A method, system, device, and medium for adaptive current protection

Abstract

The application discloses a kind of self-adapting current protection method, system, equipment and medium, belong to relay protection technical field, method includes: analog power output generates typical scene set, extracts fluctuation characteristics, constructs self-adapting model and generates dynamic reference, combines fault characteristic and outputs real-time threshold and executes protection decision.System includes data generation and scene simulation module, characteristic index extraction module, boundary parameter definition module, self-adapting model construction and reference generation module, real-time threshold output module, protection action decision module: the present application is by constructing distributed power output generation model and self-adapting setting calculation model, in steady state, linear follow-up is realized to power fluctuation using transfer function, in transient state, current limit is calculated in combination with fault ride-through mechanism.Realize that protection setting value is in real time self-adapting matching to complex operating condition, avoid the risk of distribution network protection misoperation and refusal caused by new energy output randomness, guarantee the stable operation of system.

Description

Technical Field

[0001] This invention relates to the field of relay protection technology, and specifically to a method, system, device, and medium for adaptive current protection. Background Technology

[0002] With the advancement of new power system construction, distributed power sources, represented by photovoltaic and wind power generation, are being connected to the distribution network on a large scale, transforming it from a traditional single-source radial structure to an active network with multiple power sources. While this transformation has significantly improved energy utilization efficiency, it has also profoundly changed the fault characteristics and current distribution patterns of the distribution network, making the power flow direction more complex and variable, and placing higher demands on the operating performance and adaptability of relay protection devices.

[0003] Existing distribution network current protection schemes typically employ fixed setting values ​​based on offline calculations under extreme operating conditions. This approach struggles to adapt to the highly intermittent and random characteristics of distributed generation (DG) output. Because the output power of DG exhibits significant time-varying fluctuations due to environmental and meteorological factors such as light intensity and wind speed, the load current flowing through the protection installation point varies greatly. If the setting value is set too low, malfunctions are easily caused by load current exceeding limits during high-output operation of the DG; if the setting value is set too high, insufficient sensitivity may lead to failure to operate during high-resistance or end-point faults. Furthermore, DG connected to the grid via power electronic inverters possess unique low-voltage ride-through characteristics. During grid voltage dips, their output current is limited by internal control logic, exhibiting nonlinear characteristics distinctly different from traditional synchronous generators. Traditional protection methods often overlook these differences in current characteristics under different modes of normal operation fluctuations and fault transient ride-throughs. They cannot accurately define the boundary between normal power fluctuations and short-circuit fault currents using simple fixed thresholds, making it difficult to balance protection reliability and sensitivity, thus seriously threatening the safe and stable operation of high-penetration distribution networks. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention provides a method, system, device and medium for adaptive current protection.

[0005] Therefore, the technical problem solved by this invention is that the output of distributed power sources is highly intermittent and random, resulting in drastic fluctuations in the fault characteristics of the distribution network. Traditional fixed-setting current protection is difficult to balance sensitivity and reliability, and is prone to false tripping or failure to trip.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: an adaptive current protection method, comprising: acquiring historical operating data of a distribution network and constructing a distributed generation power output model; reproducing the intermittent statistical characteristics of the distributed generation power output based on the probability distribution characteristics of the historical operating data, generating a typical scenario dataset containing multidimensional fluctuation information; analyzing the typical scenario dataset and extracting cluster centers characterizing typical fluctuating operating conditions to obtain a set of feature indicators reflecting the statistical characteristics of the output power fluctuation; and, based on the set of feature indicators, calculating the maximum fluctuation amplitude and rate of change range of the distributed generation power output under different operating conditions, defining and distinguishing between normal fluctuations and fault mutations. Feature boundary threshold parameters; based on the feature boundary threshold parameters, an adaptive setting calculation model is constructed to establish the dynamic correlation between the current protection setting threshold and the output power change, and the real-time collected output power of the distributed power source is used as an input to the adaptive setting calculation model to generate a dynamic reference value; in response to the voltage drop characteristics at the grid connection point, the current characteristics during the fault period are calculated through the adaptive setting calculation model and combined with the fault location information, and the real-time current protection setting threshold is output; a protection action decision logic is established to compare the real-time fault current with the real-time current protection setting threshold, and the protection action is triggered when the preset fault criterion is met, otherwise the normal operation of the system is maintained.

[0007] As a preferred embodiment of the adaptive current protection method of the present invention, the step of acquiring historical operating data of the distribution network and constructing a distributed power generation model, and reproducing the intermittent statistical characteristics of the distributed power output based on the probability distribution characteristics of the historical operating data to generate a typical scenario dataset containing multidimensional fluctuation information includes: preprocessing the historical operating data, constructing a sample set containing environmental correlation factors and time series factors, defining the statistical characteristics of the sample set as a target probability distribution space; configuring the distributed power generation model to form a generation mechanism that maps the space of random disturbance variables to the space of simulated power output, for outputting a virtual power sequence; defining a statistical consistency metric, mapping the virtual power sequence to the target probability distribution space, calculating the distribution deviation between the virtual power sequence and the sample set; iteratively optimizing the internal parameters of the generation mechanism, correcting the generation mechanism with the goal of minimizing the distribution deviation, until the statistical characteristics of the virtual power sequence converge to the target probability distribution space, and outputting the typical scenario dataset.

[0008] The beneficial effects of this preferred technical solution are as follows: By constructing an adversarial model architecture that includes a generation mechanism and a difference metric, this invention effectively solves the problem of scarce and unevenly distributed historical operating data samples for distributed power sources. This solution utilizes random perturbation variables to drive nonlinear mapping, enabling the generation of virtual power sequences covering extreme weather conditions and rare fluctuating operating conditions while maintaining the statistical regularity of the original data. Through iterative optimization to minimize distribution deviation, the generated typical scenario dataset possesses extremely high statistical fidelity and operating condition coverage, providing sufficient and high-quality data support for subsequent feature extraction and protection boundary definition. This significantly improves the generalization ability and robustness of the protection model when facing unseen complex fluctuating operating conditions.

[0009] As a preferred embodiment of the adaptive current protection method of the present invention, the step of parsing the typical scenario dataset and extracting cluster centers representing typical fluctuating operating conditions to obtain a set of feature indicators reflecting the statistical characteristics of the output power fluctuation includes: constructing a multidimensional feature space adapted to the typical scenario dataset, configuring a kernel density metric function to measure the local compactness of samples; traversing the scenario sample points in the multidimensional feature space, calculating the drift vector of each scenario sample point pointing to the direction of the local probability density maxima; updating the spatial coordinate position of the scenario sample points, driving the scenario sample points to continuously migrate along the drift vector until they converge to the peak region of the local probability density distribution; locking the geometric center of the peak region as the cluster center, and parsing the statistical distribution attributes of the cluster center to construct the set of feature indicators.

[0010] As a preferred embodiment of the adaptive current protection method of the present invention, the step of calculating the maximum fluctuation amplitude and rate of change range of the distributed power supply output power under different operating conditions based on the set of feature indicators, and defining the feature boundary threshold parameters that distinguish between normal fluctuations and fault mutations, includes: statistically analyzing the power amplitude dispersion and time-series rate of change distribution under typical fluctuating operating conditions represented by the set of feature indicators, and establishing a statistical confidence interval describing the probability distribution of power fluctuations; generating an upper bound envelope trajectory and a lower bound envelope trajectory covering the normal power fluctuation range based on the statistical confidence interval; analyzing the extreme value attributes and maximum change slope of the upper bound envelope trajectory and the lower bound envelope trajectory in the time domain, and generating the feature boundary threshold parameters to quantify the critical boundary between normal fluctuations and fault mutations.

[0011] As a preferred embodiment of the adaptive current protection method of the present invention, the step of constructing an adaptive setting calculation model based on the characteristic boundary threshold parameters, establishing the dynamic correlation between the current protection setting threshold and the output power change, and introducing the real-time collected distributed power output as an input quantity into the adaptive setting calculation model to generate a dynamic reference value includes: extracting statistical indicators characterizing the upper limit of normal operating condition fluctuations from the characteristic boundary threshold parameters, mapping them to a safety floating coefficient for compensating for random power fluctuations; constructing a transfer function containing a proportional following term and a fixed bias term, correcting the gain parameter of the transfer function with the safety floating coefficient, and establishing the dynamic correlation; driving the transfer function to track the real-time input of the distributed power output, calculating a protection threshold reference with dynamic avoidance characteristics; and outputting the protection threshold reference as the dynamic reference value, so that the dynamic reference value remains above the real-time fluctuation trajectory of the distributed power output during steady-state operation.

[0012] The beneficial effects of this preferred technical solution are as follows: By constructing a transfer function that includes a proportional follower term and a fixed bias term, this invention achieves real-time dynamic tracking of the protection setting value to the output power of the distributed power source. This solution extracts statistical indicators characterizing the upper limit of normal fluctuations and converts them into a safety floating coefficient, which can automatically adjust the gain parameter of the protection threshold according to the fluctuation amplitude of real-time power. This dynamic correlation mechanism breaks through the limitations of traditional fixed settings in balancing sensitivity and reliability, ensuring that the dynamic reference value always closely follows and covers the power fluctuation trajectory during steady-state operation. This not only effectively prevents maloperation caused by random fluctuations in new energy output and establishes a precise dynamic avoidance relationship, but also retains sufficient fault detection sensitivity under low-output conditions.

[0013] As a preferred embodiment of the adaptive current protection method of the present invention, the following steps are taken: In response to the voltage drop characteristics at the grid connection point, the adaptive setting calculation model is used to calculate the current characteristics during the fault period, combined with fault location information, and output a real-time current protection setting threshold. This includes: setting a mode switching threshold for the grid connection point voltage to define the steady-state following zone and the transient crossing zone of the distribution network; when the grid connection point voltage is detected to be in the steady-state following zone, the adaptive setting calculation model is driven to execute linear following logic to generate a dynamic following reference value based on the real-time distributed power output; when the grid connection point voltage is detected to have dropped to the transient crossing zone, the fault crossing response mechanism within the model is activated, and the maximum fault contribution current during the fault period is determined based on the voltage drop depth and the reactive current limit of the distributed power source; the dynamic following reference value is corrected by superimposing a preset safety margin coefficient on the maximum fault contribution current and replacing it with the current real-time current protection setting threshold.

[0014] The beneficial effects of this preferred technical solution are as follows: By defining the steady-state following region and the transient ride-through region, this invention establishes a dual-mode control logic for the low-voltage ride-through characteristics of inverter-type power supplies. When a voltage drop at the grid connection point is detected, the system can quickly activate the fault ride-through response mechanism, abandoning the linear following logic that only applies to steady state, and instead extrapolating the maximum fault contribution current unique to the fault period based on the voltage drop depth and the reactive power limit of the power supply. This mechanism fully considers the nonlinear output characteristics of active power reduction and reactive power injection of distributed power sources during fault transients. By correcting the dynamic reference value and superimposing a safety margin, it accurately calculates the protection threshold under fault conditions, effectively avoiding protection maloperation caused by sudden changes in power supply output characteristics during faults outside the fault zone, and ensuring the selectivity and safety of the distribution network during transient processes.

[0015] In a preferred embodiment of the adaptive current protection method of the present invention, the step of establishing protection action decision logic and comparing the real-time fault current with the real-time current protection setting threshold includes: collecting and parsing the real-time current vector of the distribution network, and extracting the current amplitude characteristics for fault identification; introducing the real-time current protection setting threshold as a dynamic reference, and constructing an amplitude comparison criterion for the current amplitude characteristics; when the current amplitude characteristics are detected to exceed the real-time current protection setting threshold, initiating a time-domain verification mechanism for the fault duration; after determining that the time-domain verification mechanism meets the preset time limit constraint, confirming that the fault is a permanent fault, generating a trip command to drive the circuit breaker to operate to isolate the fault area; otherwise, determining it as a transient disturbance and blocking the trip command.

[0016] To address the aforementioned technical problems, this invention also provides the following technical solution: an adaptive current protection system, comprising a data generation and scenario simulation module, used to acquire historical operating data of the distribution network and construct a distributed power generation model, reproducing the intermittent statistical characteristics of the distributed power output based on the probability distribution characteristics of the historical operating data, and generating a typical scenario dataset containing multidimensional fluctuation information; a feature index extraction module, used to parse the typical scenario dataset and extract cluster centers characterizing typical fluctuation conditions, obtaining a set of feature indices reflecting the statistical characteristics of the output power fluctuation; and a boundary parameter delineation module, used to calculate the maximum fluctuation amplitude and rate of change range of the distributed power output under different operating conditions based on the feature index set, delineating and distinguishing between normal fluctuations and fault mutations. The system comprises the following modules: a feature boundary threshold parameter; an adaptive model construction and benchmark generation module, used to construct an adaptive setting calculation model based on the feature boundary threshold parameter, establish the dynamic correlation between the current protection setting threshold and the output power change, and introduce the real-time collected distributed power output as an input to the adaptive setting calculation model to generate a dynamic benchmark value; a real-time threshold output module, used to respond to the voltage drop characteristics at the grid connection point, calculate the current characteristics during the fault period through the adaptive setting calculation model and the fault location information, and output the real-time current protection setting threshold; and a protection action decision module, used to establish protection action decision logic, compare the real-time fault current with the real-time current protection setting threshold, trigger the protection action when the preset fault criterion is met, otherwise maintain normal system operation.

[0017] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the adaptive current protection method.

[0018] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the adaptive current protection method.

[0019] The beneficial effects of this invention are as follows: It reproduces the spatiotemporal fluctuation patterns of power output using probability distribution characteristics, and defines a refined boundary distinguishing normal fluctuations from sudden faults through cluster analysis. An adaptive setting calculation model is constructed, which ensures that the protection threshold follows power fluctuations in real time during steady-state operation, establishing a dynamic avoidance relationship; during fault transients, it combines low-voltage ride-through characteristics to deduce the maximum fault contribution current and correct the setting threshold. This method achieves precise matching between protection settings and real-time operating conditions, effectively solving the problem of poor protection adaptability in high-penetration distribution networks, and significantly improving the safety and operating efficiency of the power supply system. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of an adaptive current protection method provided in one embodiment of the present invention.

[0022] Figure 2 This is a block diagram of an adaptive current protection system provided in one embodiment of the present invention.

[0023] Figure 3 This is a schematic diagram of a traditional fixed-setting protection under fault conditions in the scenario verification of this invention.

[0024] Figure 4 This is a schematic diagram of the traditional fixed setting protection under photovoltaic power output fluctuation in the scenario verification of this invention.

[0025] Figure 5 This is a schematic diagram of the adaptive current protection setting dynamic adjustment mechanism under photovoltaic power output fluctuation and fault conditions in the scenario verification of this invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0027] Example 1, referring to Figure 1 As one embodiment of the present invention, an adaptive current protection method is provided, comprising: S1: Obtain historical operating data of the distribution network and construct a distributed power generation model. Based on the probability distribution characteristics of the historical operating data, reproduce the intermittent statistical characteristics of the distributed power output and generate a typical scenario dataset containing multidimensional fluctuation information. S2: Analyze the typical scenario dataset and extract the cluster centers that characterize typical fluctuating operating conditions to obtain a set of feature indicators that reflect the statistical characteristics of the output power fluctuation; S3: Based on the set of characteristic indicators, calculate the maximum fluctuation amplitude and rate of change range of the output power of the distributed power source under different operating conditions, and define the characteristic boundary threshold parameters that distinguish between normal fluctuations and fault mutations. S4: Based on the characteristic boundary threshold parameters, construct an adaptive tuning calculation model, establish the dynamic correlation between the current protection tuning threshold and the output power change, and introduce the real-time collected distributed power output as an input into the adaptive tuning calculation model to generate a dynamic reference value. S5: In response to the voltage drop characteristics at the grid connection point, the current characteristics during the fault period are calculated by combining the adaptive setting calculation model with the fault location information, and the real-time current protection setting threshold is output. S6: Establish protection action decision logic, compare the real-time fault current with the real-time current protection setting threshold, trigger protection action when the preset fault criterion is met, otherwise maintain normal system operation.

[0028] It should be noted that traditional distribution network current protection typically uses offline setting calculations to determine fixed operating thresholds. Such fixed settings are ill-suited to the complex operating conditions following the integration of highly pervasive distributed generation (DG) power sources. Because the output power of DG is significantly affected by environmental factors, exhibiting substantial randomness and intermittent fluctuations, the load current flowing through the protection installation point varies considerably over a wide range. If the setting is configured according to the maximum operating mode, the protection may fail to operate due to insufficient sensitivity when the DG output is low. If the setting is configured according to the minimum operating mode, it is highly susceptible to maloperation due to normal load current exceeding limits during periods of high DG power generation. Furthermore, existing protection schemes often neglect the low-voltage ride-through control characteristics of DG during grid voltage dips, failing to accurately distinguish between current ramp-up caused by normal power fluctuations and short-circuit current surges caused by faults. This results in a lack of adaptive capability in differentiating between normal operation and fault states.

[0029] Therefore, to address the aforementioned problems, this invention, through steps S1 to S6, first utilizes historical data to construct a generative model and generate a typical scenario dataset. Through cluster analysis, it accurately grasps the intermittent statistical characteristics of distributed power output under different time and weather conditions, thereby extracting a set of characteristic indicators reflecting power fluctuation patterns. Based on these indicators, it calculates characteristic boundary threshold parameters that distinguish between normal fluctuations and fault mutations, and then constructs an adaptive setting calculation model capable of responding to power changes in real time. This model establishes a dynamic correlation between the current protection setting threshold and the output power. During steady-state operation, the protection threshold is set as a dynamic benchmark value that follows power fluctuations, ensuring that the protection setting is always slightly higher than the normal load current. When a voltage drop at the grid connection point is detected, the model combines fault location information and the fault ride-through characteristics of the distributed power source to accurately calculate the current characteristics during the fault period and output the real-time current protection setting threshold. The established protection action decision logic effectively overcomes the defect that fixed settings cannot adapt to source load fluctuations by comparing real-time fault current with real-time adjusted setting thresholds. It achieves real-time adaptive matching of protection settings to distributed power supply operating conditions, ensuring that the system does not malfunction under normal high power fluctuations while significantly improving the sensitivity and reliability of actions under fault conditions.

[0030] Example 2, refer to Figures 1 to 5 This is one embodiment of the present invention. Based on the previous embodiment, an adaptive current protection method is provided.

[0031] Specifically, in step S1, historical operating data of the distribution network is acquired and a distributed generation power output model is constructed. Based on the probability distribution characteristics of the historical operating data, the intermittent statistical characteristics of the distributed generation power output are reproduced, generating a typical scenario dataset containing multidimensional fluctuation information, including the following steps A1 to A4: A1: Preprocess the historical operational data to construct a sample set containing environmental correlation factors and time series factors, and define the statistical characteristics of the sample set as the target probability distribution space; First, historical power time series data of distributed generation in the distribution network are collected, and the original dataset is defined as follows. To eliminate the differences between data of different dimensions and improve the convergence speed of model training, the max-min normalization method is used to process the original data. The calculation formula is as follows: In the formula This is the raw power data. and These are the minimum and maximum values ​​in the historical data, respectively. This is the normalized sample data. Simultaneously, light intensity, ambient temperature, and time information are used as conditional vectors. This data is then concatenated with the normalized power sequence to construct a joint sample set containing both environmental and time-series factors. The statistical distribution of this sample set in the feature space is thus established as the target probability distribution space. .

[0032] A2: Configure the distributed power output generation model to form a generation mechanism that maps the space of random disturbance variables to the space of simulated power output, for outputting virtual power sequences; In this embodiment of the application, step A2 configures the distributed power output generation model to form a generation mechanism that maps the space of random perturbation variables to the space of simulated power output, and constructs a distributed power output generation model based on a generative adversarial network architecture. This model includes a generator. With discriminator Configure the generator As a generation mechanism, its input is a random noise vector that follows a standard normal distribution. ,Right now Generator Internally, it contains multiple layers of fully connected neural networks or long short-term memory networks, through parameters The defined nonlinear mapping function will transform the random noise vector Converted into a virtual power sequence, the calculation formula is expressed as follows: This virtual power sequence This refers to the generated samples in the simulated power output space, which aim to approximate the actual output characteristics of distributed power sources.

[0033] In an optional implementation, step A2 configures the distributed power output generation model to form a generation mechanism that maps the space of random disturbance variables to the space of simulated power output. The output of the virtual power sequence can be achieved by constructing a stochastic differential equation model based on physical constraints. The random disturbance variables are defined as Wiener process increments that drive the evolution of state variables. The stochastic differential equations are used to simulate the dynamic random walk trajectory of meteorological factors such as light intensity or wind speed, and the nonlinear photoelectric conversion function of photovoltaic cells or wind turbines is connected in series to map the random evolution trajectory of meteorological factors to the power domain, thereby outputting a virtual power sequence that contains both random fluctuation characteristics and strictly conforms to the physical laws of energy conversion of the equipment.

[0034] In another optional implementation, step A2 configures the distributed power output generation model to form a generation mechanism that maps the space of random perturbation variables to the space of simulated power output. The output of the virtual power sequence can also be achieved by using a decoder architecture of a time series conditional variational autoencoder. The space of random perturbation variables is regarded as a high-dimensional latent variable manifold that follows a Gaussian distribution. A decoding network with an attention mechanism is used to sample from the latent variable manifold and reconstruct the time dependence. By learning the implicit time correlation and local fluctuation trend in historical data, the abstract random perturbation variables are decoded and restored into a continuous virtual power sequence with long-range memory characteristics, ensuring that the generated sequence is highly consistent with the real data in terms of time structure.

[0035] A3: Define a statistical consistency metric, map the virtual power sequence to the target probability distribution space, and calculate the distribution deviation between the virtual power sequence and the sample set; To quantify the statistical difference between generated data and real data, a discriminant is defined. As a measure of statistical consistency. Discriminator Through parameters Construct a neural network to output the probability distribution space of the input sample belonging to the true target. The probability of distribution bias. When calculating distribution bias, the Wasserstein distance or cross-entropy loss function is used to measure the difference between the two. Define the value function. Its mathematical expression is: In the formula Represents the mathematical expectation. This indicates the discriminator's judgment result on the real sample. This represents the discriminator's judgment result on the virtual sample. The specific value of this value function reflects the distribution deviation between the virtual power sequence and the sample set.

[0036] A4: Iteratively optimize the internal parameters of the generation mechanism to correct the generation mechanism with the goal of minimizing the distribution deviation, until the statistical characteristics of the virtual power sequence converge to the target probability distribution space, and output the typical scenario dataset; A minimax game strategy is employed to iteratively optimize the model parameters. During training, the generator parameters are first fixed. Optimize discriminator parameters To maximize the value function This improves the discriminator's ability to distinguish between genuine and fake data. Then, the discriminator parameters are fixed. Optimize generator parameters To minimize the value function ,Right now The weights are updated using the gradient descent algorithm, allowing the virtual power sequence generated by the generator to gradually deceive the discriminator until the distribution deviation converges to a preset minimum value. This indicates that the generated virtual power sequence has a high degree of fit to the target probability distribution space in terms of statistical features. At this point, the iteration stops, and the trained generator is used to output a large number of virtual power samples, forming a typical scene dataset.

[0037] Specifically, in step S2, the typical scenario dataset is parsed and cluster centers representing typical fluctuating operating conditions are extracted to obtain a set of feature indicators reflecting the statistical characteristics of the output power fluctuation, including the following steps B1 to B4: B1: Construct a multidimensional feature space adapted to the typical scenario dataset, and configure a kernel density metric function to measure the local compactness of samples; The contents generated in step S1 A typical scenario dataset of samples is mapped to In Euclidean space, denoted as Each sample It includes multi-dimensional features such as active power, reactive power, and rate of change at a given moment. To accurately assess the probability density at any point in the feature space, kernel density estimation theory is introduced, and a radially symmetric Gaussian kernel function is selected. As a kernel density metric function. Setting the bandwidth parameter. The bandwidth parameter controls the smoothness and local scope of the kernel function. The resolution of cluster analysis is determined by the Gaussian kernel function, which is mathematically expressed as: In the formula The normalization constant is Let be the Euclidean distance norm.

[0038] B2: Traverse the scene sample points in the multidimensional feature space and calculate the drift vector of each scene sample point pointing to the local probability density maximum. For any given sample point in the feature space The mean shift vector is calculated using the mean shift algorithm. This vector represents the direction in which the sample point rises fastest along the probability density gradient, i.e., it points to the region of maximum local density. According to kernel density gradient estimation theory, the mean drift vector is calculated as follows: In the formula For kernel function The negative derivative profile function, i.e. This formula shows that the drift vector essentially falls within the sampling window. The weighted average of all neighboring sample points within the current point The difference in weight depends on the distance between the neighboring samples and the center point.

[0039] B3: Update the spatial coordinates of the scene sample points and drive the scene sample points to migrate continuously along the drift vector until they converge to the peak region of the local probability density distribution. Construct an iterative optimization process, using the current sample points Position according to The rules are updated to continuously move sample points towards higher-density regions in the feature space. The positional offset is calculated after each iteration. And set a convergence threshold. .when When the sample point has reached the extreme point of the local probability density function, i.e., the peak region, the iteration stops. This process will automatically merge the originally discrete power sample points in the feature space into several high-density core regions, which represent the most common typical operating modes of distributed power sources.

[0040] B4: Locate the geometric center of the peak region as the cluster center, and analyze the statistical distribution attributes of the cluster center to construct the feature index set; All sample points that converge to the same peak region are grouped into one class, and the geometric centroid of this cluster is calculated as the cluster center. The operating condition characteristics represented by each cluster center are statistically analyzed, and the maximum power fluctuation amplitude is extracted. Maximum power change rate and duration of fluctuation Key parameters, etc. These parameters are combined to form a set of feature indicators. This set quantifies the boundary behavior characteristics of distributed power sources under different typical operating conditions, providing direct data basis for subsequently determining the floating range of protection setting values.

[0041] In this embodiment of the application, step S3 calculates the maximum fluctuation amplitude and rate of change range of the distributed power supply output power under different operating conditions based on the set of characteristic indicators, and defines the characteristic boundary threshold parameters that distinguish between normal fluctuations and fault mutations, including the following steps C1 to C3: C1: Statistically analyze the power amplitude dispersion and time-series rate of change distribution under typical fluctuating conditions represented by the set of characteristic indicators, and establish the statistical confidence interval describing the probability distribution of power fluctuations; For each cluster center extracted in step S2, the dispersion of sample points within each cluster is analyzed. The expected value of the power amplitude is calculated. with standard deviation and the mathematical expectation of the rate of change of power. with standard deviation Based on the normal distribution assumption or Chebyshev's inequality, and setting a confidence level (e.g., 99.7%), the statistical confidence interval for power fluctuations is calculated. The confidence interval for the power amplitude is defined as follows: The confidence interval for the rate of change of power is defined as follows: In the formula The coverage coefficient, typically set to 3, corresponds to the confidence level. This confidence interval statistically covers the vast majority of power fluctuation samples under normal meteorological conditions, forming a probabilistic benchmark for distinguishing between normal operation and anomalous changes.

[0042] C2: Generate an upper bound envelope trajectory and a lower bound envelope trajectory that cover the normal power fluctuation range based on the statistical confidence interval; Expanding the above statistical confidence intervals along the time axis, a dynamic time-domain envelope is constructed. For any given time... Define the upper bound envelope trajectory function as The lower bound envelope locus function is In the formula This is the reference power curve under this typical operating condition. This represents the maximum permissible fluctuation deviation calculated based on the confidence interval. These two trajectories enclose a band-shaped region in the time-domain plane, which physically characterizes the maximum reasonable fluctuation range of the distributed power source's output power under normal natural conditions such as sudden changes in illumination or gusts of wind.

[0043] C3: Analyze the extreme value attributes and maximum change slope of the upper bound envelope trajectory and the lower bound envelope trajectory in the time domain, and generate the feature boundary threshold parameter to quantify the critical boundary between normal fluctuations and fault mutations; Among them, the generated upper bound envelope trajectory trajectory with lower bound envelope Perform differential analysis to calculate its slope of change in the time domain. Extract the maximum slope magnitude of change in the envelope trajectory. This value represents the fastest rate of power change under normal operating conditions. Simultaneously, it extracts the amplitude limit of the envelope trajectory. Using these two physical quantities as core elements, a feature boundary threshold parameter set is constructed. This parameter set defines the critical boundary of the system: when the real-time detected rate of power change exceeds... Or the amplitude exceeds When the range is defined, the system will determine that the current state has exceeded the normal range of environmental fluctuations, thereby identifying it as a sudden change caused by a power grid fault, and providing quantitative criteria for the subsequent activation of the adaptive protection model.

[0044] In an optional implementation, step S3 calculates the maximum fluctuation amplitude and rate of change range of the distributed power output under different operating conditions based on the set of feature indicators. The feature boundary threshold parameter that distinguishes between normal fluctuations and fault mutations can be defined by constructing a probability distribution boundary based on a Gaussian mixture model. The probability density function of multidimensional power fluctuations is reconstructed using the mean and covariance matrix in the set of feature indicators. The contour lines of the probability distribution are truncated according to a preset confidence level as the edge of the normal operation area, thereby generating a dynamic envelope surface that includes the upper and lower limits of power amplitude and the limit of rate of change. This ensures that the boundary can cover random fluctuation samples under most meteorological conditions, while excluding low-probability extreme mutation points from the normal area.

[0045] In another optional implementation, step S3 calculates the maximum fluctuation amplitude and rate of change range of the distributed power output under different operating conditions based on the set of characteristic indicators. The characteristic boundary threshold parameters that distinguish between normal fluctuations and fault mutations can also be defined by establishing a two-dimensional phase plane trajectory analysis of power amplitude and rate of change. Historical data is mapped to a phase plane coordinate system with power amplitude as the horizontal axis and power change rate as the vertical axis. The convex hull algorithm is used to calculate the smallest convex polygon that can enclose all cluster centers and their neighboring samples. The geometric vertex coordinates of the convex polygon are analyzed to determine the maximum allowable slope of change under different power levels, thereby forming a nonlinear boundary constraint that dynamically adjusts with the operating point position, improving the ability to distinguish complex fluctuation conditions.

[0046] In this embodiment of the application, in step S4, an adaptive tuning calculation model is constructed based on the characteristic boundary threshold parameter to establish the dynamic correlation between the current protection tuning threshold and the output power change. The real-time collected output power of the distributed power source is then used as an input to the adaptive tuning calculation model to generate a dynamic reference value. This includes the following steps D1 to D4: D1: Extract the statistical index representing the upper limit of normal operating condition fluctuation from the feature boundary threshold parameters, and map it to a safety floating coefficient for compensating for random power fluctuations; The feature boundary threshold parameter set generated in step S3 Calling the maximum change slope With amplitude limit To prevent malfunction of the protection system during rapid power ramp-up phases, a safety float factor is defined. The calculation logic for this coefficient is as follows: In the formula This refers to the rated capacity of the distributed power source. This is a preset reliability weighting factor, typically ranging from 0.1 to 0.2. This formula indicates that the more drastic the power fluctuations of the photovoltaic power plant according to historical data, the higher the reliability weighting factor. The larger the value, the higher the calculated safety float factor. This also increases accordingly, thus logically reserving more room for avoidance.

[0047] D2: Construct a transfer function that includes a proportional follower term and a fixed bias term, and correct the gain parameter of the transfer function with the safety float coefficient to establish the dynamic correlation. The mathematical core of establishing an adaptive tuning calculation model is the transfer function. Its expression is defined as In the formula, To measure current in real time, For the gain of the proportional follower term, This is a fixed bias term. The gain parameter is corrected using the safety float coefficient calculated in step D1, let... This transfer function establishes the dynamic correlation between the protection settings and the linear variation of the load current, where the fixed bias term... This is used to ensure that the protection setting remains at a minimum safety threshold when the current is extremely low or close to zero power output, preventing false tripping due to measurement errors.

[0048] D3: Drive the transfer function to track the real-time input of the distributed power supply output power and calculate the protection threshold benchmark with dynamic avoidance characteristics; The output power of distributed power sources is obtained in real time through the data acquisition terminal of the distribution network. And convert it into an equivalent current value. .Will The above transfer function is continuously fed in as input variables to perform real-time computation. The calculation process is performed continuously over time, ensuring that the calculated protection threshold benchmark is accurate. It closely follows the fluctuations of the actual current, much like a shadow. This following mechanism constitutes a dynamic avoidance characteristic, meaning that no matter how randomly the output of the distributed power source changes, the protection threshold can always automatically adjust its height to avoid normal load current.

[0049] D4: Output the protection threshold reference as the dynamic reference value to ensure that the dynamic reference value is always maintained above the real-time fluctuation trajectory of the distributed power supply output power during steady-state operation; The result obtained in real time from step D3 Defined as a dynamic reference value, it is output to the protection logic discrimination module. When the distribution network is in steady-state operation mode, i.e., the voltage is normal and there are no external faults, due to... It can be strictly guaranteed from the perspective of mathematical inequalities. This condition is always met. This means that the protection setpoint curve is always geometrically enveloped above the actual power fluctuation curve, forming a dynamic safety buffer, thereby completely eliminating the possibility of protection malfunctions caused by intermittent power fluctuations from new energy sources.

[0050] Step S4 aims to establish the dynamic correlation between the current protection setting threshold and changes in output power. In traditional protection schemes, refer to... Figure 3 Under ideal, stable operating conditions, the fixed setpoint (red line) is higher than the photovoltaic output current (blue line), and the system functions normally. However, when faced with significant fluctuations in photovoltaic output, the traditional solution has serious drawbacks. (Refer to...) Figure 4 During the simulation period of 0.3 to 0.5 seconds, the simulated enhanced illumination caused the photovoltaic output current to rise from 500 amps to 680 amps. Since the setpoint was fixed at 550 amps, the load current exceeded this threshold directly in approximately 0.35 seconds, leading to a false judgment that the current exceeded the threshold. To address this issue, this step drives the transfer function to track the real-time input output power of the distributed power source. (Refer to...) Figure 5 Between 0.3 and 0.5 seconds, the dynamic reference value generated by this invention is no longer a horizontal straight line, but rather adaptively adjusted based on a safety fluctuation coefficient. In the figure, the red line always remains above the blue line representing the change in photovoltaic output current, forming a dynamic safety envelope. Even if the current surges significantly, the protection threshold also increases accordingly, ensuring no false tripping occurs during the 0.3 to 0.5 second fluctuation period, thus achieving the dynamic avoidance characteristic described in step S4.

[0051] In an optional implementation, step S4 constructs an adaptive tuning calculation model based on the characteristic boundary threshold parameters, establishes the dynamic correlation between the current protection setting threshold and the output power change, and introduces the real-time collected distributed power output as an input into the adaptive tuning calculation model to generate a dynamic reference value. This can be achieved by constructing a piecewise linear gain adjustment mechanism. According to the characteristic boundary threshold parameters, the output power of the distributed power supply is divided into a low-power region, a transition region, and a rated power region. A larger fixed bias is set in the low-power region to prevent measurement noise interference, and a smaller proportional following coefficient is set in the rated power region to improve fault sensitivity under high load. The gain parameters of each region are connected by smooth interpolation, so that the generated dynamic reference value has adaptive characteristics with variable sensitivity across the entire power range.

[0052] In another optional implementation, step S4 constructs an adaptive tuning calculation model based on the characteristic boundary threshold parameters, establishes the dynamic correlation between the current protection setting threshold and the output power change, and introduces the real-time collected distributed power output as an input quantity into the adaptive tuning calculation model to generate a dynamic reference value. This can also be achieved by introducing dynamic following logic with an inertial filtering stage. A first-order low-pass filter or moving average filter is connected in series in the transfer function, and the cutoff frequency of the filter is set using the time constant index in the characteristic boundary threshold parameters. When the real-time collected distributed power output experiences high-frequency jitter, the filter can smooth out the glitches in the input signal, so that the output dynamic reference value presents a stable trajectory, preventing frequent jumps in the protection setting value caused by transient changes in illumination, and ensuring the continuity and stability of the dynamic reference value on the time axis.

[0053] Specifically, in step S5, in response to the voltage drop characteristics at the grid connection point, the current characteristics during the fault period are calculated using the adaptive setting calculation model and combined with the fault location information, and the real-time current protection setting threshold is output, including the following steps E1 to E4: E1: Set the operating mode switching threshold for the grid connection point voltage to define the steady-state following zone and transient crossing zone of the distribution network; To accurately identify the current operating state of the system, a threshold for switching operating modes is preset in the adaptive tuning calculation model. This threshold is typically taken as the nominal voltage. 0.9 times. Real-time monitoring of grid connection point voltage amplitude. When the monitoring results meet When the system is in the steady-state following region, the distributed power source mainly outputs active power. When the monitoring results meet... When the system enters the transient ride-through region, it indicates that a short-circuit fault has occurred in the power grid, causing a voltage drop. At this time, the distributed power source will switch its output characteristics according to the low voltage ride-through control strategy, changing from power source characteristics to controlled current source characteristics.

[0054] E2: When the grid connection point voltage is detected to be in the steady-state following region, the adaptive tuning calculation model is driven to execute linear following logic to generate a dynamic following reference value based on the real-time distributed power output power; Within the steady-state following region, the model maintains the transfer function relationship established in step S4. The system continuously collects the real-time output current of the distributed power source. And directly use the formula Calculate the dynamic tracking reference value. At this time, the protection logic focuses on preventing false triggering due to power fluctuations caused by sudden changes in light intensity or wind speed disturbances. The setting value serves only as a safety ceiling that fluctuates with the load level and does not involve complex fault current deduction.

[0055] E3: When the grid connection point voltage drops to the transient ride-through zone, the fault ride-through response mechanism inside the model is activated, and the maximum fault contribution current during the fault period is deduced based on the voltage drop depth and the reactive current limit of the distributed power source. In this embodiment, when step E3 detects that the grid connection point voltage has dropped to the transient ride-through zone, the fault ride-through response mechanism within the model is activated. Based on the voltage drop depth and the reactive current limit of the distributed power source, the maximum fault contribution current during the fault period is extrapolated. Once the voltage drop triggers the transient ride-through logic, the model immediately calls the low-voltage ride-through algorithm. First, the voltage drop depth is calculated. Calculate the reactive current component that needs to be injected by distributed generation based on grid connection standards. Its calculation formula is In the formula This is the reactive power support factor. Also, the inverter's capacity limitations are considered. The maximum allowable active current component during a fault is calculated. Based on these two orthogonal components, the maximum fault contribution current that distributed generation sources may provide to the grid during fault ride-through is calculated by vector synthesis. This value accurately reflects the power supply's maximum output capability at the moment of failure.

[0056] In an optional implementation, when step E3 detects that the grid connection point voltage has dropped to the transient ride-through zone, the fault ride-through response mechanism inside the model is activated. Based on the voltage drop depth and the reactive current limit of the distributed power source, the maximum fault contribution current during the fault period can be estimated by searching a pre-set standardized low voltage ride-through characteristic curve library. The grid connection point voltage drop depth is discretized into several levels of indexes, and the reactive current injection ratio coefficient and active current reduction coefficient corresponding to the local grid connection guidelines are quickly matched in the local database. The estimated current vector magnitude under different drop depths is obtained directly by looking up the table, thereby avoiding complex real-time floating-point calculations and significantly improving the response speed and calculation real-time performance to grid voltage dips.

[0057] In another optional implementation, when step E3 detects that the grid connection point voltage drops to the transient ride-through zone, the fault ride-through response mechanism inside the model is activated. Based on the voltage drop depth and the reactive current limit of the distributed power source, the maximum fault contribution current during the fault period can be deduced by constructing an instantaneous power balance analysis model based on positive and negative sequence separation. The positive sequence component and negative sequence component at the moment of voltage drop are extracted using software phase-locked loop technology. The positive sequence active current command and reactive current command limits of the inverter under the target of suppressing negative sequence overcurrent are calculated respectively. The positive and negative sequence current components are superimposed by vector synthesis technology to reconstruct the three-phase current envelope during the fault period, thereby accurately deducing the peak value of the maximum fault contribution current under asymmetrical fault conditions such as single-phase grounding or two-phase short circuit.

[0058] E4: Correct the dynamic following reference value, and replace the maximum fault contribution current with a preset safety margin coefficient as the current real-time current protection setting threshold. To ensure that the protection device does not malfunction in the event of an external fault, the protection threshold must be raised above the maximum short-circuit contribution capability of the distributed power source. Therefore, the linear follower value in S4 is no longer used; instead, a reliability braking factor is introduced. (A value of 1.1 to 1.2 can be used), the maximum fault contribution current calculated in step E3 is amplified and corrected, and the final real-time setting threshold is calculated. The system will The current effective setting is immediately replaced, thereby achieving a seamless switch of the protection logic from steady-state avoidance mode to transient braking mode, ensuring selectivity under complex fault conditions.

[0059] Specifically, in step S6, a protection action decision logic is established, which compares the real-time fault current with the real-time current protection setting threshold. When the preset fault criterion is met, the protection action is triggered; otherwise, the system is maintained in normal operation. This includes the following steps F1 to F4: F1: Collect and analyze the real-time current vector of the distribution network, and extract the current amplitude characteristics for fault identification; A high-frequency sampling module is used to discretize the current in the distribution network lines to obtain a discrete-time series. A full-cycle Fourier transform algorithm is then used to perform spectral analysis on the sampled series, filtering out higher harmonic components and DC components, and extracting the real part of the fundamental current. According to the formula Calculate the fundamental amplitude of the current, and then calculate the fundamental amplitude. The core current amplitude characteristic was established for subsequent fault diagnosis. This characteristic can accurately reflect the changes in power frequency electrical quantities after a fault occurs.

[0060] F2: Introduce the real-time current protection setting threshold as a dynamic reference to construct an amplitude comparison criterion for the current amplitude characteristics; The current real-time current protection setting threshold is retrieved from the output of step S5. This threshold is a dynamic variable that changes in real time with the system operating conditions. In steady state, it is a fluctuating value that follows the power; in transient state, it is the corrected limit value for the fault contribution. The mathematical expression for constructing the amplitude comparison criterion is as follows: This criterion defines the activation conditions of the protection device. That is, the system will determine that there may be a fault risk only when the amplitude of the fundamental current detected in real time physically exceeds the dynamic safety threshold calculated by the model, thereby activating the subsequent logic process.

[0061] F3: When the current amplitude characteristic is detected to exceed the real-time current protection setting threshold, the time-domain verification mechanism for the fault duration is initiated. Set a de-shake timer. With the preset time limit constraint threshold At that moment The amplitude comparison criterion was detected to be met, that is At that time, the timer begins to accumulate. The purpose of this time-domain verification mechanism is to filter out microsecond- or millisecond-level instantaneous inrush currents caused by lightning overvoltage, large motor startup, or load switching, so as to prevent the protection device from malfunctioning due to non-faulty short-term disturbances.

[0062] F4: After determining that the time-domain verification mechanism meets the preset time limit constraint, confirm that the fault is a permanent fault, generate a trip command to drive the circuit breaker to act and isolate the fault area; otherwise, determine that it is a transient disturbance and block the trip command. The system continuously monitors the timer status. If throughout the entire timing cycle... Within this timeframe, if the amplitude comparison criterion remains valid—that is, if the fault current persists and exceeds the threshold—the anomaly is determined to be a permanent fault, and the protection device immediately generates a trip pulse signal to drive the circuit breaker to trip and disconnect the faulty line. Conversely, if the timer reaches [a certain threshold], [the fault will not be determined]. Previously, the real-time current amplitude fell below the threshold, i.e. If the anomaly is determined to be a transient disturbance, the system automatically resets the timer and blocks the trip command to maintain the normal power supply of the distribution network, thereby achieving selectivity and reliability of the protection action.

[0063] Refer again Figure 3 In step S6, a real fault is simulated when the system has run for 0.6 seconds. In the traditional fixed-value mode, the fault causes a current surge, triggering fault detection at 0.6 seconds. (Compare with reference...) Figure 5At 0.6 seconds, although the setting threshold of this invention had been rising with the increasing power and had increased to approximately 750 amperes, when a fault occurred at 0.6 seconds, the fault current instantaneously exceeded this dynamic threshold. At this point, the protection logic determined that the fault criterion was met. Figure 5 The diagram clearly shows the current curve at the moment of the fault, the threshold curve after the breakdown adjustment is terminated, indicating that the present invention does not sacrifice the ability to detect real faults while ensuring that fluctuations do not cause false tripping, and successfully triggers the protection action.

[0064] Example 3 illustrates an adaptive current protection method. It should be noted that the technical solution of this adaptive current protection system and the technical solution of the adaptive current protection method described above belong to the same concept. Details not described in detail in this embodiment can be found in the description of the adaptive current protection method described above.

[0065] This embodiment also provides an adaptive current protection system, including: The data generation and scenario simulation module is used to acquire historical operating data of the distribution network and construct a distributed power generation model. Based on the probability distribution characteristics of the historical operating data, it reproduces the intermittent statistical characteristics of the output power of the distributed power source and generates a typical scenario dataset containing multidimensional fluctuation information. The feature index extraction module is used to parse the typical scenario dataset and extract the cluster centers that characterize typical fluctuating conditions, thereby obtaining a set of feature indices that reflect the statistical characteristics of the output power fluctuation. The boundary parameter definition module is used to calculate the maximum fluctuation amplitude and rate of change range of the output power of the distributed power source under different operating conditions based on the set of feature indicators, and to define the feature boundary threshold parameters that distinguish between normal fluctuations and fault mutations. The adaptive model construction and benchmark generation module is used to construct an adaptive tuning calculation model based on the feature boundary threshold parameters, establish the dynamic correlation between the current protection setting threshold and the output power change, and introduce the real-time collected output power of the distributed power source as an input to the adaptive tuning calculation model to generate a dynamic benchmark value. The real-time threshold output module is used to respond to the voltage drop characteristics at the grid connection point, calculate the current characteristics during the fault period through the adaptive setting calculation model and the fault location information, and output the real-time current protection setting threshold. The protection action decision module is used to establish protection action decision logic, compare the real-time fault current with the real-time current protection setting threshold, and trigger protection action when the preset fault criterion is met; otherwise, it maintains normal system operation.

[0066] This embodiment also provides an electronic device applicable to an adaptive current protection method, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the adaptive current protection method proposed in the above embodiment.

[0067] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements an adaptive current protection method as described in the above embodiments.

[0068] The storage medium proposed in this embodiment and the method for implementing adaptive current protection proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0069] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0070] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for adaptive current protection, characterized in that, include: Historical operation data of the distribution network is acquired and a distributed power generation model is constructed. Based on the probability distribution characteristics of the historical operation data, the intermittent statistical characteristics of the output power of the distributed power source are reproduced, and a typical scenario dataset containing multidimensional fluctuation information is generated. The typical scenario dataset is analyzed and cluster centers representing typical fluctuating operating conditions are extracted to obtain a set of feature indicators reflecting the statistical characteristics of the output power fluctuation. Based on the set of characteristic indicators, calculate the maximum fluctuation amplitude and rate of change range of the output power of the distributed power source under different operating conditions, and define the characteristic boundary threshold parameters that distinguish between normal fluctuations and fault mutations. Based on the characteristic boundary threshold parameters, an adaptive tuning calculation model is constructed to establish the dynamic correlation between the current protection tuning threshold and the output power change, and the real-time collected output power of the distributed power source is used as an input to the adaptive tuning calculation model to generate a dynamic reference value. In response to the voltage drop at the grid connection point, the adaptive setting calculation model is used to calculate the current characteristics during the fault period by combining the fault location information, and the real-time current protection setting threshold is output. Establish a protection action decision logic, compare the real-time fault current with the real-time current protection setting threshold, and trigger the protection action when the preset fault criterion is met; otherwise, maintain normal system operation.

2. The adaptive current protection method as described in claim 1, characterized in that: The process of acquiring historical operating data of the distribution network and constructing a distributed generation power generation model, reproducing the intermittent statistical characteristics of the distributed generation power output based on the probability distribution characteristics of the historical operating data, and generating a typical scenario dataset containing multidimensional fluctuation information includes: The historical operational data is preprocessed to construct a sample set containing environmental correlation factors and time series factors, and the statistical characteristics of the sample set are defined as the target probability distribution space. Configure the distributed power output generation model to form a generation mechanism that maps the space of random disturbance variables to the space of simulated power output, for outputting virtual power sequences; Define a statistical consistency metric, map the virtual power sequence to the target probability distribution space, and calculate the distribution deviation between the virtual power sequence and the sample set; The internal parameters of the generation mechanism are iteratively optimized to minimize the distribution deviation, and the generation mechanism is corrected until the statistical characteristics of the virtual power sequence converge to the target probability distribution space, and the typical scenario dataset is output.

3. The adaptive current protection method as described in claim 2, characterized in that: The process of parsing the typical scenario dataset and extracting cluster centers representing typical fluctuating operating conditions yields a set of feature indicators reflecting the statistical characteristics of the output power fluctuations, including: Construct a multidimensional feature space adapted to the typical scenario dataset, and configure a kernel density metric function to measure the local compactness of samples; Traverse the scene sample points in the multidimensional feature space and calculate the drift vector of each scene sample point pointing to the local probability density maximum. Update the spatial coordinates of the scene sample points and drive the scene sample points to migrate continuously along the drift vector until they converge to the peak region of the local probability density distribution. The geometric center of the peak region is identified as the cluster center, and the statistical distribution attributes of the cluster center are analyzed to construct the feature index set.

4. The adaptive current protection method as described in claim 3, characterized in that: The calculation of the maximum fluctuation amplitude and rate of change range of the distributed power supply output power under different operating conditions based on the set of characteristic indicators, and the definition of the characteristic boundary threshold parameters that distinguish between normal fluctuations and fault mutations, include: The power amplitude dispersion and time-series rate of change distribution under typical fluctuating operating conditions represented by the set of characteristic indicators are statistically analyzed to establish a statistical confidence interval describing the probability distribution of power fluctuations. Based on the statistical confidence interval, an upper bound envelope trajectory and a lower bound envelope trajectory covering the normal power fluctuation range are generated; The extreme value attributes and maximum change slope of the upper bound envelope trajectory and the lower bound envelope trajectory in the time domain are analyzed to generate the feature boundary threshold parameter to quantify the critical boundary between normal fluctuations and fault mutations.

5. The adaptive current protection method as described in claim 1, characterized in that: The step of constructing an adaptive tuning calculation model based on the characteristic boundary threshold parameters, establishing the dynamic correlation between the current protection tuning threshold and the output power change, and introducing the real-time collected distributed power output as an input into the adaptive tuning calculation model to generate a dynamic reference value includes: Extract the statistical index representing the upper limit of fluctuation under normal operating conditions from the feature boundary threshold parameters, and map it to a safety floating coefficient for compensating for random power fluctuations; Construct a transfer function that includes a proportional follower term and a fixed bias term, and correct the gain parameter of the transfer function with the safety float coefficient to establish the dynamic correlation. The transfer function is driven to track the real-time input output power of the distributed power source, and a protection threshold benchmark with dynamic avoidance characteristics is calculated. The protection threshold reference is output as the dynamic reference value, so that the dynamic reference value remains above the real-time fluctuation trajectory of the distributed power supply output power during steady-state operation.

6. The adaptive current protection method as described in claim 5, characterized in that: The response to voltage dips at the grid connection point, using the adaptive setting calculation model and fault location information, calculates the current characteristics during the fault period and outputs real-time current protection setting thresholds, including: Set the operating mode switching threshold for the grid connection point voltage to define the steady-state following zone and transient crossing zone of the distribution network; When the grid connection point voltage is detected to be in the steady-state following region, the adaptive tuning calculation model is driven to execute linear following logic to generate a dynamic following reference value based on the real-time distributed power output. When the grid connection point voltage drops to the transient ride-through zone, the fault ride-through response mechanism inside the model is activated, and the maximum fault contribution current during the fault period is determined based on the voltage drop depth and the reactive current limit of the distributed power source. The dynamic tracking reference value is corrected by replacing the maximum fault contribution current with a preset safety margin coefficient and then replacing it with the current real-time current protection setting threshold.

7. The adaptive current protection method as described in claim 1, characterized in that: The establishment of the protection action decision logic, which compares the real-time fault current with the real-time current protection setting threshold, includes: Collect and analyze the real-time current vector of the distribution network, and extract the current amplitude characteristics for fault identification; The real-time current protection setting threshold is introduced as a dynamic reference to construct an amplitude comparison criterion for the current amplitude characteristics; When the current amplitude characteristic is detected to exceed the real-time current protection setting threshold, a time-domain verification mechanism for the fault duration is initiated. After determining that the time-domain verification mechanism meets the preset time limit constraint, the fault is confirmed to be a permanent fault, and a trip command is generated to drive the circuit breaker to operate and isolate the fault area; otherwise, it is determined to be a transient disturbance and the trip command is blocked.

8. A system for adaptive current protection, employing an adaptive current protection method as described in any one of claims 1 to 7, characterized in that, include: The data generation and scenario simulation module is used to acquire historical operating data of the distribution network and construct a distributed power generation model. Based on the probability distribution characteristics of the historical operating data, it reproduces the intermittent statistical characteristics of the output power of the distributed power source and generates a typical scenario dataset containing multidimensional fluctuation information. The feature index extraction module is used to parse the typical scenario dataset and extract the cluster centers that characterize typical fluctuating conditions, thereby obtaining a set of feature indices that reflect the statistical characteristics of the output power fluctuation. The boundary parameter definition module is used to calculate the maximum fluctuation amplitude and rate of change range of the output power of the distributed power source under different operating conditions based on the set of feature indicators, and to define the feature boundary threshold parameters that distinguish between normal fluctuations and fault mutations. The adaptive model construction and benchmark generation module is used to construct an adaptive tuning calculation model based on the feature boundary threshold parameters, establish the dynamic correlation between the current protection setting threshold and the output power change, and introduce the real-time collected output power of the distributed power source as an input to the adaptive tuning calculation model to generate a dynamic benchmark value. The real-time threshold output module is used to respond to the voltage drop characteristics at the grid connection point, calculate the current characteristics during the fault period through the adaptive setting calculation model and the fault location information, and output the real-time current protection setting threshold. The protection action decision module is used to establish protection action decision logic, compare the real-time fault current with the real-time current protection setting threshold, and trigger protection action when the preset fault criterion is met; otherwise, it maintains normal system operation.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the adaptive current protection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the adaptive current protection method according to any one of claims 1 to 7.

Images