CN-121978448-A - Flying capacitor converter fault identification method and device based on auxiliary winding and computer readable storage medium

CN121978448ACN 121978448 ACN121978448 ACN 121978448ACN-121978448-A

Abstract

The invention relates to a method, a device and a computer readable storage medium for identifying faults of a flying capacitor converter based on an auxiliary winding, wherein the method comprises the steps of obtaining an inductance voltage signal through induction of the auxiliary winding and synchronously sampling by taking a carrier period as a window; the method comprises the steps of synchronously extracting direct current components of inductance voltage and harmonic complex components at carrier frequency in each window, taking the phase angle of the harmonic complex components as a fault unit positioning feature, taking the direct current components as fault type distinguishing features, constructing two-dimensional judging points, matching the judging points with a fault feature mapping relation pre-established based on a harmonic vector symmetrical offset principle under carrier phase shift modulation, and realizing accurate output of the fault unit and the type. According to the invention, through non-invasive sensing and dual-feature fusion analysis, fault positioning accuracy and type identification capability are remarkably improved, and anti-interference robustness and real-time diagnosis reliability are enhanced by utilizing a continuous period confirmation strategy.

Inventors

LING JIAYAO
Tang sai
HANG LIJUN
HE YUANBIN
Pang Feiyang
SONG LIJUN

Assignees

杭州电子科技大学

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (10)

1. A flying capacitor converter fault identification method based on auxiliary windings is characterized by comprising the following steps of; S1, synchronously sampling inductance voltage signals of the converter, and taking a carrier period of the inductance voltage signals as a calculation window; S2, synchronously extracting a direct current component value and a harmonic complex component at a carrier frequency from the sampling signal in each calculation window; S3, constructing a judgment point P (k) by taking the phase angle of the harmonic complex component as a fault unit positioning feature and the direct current component value as a fault type judging feature; And S4, matching the judging points with a pre-stored fault characteristic mapping relation, and outputting the number and the fault type of the fault switch unit according to a matching result, wherein the establishment of the fault characteristic mapping relation is based on a symmetry counteracting principle of harmonic vectors of each switch unit under carrier phase shifting modulation, and when any switch unit is in fault, the symmetry breaking causes the phase angle of the harmonic complex component to present the characteristic related to the modulation phase of the unit.
2. The method of claim 1, wherein the inductor voltage signal is obtained by induction of an auxiliary winding provided on a magnetic core of a magnetic element of the transformer.
3. The method according to claim 1, wherein in step S2, the extracting the dc component and the harmonic complex component is specifically implemented by performing a discrete fourier transform on a sampling sequence with a carrier period as a window.
4. A method of identifying a fault in a flying capacitor converter based on an auxiliary winding as claimed in claim 3, further comprising, in step S2; Envelope analysis is performed on the sampled signal to extract an envelope spectrum and to identify an early aging state of the switching device based on changes in specific sideband energy in the envelope spectrum.
5. The auxiliary winding-based flying capacitor converter fault identification method of claim 4, wherein the fault types at least include short circuit faults, open circuit faults and early aging faults; Wherein the short circuit fault and the open circuit fault are distinguished by positive and negative offset of the direct current component, and the early aging fault is identified by low-frequency sideband characteristics of the envelope spectrum; And when the phase angle of the harmonic complex component at the carrier frequency is judged to be in any one state of short-circuit fault, open-circuit fault and early aging fault, the phase angle of the harmonic complex component at the carrier frequency and the modulation phase of each switch unit are in a linear mapping relation, and a target angle partition corresponding to the fault unit is constructed according to the linear mapping relation and is used for judging the matching of the point abscissa.
6. The auxiliary winding-based flying capacitor converter fault identification method of claim 1, further comprising a fault verification step; If the judging point of the first calculation window does not definitely fall into the fault area, continuing to analyze the judging point of at least one subsequent period, and taking the continuous judging point stably falling into the same fault area as the basis of final fault confirmation.
7. The auxiliary winding-based flying capacitor converter fault identification method according to claim 1, wherein the flying capacitor converter is a single-phase N-level flying capacitor buck converter, and at least comprises N-1 switch units and N-2 flying capacitors.
8. The auxiliary winding-based flying capacitor converter fault identification method according to claim 7, wherein when the level number N of the converter is greater than 3, extracting a higher harmonic characteristic ratio as a third dimension coordinate, and expanding the decision point P (k) into a three-dimension characteristic vector for cluster analysis to distinguish fault attribution of different switch units.
9. An apparatus for implementing the auxiliary winding-based flying capacitor converter fault identification method of any one of claims 1 to 8, comprising; The auxiliary winding is arranged on the magnetic core of the magnetic element of the converter and is used for acquiring an induction voltage signal reflecting the high-frequency characteristic of the induction voltage; the sampling circuit is used for converting the induced voltage signal into a digital sequence; the processor is used for executing Hilbert transformation, discrete Fourier transformation, characteristic spectral line extraction, distinguishing point construction, lookup table matching and safety margin and continuous period confirmation logic; and a memory for storing a fault diagnosis lookup table, a discriminant rule, and higher order harmonic feature ratio data at N > 3.
10. A computer-readable storage medium, characterized in that the medium stores a computer program which, when executed by a processor, carries out the method steps of any one of claims 1 to 8.

Description

Flying capacitor converter fault identification method and device based on auxiliary winding and computer readable storage medium Technical Field The present application relates to the field of power electronic converter fault diagnosis technologies, and in particular, to a method and apparatus for identifying a flying capacitor converter fault based on an auxiliary winding, and a computer readable storage medium. Background The multi-level flying capacitor DC-DC converter comprises a plurality of switch units and a flying capacitor, and the voltage stress unbalance and the system reliability reduction can be caused by the occurrence of short circuit or open circuit faults of any switch device. The existing diagnosis scheme often depends on additional voltage/current sensing and complex online calculation, and has the problems of high hardware cost, difficult deployment, insufficient anti-interference capability, unstable early fault decision and the like. Fault diagnosis is typically accomplished in the prior art by monitoring inductor current or individual flying capacitor voltages. One common solution is to equip each switching unit with a dedicated voltage and current sensor, which can realize positioning, but significantly increases hardware cost and volume, and limits diagnostic accuracy due to the sampling bandwidth and anti-interference capability of the sensor in a high-frequency switching environment. In another scheme, frequency domain characteristics of inductance voltage are utilized for analysis, but in practical application, conventional frequency domain analysis often depends on strict carrier synchronization, and for hidden faults such as driving delay or weak parameter drift caused by early aging of a switching device, characteristic signals are easily submerged by strong carrier interference, so that diagnosis sensitivity is insufficient, and early warning is difficult to realize at the initial stage of the fault. In chinese patent CN107306083B, a voltage balance control device and a voltage balance control method for a flying capacitor are disclosed, in which a current direction estimating unit is used to obtain a voltage variation of any selected flying capacitor from a flying capacitor multi-level conversion circuit, and receive feedback signals of two adjacent switches of the selected flying capacitor, and perform an average or cumulative operation on the feedback signals generated in a regulation period to output an operation result, and a sign obtained by multiplying and/or dividing the voltage variation and the operation result is taken as an estimate of the current direction. In Chinese patent CN111983524B, a transformer winding fault evaluation method based on oscillatory wave time-frequency transformation is disclosed, the time-frequency characteristic is obtained through Hilbert transformation of an oscillatory wave signal after denoising, the standard deviation variance coefficient of the peak and the trough of a time-frequency characteristic curve is calculated to judge the fault type of a transformer winding, then the center of mass of the peak and the trough of a selected frequency band is obtained through binarization processing of the time-frequency characteristic curve under normal and fault conditions and the graph surrounded by the abscissa axis, finally the deviation index of the center of mass coordinate is provided, and the fault degree of the transformer winding is evaluated. In the above scheme, the former is mainly based on feedback signals to realize balance control and fault prevention on capacitor voltage, while the former is helpful to maintain the normal working point of flying capacitor, but is not extracted and diagnosed for early fault characteristics of a switching device, the dependent operation feedback signals are easy to be interfered under strong switching noise and can not effectively distinguish fault types and degrees, while the latter adopts Hilbert transformation to process signals, but the application object and fault mechanism are different from high-frequency switch faults in a multi-level switch converter, and when the method is directly transplanted into the high-frequency and multi-level flying capacitor converter, the problems of high switching frequency, complex signal modulation, dense interference components and the like are faced. Further, after the two schemes are combined and examined, although state monitoring based on internal logic (such as capacitor voltage balance degree and switching time sequence consistency) and fault feature extraction based on feature enhancement can be combined, a richer fault feature set is formed. The characteristics are analyzed and classified on line by means of a pattern recognition or machine learning algorithm (such as a support vector machine and a lightweight neural network), so that the leap from 'simple voltage unbalance alarming' to 'accurate fault type (such as specific switch short circuit,