Electric power protection system capable of rapidly isolating short-circuit fault

Abstract

The invention relates to an electric power protection system for quickly isolating a short-circuit fault, which can effectively capture and identify the short-circuit fault within millisecond-level time at the initial stage of the fault through the combination of the current change rate and margin / kurtosis, thereby shortening the fault detection response time and reducing the fault misjudgment and missed judgment probability. In a normal mode, simple judgment is realized only by using a current change rate threshold, so that the processing burden in a non-fault period is greatly reduced; only when an obvious sudden current change occurs, double criteria of a margin factor and a kurtosis ratio of an enhancement mode are activated to carry out deep identification, so that an excessive protection behavior is effectively avoided; a trip control mode combining two-state / fuzzy logic is adopted, a trip switch is accurately matched through an incidence matrix and a fault direction vector, rapid isolation is achieved, the risk of further diffusion of faults can be reduced, mistaken disconnection of irrelevant nodes is avoided, and the minimum power failure range is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid fault protection, and in particular to a power protection system for rapidly isolating short-circuit faults. Background Art

[0002] With the continuous expansion of the scale of power systems and the widespread access of distributed power sources in distribution networks, how to achieve fast, stable and highly fault-tolerant power protection after a fault occurs has become a key issue that needs to be urgently addressed in the field of power engineering.

[0003] Especially in diverse scenarios such as flexible DC distribution networks and conventional AC distribution networks, fault response speed, detection accuracy, and system robustness are of great significance for ensuring the continuity and security of power supply.

[0004] In the fault protection of flexible DC distribution networks, existing technologies can effectively identify the sudden change characteristics of fault currents by configuring current-limiting reactors at both ends of the positive and negative lines and combining them with high-speed sampling and filtering processing. Criteria such as margin factors and kurtosis coefficients can be used to determine the presence of faults within and outside the area and the fault polarity. At the same time, based on the operating characteristics of conventional distribution networks, studies have proposed using matrix methods to dynamically correct the topology and fault information of the distribution network, thereby achieving rapid fault location and isolation.

[0005] These two approaches show certain advantages in the cases of sharp increase in short-circuit current and complex multi-node topology, but a single method still has problems such as insufficient response flexibility and susceptibility to external conditions when facing complex working conditions such as noise interference, multiple concurrent faults, frequent changes in network topology, and strong random disturbances.

[0006] Since both flexible DC and traditional distribution networks may experience large fault current shocks within milliseconds, and are affected by factors such as multi-source grid-connected harmonics, communication delays, and changes in branch topology, existing technologies are still insufficient in capturing sudden changes in signals at high speed, distinguishing fault attributes, and dynamically isolating faults. The main manifestations are: it is difficult to balance rapidity and high anti-interference capabilities, the fault tolerance mechanism is weak, and there is a strong reliance on single-end or dual-end communication, which can easily lead to a decrease in detection accuracy and protection reliability under extreme working conditions (such as multiple concurrent faults or highly noisy environments). Therefore, there is an urgent need for a comprehensive power protection system that can integrate high-speed fault judgment, enhanced filtering, fault-tolerant logic, and adaptive control to meet the fault self-recovery needs of modern power grids in multiple dimensions and scenarios. Summary of the Invention

[0007] Aiming at the problems of redundant complexity, high cost and susceptibility of judgment criteria to interference and resulting in malfunction in existing technologies, a power protection system with rapid isolation of short-circuit faults is proposed.

[0008] The technical solution of the present invention is: a power protection system for rapidly isolating short-circuit faults, comprising a detection and acquisition module, a data processing module, a fault identification module and a fault isolation execution module;

[0009] The detection and acquisition module collects the current and voltage signals of the flexible DC distribution line and transmits them to the data processing module via optical fiber;

[0010] The data processing module is used to perform a first-level filtering on the received signal to achieve fast and smooth processing, and obtain pre-processed data under normal working conditions. When the pre-processed data is abnormal, that is, exceeds the data threshold under normal working conditions, a second-level filtering is triggered to obtain pre-processed data and predicted data under suspected fault conditions. The processed data is sent to the fault identification module;

[0011] The fault identification module, based on the processed data obtained by the data processing module, triggers fault identification through a combination of the current change rate criterion and the current root mean square change rate criterion. Once fault identification is triggered, the margin factor criterion is used to evaluate the dynamic fluctuation amplitude of the voltage monitoring point in a short period of time to locate the fault section. At the same time, the kurtosis ratio criterion is used to determine the fault polarity and distinguish between positive, negative, and inter-pole faults.

[0012] The fault isolation execution module receives the judgment result of the fault identification module and executes fault isolation according to the judgment result.

[0013] Preferably, the detection and acquisition module includes a high-sensitivity Hall current sensor connected in series on the positive and negative busbars of the current-limiting inductor series circuit, and a high-precision voltage sensor arranged at each end of the inductor; the sampling frequency is set to 12-15kHz, and anti-interference optical fiber transmission is adopted.

[0014] Preferably, the first-level filtering adopts a Savitzky–Golay filter to perform low-order polynomial least squares fitting in a narrow sliding window, which on the one hand locally averages the random noise and on the other hand completely retains the 0th to pth order derivatives of the signal, thereby smoothing the waveform without weakening the spike mutation characteristics.

[0015] Preferably, the secondary filtering, i.e., the sampled improved Kalman filter, performs more in-depth processing on the voltage and current signals. The specific method is as follows: the state space model of the system signal is expressed as:

[0016] Equation of state:

[0017] X k =F k X k-1 +w k-1 Measurement equation:

[0018] Z k =Hk X k +v k

[0019] Where, X k is the system state vector at the kth moment, including the true values ​​of the voltage and current signals; F k is the state transfer matrix, representing the change of state from time k-1 to time k; Z k is the observation signal vector at the kth moment, that is, the measurement value after the first level of filtering preprocessing; H k is the observation matrix, which maps the true state to the measurement space; w k-1 、v k are process noise and observation noise respectively, assuming that they satisfy Gaussian distribution and are independent of each other: w k-1 ~N(0,Q k-1 ),v k ~N(0,R k ), where Q k-1 is the covariance matrix of process noise; R k is the covariance matrix of the observation noise;

[0020] The improved Kalman filter calculates and updates the prior error covariance matrix P in real time kk-1 and the Kalman gain K k , to achieve state prediction and correction:

[0021] Prediction steps:

[0022]

[0023] in is the prior state estimate at time k, is the posterior result at time k-1;

[0024] F k is the state transfer matrix;

[0025] P kk-1 is the prior error covariance at time k, P k-1k-1 is the posterior error covariance of the previous moment;

[0026] Q k-1 is the process noise covariance;

[0027] Update steps:

[0028]

[0029] P kk =(IK k H k )P kk-1

[0030] where K k is the Kalman gain, which weighs the reliability of prediction and observation, is the transpose of the observation matrix, R k is the measurement noise covariance matrix;

[0031] and P kk is the posterior estimate and error covariance after observation correction; H k is the observation matrix, I is the identity matrix;

[0032] By adjusting the measurement noise covariance R k and process noise covariance Q k When arc interference, strong EMI or sensor temperature drift occurs, the filter automatically reduces the weight of the observation to keep the estimation stable; the sensitivity is quickly restored after the interference subsides.

[0033] Preferably, the combined trigger mechanism is:

[0034]

[0035] On fast roads, the current change rate |di / dt| is triggered immediately when it exceeds the current change rate threshold K1, and the response time is <0.2ms;

[0036] Backup path: If the current change rate |di / dt| does not exceed the current change rate threshold K1, monitor the current RMS change rate Δrms for M = 2 to 3 consecutive windows; once all reach the current RMS change rate threshold K2, the protection is also triggered.

[0037] Preferably, the current change rate threshold K1 is obtained by the following five steps:

[0038] 1) Short circuit steep slope estimation:

[0039] According to the line rated phase voltage U nom and the minimum equivalent inductance L min Estimating the upper limit of the short circuit initial slope

[0040] Take K rated 30% to 50% is used as the sensitivity benchmark for high-resistance grounding faults;

[0041] 2) Calculate the noise margin:

[0042] Statistical 24h normal operation waveform current change rate standard deviation σ di / dt , and set

[0043] Noise margin K noise =k σ σ di / dt, k σ＝ 6 to 8 to ensure that the probability of false operation is less than 0.1%;

[0044] 3) Dynamic range of the measurement chain:

[0045] Combined with transformer or Hall sensor full scale I FS and ADC sampling period T s , calculate the upper limit of the observable slope of the measurement chain If K sensor If it is lower than other items, it can be compensated by increasing the sampling rate or changing the sensor;

[0046] 4) Calculate the high resistance fault sensitivity coefficient:

[0047] Take K HIF =βK rated ,β=0.3~0.5, to take into account the capture of high ground resistance or arc high resistance fault;

[0048] 5) Threshold synthesis

[0049] The final threshold is the largest of the three, plus a design margin of ε = 10% to 20%:

[0050] K1=(max{K noise ,K HIF ,K sensor})(1+ε)

[0051] Finally, the current change rate threshold K1 is obtained.

[0052] Preferably, the current root mean square change rate threshold K2 is obtained by:

[0053] The root mean square change rate is the change in the effective value of the current in the kth sliding window I rms (k) minus the change in the effective value of the current in the k-1th sliding window I rms (k-1), and divided by the window width T w , and then normalize the amplitude increment to the rate of change, specifically:

[0054]

[0055] Among them, the effective value of current

[0056]

[0057] N is the number of sampling points involved in the calculation, is the instantaneous value of the current at the kN+jth sampling point.

[0058] Statistical standard deviation of the effective current value under normal working conditions σ Irms , take the current RMS change rate threshold K2 = kσ σ Irms , k σ =4~6.

[0059] The beneficial effects of the present invention are as follows: the power protection system for rapid isolation of short-circuit faults of the present invention, through the combination of current change rate and margin / kurtosis, the system can effectively capture and identify short-circuit faults within milliseconds at the initial stage of the fault, which not only shortens the fault detection response time, but also reduces the probability of fault misjudgment and missed judgment; in normal mode, only the current change rate threshold is used to achieve simple judgment, which greatly reduces the processing burden in the non-fault period; only when a significant current mutation occurs, the margin factor and kurtosis ratio dual criteria of the enhanced mode are activated for in-depth identification, effectively avoiding excessive protection behavior; a tripping control method combining binary / fuzzy logic is adopted, and the tripping switch is accurately matched with the fault direction vector through the correlation matrix and rapid isolation is achieved, which can not only reduce the risk of further spread of the fault, but also avoid the erroneous disconnection of irrelevant nodes, and achieve a minimized power outage range. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Schematic diagram of the overall structure of the system of the present invention;

[0061] Figure 2 This is a two-stage filtering flow chart for data processing of the present invention;

[0062] Figure 3 This is a fault identification dual-mode criterion logic diagram of the present invention;

[0063] Figure 4 This is a flow chart of the fault isolation execution module of the present invention. DETAILED DESCRIPTION

[0064] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0065] This paper proposes a high-response short-circuit fault self-restoring power protection system with an optimized strategy. By adopting dual-mode dynamic criterion switching technology and an adaptive fault isolation mechanism, the protection system's economy, reliability, and response efficiency are significantly improved. Specifically, it includes the following modules and functional implementation methods:

[0066] The present invention adopts modular design, such as Figure 1 As shown in the figure, the system mainly includes a detection and acquisition module, a data processing module, a fault identification module and a fault isolation execution module. Each module realizes efficient and reliable overall operation of the protection system through clear functional division of labor.

[0067] The detection and acquisition module collects the current and voltage signals of the flexible DC distribution line and transmits them to the data processing module via optical fiber;

[0068] The data processing module is used to perform a primary filtering on the received signal to achieve fast and smooth processing, obtain pre-processed data under normal working conditions, and trigger a secondary filtering when the pre-processed data is abnormal (exceeds the data threshold under normal working conditions) to obtain pre-processed data and predicted data under suspected fault conditions. The processed data is sent to the fault identification module;

[0069] The fault identification module, based on the acquired processed data, determines whether fault identification is triggered by the current change rate. Once fault identification is triggered, it uses the margin factor criterion to evaluate the dynamic fluctuation amplitude of the voltage monitoring point in a short period of time to locate the fault section. The kurtosis ratio criterion determines the fault polarity and distinguishes between positive, negative, and inter-pole faults.

[0070] The fault isolation execution module receives the judgment result of the fault identification module and executes fault isolation according to the judgment result.

[0071] The detection and acquisition module adopts an optimized high-speed data sampling design with a sampling frequency set to 12-15kHz. Compared with traditional solutions, it effectively reduces the burden of data processing while ensuring sufficient response speed and accuracy.

[0072] Specifically, the system connects high-sensitivity Hall effect current sensors to the positive and negative busbars of the current-limiting reactor series circuit, and places a high-precision voltage sensor at each end of the reactor. These two sensors sample synchronously, and the measured current and voltage data are transmitted via optical fiber to a data processing module, providing accurate and reliable input data for filtering, current rate of change, margin, and kurtosis analysis.

[0073] During specific implementation, current and voltage data are transmitted via optical fiber to the central processing unit located in the protection and measurement control screen in the main control room, avoiding the interference risk of traditional copper wire transmission and further improving the quality of the collected signals.

[0074] The present invention constructs a processing strategy with dual filtering and redundant fault tolerance capabilities in the data processing module.

[0075] The system uses a two-stage filtering system to process basic noise and abnormal disturbances respectively:

[0076] When the protection system of the present invention is in normal operation, it uses Savitzky-Golay filter as the first-level filtering method to quickly smooth the voltage and current signals collected in real time under normal working conditions. Figure 2 The two-stage filtering flow chart of data processing is shown in the following figure. In the specific implementation process, for any sampling signal sequence:

[0077] {x i}={x1,x2,...,x n}

[0078] Select a sliding window (the length is an odd number 2m+1, where m is a natural number) and perform a low-order (usually 2nd to 5th order) polynomial fitting on the signal data points in each window:

[0079]

[0080] Where, is the smoothed value after fitting within the window; p is the polynomial order, satisfying p<2m+1; a0, a1, ..., a p is the polynomial fitting coefficient; c is the window center position value.

[0081] Fit and calculate the window center point C position value by least squares method The filtered value The fitting filter value of the center point C position value Replace the original measurement value and realize signal smoothing.

[0082] The Savitzky–Golay filter employed in this paper performs a low-order polynomial least-squares fit within a narrow sliding window, locally averaging random noise while fully preserving the signal's 0th to pth order derivatives. This smoothes the waveform without diminishing spikes or sudden changes. This method effectively preserves both spikes and sudden changes in the signal, making it particularly suitable for capturing and denoising transients in the early stages of faults in power systems.

[0083] When the pre-processed data shows an anomaly (exceeding the data threshold under normal operating conditions), the present invention can be determined as a suspected fault state. At this time, the secondary filtering method, namely the improved Kalman filter, is activated to perform more in-depth processing of the voltage and current signals. In the specific implementation process, the state space model of the system signal is expressed as:

[0084] Equation of state:

[0085] X k =F k X k-1 +w k-1

[0086] Measurement equation:

[0087] Z k =H k H k +v k

[0088] Where, X kis the system state vector at the kth moment (such as the true value of the voltage and current signals); F k is the state transfer matrix, representing the change of state from time k-1 to time k; Z k is the observed signal vector at the kth moment (i.e., the measured value after the first level of filtering preprocessing); H k is the observation matrix, which maps the true state to the measurement space; w k-1 、v k are process noise and observation noise respectively, assuming that they satisfy Gaussian distribution and are independent of each other: w k-1 ~N(0,Q k-1 ), v k ~N(0,R k ), where Q k-1 is the covariance matrix of process noise; R k is the covariance matrix of the observation noise;

[0089] The improved Kalman filter calculates and updates the prior error covariance matrix P in real time kk-1 and the Kalman gain K k , to achieve state prediction and correction:

[0090] Prediction steps:

[0091]

[0092] in is the prior state estimate at time k, is the posterior result at time k-1;

[0093] F k is the state transfer matrix;

[0094] P kk-1 is the prior error covariance at time k, P k-1k-1 is the posterior error covariance of the previous moment;

[0095] Q k-1 is the process noise covariance;

[0096] Update steps:

[0097]

[0098] P kk =(IK k H k )P kk-1

[0099] where K k is the Kalman gain, which weighs the reliability of prediction and observation, is the transpose of the observation matrix, Rk is the measurement noise covariance matrix;

[0100] and P kk is the posterior estimate and error covariance after observation correction; H k is the observation matrix, I is the identity matrix;

[0101] The present invention adjusts the measurement noise covariance R k and process noise covariance Q k When arc interference, strong EMI or sensor temperature drift occurs, the filter can automatically reduce the weight of the observation to keep the estimation stable; the sensitivity is quickly restored after the interference subsides.

[0102] Through the above process, when the power system experiences severe disturbances or fault conditions, the secondary filtering module can realize real-time prediction and dynamic adjustment of signals, effectively dealing with signal distortion problems in strong interference and high noise environments of the power system, and providing accurate and stable data information for fault identification.

[0103] The core starting criterion of the fault identification module is the current change rate:

[0104]

[0105] In combination with the strategy of distinguishing between lightning strikes and transient interference, the present invention introduces dual-criteria combinational logic:

[0106] Set the current change rate threshold: K1 = 90A / ms; this current change rate threshold is for a typical 35kV, 630A ring main unit line. For feeders with rated parameters different from those of a typical 35kV ring main unit line, the present invention uses the following five steps to set the current change rate threshold K1.

[0107] 1. Estimation of short circuit steep slope

[0108] According to the line rated phase voltage U nom and the minimum equivalent inductance L min Estimating the upper limit of the short circuit initial slope

[0109] Take K rated 30% to 50% of the sensitivity is used as the sensitivity benchmark for high-resistance grounding faults.

[0110] 2. Calculating the Noise Margin

[0111] Statistical 24h normal operation waveform current change rate standard deviation σ di / dt , and set

[0112] Noise margin K noise =k σ σ di / dt , kσ＝ 6 to 8 to ensure that the probability of false operation is less than 0.1%.

[0113] 3. Dynamic range of the measurement chain

[0114] Combined with mutual inductor (or Hall sensor) full scale I FS and ADC sampling period T s , calculate the upper limit of the observable slope of the measurement chain If K sensor If it is lower than other items, it can be compensated by increasing the sampling rate or changing the sensor.

[0115] 4. Calculate the high resistance fault sensitivity coefficient and take K HIF =βK rated ,β=0.3~0.5, to take into account the capture of high ground resistance or high arc resistance fault.

[0116] 5. Threshold Comprehensive

[0117] The final threshold is the largest of the three, plus a design margin of ε = 10% to 20%:

[0118] K1=(max{K noise ,K HIF ,K sensor})(1+ε)

[0119] Finally, the current change rate threshold K1 is obtained.

[0120] The current root mean square change rate Δrms is also introduced as an auxiliary criterion to construct a combined trigger mechanism.

[0121] The root mean square change rate Δrms is the change in the effective value of the current within adjacent sliding windows. It reflects the speed of change at the energy / amplitude level and is used to compensate for the problem of insufficient sensitivity to slow rise, high resistance or transformer saturation faults when relying solely on the instantaneous current change rate.

[0122] The root mean square change rate in the present invention is the change in the effective value of the current in the kth sliding window I rms (k) minus the change in the effective value of the current in the k-1th sliding window I rms (k-1), and divided by the window width T w , and then normalize the amplitude increment to the rate of change, specifically:

[0123]

[0124] Among them, the effective value of current

[0125]

[0126] N is the number of sampling points involved in the calculation, is the instantaneous value of the current at the kN+jth sampling point.

[0127] Statistical standard deviation of the effective current value under normal working conditions σ Irms , take the current RMS change rate threshold K2 = k σ σ Irms , k σ＝ 4-6

[0128] The fault start logic is

[0129]

[0130] Fast road |di / dt| exceeds the threshold and triggers immediately, with a response time of <0.2ms.

[0131] Backup path: If |di / dt| does not exceed the threshold, monitor Δrms for M = 2 to 3 consecutive windows; once they all reach K2, the protection is triggered.

[0132] Reset condition: When |di / dt|<0.3K1 and Δrms<0.3K2 for t reset =50~100ms, Start_Flag is automatically cleared.

[0133] Drawing on the fault-tolerant concept of the matrix method, the system sets up an independent dual-channel sampling structure at each end of the current-limiting reactor to achieve the following functions:

[0134] All eigenvalues ​​(margin, kurtosis) are calculated and cross-checked in parallel under different paths;

[0135] Once packet loss or abnormal waveform occurs at the detection point, the system can switch to the redundant channel to continue feature extraction;

[0136] In the event of multiple-point failures or node signal loss, the system can construct an interpolation / prediction model to compensate for the missing amount.

[0137] Through the above processing strategy, the data processing module not only has anti-interference stability, but also supports the accurate extraction and real-time identification of fault mutation characteristics, meeting the response requirements of the high-speed protection system.

[0138] like Figure 3 As shown in the fault identification dual-mode judgment logic diagram, the fault identification module adopts dual-mode judgment, taking into account the judgment requirements of normal and fault states, and at the same time deeply explores and transforms the possible adverse factors in the system, further improving the accuracy and anti-interference ability of fault identification.

[0139] Normal mode: fast current change rate criterion

[0140] Under normal operation, the system mainly relies on the current change rate criterion as follows:

[0141]

[0142] When load starts and stops or small fluctuations are within the normal range, this criterion will not easily trigger protection action. Only when the current change rate significantly exceeds the threshold will the subsequent enhancement mode be activated to ensure that excessive protection does not occur during normal operation.

[0143] Enhanced mode: dual criteria of margin factor and kurtosis ratio,

[0144] Once the current change rate reaches or exceeds the threshold, the system automatically switches to the enhanced judgment mode and confirms the fault section and polarity by calculating the margin factor and kurtosis ratio.

[0145] The margin factor criterion is used to evaluate the dynamic fluctuation amplitude of a specific voltage monitoring point in a short period of time, thereby locating the fault section.

[0146] Define the margin factor as follows:

[0147]

[0148] Where, u j (i) is the i-th adaptive Kalman residual voltage sample value of the j-th monitoring point in the current statistical window, which has the characteristics of zero mean and constant variance, and N is the number of sampling points.

[0149] This margin factor measures the relative volatility of the voltage waveform. The larger the value, the more severe the voltage disturbance at that point and the closer the fault is to that point. The point with the maximum value is the fault section.

[0150] The kurtosis ratio criterion is used to determine the fault polarity. By comparing the kurtosis coefficients of the positive and negative voltage waveforms, it can distinguish between positive pole faults, negative pole faults, or inter-pole faults. The definition is as follows:

[0151] Kurtosis ratio threshold Where, Kurt(.) represents the kurtosis operation function, u Lpa 、u Lna The transient voltage data at both ends of the positive and negative reactors are respectively, and the filtering results are output by the data processing module. When K3>1.3, it is a positive pole fault, when K3<0.78, it is a negative pole fault, and 0.78≤K3≤1.3, it is an inter-pole fault.

[0152] Noise and harmonics serve as auxiliary positioning tools. When the system detects noise or harmonic peaks of a specific amplitude or frequency band, it assists in correcting the segment determination of the margin factor based on their changes in timing before and after the fault occurs, thereby improving anti-interference capabilities.

[0153] After detecting an abnormally large current, the system automatically collects characteristic parameters of the initial transient waveform (such as waveform climbing speed, distortion coefficient, etc.) and uses them in combination with the kurtosis ratio.

[0154] The delay window is used for deep calculations, taking advantage of the time gap between fault action and circuit breaker tripping to perform more complex feature extraction and secondary criterion verification, making full use of system delays to improve decision reliability.

[0155] like Figure 4 As shown, the fault isolation execution module adopts optimized trip control logic.

[0156] In normal state, the system adopts binary logic (closed / open) to reduce the complexity of judgment.

[0157] When the fault mode is triggered, the system automatically switches to the fuzzy logic mode and establishes the trip information vector through the correlation matrix A and the fault direction vector G:

[0158] QPD=A·G

[0159] The correlation matrix provides weights, and the direction vector provides fault direction. The two are multiplied to obtain the fuzzy confidence. After thresholding and packaging, the final trip command is obtained.

[0160] Specifically, the behavior of the correlation matrix A is the fault criterion, and the columns are protected elements, where element a pq is the weight of criterion p on component q;

[0161] The fault direction vector G is {-1, 0, 1}, where 1 indicates that the fault is in the forward direction of element q, -1 indicates the reverse direction, and 0 indicates uncertainty.

[0162] This vector is used to make tripping decisions, clarify the tripping path, effectively reduce the impact of unknown states, and avoid false operations or tripping delays.

[0163] The proposed protection system solution utilizes innovative dual-mode dynamic criteria switching, a "reverse utilization" technology strategy, and adaptive tripping logic and reclosing mechanisms to ensure absolute reliability in fault conditions while minimizing system complexity and redundancy during normal operation. By fully leveraging existing unfavorable factors in the system, it further enhances criteria reliability and fault isolation accuracy at critical moments. The overall design is scientifically sound, with a clear and innovative technical approach, making it highly valuable for widespread application.

[0164] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make various modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.

Claims

1. A power protection system for rapid isolation of short-circuit faults, characterized in that: It includes detection and acquisition module, data processing module, fault identification module and fault isolation execution module; The detection and acquisition module collects the current and voltage signals of the flexible DC distribution line and transmits them to the data processing module via optical fiber; The data processing module is used to perform a first-level filtering on the received signal to achieve fast and smooth processing, and obtain pre-processed data under normal working conditions. When the pre-processed data is abnormal, that is, exceeds the data threshold under normal working conditions, a second-level filtering is triggered to obtain pre-processed data and predicted data under suspected fault conditions. The processed data is sent to the fault identification module; The fault identification module, based on the processed data obtained by the data processing module, triggers fault identification through a combination of the current change rate criterion and the current root mean square change rate criterion. Once fault identification is triggered, the margin factor criterion is used to evaluate the dynamic fluctuation amplitude of the voltage monitoring point in a short period of time to locate the fault section. At the same time, the kurtosis ratio criterion is used to determine the fault polarity and distinguish between positive, negative, and inter-pole faults. The fault isolation execution module receives the judgment result of the fault identification module and executes fault isolation according to the judgment result.

2. The power protection system for rapid isolation of short-circuit faults according to claim 1, characterized in that: The detection and acquisition module includes a high-sensitivity Hall current sensor connected in series to the positive and negative busbars of the current-limiting reactor series circuit, and a high-precision voltage sensor arranged at each end of the reactor; the sampling frequency is set to 12-15kHz, and anti-interference optical fiber transmission is adopted.

3. The power protection system for rapid isolation of short-circuit faults according to claim 1, characterized in that: The first-stage filtering adopts Savitzky–Golay filter and performs low-order polynomial least squares fitting in a narrow sliding window. On the one hand, it locally averages the random noise, and on the other hand, it completely retains the 0th to pth order derivatives of the signal, thereby smoothing the waveform without weakening the spike mutation characteristics.

4. The power protection system for rapid isolation of short-circuit faults according to claim 3, characterized in that: The secondary filtering is a sampled improved Kalman filter, which processes the voltage and current signals more deeply. The specific method is as follows: The state space model of the system signal is expressed as: Equation of state: X k =F k X k-1 +w k-1 Measurement equation: Z k =H k X k +v k Where, X k is the system state vector at the kth moment, including the true values ​​of the voltage and current signals; F k is the state transfer matrix, representing the change of state from time k-1 to time k; Z k is the observation signal vector at the kth moment, that is, the measurement value after the first level of filtering preprocessing; H k is the observation matrix, which maps the true state to the measurement space; w k-1 、v k are process noise and observation noise respectively, assuming that they satisfy Gaussian distribution and are independent of each other: w k-1 ~N(0,Q k-1 ),v k ~N(0,R k ), where Q k-1 is the covariance matrix of process noise; R k is the covariance matrix of the observation noise; The improved Kalman filter calculates and updates the prior error covariance matrix P in real time kk-1 and the Kalman gain K k , to achieve state prediction and correction: Prediction steps: in is the prior state estimate at time k, is the posterior result at time k-1; F k is the state transfer matrix; P kk-1 is the prior error covariance at time k, P k-1k-1 is the posterior error covariance of the previous moment; Q k-1 is the process noise covariance; Update steps: P kk =(I-K k H k )P kk-1 where K k is the Kalman gain, which weighs the reliability of prediction and observation, is the transpose of the observation matrix, R k is the measurement noise covariance matrix; and P kk is the posterior estimate and error covariance after observation correction; H k is the observation matrix, I is the identity matrix; By adjusting the measurement noise covariance R k and process noise covariance Q k When arc interference, strong EMI or sensor temperature drift occurs, the filter automatically reduces the weight of the observation to keep the estimation stable; the sensitivity is quickly restored after the interference subsides.

5. The power protection system for rapid isolation of short-circuit faults according to claim 1, characterized in that: The combined trigger mechanism is On fast roads, the current change rate |di / dt| is triggered immediately when it exceeds the current change rate threshold K1, and the response time is <0.2ms; Backup path: If the current change rate |di / dt| does not exceed the current change rate threshold K1, monitor the current RMS change rate Δrms for M = 2 to 3 consecutive windows; once all reach the current RMS change rate threshold K2, the protection is also triggered.

6. The power protection system for rapid isolation of short-circuit faults according to claim 5, characterized in that: The current change rate threshold K1 is obtained by the following five steps: 1) Short circuit steep slope estimation: According to the line rated phase voltage U nom With the minimum equivalent inductance L min Estimating the upper limit of the short circuit initial slope Take K rated 30% to 50% is used as the sensitivity benchmark for high-resistance grounding faults; 2) Calculate the noise margin: Statistical 24h normal operation waveform current change rate standard deviation σ di / dt , and set Noise margin K noise =k σ σ di / dt , k σ＝ 6 to 8 to ensure that the probability of false operation is less than 0.1%; 3) Dynamic range of the measurement chain: Combined with transformer or Hall sensor full scale I FS and ADC sampling period T s , calculate the upper limit of the observable slope of the measurement chain If K sensor If it is lower than other items, it can be compensated by increasing the sampling rate or changing the sensor; 4) Calculate the high resistance fault sensitivity coefficient: Take K HIF =βK rated ,β=0.3~0.5, to take into account the capture of high ground resistance or arc high resistance fault; 5) Threshold synthesis The final threshold is the largest of the three, plus a design margin of ε = 10% to 20%: K1=(max{K noise ,K HIF ,K sensor })(1+e) Finally, the current change rate threshold K1 is obtained.

7. The power protection system for rapid isolation of short-circuit faults according to claim 5 or 6, characterized in that: The current root mean square change rate threshold K2 is obtained by: The root mean square change rate is the change in the effective value of the current in the kth sliding window I rms (k) minus the change in the effective value of the current in the k-1th sliding window I rms (k-1), and divided by the window width T w , and then normalize the amplitude increment to the rate of change, specifically: Among them, the effective value of current N is the number of sampling points involved in the calculation, is the instantaneous value of the current at the kN+jth sampling point. Statistical standard deviation of the effective current value under normal working conditions σ Irms , take the current RMS change rate threshold K2 = k σ σ Irms , k σ＝ 4~6.

Images