Method and system for monitoring internal fault impedance characteristics of high-voltage parallel capacitor

Abstract

The invention discloses a high-voltage parallel capacitor internal fault impedance characteristic monitoring method and a high-voltage parallel capacitor internal fault impedance characteristic monitoring system. The impedance characteristic monitoring method comprises the following steps: acquiring voltage and current signals; after anti-aliasing filtering and analog-to-digital conversion are carried out, a quasi-synchronous sampling algorithm is adopted for processing, and a sine amplitude and a cosine amplitude of a fundamental wave are calculated in a weight coefficient weighted average mode; fundamental waves and multiple main harmonic components are separated based on a fast Fourier transform algorithm, and complex impedance values under the fundamental waves and all harmonic frequencies are calculated; constructing a multi-state impedance characteristic criterion library, performing longitudinal trend comparison and transverse inter-phase comparison on the impedance value calculated in real time and the criterion library, and extracting fault characteristic parameters; according to the method, the random forest classifier after offline training is adopted to recognize the fault type, corresponding response is triggered according to the fault recognition result, and full-state, high-sensitivity and rapid online monitoring and protection of a high-voltage parallel capacitor branch from normal operation to various internal faults and resonance risks are achieved.

Description

Technical Field

[0001] This invention relates to the field of impedance characteristic monitoring technology, and in particular to a method and system for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor. Background Technology

[0002] High-voltage parallel capacitors are core components of reactive power compensation, and their safe operation is crucial. Currently used unbalanced protection methods (such as open-delta voltage protection and bridge differential current protection) have significant drawbacks: they cannot identify symmetrical faults in the capacitor bank, nor can they effectively protect series reactors. When an inter-turn short circuit occurs in a reactor, it often escalates into a serious accident. Furthermore, capacitor aging, self-healing, and reactor insulation damage can cause branch impedance characteristics to gradually deviate from design values, potentially triggering harmonic resonance and causing equipment damage. Existing protection methods lack early warning capabilities for such risks.

[0003] Although impedance protection was studied in the early stages, it failed to be practical due to insufficient measurement accuracy. In recent years, although online monitoring solutions have emerged, most of them require modification of primary equipment or the addition of sensors, and their diagnostic capabilities for reactor faults and resonance risks are limited. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method and system for monitoring the internal fault impedance characteristics of high-voltage parallel capacitors. This system can monitor branch impedance characteristics online with high precision, comprehensively identify various faults, and provide early warning of resonance risks, thus filling the gap in traditional protection.

[0005] To achieve the above objectives, the present invention provides a method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor, comprising the following steps:

[0006] S1 collects the signals of capacitor terminal voltage, series reactor terminal voltage and branch current of the parallel capacitor branch; S2, after performing anti-aliasing filtering and analog-to-digital conversion on the acquired signal, uses a quasi-synchronous sampling algorithm to process it, and calculates the fundamental sine amplitude and cosine amplitude of the voltage and current signals by weighted averaging using weight coefficients; S3, based on the fast Fourier transform algorithm, separates the fundamental wave and the 3rd, 5th and 7th main harmonic components, and calculates the complex impedance values ​​of the fundamental wave and each harmonic frequency. S4. Construct a multi-state impedance characteristic criterion library that includes normal state, fault occurrence state and accident expansion critical point. The criterion library includes voltage distribution ratio criterion, impedance change criterion, and harmonic impedance characteristic offset criterion. Compare the real-time calculated complex impedance value with the criterion library in the longitudinal trend and in the horizontal phase comparison to extract fault characteristic parameters. S5. Based on the fault feature parameters, a random forest classifier trained offline is used to identify the fault types, including capacitor unit symmetry faults, series reactor inter-turn short circuit faults, discharge coil faults, and branch resonance faults. S6, based on the fault identification results, triggers graded alarm signals or protection trip commands, and uploads the fault data to the monitoring backend.

[0007] Preferably, in step S2, the weight coefficients of the quasi-synchronous sampling algorithm Based on the type of quadrature formula, the number of samples per single signal period, and the number of sampling periods, these values ​​are pre-calculated and stored. The real and imaginary parts of the voltage and current signals are then weighted once to obtain the fundamental sine amplitude *b* and the cosine amplitude *a*. The calculation formula is as follows: ; ; in, Sampling time The signal value, ω is the angular frequency.

[0008] Preferably, in step S3, the complex impedance value includes a resistive component and a reactance component; The aforementioned resistance component is expressed as: ; The reactance component is expressed as: ; in, , These are the fundamental cosine amplitude and the sine amplitude of the voltage signal, respectively. , These are the fundamental cosine amplitude and sine amplitude of the current signal, respectively.

[0009] Preferably, the method for establishing the voltage distribution ratio criterion in step S4 is as follows: The ratio of capacitor terminal voltage to series reactor terminal voltage at the fundamental frequency ,in, This represents the fundamental effective value of the capacitor terminal voltage. This represents the fundamental effective value of the voltage across the series reactor; multiple state thresholds are based on the design reactance rate of the series reactor. And capacitor bank tuning frequency setting.

[0010] Preferably, for capacitor banks designed to suppress 5th and higher harmonics with a reactance of 5%, the warning threshold for the fundamental voltage ratio is set to 24.19, and the tripping threshold is set to 24.9.

[0011] Preferably, the impedance change criterion in step S4 is calculated based on the single-phase impedance formula of the capacitor bank; for each phase... n Each series segment consists of a series segment, and each series segment is composed of a series segment. m The single-phase impedance of a capacitor bank consisting of capacitors connected in parallel under normal conditions is expressed as: ; in, This refers to the capacitance value of a single capacitor. Angular frequency; When a single capacitor in a capacitor bank experiences a breakdown fault, the other capacitors in the same parallel section will discharge to the faulty capacitor, increasing the current flowing through it. The fuse will then blow, disconnecting the faulty capacitor. Based on the series-parallel relationship after the fault, the impedance of the capacitor bank at this point can be obtained, expressed as: ; The change in reactance of the capacitor bank before and after a fault is expressed as follows: ; The relative change in the reactance of the capacitor bank after a single capacitor breakdown fault is expressed as follows: ; The impedance change criterion is based on Set multi-level early warning thresholds within the range of 2.7% to 20%.

[0012] Preferably, the method for establishing the harmonic impedance characteristic offset criterion in step S4 is as follows: Under normal operating conditions of the capacitor bank, the reference harmonic impedance value corresponding to the main harmonic is acquired and stored; Real-time calculation of harmonic impedance values ​​at the main harmonic frequencies; Calculate the resistance offset coefficient and reactance offset coefficient of each harmonic impedance; Set the resistance and reactance offset thresholds for each harmonic frequency; When the resistance offset coefficient or reactance offset coefficient of any harmonic frequency exceeds its corresponding threshold, it is determined that the harmonic impedance characteristic has shifted.

[0013] Preferably, the fault characteristic parameters in step S4 include: voltage distribution ratio, impedance change, harmonic impedance offset, and phase offset.

[0014] Preferably, the random forest classifier in step S5 contains more than 500 decision trees, and the input feature dimensions include: voltage distribution ratio, impedance change, harmonic impedance offset, and phase offset.

[0015] This invention also provides a monitoring system for the internal fault impedance characteristics of a high-voltage parallel capacitor, employing the aforementioned method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor, comprising: The signal acquisition module is used to acquire signals of capacitor terminal voltage, series reactor terminal voltage, and branch current in the parallel capacitor branch. The signal processing module is used to perform anti-aliasing filtering and analog-to-digital conversion on the acquired signals, and to calculate the fundamental sine amplitude and cosine amplitude of the voltage and current signals using a quasi-synchronous sampling algorithm. The quasi-synchronous sampling algorithm uses pre-stored weighting coefficients for weighted averaging. The impedance calculation module is used to separate the fundamental wave and the 3rd, 5th and 7th major harmonic components based on the fast Fourier transform algorithm, and to calculate the complex impedance values ​​of the fundamental wave and each harmonic frequency. The fault feature extraction module constructs a multi-state impedance characteristic criterion library. The multi-state includes normal state, fault occurrence state, and accident expansion critical point. The criterion library includes voltage distribution ratio criterion, impedance change criterion, and harmonic impedance characteristic offset criterion. This module is used to compare the real-time calculated complex impedance value with the criterion library in a longitudinal trend comparison and a horizontal phase comparison to extract fault feature parameters. The fault identification module includes an offline-trained random forest classifier for identifying fault types based on the fault feature parameters. The fault types include capacitor unit symmetry faults, series reactor inter-turn short-circuit faults, discharge coil faults, and branch resonance faults. The output and communication module is used to trigger graded alarm signals or protection trip commands based on the fault identification results, and upload the fault data to the monitoring backend.

[0016] The present invention employs the above-mentioned method and system for monitoring the internal fault impedance characteristics of high-voltage parallel capacitors, and has the following beneficial effects: (1) This invention solves the blind spots of traditional protection: Traditional unbalanced protection (such as open delta and bridge differential protection) cannot identify symmetrical faults in capacitor banks, nor can it effectively protect against inter-turn short circuits in series reactors. This invention, through impedance characteristic analysis, can simultaneously identify symmetrical faults in capacitor units, inter-turn short circuits in reactors, discharge coil faults, and branch resonance risks, thus achieving full-state monitoring.

[0017] (2) This invention achieves high-precision extraction of the impedance components of the fundamental wave and the main harmonics (3rd, 5th, and 7th) by weighted average of weight coefficients and fast Fourier transform, thereby improving the sensitivity and reliability of fault detection.

[0018] (3) The present invention constructs a multi-state criterion library including voltage distribution ratio criterion, impedance change criterion, and harmonic impedance characteristic offset criterion. Through longitudinal trend comparison and horizontal phase comparison, abnormal impedance changes can be detected early, and early warning protection can be achieved.

[0019] (4) This invention introduces a random forest classifier: using a random forest model with more than 500 decision trees and inputting multidimensional features, it realizes automatic identification and classification of fault types, improving the accuracy and intelligence level of diagnosis. Attached Figure Description

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] Figure 1 This is a flowchart of a method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to the present invention. Figure 2 This is a diagram illustrating the composition of a high-voltage parallel capacitor internal fault impedance characteristic monitoring system according to the present invention. Detailed Implementation

[0022] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0023] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0024] Example 1 This embodiment describes a method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor, such as... Figure 1 As shown, the steps are as follows: S1: Collects signals of capacitor terminal voltage, series reactor terminal voltage, and branch current of the parallel capacitor branch.

[0025] Signal acquisition must achieve strictly synchronous sampling, using the same clock source to trigger analog-to-digital conversion, ensuring that voltage and current signals are captured at the same time to eliminate phase errors introduced by sampling time differences.

[0026] S2: After anti-aliasing filtering and analog-to-digital conversion of the acquired signal, a quasi-synchronous sampling algorithm is used to process it, and the fundamental sine amplitude and cosine amplitude of the voltage and current signals are calculated by weighted averaging using weighted coefficients.

[0027] Weighting coefficients of the quasi-synchronous sampling algorithm Based on the type of quadrature formula, the number of samples per single signal period, and the number of sampling periods, these values ​​are pre-calculated and stored. A single weighted summation of the real and imaginary parts of the voltage and current signals yields the fundamental sine amplitude *b* and the cosine amplitude *a*. The calculation formula is as follows: ; ; in, Sampling time The signal value, ω is the angular frequency.

[0028] S3: Based on the Fast Fourier Transform algorithm, the fundamental wave and the 3rd, 5th and 7th major harmonic components are separated, and the complex impedance values ​​of the fundamental wave and each harmonic frequency are calculated.

[0029] The complex impedance value includes a resistive component and a reactance component. The aforementioned resistance component is expressed as: ; The reactance component is expressed as: ; in, , These are the fundamental cosine amplitude and the sine amplitude of the voltage signal, respectively. , These are the fundamental cosine amplitude and sine amplitude of the current signal, respectively.

[0030] S4: Construct a multi-state impedance characteristic criterion library that includes normal state, fault occurrence state and fault expansion critical point. The criterion library includes voltage distribution ratio criterion, impedance change criterion, and harmonic impedance characteristic offset criterion. The real-time calculated complex impedance value is compared with the criterion library in the longitudinal trend and in the horizontal phase comparison to extract fault characteristic parameters.

[0031] Capacitive reactive power compensation devices used in substations typically employ a single-star connection with a symmetrical three-phase structure. Each phase of the capacitor bank contains multiple series segments, and each series segment consists of several capacitors connected in parallel.

[0032] The method for establishing the voltage distribution ratio criterion is as follows: The ratio of capacitor terminal voltage to series reactor terminal voltage at the fundamental frequency ,in, This represents the fundamental effective value of the capacitor terminal voltage. This represents the fundamental effective value of the voltage across the series reactor; multiple state thresholds are based on the design reactance rate of the series reactor. And capacitor bank tuning frequency setting.

[0033] For capacitor banks designed to suppress harmonics above the 5th order and with a reactance of 5%, the warning threshold for the fundamental voltage ratio is set to 24.19, and the tripping threshold is set to 24.9.

[0034] The impedance change criterion is calculated based on the single-phase impedance formula of a capacitor bank; for each phase... n Each series segment consists of a series segment, and each series segment is composed of a series segment. m The single-phase impedance of a capacitor bank consisting of capacitors connected in parallel under normal conditions is expressed as: ; When a single capacitor in a capacitor bank experiences a breakdown fault, the other capacitors in the same parallel section will discharge to the faulty capacitor, increasing the current flowing through it. The fuse will then blow, disconnecting the faulty capacitor. Based on the series-parallel relationship after the fault, the impedance of the capacitor bank at this point can be obtained, expressed as: ; The change in reactance of the capacitor bank before and after a fault is expressed as follows: ; The relative change in the reactance of the capacitor bank after a single capacitor breakdown fault is expressed as follows: ; The impedance change criterion is based on Set multi-level early warning thresholds within the range of 2.7% to 20%.

[0035] The method for establishing the harmonic impedance characteristic offset criterion is as follows: Under normal operating conditions of the capacitor bank, the reference harmonic impedance value corresponding to the main harmonic is acquired and stored. The reference harmonic impedance value includes the resistive component and the reactance component. Real-time calculation of harmonic impedance values ​​at major harmonic frequencies; Calculate the resistance offset coefficient and reactance offset coefficient of each harmonic impedance; Set the resistance and reactance offset thresholds for each harmonic frequency; When the resistance offset coefficient or reactance offset coefficient of any harmonic frequency exceeds its corresponding threshold, it is determined that the harmonic impedance characteristic has shifted.

[0036] Vertical trend comparison continuously tracks the change trajectory of the impedance parameter of the same branch over time, and adjusts the voltage distribution ratio. The impedance is continuously compared with preset thresholds in the criterion library, and the current impedance change is calculated and compared with multi-level thresholds for impedance change criteria in the criterion library. This analysis is then used to further analyze the data. By analyzing the rate of change, approximation speed, and trend pattern of values ​​relative to threshold boundaries, the longitudinal trend characteristics of voltage distribution ratios and impedance changes are extracted. This effectively captures progressive defects such as capacitor aging and slow inter-turn short circuits in reactors, enabling early fault warning. Lateral phase-to-phase comparison simultaneously compares the symmetry of the three-phase branches and calculates the voltage ratio of each phase. By considering the relative positional difference and phase dispersion with respect to the threshold, and based on the harmonic impedance characteristic offset criteria in the criterion library, the resistance offset coefficient and reactance offset coefficient of each harmonic impedance are calculated and compared with the preset offset threshold to determine the harmonic impedance offset and phase offset. This enables rapid and sensitive identification of sudden asymmetrical faults such as single capacitor breakdown and differentiation from system-side disturbances. This collaborative analysis mechanism deeply integrates the state evolution in the time dimension with the balance relationship in the spatial dimension, extracting and outputting fault feature parameters. The fault feature parameters include: voltage distribution ratio (including its longitudinal trend characteristics and lateral deviation characteristics), impedance change, harmonic impedance offset, and phase offset. This not only significantly improves the sensitivity and reliability of fault detection but also provides high-value, high-discrimination input features for subsequent random forest classifiers to accurately distinguish fault types, thus forming a complete closed loop from state perception to intelligent diagnosis.

[0037] S5: Based on fault feature parameters, a random forest classifier trained offline is used to identify fault types, including capacitor unit symmetry faults, series reactor inter-turn short circuit faults, discharge coil faults, and branch resonance faults.

[0038] The random forest classifier contains 500 decision trees, and the input feature dimensions include: voltage distribution ratio, impedance change, harmonic impedance offset, and phase offset.

[0039] Training and testing data are derived from three sources: electromagnetic transient simulations (such as PSCAD / EMTP platforms), measured data from laboratory prototype platforms, and some historical fault waveform recordings from the field (after anonymization), ensuring the diversity and authenticity of the data sources. A sample dataset of over 10,000 sets was constructed, comprehensively covering the main fault types: capacitor unit symmetry faults, series reactor inter-turn short-circuit faults (including inter-turn short-circuit faults of varying degrees), discharge coil faults, and branch resonance faults. Each type of fault sample includes variants with different severity levels and different initial operating conditions (such as different background harmonics and voltage fluctuations) to improve the model's generalization ability.

[0040] By performing grid search and cross-validation on key hyperparameters such as the number of decision trees, maximum depth, and feature subset size, the optimal parameter combination was determined, and the final model configuration was a random forest containing 500 decision trees.

[0041] A stratified sampling strategy was adopted, dividing the dataset into a training set (7,000 groups) and an independent test set (3,000 groups) in a 7:3 ratio to ensure that the sample ratio of various faults in the training set and the test set is consistent with that in the original dataset.

[0042] S6: Based on the fault identification results, trigger graded alarm signals or protection trip commands, and upload the fault data to the monitoring backend.

[0043] To further verify the effectiveness of the method of this invention, performance tests were conducted on a hybrid test platform combining a high-voltage, high-current laboratory with a real-time digital simulation system. This platform is designed to simulate real electrical environments and signal conditions, while possessing precise and controllable fault injection capabilities.

[0044] Specifically, the RTDS real-time simulation system simulates a 10kV distribution network; Fault injection unit: Programmable resistor box to simulate faults; Standard measuring equipment: Fluke 6105A power standard source; Environmental simulation chamber: temperature and humidity are controllable (-40℃~+85℃); The performance test results are shown in Table 1 below.

[0045] Table 1 Basic Performance Test Results

[0046] As shown in the table above, the measurement accuracy of 0.2% and the impedance calculation error of 0.52% provide a solid foundation for the accurate extraction of fault characteristics and the reliable comparison of multi-state criteria, greatly reducing the possibility of false alarms and missed alarms. The maximum fault response time of 86ms indicates that the system can complete diagnosis and issue commands in a very short time after a fault occurs, which is crucial for suppressing faults that may deteriorate rapidly, such as inter-turn short circuits in series reactors, and preventing the escalation of accidents. Therefore, the method of this invention performs excellently in both accuracy and speed, overcoming the limitations of traditional protection in terms of accuracy or response, and achieving accurate, rapid, and intelligent monitoring of internal faults and resonance risks in high-voltage parallel capacitors.

[0047] Example 2 This embodiment is a monitoring system for the internal fault impedance characteristics of a high-voltage parallel capacitor, employing the monitoring method for the internal fault impedance characteristics of a high-voltage parallel capacitor described in Embodiment 1. Figure 2 As shown, it consists of a signal acquisition module, a signal processing module, an impedance calculation module, a fault feature extraction module, a fault identification module, and an output and communication module.

[0048] The signal acquisition module is used to acquire signals of capacitor terminal voltage, series reactor terminal voltage and branch current of the parallel capacitor branch; it is used to implement step S1 in embodiment 1, which will not be described again here.

[0049] The signal processing module is used to perform anti-aliasing filtering and analog-to-digital conversion on the acquired signal, and to calculate the fundamental sine amplitude and cosine amplitude of the voltage and current signals using a quasi-synchronous sampling algorithm. The quasi-synchronous sampling algorithm uses pre-stored weighting coefficients for weighted averaging. The weighting coefficients are pre-calculated based on the type of quadrature formula, the number of samples per signal period, and the number of sampling periods. This module is used to implement step S2 in Example 1, and will not be described in detail here.

[0050] The impedance calculation module is used to separate the fundamental wave and the 3rd, 5th and 7th major harmonic components based on the fast Fourier transform algorithm, and to calculate the complex impedance values ​​of the fundamental wave and each harmonic frequency. The complex impedance values ​​include resistive and reactive components. It is used to implement step S3 in embodiment 1, which will not be described again here.

[0051] The fault feature extraction module constructs a multi-state impedance characteristic criterion library, which includes normal state, fault occurrence state, and fault expansion critical point, and includes voltage distribution ratio criterion, impedance change criterion, and harmonic impedance characteristic offset criterion. This module is used to compare the real-time calculated impedance value with the criterion library in a longitudinal trend and a transverse phase comparison to extract fault feature parameters. It is used to implement step S4 in embodiment 1, which will not be repeated here.

[0052] The fault identification module includes an offline-trained random forest classifier for identifying fault types based on the fault feature parameters. The fault types include capacitor unit symmetry faults, series reactor inter-turn short-circuit faults, discharge coil faults, and branch resonance faults. It is used to implement step S5 in Embodiment 1, which will not be described again here.

[0053] The output and communication module is used to trigger graded alarm signals or protection trip commands based on the fault identification results, and upload the fault data to the monitoring backend; it is used to implement step S6 in embodiment 1, which will not be described again here.

[0054] It should be noted that each module in the aforementioned high-voltage parallel capacitor internal fault impedance characteristic monitoring system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module. For specific limitations regarding the high-voltage parallel capacitor internal fault impedance characteristic monitoring system, please refer to the limitations of the high-voltage parallel capacitor internal fault impedance characteristic monitoring method described above; both have the same function and role, and will not be repeated here.

[0055] Therefore, the present invention adopts the above-mentioned method and system for monitoring the internal fault impedance characteristics of high-voltage parallel capacitors. Through high-precision synchronous acquisition, fundamental-harmonic complex impedance calculation based on quasi-synchronous sampling and FFT, fault feature extraction based on multi-criteria fusion, and intelligent classification and identification using random forest, it realizes full-state, highly sensitive, and rapid online monitoring and protection of high-voltage parallel capacitor branches from normal operation to various internal faults (including symmetrical faults) and resonance risks.

[0056] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor, characterized in that, Including the following steps: S1 collects the signals of capacitor terminal voltage, series reactor terminal voltage and branch current of the parallel capacitor branch; S2, after performing anti-aliasing filtering and analog-to-digital conversion on the acquired signal, uses a quasi-synchronous sampling algorithm to process it, and calculates the fundamental sine amplitude and cosine amplitude of the voltage and current signals by weighted averaging with weight coefficients; S3, based on the fast Fourier transform algorithm, separates the fundamental wave and the 3rd, 5th and 7th main harmonic components, and calculates the complex impedance values ​​of the fundamental wave and each harmonic frequency. S4. Construct a multi-state impedance characteristic criterion library that includes normal state, fault occurrence state and accident expansion critical point. The criterion library includes voltage distribution ratio criterion, impedance change criterion, and harmonic impedance characteristic offset criterion. Compare the real-time calculated complex impedance value with the criterion library in the longitudinal trend and in the horizontal phase comparison to extract fault characteristic parameters. S5. Based on the fault feature parameters, a random forest classifier trained offline is used to identify the fault types, including capacitor unit symmetry faults, series reactor inter-turn short circuit faults, discharge coil faults, and branch resonance faults. S6, based on the fault identification results, triggers graded alarm signals or protection trip commands, and uploads the fault data to the monitoring backend.

2. The method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 1, characterized in that, In step S2, the weight coefficients of the quasi-synchronous sampling algorithm Based on the type of quadrature formula, the number of samples per single signal period, and the number of sampling periods, these values ​​are pre-calculated and stored. The real and imaginary parts of the voltage and current signals are then weighted once to obtain the fundamental sine amplitude *b* and the cosine amplitude *a*. The calculation formula is as follows: ； ； in, Sampling time The signal value, ω is the angular frequency.

3. The method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 1, characterized in that, In step S3, the complex impedance value includes a resistive component and a reactance component. The resistance component is expressed as: ； The reactance component is expressed as: ； in, , These are the fundamental cosine amplitude and the sine amplitude of the voltage signal, respectively. , These are the fundamental cosine amplitude and sine amplitude of the current signal, respectively.

4. The method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 1, characterized in that, In step S4, the method for establishing the voltage distribution ratio criterion is as follows: The ratio of capacitor terminal voltage to series reactor terminal voltage at the fundamental frequency ,in, This is the effective value of the fundamental frequency of the capacitor terminal voltage. This represents the fundamental effective value of the voltage across the series reactor; multiple state thresholds are based on the design reactance rate of the series reactor. And capacitor bank tuning frequency setting.

5. The method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 4, characterized in that, For capacitor banks designed to suppress harmonics above the 5th order and with a reactance of 5%, the warning threshold for the fundamental voltage ratio is set to 24.19, and the tripping threshold is set to 24.

9.

6. The method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 1, characterized in that, The impedance change criterion in step S4 is calculated based on the single-phase impedance formula of the capacitor bank; for each phase... n Each series segment consists of a series segment, and each series segment is composed of a series segment. m The single-phase impedance of a capacitor bank consisting of capacitors connected in parallel under normal conditions is expressed as: ； in, This refers to the capacitance value of a single capacitor. Angular frequency; When a single capacitor in a capacitor bank experiences a breakdown fault, the other capacitors in the same parallel section will discharge to the faulty capacitor, increasing the current flowing through it. The fuse will then blow, disconnecting the faulty capacitor. Based on the series-parallel relationship after the fault, the impedance of the capacitor bank at this time can be expressed as: ； The change in reactance of the capacitor bank before and after a fault is expressed as follows: ； The relative change in the reactance of the capacitor bank after a single capacitor breakdown fault is expressed as follows: ； The impedance change criterion is based on Set multi-level early warning thresholds within the range of 2.7% to 20%.

7. The method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 6, characterized in that, In step S4, the method for establishing the harmonic impedance characteristic offset criterion is as follows: Under normal operating conditions of the capacitor bank, the reference harmonic impedance value corresponding to the main harmonic is acquired and stored; Real-time calculation of harmonic impedance values ​​at the main harmonic frequencies; Calculate the resistance offset coefficient and reactance offset coefficient of each harmonic impedance; Set the resistance and reactance offset thresholds for each harmonic frequency; When the resistance offset coefficient or reactance offset coefficient of any harmonic frequency exceeds its corresponding threshold, it is determined that the harmonic impedance characteristic has shifted.

8. The method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 1, characterized in that, In step S4, the fault characteristic parameters include: voltage distribution ratio, impedance change, harmonic impedance offset, and phase offset.

9. A method for monitoring the internal fault impedance characteristics of a high-voltage parallel capacitor according to claim 8, characterized in that, In step S5, the random forest classifier contains more than 500 decision trees, and the input feature dimensions include: voltage distribution ratio, impedance change, harmonic impedance offset, and phase offset.

10. A monitoring system for the internal fault impedance characteristics of a high-voltage parallel capacitor, employing the monitoring method for the internal fault impedance characteristics of a high-voltage parallel capacitor as described in any one of claims 1-9, characterized in that, include: The signal acquisition module is used to acquire signals of capacitor terminal voltage, series reactor terminal voltage, and branch current in the parallel capacitor branch. The signal processing module is used to perform anti-aliasing filtering and analog-to-digital conversion on the acquired signals, and to calculate the fundamental sine amplitude and cosine amplitude of the voltage and current signals using a quasi-synchronous sampling algorithm. The quasi-synchronous sampling algorithm uses pre-stored weighting coefficients for weighted averaging. The impedance calculation module is used to separate the fundamental wave and the 3rd, 5th and 7th major harmonic components based on the fast Fourier transform algorithm, and to calculate the complex impedance values ​​of the fundamental wave and each harmonic frequency. The fault feature extraction module is used to construct a multi-state impedance characteristic criterion library. The multi-state includes normal state, fault occurrence state and accident expansion critical point. The criterion library includes voltage distribution ratio criterion, impedance change criterion and harmonic impedance characteristic offset criterion. This module is used to compare the real-time calculated complex impedance value with the criterion library in the longitudinal trend and in the horizontal phase comparison to extract fault feature parameters. The fault identification module includes an offline-trained random forest classifier for identifying fault types based on the fault feature parameters. The fault types include capacitor unit symmetry faults, series reactor inter-turn short-circuit faults, discharge coil faults, and branch resonance faults. The output and communication module is used to trigger graded alarm signals or protection trip commands based on the fault identification results, and upload the fault data to the monitoring backend.

Images