DC system ground fault detection method and system

Abstract

The invention discloses a direct-current system ground fault detection method and system, and the method comprises the steps: respectively connecting resistors with equal resistance values between a positive bus and a negative bus of a direct-current system and the ground to form a symmetric resistance network, and collecting positive / negative voltage to ground and the grounding current of each branch in real time; extracting a high-frequency component through wavelet transform and energy entropy analysis to generate a frequency domain mark; performing wavelet transformation and energy entropy calculation on the grounding current signal, and extracting a high-frequency detail component; and a hierarchical collaborative decision is executed, immediate tripping is triggered by double marks, a delay recheck mechanism is started by single mark triggering, tripping is executed after continuous triggering, and a fault is positioned and reported. According to the technical system, the comprehensive identification speed of the grounding fault of the direct-current system is improved, full-process closed-loop protection of rapid sensing and identification, reliable decision making and accurate positioning is formed, and the operation safety of the direct-current power distribution system is improved.

Description

Technical Field

[0001] The present invention relates to the field of power detection technology, and in particular to a method and system for detecting ground faults in a DC system. Background Art

[0002] Traditional DC grounding protection methods primarily include the resistance bridge method, harmonic injection method, and transient analysis method. However, these methods all have limitations: the resistance bridge method relies on the symmetry of a balanced bridge, cannot detect high-resistance grounding faults, and is susceptible to interference from the system's ground capacitance; the harmonic injection method requires an additional signal source, which may affect system stability. The transient analysis method has high requirements for sampling rate and algorithm real-time performance and lacks an accurate mathematical model. The shortcomings of traditional DC grounding protection methods limit the ability to quickly identify DC system grounding faults, detect high-resistance faults, and locate faults. Therefore, a more efficient and reliable solution is needed to address the complex scenarios of DC system grounding faults. Summary of the Invention

[0003] In order to overcome the shortcomings of the above-mentioned traditional DC grounding protection method, which limits the ability to quickly identify DC system grounding faults, detect high-resistance faults, and locate faults, the present invention provides a DC system grounding fault detection method and system.

[0004] In order to solve the above problems, the present invention is implemented according to the following technical solutions:

[0005] A method for detecting ground faults in a DC system according to the first aspect of the present invention comprises the following steps: Step 1: connecting resistors of equal resistance between the positive and negative busbars of the DC system and the ground to form a symmetrical resistor network, wherein the symmetrical resistor network has a star-shaped symmetrical topology; Step 2: collecting the positive and negative voltage signals of the symmetrical resistor network and the ground current signals of each branch; calculating the voltage deviation to ground based on the positive and negative voltage signals, and comparing them with a preset threshold to generate a time domain fault mark; performing wavelet transform and energy entropy calculation on the ground current signals of each branch to extract high-frequency detail components; calculating the energy entropy of the detail coefficients at each scale, and comparing them with a dynamic threshold to generate a frequency domain fault mark; Step 3: performing hierarchical collaborative fault judgment and fault decision-making: when the time domain fault mark and the frequency domain fault mark are triggered at the same time, tripping is performed immediately; if only a single mark is triggered, a delayed review mechanism is started, and the fault is confirmed and tripping is performed after only a single mark is continuously triggered; Step 4: instructing to cut off the fault branch and reporting the type and location of the fault.

[0006] Preferably, the symmetrical resistor network quantifies the voltage deviation to ground as follows: ΔV=V + -V - =V dc ; Where ΔV is the voltage difference to ground, V +is the voltage of the positive pole of the symmetrical network to ground, V - is the voltage of the negative pole of the symmetrical network to ground, V dc is the DC bus voltage.

[0007] Preferably, the wavelet transform and energy entropy calculation convolute the ground current signal through a set of scaled and translated wavelet functions to obtain local characteristics of the signal at different scales and times; the wavelet transform separates the high-frequency details and low-frequency approximation of the signal through multi-scale decomposition.

[0008] Preferably, in the wavelet transform, the energy entropy can be used to evaluate the complexity of the signal by calculating the distribution of the wavelet transform coefficients; the energy entropy calculation formula is: Among them, S is energy entropy, j is the number of layers, E j is the detail coefficient energy E of the jth layer j , E total is the total energy.

[0009] Preferably, the fault decision includes: the time domain fault criterion is ΔV>V_th, and the duration is >20ms; wherein ΔV is the voltage difference to ground, and V_th is the first threshold; the frequency domain fault criterion is S>S_th, and the high-frequency component energy proportion is >30%; wherein S is the energy entropy, and S_th is the second threshold; the delayed review mechanism is that when the time domain fault criterion and the frequency domain fault criterion are met at the same time, the circuit breaker trips immediately; when only the time domain fault criterion or the frequency domain fault criterion is met, a secondary verification is performed after a delay of 10ms.

[0010] On the other hand, the present invention provides a DC system grounding fault detection system, which includes: a symmetrical resistor network module for constructing a symmetrical resistor network; a sensor module for detecting the positive and negative pole voltages to ground and the current of each branch grounding wire; a signal adjustment module for suppressing high-frequency noise and eliminating AC coupling interference; a control module, which includes a data acquisition unit, a time domain processing module and a frequency domain processing module; and a relay module, which is used for fault isolation and alarm.

[0011] Preferably, the symmetrical resistor network meets the following conditions: the resistance range is 10kΩ-100kΩ, the power level is 1.5 times the rated value; metal film resistors are used and encapsulated in a shielding box, and the temperature drift coefficient is <50ppm / ℃; the ground capacitor and the symmetrical resistor network form a low-pass filter.

[0012] Preferably, the sensor module includes a voltage sensor and a current sensor; the signal conditioning module includes a low-pass filter and a power frequency notch filter; and the sampling rate of the data acquisition unit is ≥200 kS / s.

[0013] Compared with existing technologies, the present invention achieves the following benefits: By constructing a star-shaped symmetrical monitoring topology through a symmetrical resistor network, the sensitivity of detecting positive and negative voltage deviations relative to ground is increased by over 40%, addressing the large blind spot detection issue of traditional voltage balancing methods for single-point ground faults. A dual-domain detection technique, combining wavelet transform with dynamic energy entropy analysis, precisely extracts high-frequency detail components through multi-scale decomposition. Combined with a dynamic threshold algorithm, this technology improves recognition accuracy from less than 70% with traditional methods to 95%. A hierarchical collaborative decision-making mechanism innovatively integrates time-domain threshold triggering with frequency-domain energy mutation signatures. A dual-flag verification mechanism reduces the false operation rate to below 0.5%, while a delayed review strategy ensures reliability when triggering with a single signature. Combined with synchronous branch current monitoring and a topology analysis algorithm, the faulty branch can be accurately located in a short period of time, with a positioning error of less than 0.5 meters. This technical system accelerates the comprehensive identification speed of DC system ground faults by over 60% compared to traditional methods, creating a full-process closed-loop protection system with rapid perception and identification, reliable decision-making, and precise positioning, thereby enhancing the operational safety of DC distribution systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein:

[0015] Figure 1 It is a schematic diagram of the steps of a DC system grounding fault detection method of the present invention; DETAILED DESCRIPTION

[0016] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0017] like Figure 1 As shown, a method for detecting a ground fault in a DC system according to the first aspect of the present invention comprises the following steps:

[0018] Step 1: Connect resistors of equal resistance between the positive and negative busbars of the DC system and the ground to form a symmetrical resistor network, which has a star-shaped symmetrical topology. Step 2: Collect the positive and negative voltage signals of the symmetrical resistor network and the grounding current signals of each branch. Calculate the voltage deviation to the ground based on the positive and negative voltage signals, and compare them with the preset threshold to generate a time domain fault mark. Perform wavelet transform and energy entropy calculation on the grounding current signal of each branch to extract the high-frequency detail component. Calculate the energy entropy of the detail coefficient of each scale, compare it with the dynamic threshold, and generate a frequency domain fault mark. Step 3: Perform hierarchical collaborative fault judgment and fault decision-making: when the time domain fault mark and the frequency domain fault mark are triggered at the same time, tripping is executed immediately. If only a single mark is triggered, start the delayed review mechanism, and confirm the fault and trip after only a single mark is continuously triggered. Step 4: Instruct to cut off the fault branch and report the type and location of the fault.

[0019] In one embodiment, a symmetrical resistor network is used to quantify the voltage deviation to ground, thereby achieving preliminary fault detection.

[0020] Assume that the voltage of the positive pole of the DC system to the ground is V+ and the voltage of the negative pole of the DC system to the ground is V-, and the capacitance of the system to the ground is C.

[0021] The symmetrical resistor network consists of a positive resistor to ground R+ and a negative resistor to ground R-, and satisfies R + =R - =R eq .

[0022] In one embodiment, under normal conditions, when the system is symmetrical, the voltages of the positive and negative electrodes to ground satisfy:

[0023] At this time, the voltage difference to ground ΔV=V + -V - =2V + =2V - =V dc , but the total leakage current of the system is zero.

[0024] In the fault state: Assuming that the positive electrode is grounded through the fault resistance, the equivalent circuit equation is: Where Rf is the fault resistance.

[0025] Substitute the above equivalent circuit equation into R + =R - =R eq and V + -V - =V dc The solution is:

[0026]

[0027] After simplifying the above formula, we can obtain: The voltage of the positive pole of the symmetrical network to the ground: Symmetrical network negative pole to ground voltage: Voltage difference to ground: ΔV = V + -V - =V dc ; Among them, V dc is the DC bus voltage.

[0028] In a preferred embodiment, the relationship between the symmetrical resistor network and the ground voltage deviation is: ΔV = V + -V - =V dc ; Where ΔV is the voltage difference to ground, V + is the voltage between the positive pole of the symmetrical network and the ground, V- is the voltage between the negative pole of the symmetrical network and the ground, V dc Where: ΔV = V + -V - =V dc Heng is established.

[0029] The parameter design and sensitivity analysis when constructing a star symmetrical topology are as follows: Fault resistance R f What changes is the ground voltage V + and V - The absolute value, not the difference.

[0030] When in low resistance state: R f < <R eq ; Symmetrical network positive pole to ground voltage:

[0031] When in high resistance state: R f >>R eq ; Symmetrical network negative pole to ground voltage:

[0032] Calculate power loss: Symmetrical resistor R eq Need to meet Among them, P loss is the power loss, P rated is the rated power.

[0033] The symmetrical resistor network and the ground capacitor C form a low-pass filter. The calculation formula for anti-capacitive interference is: fc>10f noise , where f noise is the system noise frequency, and fc is the cutoff frequency.

[0034] In a preferred embodiment, the wavelet transform and energy entropy calculation convolute the ground current signal through a set of scaled and translated wavelet functions to obtain the local characteristics of the signal at different scales and times; the wavelet transform separates the high-frequency details and low-frequency approximation of the signal through multi-scale decomposition.

[0035] It is understandable that the wavelet transform is a time-frequency analysis tool that can decompose a signal into frequency components at multiple scales for localized analysis in time and frequency. Compared with the traditional Fourier transform, the wavelet transform includes the continuous wavelet transform and the discrete wavelet transform. The wavelet transform can provide the instantaneous frequency characteristics of the signal, and therefore has advantages when processing signals with sudden or transient changes, such as ground fault signals. Energy entropy is an indicator used to quantify the complexity of a signal, and is particularly suitable for detecting transient changes in signals. In the wavelet transform, energy entropy can be used to evaluate the complexity of the signal by calculating the distribution of wavelet transform coefficients. The application of energy entropy in fault detection can determine whether abnormal changes have occurred in the signal, thereby improving the accuracy and efficiency of judgment.

[0036] In a preferred embodiment, in the wavelet transform, the energy entropy can be used to evaluate the complexity of the signal by calculating the distribution of the wavelet transform coefficients; the energy entropy calculation formula is: Among them, S is energy entropy, j is the number of layers, E j is the detail coefficient energy E of the jth layer j , E total is the total energy.

[0037] The local features of the signal at different scales and time positions are obtained by convolution through a set of scaled and translated wavelet functions. This process is called continuous wavelet transform. The mathematical expression of continuous wavelet transform is as follows:

[0038] Where W(a,b) is a set of scaled and translated wavelet functions (a, b) in continuous wavelet transform, and x(t) is the signal.

[0039] Wavelet base selection and multi-resolution analysis: Ground fault transient signals contain high-frequency components (such as the rapid rising edge of arc discharge), while normal noise is mostly low-frequency or stationary. Wavelet transforms separate the signal's high-frequency details (Detail Coefficients) from its low-frequency approximations (Approximation Coefficients) through multi-scale decomposition.

[0040] In one embodiment, the preferred wavelet basis is db4, which has compact support and orthogonality, and the waveform matches the rising edge of the fault transient signal; the preferred decomposition level J=5: taking into account both high-frequency resolution and computational efficiency.

[0041] The discrete wavelet transform decomposes the signal into a set of discrete high-frequency and low-frequency sub-bands. Through multi-level decomposition, the discrete wavelet transform can extract the multi-scale information of the signal, thereby effectively processing and analyzing the mutation characteristics in the signal. x(t) is decomposed into the approximate coefficient A J And detail coefficients D1, D2, ..., DJ: the calculation formula is

[0042] Energy entropy is calculated as the energy E of the detail coefficient of the jth layer j =∑ k |D j (K)| 2 , total energy The energy entropy S reflects the degree of disorder of energy distribution and is calculated as follows: During a fault, the proportion of high-frequency energy increases and S increases significantly.

[0043] Under normal circumstances, the energy entropy of a signal is low because the signal changes smoothly and regularly. However, when a ground fault occurs, the system's current or voltage will experience sudden changes or fluctuations, which manifests as a high concentration of energy in the wavelet transform coefficients. The calculated energy entropy will also rise rapidly, indicating that an abnormal fault has occurred in the system.

[0044] In a preferred embodiment, the fault decision includes: the time domain fault criterion is ΔV>V_th, and the duration is >20ms; wherein ΔV is the voltage difference to ground, and V_th is the first threshold; the frequency domain fault criterion is S>S_th, and the high-frequency component energy proportion is >30%; wherein S is the energy entropy, and S_th is the second threshold; the delayed review mechanism is that when the time domain fault criterion and the frequency domain fault criterion are simultaneously met, the circuit breaker trips immediately; when only the time domain fault criterion or the frequency domain fault criterion is met, a secondary verification is performed after a delay of 10ms.

[0045] It is understandable that under normal conditions, the entropy value S of historical data under normal conditions is calculated. normal The mean μ and standard deviation σ of . Set S th =μ-4σ, ensuring false alarm rate <0.1%.

[0046] The process of generating and triggering a time domain fault flag is as follows:

[0047] S10, reading voltage signals of the DC system positive electrode voltage to ground (V+) and the DC system negative electrode voltage to ground (V-).

[0048] S20, by formula and Get ΔV + and ΔV-voltage difference to ground, ΔV + is the voltage difference between the positive electrode and the ground, and ΔV- is the voltage difference between the negative electrode and the ground.

[0049] S30, compare ΔV + / ΔV- and ΔV th .

[0050] Perform primary fault judgment based on the or Trigger the primary fault warning; the secondary fault judgment criterion is energy entropy S>S th , confirm the fault;

[0051] The processing time is less than 20ms, where ΔV th is the voltage excursion threshold; S th is the energy entropy threshold.

[0052] S40: If the limit is exceeded, the time domain fault flag is triggered.

[0053] The voltage deviation ΔV to ground can be calculated in real time to execute the primary judgment criteria.

[0054] The process of generating frequency domain fault flag and triggering is as follows:

[0055] S11, cache 1ms window current signal.

[0056] S12, perform 5-layer db4 wavelet decomposition.

[0057] S13, extract 5 layers of detail coefficients.

[0058] S14, calculate the energy entropy S.

[0059] S15, compare S with the threshold S th .

[0060] S16, if exceeded, triggers the frequency domain fault flag.

[0061] Perform wavelet transform and energy entropy calculation to realize high resistance fault identification.

[0062] On the other hand, the present invention provides a DC system grounding fault detection system, which includes: a symmetrical resistor network module for constructing a symmetrical resistor network; a sensor module for detecting the positive and negative pole voltages to ground and the current of each branch grounding wire; a signal adjustment module for suppressing high-frequency noise and eliminating AC coupling interference; a control module, which includes a data acquisition unit, a time domain processing module and a frequency domain processing module; and a relay module, which is used for fault isolation and alarm.

[0063] It is understandable that the symmetrical resistor network module is encapsulated in a shielded box to avoid electromagnetic interference. Low-temperature drift resistors (such as metal film resistors with a temperature drift coefficient of <50ppm / ℃) are selected.

[0064] Furthermore, the symmetrical resistor network meets the following conditions: the resistance range is 10kΩ-100kΩ, the power level is 1.5 times the rated value; metal film resistors are used and encapsulated in a shielding box, and the temperature drift coefficient is <50ppm / ℃; the ground capacitor and the symmetrical resistor network form a low-pass filter.

[0065] Furthermore, the sensor module includes a voltage sensor and a current sensor; the signal conditioning module includes a low-pass filter and an industrial frequency trap; the sampling rate of the data acquisition unit is ≥200kS / s. The sensor module is provided with a voltage sensor and a current sensor for detecting the voltage between the positive and negative poles and the ground, and the current of each branch grounding wire. The voltage sensor is installed at the positive pole and the negative pole to the ground, and the current sensor is installed at the grounding wire of each branch to locate the fault branch. In one embodiment, the data acquisition unit in the control module is an AD converter, the AD converter has a resolution of 16 bits (such as AD7606), and a sampling rate of 200kS / s, which can meet the requirements of wavelet transform analysis. And synchronous sampling is performed: the positive and negative pole voltage signals are synchronously collected to reduce phase errors.

[0066] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Therefore, any modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A method for detecting a ground fault in a DC system, characterized in that: The following steps are involved: Step 1: Connect resistors of equal resistance between the positive and negative busbars of the DC system and the ground to form a symmetrical resistor network, wherein the symmetrical resistor network has a star-shaped symmetrical topology. Step 2: collecting the positive and negative voltage signals of the symmetrical resistor network and the ground current signals of each branch; Calculate the voltage deviation to ground based on the positive and negative voltage signals, compare it with the preset threshold, and generate a time domain fault flag; Perform wavelet transform and energy entropy calculation on the grounding current signals of each branch to extract high-frequency detail components; calculate the energy entropy of detail coefficients at each scale, compare it with the dynamic threshold, and generate frequency domain fault signs; Step 3: Conduct hierarchical collaborative fault diagnosis and decision-making: When the time domain fault flag and the frequency domain fault flag are triggered simultaneously, tripping is performed immediately; If only a single flag is triggered, the delayed review mechanism is activated, and the fault is confirmed and the circuit breaker trips after only a single flag is continuously triggered; Step 4: Instruct to cut off the fault branch and report the type and location of the fault.

2. A DC system ground fault detection method according to claim 1, characterized in that: The relationship between the symmetrical resistor network and the ground voltage deviation is: ΔV = V + -V - =V dc ; Where ΔV is the voltage difference to ground, V + is the voltage of the positive pole of the symmetrical network to ground, V - is the voltage of the negative pole of the symmetrical network to ground, V dc is the DC bus voltage.

3. A DC system ground fault detection method according to claim 1, characterized in that: The wavelet transform and energy entropy calculation are used to convolve the ground current signal through a set of scaled and translated wavelet functions to obtain local features of the signal at different scales and times; The wavelet transform separates the high-frequency details and low-frequency approximations of the signal through multi-scale decomposition.

4. A DC system ground fault detection method according to claim 1, characterized in that: In the wavelet transform, the energy entropy can be used to evaluate the complexity of the signal by calculating the distribution of the wavelet transform coefficients; The energy entropy calculation formula is: Among them, S is energy entropy, j is the number of layers, E j is the detail coefficient energy E of the jth layer j , E total is the total energy.

5. A DC system ground fault detection method according to claim 1, characterized in that: The fault decision-making includes: The time domain fault criterion is ΔV>V_th and the duration is >20ms; Wherein, ΔV is the voltage difference to ground, and V_th is the first threshold; The frequency domain fault judgment criteria are S>S_th, and the high-frequency component energy ratio>30%; Where S is the energy entropy, S_th is the second threshold; The delayed review mechanism is that when the time domain fault criterion and the frequency domain fault criterion are simultaneously met, the circuit breaker trips immediately; when only the time domain fault criterion or the frequency domain fault criterion is met, a secondary verification is performed after a delay of 10ms.

6. A DC system ground fault detection system, based on the DC system ground fault detection method according to any one of claims 1 to 5, characterized in that: The fault detection system comprises: Symmetrical resistor network module, used to build a symmetrical resistor network; Sensor module, used to detect the positive and negative voltages to ground and the grounding currents of each branch; Signal conditioning module, used to suppress high-frequency noise and eliminate AC coupling interference; A control module, comprising a data acquisition unit, a time domain processing module, and a frequency domain processing module; Relay module, the relay module is used for fault isolation and alarm.

7. A DC system ground fault detection system according to claim 6, characterized in that: The symmetrical resistor network meets the following conditions: The resistance range is 10kΩ-100kΩ, and the power level is 1.5 times the rated value; Use metal film resistors and encapsulate them in a shielded box, with a temperature drift coefficient of <50ppm / ℃; The capacitor to ground and the symmetrical resistor network form a low-pass filter.

8. A DC system ground fault detection system according to claim 1, characterized in that: The sensor module includes a voltage sensor and a current sensor; The signal conditioning module includes a low-pass filter and a power frequency notch filter; The sampling rate of the data acquisition unit is ≥200 kS / s.

Images