis. During normal operation of the converter, the decision point... Located in the normal operating range; when any switching device fails, the decision point... The fault will migrate along the direction corresponding to the fault type and location, and enter the corresponding fault area, thereby achieving fault identification and location. Furthermore, in the three-level structure, the fault attribution of different switching units (or different devices) can be distinguished based on the angular partitioning of the horizontal axis.

[0067] For example, using a preset angle threshold ( As a left-right boundary, the left and right sides can be corresponding to the first and second switching units respectively, thereby realizing the location of the fault unit; at the same time, based on the vertical axis The sign and amplitude changes of the signal can distinguish between short-circuit and open-circuit faults. Therefore, the decision points for short-circuit / open-circuit faults of different switching devices under three-level conditions can be obtained. The target value is determined and a three-level fault decision table as shown in Table 2 is formed. During operation, when the real-time calculated value is... When the result matches the target decision point (or its tolerance neighborhood) in Table 2, the corresponding fault type and fault location result will be output.

[0068] Table 2 - Faults in each switching unit of a three-level flying capacitor buck converter coordinate values

[0069] In the three-level application example (N = 3), the three-level fault decision table (as shown in Table 2) can be considered a special case of the aforementioned N-level general lookup table under the condition of N = 3. During operation, the decision points are calculated in real time. By matching the target coordinates with the decision table / lookup table, the determination result of the fault switch unit and the fault type can be output.

[0070] S4. Match the decision point with the pre-stored fault feature mapping relationship, and output the fault switch unit number and fault type based on the matching result. The establishment of the fault feature mapping relationship is based on the principle of symmetrical cancellation of harmonic vectors of each switch unit under carrier phase shift modulation. When any switch unit fails, the symmetry is broken, causing the phase angle of the harmonic complex component to exhibit characteristics related to the modulation phase of that unit.

[0071] For example, in an N-level converter, the modulation phase of each switching unit is fixed. When a switching unit fails, the phase angle of the harmonic complex components will shift towards the modulation phase of the failed unit. Specifically, within the first calculation window after a fault occurs, the sampled signal may simultaneously contain both normal and fault segments, thus causing the decision point corresponding to that window to exhibit transitional characteristics (also known as the "window period"). Figure 5 As shown, the converter exhibits a temporal process from normal operation to fault occurrence, fault identification, and fault location: when the fault occurs between the 0th DFT calculation time and the 1st DFT calculation time, the decision point obtained at the 0th DFT time... The decision point is usually located near the origin of the coordinate system; while the decision point obtained at the first DFT time step... The location of the fault will be determined by factors such as the relative occurrence time of the fault, the duty cycle, the type of fault, and the location of the fault.

[0072] In one optional implementation, the fault types include at least short-circuit faults, open-circuit faults, and early aging faults. Short-circuit faults and open-circuit faults are distinguished by the positive or negative offset of the DC component. For example, a short-circuit fault may cause a significant increase in the DC component, while an open-circuit fault may cause a decrease in the DC component or make it negative. Early aging faults are identified by the low-frequency sideband characteristics of the envelope spectrum. When a switching device ages prematurely, its switching characteristics change, resulting in specific low-frequency sideband energies in the envelope spectrum.

[0073] Furthermore, when the fault is determined to be in any of the following states: short-circuit fault, open-circuit fault, or early aging fault, the phase angle of the harmonic complex component at the carrier frequency has a linear mapping relationship with the modulation phase of each switching unit. Based on this linear mapping relationship, a target angle partition corresponding to the fault unit is constructed for matching the abscissa of the determination point. By ensuring that the determination point P(k) falls within the predefined fault region, the number and fault type of the faulty switching unit can be determined.

[0074] In one optional implementation, a fault confirmation step is also included; if the decision point of the first calculation window does not clearly fall into the fault region, the decision points of at least one subsequent cycle are analyzed, and the final fault confirmation is based on the consistent occurrence of consecutive decision points falling into the same fault region. This effectively avoids misjudgments caused by transient interference or noise, and improves the reliability of fault diagnosis.

[0075] To further illustrate the impact of the relative time of the fault occurrence on the decision result of the first cycle, in another preferred embodiment, the relative time of the fault is defined as... The following example illustrates the issue using a short-circuit or open-circuit fault in switch unit SW1. Figure 6 and Figure 7 As shown, when a short-circuit fault occurs in SW1, the relative time of the fault is... The relative order with respect to the preset reference point (e.g., the intersection point I of the modulated wave and the carrier wave) will affect the estimated value of the DC component of the inductor voltage in the first cycle. It can satisfy the segmentation relationship: ; Where T is the carrier period. Similarly, as Figure 8 and Figure 9 As shown, when an open circuit fault occurs in SW1, Changes will also lead to It exhibits segmented characteristics and can satisfy segmented relationships: ; For faults in switching unit SW2, the same analysis method as for SW1 can be used to obtain the corresponding patterns. To avoid repetition, it will not be elaborated here. It should be understood that in certain... Under the given conditions, even if the converter has failed, the first calculation cycle will still yield the following values. It may still be within the normal range or close to the origin, making the fault characteristics inconspicuous within that cycle. To address this, this embodiment employs the aforementioned continuous cycle confirmation strategy, utilizing decision points obtained from subsequent calculation cycles (e.g., This confirmation process allows for a thorough assessment of the converter's health status and enables fault identification and location.

[0076] In summary, the carrier phase-shift modulation principle in this invention is configured as follows: the phase angle of the harmonic complex component at the carrier frequency and the modulation phase of the switching unit that has lost its modulation function under the fault state and its adjacent healthy switching unit have a deterministic mapping determined by the circuit topology and modulation rules; this mapping relationship ensures that when different switching units fail, their corresponding harmonic complex component phase angles occupy different preset angle intervals on the complex plane. Furthermore, this influence mechanism explains why the introduction of a continuous periodic confirmation strategy is crucial in this invention. Figures 5-9 As shown, since the fault may occur at any time within the calculation window, the sampling signal of the first window is essentially a mixture of normal segments and fault segments. This leads to uncertainty in the position of the first decision point P(1), which may fall into the normal region, the uncertain transition region, or, although it falls into the fault region, its features are not yet significant.

[0077] The continuous periodic confirmation strategy is as follows: in the subsequent calculation window where the fault persists, the sampled signal will fully reflect the fault state, so that the corresponding decision points P(2), P(3), etc., can stably and clearly fall into the specific fault region predefined by the deterministic mapping relationship. By requiring the decision points of multiple consecutive periods to consistently point to the same fault region, occasional misjudgments caused by instantaneous interference, sampling noise, or the initial transition period of the fault are effectively filtered out, thereby achieving rapid capture and locking of the fault while ensuring diagnostic accuracy.

[0078] Therefore, the method described in this invention utilizes the strong correlation between the physical changes in the harmonic phase angle and the DC component at the carrier frequency caused by the fault to construct a highly discriminative decision point. Through a continuous periodic confirmation mechanism, it cleverly resolves the initial diagnostic ambiguity caused by the randomness of the fault occurrence time. This enables the method to achieve rapid, reliable, and accurate online identification and location of fault types such as short circuits, open circuits, and early aging in the switching unit of the flying capacitor converter in complex actual operating environments.

[0079] The present invention also proposes an apparatus for implementing a fault identification method for a flying capacitor converter based on an auxiliary winding, comprising: An auxiliary winding disposed on the magnetic core of the converter's magnetic element is used to acquire an induced voltage signal that reflects the high-frequency characteristics of the inductor voltage. A sampling circuit is used to convert the induced voltage signal into a digital sequence; The processor is used to execute Hilbert transform, discrete Fourier transform, feature spectral line extraction, discrimination point construction, lookup table matching, and safety margin and continuous period confirmation logic. And a memory for storing fault diagnosis lookup tables, discrimination rules, and higher harmonic characteristic ratio data when N>3.

[0080] In one embodiment, the processor further includes a three-dimensional feature vector construction module for extracting the amplitude ratio or phase difference of the second or third type of carrier harmonics when N > 3, and expanding the discrimination points into three-dimensional feature vectors for cluster analysis.

[0081] Furthermore, this invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned method for fault identification of a flying capacitor converter based on an auxiliary winding. When executed, the computer program may include the flow of the respective embodiments of the aforementioned methods for fault identification of flying capacitor converters based on auxiliary windings. Any references to memory, storage, databases, or other media used in this application and in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0082] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method for fault identification of a flying capacitor converter based on an auxiliary winding that includes a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method for fault identification of a flying capacitor converter based on an auxiliary winding. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of additional identical elements in the process, apparatus, article, or method for fault identification of a flying capacitor converter based on an auxiliary winding that includes that element.

[0083] The above description is only a preferred embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural changes made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A fault identification method for a flying capacitor converter based on an auxiliary winding, characterized in that, Includes the following steps; S1. The inductor voltage signal of the converter is synchronously sampled, and its carrier period is used as the calculation window; S2. Within each calculation window, a DC component value and a harmonic complex component located at the carrier frequency are synchronously extracted from the sampled signal. S3. Using the phase angle of the harmonic complex component as the fault unit location feature and the DC component value as the fault type discrimination feature, construct a decision point. P ( k ); S4. Match the decision point with the pre-stored fault feature mapping relationship, and output the fault switch unit number and fault type according to the matching result; wherein, the establishment of the fault feature mapping relationship is based on the principle of symmetrical cancellation of harmonic vectors of each switch unit under carrier phase shift modulation; when any switch unit fails, the symmetry is destroyed, causing the phase angle of the harmonic complex component to exhibit characteristics related to the modulation phase of the unit.

2. The method for fault identification of a flying capacitor converter based on an auxiliary winding according to claim 1, characterized in that, The inductor voltage signal is obtained by induction through an auxiliary winding located on the magnetic core of the converter's magnetic element.

3. The method for fault identification of a flying capacitor converter based on an auxiliary winding according to claim 1, characterized in that, In step S2, the extraction of the DC component and harmonic complex component is specifically achieved by performing a discrete Fourier transform on the sampling sequence with the carrier period as the window.

4. The method for fault identification of a flying capacitor converter based on an auxiliary winding according to claim 3, characterized in that, Step S2 also includes; Envelope analysis is performed on the sampled signal to extract the envelope spectrum, and the early aging state of the switching device is identified based on the changes in the energy of specific sidebands in the envelope spectrum.

5. The method for fault identification of a flying capacitor converter based on an auxiliary winding according to claim 4, characterized in that, The fault types include at least short-circuit faults, open-circuit faults, and early aging faults. The short-circuit fault and the open-circuit fault are distinguished by the positive and negative offset of the DC component, and the early aging fault is identified by the low-frequency sideband features of the envelope spectrum. Furthermore, when it is determined that the fault is in any of the following states: short circuit fault, open circuit fault, or early aging fault, the phase angle of the harmonic complex component at the carrier frequency is linearly mapped to the modulation phase of each switching unit; and a target angle partition corresponding to the fault unit is constructed based on the linear mapping relationship for matching the horizontal coordinate of the discrimination point.

6. The method for fault identification of a flying capacitor converter based on an auxiliary winding according to claim 1, characterized in that, It also includes a fault confirmation step; If the decision point of the first calculation window does not clearly fall into the fault area, the decision points of at least one subsequent cycle will be analyzed, and the final fault confirmation will be based on the fact that consecutive decision points stably fall into the same fault area.

7. The method for fault identification of a flying capacitor converter based on an auxiliary winding according to claim 1, characterized in that, The flying capacitor converter is a single-phase N-level flying capacitor buck converter, which includes at least N-1 switching units and N-2 flying capacitors.

8. The method for fault identification of a flying capacitor converter based on an auxiliary winding according to claim 7, characterized in that, When the converter level N is greater than 3, the higher-order harmonic characteristic ratio is extracted as the third-dimensional coordinate, and the decision point is... P ( k The feature vectors are expanded into three-dimensional feature vectors for cluster analysis to distinguish the fault attribution of different switching units.

9. An apparatus for implementing the fault identification method for a flying capacitor converter based on an auxiliary winding as described in any one of claims 1 to 8, characterized in that, include; An auxiliary winding disposed on the magnetic core of the converter's magnetic element is used to acquire an induced voltage signal that reflects the high-frequency characteristics of the inductor voltage. A sampling circuit is used to convert the induced voltage signal into a digital sequence; The processor is used to execute Hilbert transform, discrete Fourier transform, feature spectral line extraction, discrimination point construction, lookup table matching, and safety margin and continuous period confirmation logic. And a memory for storing fault diagnosis lookup tables, discrimination rules, and higher harmonic characteristic ratio data when N>3.

10. A computer-readable storage medium, characterized in that, The medium stores a computer program that, when executed by a processor, performs the steps of the method according to any one of claims 1 to 8.

Images