Island division and fault reconstruction method and system for power distribution system

Abstract

The invention provides a power distribution system island division and fault reconstruction method and system. The method comprises the steps of obtaining a topological structure, load characteristic data and renewable energy output characteristic data of a power distribution system, and generating a mathematical model of the power distribution system; performing optimization calculation based on a mixed meta-heuristic algorithm framework and an adaptive step length gradient descent method; performing fuzzy binning processing on the optimization result to obtain a calibrated reconstruction optimization result; and finally, carrying out island division and fault reconstruction based on a calibration result. According to the method, the uncertainty influence of renewable energy sources can be effectively processed, and the accuracy and reliability of island division and fault reconstruction of the power distribution system are improved.

Description

Technical Field

[0001] This application relates to the field of power distribution system technology, and in particular to a method and system for islanding and fault reconfiguration in power distribution systems. Background Technology

[0002] With the continuous development of power systems, the scale of distribution networks is expanding and their structures are becoming increasingly complex. At the same time, the large-scale integration of renewable energy sources presents new challenges to the safe and stable operation of distribution systems. When a distribution system failure occurs, timely islanding and fault reconfiguration are necessary to ensure reliable power supply.

[0003] Currently, common power distribution system fault reconfiguration methods are mainly based on heuristic or optimization algorithms. These methods search for the optimal reconfiguration scheme by constructing objective functions and constraints. For example, intelligent optimization methods such as genetic algorithms and particle swarm optimization are used for power distribution network reconfiguration optimization.

[0004] Some technical solutions introduce distribution system reconfiguration methods that consider uncertainties, describing the output characteristics of renewable energy through probabilistic models and taking their impact into account during the optimization process. This method evaluates the reliability of system reconfiguration schemes by establishing probabilistic constraint models.

[0005] However, existing technologies have the following problems: First, they do not adequately consider the uncertainties of renewable energy and lack effective risk quantification indicators; second, the optimization algorithms are prone to getting trapped in local optima and have low convergence efficiency; and third, the error analysis and calibration of the optimization results are not perfect, which affects the reliability of the reconstruction scheme. Summary of the Invention

[0006] In view of this, this application provides a method and system for islanding and fault reconfiguration in a power distribution system, which solves the problems in the prior art that the uncertainty impact of renewable energy is not fully considered, the optimization algorithm is prone to getting trapped in local optima, and the error analysis and calibration of the optimization results are not perfect.

[0007] This application provides a method for islanding and fault reconfiguration in a power distribution system, including:

[0008] The topology, load characteristics, and renewable energy output characteristics of the power distribution system are acquired. A basic model of the power distribution system is established based on node-branch relationships and power flow constraints. The Conditional Value at Risk (CVaR) index is introduced to quantify the impact of renewable energy uncertainties, and a mathematical model of the power distribution system is generated.

[0009] Using the mathematical model of the power distribution system, optimization calculations are performed based on the hybrid meta-heuristic algorithm framework and the gradient descent method with adaptive step size to obtain the initial reconstruction optimization results;

[0010] The initial reconstruction optimization results are subjected to fuzzy binning. The data samples are dynamically binned based on the fuzzy clustering algorithm. The error statistical characteristics of each bin interval are calculated to obtain the calibrated reconstruction optimization results.

[0011] Based on the reconfiguration optimization results after calibration, the islanding conditions are judged by comprehensively considering power capacity, load demand and network constraints, the reconfiguration operation sequence is optimized, the final islanding and fault reconfiguration scheme is generated, and the islanding and fault reconfiguration of the power distribution system are performed based on the islanding and fault reconfiguration scheme.

[0012] Acquire topology, load characteristic data, and renewable energy output characteristic data of the distribution system. Establish a basic model of the distribution system based on node-branch relationships and power flow constraints. Introduce the Conditional Value at Risk (CVaR) index to quantify the impact of renewable energy uncertainties, generating a mathematical model of the distribution system, including:

[0013] Based on the topology and load characteristic data of the power distribution system, node power injection balance equations and voltage constraints are established to obtain the power flow model of the power distribution system.

[0014] For the renewable energy output characteristic data, the Beta distribution is used to describe the probability density function of photovoltaic output, and the Weibull distribution is used to describe the probability distribution of wind power output, so as to obtain the probabilistic characteristic model of renewable energy.

[0015] By combining the power flow model of the power distribution system and the probabilistic characteristic model of the renewable energy source, and using linearization to transform the conditional value at risk (CVaR) into a linear constraint, the mathematical model of the power distribution system is obtained.

[0016] Based on the topology and load characteristic data of the power distribution system, node power injection balance equations and voltage constraints are established to obtain the power flow model of the power distribution system, including:

[0017] Obtain the node admittance matrix elements, node voltage phase angle difference data, and ±5% constraint range of node voltage amplitude. Calculate the active and reactive power injection at the node based on the node power injection balance equation to obtain the power balance equation set.

[0018] By applying the voltage constraint condition to the power balance equations, the power flow model of the power distribution system is obtained.

[0019] Using the mathematical model of the power distribution system, optimization calculations are performed based on a hybrid meta-heuristic algorithm framework and an adaptive step-size gradient descent method to obtain initial reconstruction optimization results, including:

[0020] The optimization population is initialized based on the Latin hypercube sampling method, a penalty function containing the default degree is constructed, and a hybrid metaheuristic optimization framework is established.

[0021] Iterative optimization is performed using the hybrid meta-heuristic optimization framework and gradient descent method. The search step size is dynamically adjusted according to the changes in the objective function to obtain the initial reconstruction optimization result.

[0022] Iterative optimization is performed using the aforementioned hybrid meta-heuristic optimization framework and gradient descent method. The search step size is dynamically adjusted according to changes in the objective function to obtain the initial reconstruction optimization results, including:

[0023] Calculate the continuous change in an individual's position, and determine that local convergence has been achieved when the change is less than a preset threshold, thus obtaining the local convergence result.

[0024] Calculate the population diversity index. When the population diversity is below the threshold and the optimal solution has not improved for five consecutive generations, global convergence is determined, and the initial reconstruction optimization result is obtained.

[0025] The initial reconstruction optimization results are subjected to fuzzy binning. Based on a fuzzy clustering algorithm, the data samples are dynamically binned, and the error statistical characteristics of each bin interval are calculated to obtain the calibrated reconstruction optimization results, including:

[0026] The initial reconstruction optimization results are subjected to z-score standardization to extract feature parameters such as load change rate, voltage deviation and power factor, resulting in a standardized dataset.

[0027] The standardized dataset is dynamically binned using the fuzzy C-means clustering algorithm. The optimal number of bins is determined based on the silhouette coefficient, and the sample membership degree is calculated to obtain the fuzzy binning result.

[0028] The error statistics and conditional error correction functions of each bin interval in the fuzzy binning result are calculated to obtain the calibrated reconstruction optimization result.

[0029] Calculate the error statistics and conditional error correction function for each bin interval in the fuzzy binning result to obtain the calibrated reconstruction optimization result, including:

[0030] Calculate the mean and standard deviation of the error for each bin interval, construct an error correction model, and obtain the initial correction coefficients;

[0031] The initial correction coefficients are dynamically updated using an exponential smoothing method to obtain the calibrated reconstruction optimization result.

[0032] Based on the calibrated reconfiguration optimization results, considering power capacity, load demand, and network constraints, islanding conditions are determined, the reconfiguration operation sequence is optimized, and the final islanding and fault reconfiguration scheme is generated, including:

[0033] Calculate the matching degree evaluation index between distributed power sources and loads, and perform power balance constraints and network connectivity analysis to obtain the judgment results of islanding conditions;

[0034] An improved dynamic programming algorithm is used to design a switch operation sequence, and an operation cost penalty term is introduced to obtain an optimized reconstruction operation sequence.

[0035] The optimized reconfiguration operation sequence is evaluated for reliability based on Monte Carlo simulation, and the power supply reliability index is calculated to obtain the final islanding and fault reconfiguration scheme.

[0036] The optimized reconfiguration operation sequence is evaluated for reliability based on Monte Carlo simulation, and power supply reliability indices are calculated to obtain the final islanding and fault reconfiguration scheme, including:

[0037] Scenario samples containing renewable energy output and load demand are generated. These scenario samples are randomly sampled based on Beta and Weibull distributions. The system average outage frequency index (SAIFI) and the system average outage duration index (SAIDI) are calculated to obtain the reliability assessment results.

[0038] Confidence interval analysis is performed on the reliability assessment results to evaluate statistical reliability and obtain the final islanding and fault reconstruction scheme.

[0039] This application also provides a power distribution system islanding and fault reconfiguration system, including:

[0040] The mathematical model building module is used to acquire the topology, load characteristic data and renewable energy output characteristic data of the power distribution system, establish the basic model of the power distribution system based on the node-branch relationship and power flow constraints, and introduce the Conditional Value at Risk (CVaR) index to quantify the impact of renewable energy uncertainty and generate the mathematical model of the power distribution system.

[0041] The fault reconfiguration optimization module is used to perform optimization calculations based on the mathematical model of the power distribution system, using a hybrid meta-heuristic algorithm framework and an adaptive step-size gradient descent method, to obtain the initial reconfiguration optimization results.

[0042] The error estimation module is used to perform fuzzy binning on the initial reconstruction optimization results, dynamically binning the data samples based on the fuzzy clustering algorithm, calculating the error statistical characteristics of each bin interval, and obtaining the calibrated reconstruction optimization results.

[0043] The scheme generation module is used to determine the islanding conditions based on the calibrated reconfiguration optimization results, taking into account power capacity, load demand and network constraints, optimize the reconfiguration operation sequence, generate the final islanding and fault reconfiguration scheme, and perform islanding and fault reconfiguration of the power distribution system based on the islanding and fault reconfiguration scheme.

[0044] This application embodiment also provides a computer device, the computer device comprising:

[0045] At least one processor; and,

[0046] A memory communicatively connected to the at least one processor; wherein,

[0047] The memory stores instructions that can be executed by the at least one processor, which enables the at least one processor to perform the above-described power distribution system islanding and fault reconfiguration method.

[0048] This application also provides a computer-readable storage medium that stores computer instructions for causing a computer to execute the above-described method for islanding and fault reconfiguration of a power distribution system.

[0049] This application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the above-described method for islanding and fault reconfiguration of a power distribution system.

[0050] This application has the following technical effects:

[0051] 1. By introducing the Conditional Value at Risk (CVaR) index, the impact of uncertainty in renewable energy is effectively quantified, thus improving the accuracy of the model;

[0052] 2. By adopting a hybrid meta-heuristic algorithm framework and an adaptive step-size gradient descent method, the convergence efficiency of the optimization algorithm is improved, and local optima are avoided.

[0053] 3. The dynamic binning method based on fuzzy clustering enables accurate calibration of optimization results and improves the reliability of the reconstruction scheme;

[0054] 4. The reliability of the reconstruction scheme was evaluated through Monte Carlo simulation, which ensured the practicality and effectiveness of the scheme. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0056] Figure 1 A flowchart illustrating the power distribution system islanding and fault reconfiguration method provided in this application embodiment;

[0057] Figure 2 This is a schematic diagram illustrating the construction process of the mathematical model of the power distribution system in the embodiments of this application;

[0058] Figure 3 This is a schematic diagram of the optimization process of the hybrid metaheuristic algorithm framework in the embodiments of this application;

[0059] Figure 4 This is a schematic diagram of the fuzzy binning process in an embodiment of this application;

[0060] Figure 5 This is a schematic diagram illustrating the process of generating the island partitioning and fault reconstruction scheme in the embodiments of this application. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0062] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0063] In this document, the term "and / or" merely describes a relationship, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0064] like Figure 1 As shown in the figure, this application provides a method for islanding and fault reconfiguration of a power distribution system, including:

[0065] S1: Obtain the topology, load characteristics, and renewable energy output characteristics of the power distribution system. Establish a basic model of the power distribution system based on node-branch relationships and power flow constraints. Introduce the Conditional Value at Risk (CVaR) index to quantify the impact of renewable energy uncertainties and generate a mathematical model of the power distribution system.

[0066] In constructing a mathematical model of a power distribution system, the first step is to comprehensively acquire the system's basic data. Topology data includes physical characteristics such as the node connections, line parameters, and transformer parameters of the distribution network. This data can typically be obtained from a Geographic Information System (GIS) or asset management system for the power distribution system. For each line, its starting and ending node numbers, impedance parameters, rated capacity, and other characteristic parameters need to be recorded. Transformer data includes parameters such as rated capacity, impedance, and turns ratio.

[0067] Obtaining load characteristic data requires considering the electricity consumption characteristics of different types of users. Residential users typically exhibit a clear daily periodicity in their load, with significant seasonal variations. Industrial users' load characteristics are closely related to production shifts and may exhibit large power fluctuations. Commercial users' electricity consumption patterns are primarily influenced by business hours. Statistical analysis of historical load data can extract typical load curves and fluctuation characteristics for various user types, which can be described using load forecasting models.

[0068] Obtaining renewable energy output characteristic data requires categorizing different types of renewable energy. For photovoltaic power generation, historical power generation data, irradiance data, and other environmental factor information need to be collected, as these data exhibit significant meteorological dependence and periodicity. Wind power output data, on the other hand, is mainly related to meteorological conditions such as wind speed and direction, and its randomness and intermittency are more pronounced. By analyzing this historical data, corresponding probability distribution models can be established to describe its output characteristics.

[0069] When establishing the basic model of a power distribution system, a network topology matrix needs to be constructed based on the node-branch relationships. This matrix describes the connection relationships between various nodes in the system and is the foundation for power flow calculations. Simultaneously, power flow constraints need to be considered, including node power balance constraints, line capacity constraints, and voltage constraints. Node power balance constraints ensure that the injected power and outflow power at each node are equal; line capacity constraints ensure that the line load does not exceed the rated value; and voltage constraints ensure that the node voltage remains within the allowable range.

[0070] To accurately quantify the uncertainties surrounding renewable energy, this application innovatively introduces the Conditional Value at Risk (CVaR) metric. As a typical risk measure, CVaR effectively characterizes the tail risk of renewable energy output. By setting appropriate confidence levels, CVaR can assess the risks the system may face under extreme conditions. This approach not only considers the expected value of renewable energy output but also focuses on the potential impact of its fluctuations on the system.

[0071] During model construction, various operational constraints also need to be considered. For example, the radial operation constraint requires the system to maintain a radial topology during normal operation; the power supply radius constraint limits the power supply range of the distribution network; and the equipment switching frequency constraint considers the service life of the switching equipment. These constraints together constitute the constraint set of the mathematical model of the power distribution system.

[0072] Through the above steps, a complete mathematical model of the power distribution system is finally formed. This model includes both the static characteristics of the system (such as network topology and equipment parameters) and reflects the dynamic characteristics of the system (such as load changes and fluctuations in renewable energy output). The accuracy of the model directly affects the effectiveness of subsequent optimization calculations; therefore, special attention must be paid to the accuracy and completeness of the data during the model construction process.

[0073] The advantages of this modeling method are: first, it comprehensively considers various characteristics and constraints of the power distribution system; second, by introducing the CVaR index, it innovatively solves the problem of quantifying the uncertainty of renewable energy; and finally, the model has strong practicality and scalability, and can adapt to power distribution systems of different sizes and types.

[0074] Among them, such as Figure 2 As shown, S1 specifically includes:

[0075] S1.1: Based on the topology and load characteristic data of the power distribution system, establish the node power injection balance equation and voltage constraints to obtain the power flow model of the power distribution system;

[0076] In a specific embodiment of this application, the construction process of the power distribution system mathematical model first requires establishing a power flow model of the power distribution system (step S1.1). Specifically, based on the topology information of the power distribution system, each element Yij of the node admittance matrix Y can be obtained, where i and j represent any two nodes in the power distribution system. For each node i, its voltage phase angle difference δi and node voltage amplitude Vi need to meet the operating constraints. Generally, the allowable fluctuation range of the voltage amplitude is ±5%, i.e., 0.95≤Vi≤1.05. Based on the node voltage and admittance matrix, the node power injection balance equation can be established, and the active power injection Pi and reactive power injection Qi of node i can be calculated. It should be noted that the power injection balance equation needs to consider the power demand characteristics of the load nodes and the output characteristics of distributed power sources. By combining the power balance equations of all nodes and applying voltage constraints, the complete power flow model of the power distribution system can be obtained.

[0077] S1.2: For the renewable energy output characteristic data, the probability density function of photovoltaic output is described by the Beta distribution, and the probability distribution of wind power output is described by the Weibull distribution, so as to obtain the probabilistic characteristic model of renewable energy.

[0078] Secondly, to accurately describe the uncertainty characteristics of renewable energy (step S1.2), this application employs different probability distribution models to characterize the output characteristics of photovoltaic and wind power. For example, for photovoltaic output, its probability density function can be described by a Beta distribution, with the probability density function f(x; α, β) = xα⁻¹(1-x)β⁻¹ / B(α, β), where α and β are shape parameters, and B(α, β) is the Beta function. By analyzing historical data, the most suitable values ​​of α and β can be estimated. For wind power output, a Weibull distribution is used, with its probability density function f(x; k, c) = (k / c)(x / c)k⁻¹exp[-(x / c)k], where k is the shape parameter and c is the scale parameter. The probability distribution parameters of wind power output can also be estimated by fitting historical data. In summary, by establishing these two probability distribution models, the stochastic output characteristics of renewable energy can be effectively characterized.

[0079] S1.3: Combine the power flow model of the power distribution system and the probabilistic characteristic model of the renewable energy, and use linearization to transform the conditional value of risk (CVaR) into a linear constraint to obtain the mathematical model of the power distribution system.

[0080] Finally, to incorporate the uncertainty of renewable energy into the optimization decision-making of the distribution system (step S1.3), this application introduces Conditional Value at Risk (CVaR) as a risk quantification index. First, the terms involving renewable energy output in the power flow model of the distribution system are replaced with their probabilistic characteristic models. Second, considering that the nonlinear characteristics of the CVaR index may increase the difficulty of solving the problem, this application uses a piecewise linearization method to transform the CVaR constraints into a set of linear constraints. Specifically, given a confidence level α (usually taken as 0.95), the probability distribution interval of renewable energy output is divided into several sub-intervals, and the original nonlinear function is approximated by a linear function within each sub-interval. In this way, by introducing auxiliary variables, the CVaR constraints can be transformed into a set of linear inequality constraints. It should be noted that the number of sub-intervals needs to be balanced between computational accuracy and solution efficiency. After the above processing, a mathematical model of the distribution system incorporating the uncertainty of renewable energy is finally obtained.

[0081] Specifically, S1.1 includes:

[0082] S1.1.1: Obtain the node admittance matrix elements, node voltage phase angle difference data, and the ±5% constraint range of the node voltage amplitude value. Calculate the active and reactive power injection at the node based on the node power injection balance equation to obtain the power balance equation set.

[0083] In step S1.1.1, it is first necessary to obtain the node admittance matrix elements of the system in detail. Specifically, for any two nodes i and j in the distribution system, their admittance matrix element Yij can be calculated based on the line parameters. For example, for the conductor between adjacent nodes i and j, its admittance can be calculated using the conductor's resistance and reactance parameters: Yij = 1 / (Rij + jXij), where Rij and Xij represent the resistance and reactance values ​​of the conductor between nodes i and j, respectively. In addition, it is also necessary to obtain the node voltage phase angle difference data δij, which represents the voltage phase difference between nodes i and j. In practical applications, the voltage phase angle of a reference node (such as a substation node) is usually set to 0 degrees, and the voltage phase angles of other nodes are obtained through measurement or calculation. The voltage amplitude value Vi of each node in the system needs to meet the requirements of the power system operation specifications, generally requiring its variation range to be controlled within ±5% of the nominal value, i.e., 0.95pu≤Vi≤1.05pu. Based on the above parameters and the nodal power injection balance equation, the active power injection Pi and reactive power injection Qi of each node can be calculated. It should be noted that for load nodes, the power injection is negative, representing power consumption; for generator nodes, the power injection is positive, representing power supplied to the system.

[0084] S1.1.2: Apply the voltage constraint condition to the power balance equation set to obtain the power flow model of the power distribution system.

[0085] In step S1.1.2, voltage constraints need to be applied to the above power balance equations. First, for each node i, its power balance equation can be expressed as: Pi=Vi∑

[0086] The equations (Vj(Gijcosδij+Bijsinδij)) and Qi=Vi∑(Vj(Gijsinδij-Bijcosδij)) are used, where Gij and Bij are the conductance and susceptance components of the nodal admittance matrix, respectively. These equations must simultaneously satisfy the nodal voltage constraint: 0.95≤Vi≤1.05. Furthermore, line capacity constraints must be considered, meaning the apparent power flowing through any line cannot exceed its rated capacity. By comprehensively considering these constraints, a complete power flow model for the distribution system can be obtained. It is important to note that in practical applications, power flow calculations often employ iterative methods, such as the Newton-Raphson method, which requires considering convergence requirements in the model. This power flow model will serve as the basis for subsequent optimization calculations, providing crucial technical support for islanding and fault reconfiguration in distribution systems.

[0087] S2: Using the mathematical model of the power distribution system, optimization calculations are performed based on the hybrid meta-heuristic algorithm framework and the gradient descent method with adaptive step size to obtain the initial reconstruction optimization results;

[0088] After obtaining the mathematical model of the power distribution system, the core task of optimization calculation is to find the optimal reconfiguration scheme. This application's embodiment employs a hybrid metaheuristic algorithm framework combined with an adaptive step-size gradient descent method. This innovative combination fully leverages the advantages of both algorithms, ensuring global search capability while improving local convergence efficiency.

[0089] The hybrid metaheuristic algorithm framework is designed based on the synergy of multiple optimization strategies. First, Latin hypercube sampling is used for population initialization. This method ensures that the initial solutions are uniformly distributed in the solution space, effectively avoiding sampling blind spots that may arise from random initialization. During initialization, each individual represents a potential reconstruction scheme, including decision variables such as switch state combinations and load transfer strategies. Simultaneously, to ensure the feasibility of the initial solutions, constraint checks are performed on each individual to ensure that it meets the basic operational requirements of the system.

[0090] The algorithm employs a multi-level optimization strategy in its search process. At the global search level, diverse crossover and mutation operators are designed to enhance population diversity. Crossover operations mainly include single-point crossover, multi-point crossover, and uniform crossover, with different crossover strategies suitable for different decision variable characteristics. Mutation operations, through random perturbation, help individuals escape local optima. In particular, the embodiments of this application consider the topological constraints of the power distribution system during the mutation process, ensuring that the mutated solution still meets the system's operational requirements.

[0091] At the local search level, an innovative adaptive step-size gradient descent method is introduced. Traditional fixed step-size strategies often struggle to balance convergence speed and accuracy, while the adaptive step-size mechanism dynamically adjusts the step size based on the current search state. When the objective function value decreases rapidly, a larger step size is used to accelerate convergence; when approaching the optimal solution, the step size is automatically reduced to improve accuracy. The step size adjustment is based on multiple factors, including the gradient information of the objective function, the effect of the current iteration, and the degree of constraint violation.

[0092] To improve the practicality of the algorithm, this application also designs a dedicated constraint handling mechanism. Firstly, the penalty function is designed by introducing a dynamic penalty factor, which transforms the degree of constraint violation into a penalty term in the objective function. The magnitude of the penalty factor is dynamically adjusted during the iteration process, being relatively lenient in the early stages of the search to expand the feasible solution space, and gradually increasing in the later stages to ensure the feasibility of the final solution.

[0093] The convergence criteria employ a multi-criteria strategy. First, there's a local convergence criterion, which determines local convergence by monitoring continuous changes in individual positions. Second, there's a global convergence criterion, primarily observing population diversity and the improvement in the optimal solution. When population diversity falls below a threshold and the optimal solution fails to improve significantly over multiple generations, the algorithm is considered to have achieved global convergence.

[0094] The algorithm's optimization process also includes an adaptive parameter adjustment mechanism. For example, the crossover and mutation probabilities are dynamically adjusted according to the population's evolutionary state, prioritizing exploration in the early stages and focusing more on local improvements in the later stages. Simultaneously, to balance computational efficiency and optimization effectiveness, a dynamic population size adjustment strategy is designed, which adaptively adjusts the population size based on the problem's scale and complexity.

[0095] Finally, the embodiments of this application also consider the feasibility requirements in practical engineering. During the optimization process, engineering constraints such as operation count limits and device switching time intervals are introduced to ensure that the final refactoring scheme has strong practical significance. Simultaneously, a solution repair mechanism is designed; when an infeasible solution is generated during the search, it can be repaired into a feasible solution through heuristic rules, improving the robustness of the algorithm.

[0096] The main advantages of this optimization method are: first, the hybrid algorithm framework provides powerful global search capabilities; second, the adaptive step size mechanism significantly improves the convergence efficiency of the algorithm; and finally, multi-level constraint handling ensures the practicality of the solution. In this way, high-quality initial reconstruction optimization results can be effectively obtained, laying a solid foundation for subsequent scheme optimization.

[0097] Among them, such as Figure 3 As shown, S2 specifically includes:

[0098] S2.1: The optimization population is initialized based on the Latin hypercube sampling method, a penalty function containing the default degree is constructed, and a hybrid metaheuristic optimization framework is established.

[0099] After obtaining the mathematical model of the power distribution system, a hybrid metaheuristic optimization framework needs to be established for solution (step S2.1). Specifically, the Latin hypercube sampling method is first used to initialize the optimization population. This sampling method ensures that the initial population is uniformly distributed in the solution space, improving search efficiency. During initialization, each individual represents a possible reconfiguration scheme, including decision variables such as switch states and island partitioning. It should be noted that, to ensure the feasibility of the solution, the radial operation requirements and power supply radius constraints of the power distribution system need to be considered when constructing the initial population. In addition, this application constructs a penalty function containing a violation degree to evaluate the quality of the reconfiguration scheme. This penalty function not only considers the objective function value (such as the power supply reliability index) but also introduces a penalty term for the degree of constraint violation, including power balance constraints, voltage constraints, and network radial constraints. By reasonably setting the penalty coefficient, the search direction can be effectively guided during the optimization process, avoiding the generation of infeasible solutions.

[0100] S2.2: Iterative optimization is performed using the hybrid meta-heuristic optimization framework and gradient descent method. The search step size is dynamically adjusted according to the changes in the objective function to obtain the initial reconstruction optimization result.

[0101] After establishing the optimization framework, iterative optimization is performed using a hybrid meta-heuristic algorithm and gradient descent (step S2.2). First, the hybrid meta-heuristic framework combines various optimization operators, including local search based on a greedy strategy and global exploration based on probabilistic selection. In each iteration, different optimization operators are dynamically selected based on the fitness value of each individual to balance the algorithm's exploration and utilization capabilities. Second, a gradient descent method with an adaptive step size is introduced in the local search phase. Specifically, the search step size is dynamically adjusted based on the changes in the objective function near the current solution. When the objective function decreases rapidly, a larger step size is used to accelerate convergence; when approaching the optimal solution, the step size is reduced to improve accuracy. This adaptive mechanism effectively improves the algorithm's convergence efficiency. It should be noted that, considering the discrete nature of the problem, appropriate continuous processing is required when applying the gradient descent method, and the results are mapped back to the discrete space after the search is completed. Through the above optimization process, the initial reconstructed optimization results can be obtained.

[0102] Specifically, S2.2 includes:

[0103] S2.2.1: Calculate the continuous change in the individual's position. When it is less than a preset threshold, it is determined that local convergence has been achieved, and the local convergence result is obtained.

[0104] During optimization, to accurately determine whether the algorithm has reached convergence, analysis is needed from both local and global perspectives. The first step is to assess local convergence. Specifically, in each iteration, the continuous change in the position of each individual in the population needs to be calculated. This change reflects the individual's movement in the solution space, including changes in switching states and power distribution. When the continuous change of an individual is less than a preset threshold, it indicates that the individual has stabilized in a local region. It's important to note that the selection of this preset threshold requires a trade-off between convergence accuracy and computational efficiency. If the threshold is set too small, it may increase unnecessary iterations; if the threshold is too large, it may lead to premature convergence. In practical applications, a suitable threshold can be determined experimentally based on the system size and the characteristics of the optimization objective. When the change in an individual's position remains consistently below the threshold, local convergence can be considered achieved, and the current local convergence result is recorded.

[0105] S2.2.2: Calculate the population diversity index. When the population diversity is lower than the threshold and the optimal solution has not improved for five consecutive generations, global convergence is determined, and the initial reconstruction optimization result is obtained.

[0106] At this stage, the global convergence status of the optimization algorithm is mainly evaluated by calculating the population diversity index. The population diversity index reflects the degree of difference between individuals in the current population and can be measured from multiple dimensions, such as the distribution range of decision variables and the distance between individuals. When the population diversity index is below a set threshold, it indicates that the population tends to concentrate and the search space has been fully explored. In addition, the evolution of the optimal solution needs to be observed. If, during five consecutive iterations, the optimal solution has not significantly improved (i.e., the improvement in the objective function value is less than the set accuracy requirement), and the population diversity is also at a low level, then the algorithm can be considered to have reached global convergence. It should be noted that the "five generations" setting here is based on empirical values ​​and can be appropriately adjusted according to the specific characteristics of the problem in practical applications. When both the judgment conditions of population diversity and optimal solution stability are met, the optimization algorithm can be considered to have reached global convergence, and the current optimization result is used as the initial reconstruction optimization result.

[0107] S3: Perform fuzzy binning on the initial reconstruction optimization results, dynamically bin the data samples based on the fuzzy clustering algorithm, calculate the error statistical characteristics of each bin interval, and obtain the calibrated reconstruction optimization results;

[0108] After obtaining the initial reconfiguration optimization results, fuzzy binning and error calibration are required to improve the reliability and accuracy of the scheme. This process begins with data preprocessing of the optimization results. Through z-score standardization, characteristic parameters with different dimensions are transformed to a unified scale space, eliminating the influence of dimensional differences. Key characteristic parameters include system operating indicators such as load change rate, voltage deviation, and power factor, which comprehensively reflect the operating status of the power distribution system.

[0109] In the fuzzy binning process, this application innovatively employs a fuzzy C-means clustering algorithm for dynamic binning. Compared to traditional fixed binning methods, dynamic binning can better adapt to the distribution characteristics of the data. The number of bins is determined using the silhouette coefficient as an evaluation index; by trying different numbers of bins multiple times, the scheme with the optimal silhouette coefficient is selected. During the clustering process, each data sample is assigned a membership value, reflecting its degree of belonging to different bins. This fuzzy classification method can better handle uncertainties at data boundaries.

[0110] The calculation of error statistics employs a hierarchical analysis method. First, basic statistics, including the mean error and standard deviation, are calculated for each bin. These statistics reflect the central tendency and dispersion of the data within the current bin. Second, a conditional error correction function is constructed, which considers the membership information of the data samples and can adaptively adjust the error correction intensity according to actual operating conditions. Finally, an exponential smoothing method is used to dynamically update the correction coefficients, ensuring the stability and adaptability of the calibration process.

[0111] Among them, such as Figure 4 As shown, S3 specifically includes:

[0112] S3.1: Perform z-score standardization on the initial reconstruction optimization results, extract feature parameters such as load change rate, voltage deviation and power factor, and obtain a standardized dataset;

[0113] Specifically, the initial reconstruction and optimization results are z-score standardized to transform data with different dimensions into a unified scale space. During data feature extraction, three key parameters are emphasized: load change rate, voltage deviation, and power factor. The load change rate reflects the dynamic changes in system load over time, requiring consideration of electricity consumption patterns for different time periods and user types. Voltage deviation characterizes the degree to which node voltage deviates from its rated value, directly affecting power supply quality. The power factor reflects the economic operating level of the system. Furthermore, other feature parameters can be added based on actual needs. Standardization eliminates the influence of different dimensions among feature parameters, making subsequent cluster analysis more accurate and reliable. It should be noted that outlier handling is crucial during standardization; methods such as median substitution can be used for data correction.

[0114] S3.2: The standardized dataset is dynamically binned using the fuzzy C-means clustering algorithm. The optimal number of bins is determined based on the silhouette coefficient, and the sample membership degree is calculated to obtain the fuzzy binning result.

[0115] Next, fuzzy C-means clustering is used to dynamically bin the standardized dataset. First, the optimal number of bins needs to be determined. This application uses the silhouette coefficient as the evaluation metric, and by trying different numbers of bins multiple times, the scheme with the largest silhouette coefficient is selected as the final number of bins. During the clustering process, each data sample is assigned a membership value, representing its degree of belonging to each bin. Compared with traditional hard binning methods, fuzzy clustering can better handle fuzziness at data boundaries and improve the rationality of the binning results. It is important to note that when calculating sample membership, the physical meaning of the data features should be considered, and different weighting coefficients can be set for different feature parameters. In this way, a fuzzy binning result containing sample membership information is finally obtained.

[0116] S3.3: Calculate the error statistics and conditional error correction function of each bin interval in the fuzzy binning result to obtain the calibrated reconstruction optimization result.

[0117] Finally, for the fuzzy binning results, it is necessary to calculate the error statistical characteristics of each bin interval (step S3.3). First, statistical analysis is performed on the data samples within each bin interval, focusing on calculating statistical quantities such as the mean and standard deviation of the error. These statistical quantities reflect the error distribution characteristics of the current reconstruction optimization results under different operating conditions. Second, based on these statistical characteristics, a conditional error correction function is constructed. This function considers the membership information of the data samples and can adaptively adjust the error correction intensity according to the actual operating conditions. In this way, the calibrated reconstruction optimization results are finally obtained. This result not only retains the main characteristics of the original optimization but also incorporates the correction information from the error analysis, improving the reliability and practicality of the optimization results.

[0118] Specifically, S3.3 includes:

[0119] S3.3.1: Calculate the mean and standard deviation of the error for each bin interval, construct the error correction model, and obtain the initial correction coefficients;

[0120] In the error analysis phase, the first step is to calculate detailed error statistics for each bin interval. Specifically, for each bin interval, the mean error is calculated, reflecting the overall deviation level of the reconstruction optimization results within that interval. Simultaneously, the standard deviation is calculated to characterize the dispersion of the error distribution. These two basic statistics together constitute the fundamental characteristics of the error distribution. It should be noted that when calculating these statistics, the membership degree of a sample within the bin must be considered; that is, a weighted calculation method is used to give samples with higher membership degrees greater influence. Based on the above statistical characteristics, an error correction model is constructed. This model can adaptively generate correction coefficients according to the current operating conditions. The calculation of the initial correction coefficients needs to comprehensively consider the mean error, standard deviation, and the acceptable error range of the system, and normalization processing is used to ensure the rationality of the correction coefficients.

[0121] S3.3.2: The initial correction coefficients are dynamically updated using the exponential smoothing method to obtain the reconstructed optimization results after calibration.

[0122] To enable dynamic adaptation in error correction, this application employs an exponential smoothing method to dynamically update the initial correction coefficients. Specifically, after each application of the correction coefficients, evaluation and adjustments are made based on the actual results. The exponential smoothing method assigns greater weight to recent data, allowing the correction process to respond promptly to changes in system state. During the update process, an appropriate smoothing coefficient needs to be set, which determines the weight allocation between old and new data. A larger smoothing coefficient makes the system more sensitive to new data, and the correction effect adjusts rapidly with system changes; a smaller smoothing coefficient provides better smoothing and avoids drastic fluctuations during the correction process. Through this dynamic update mechanism, the calibrated reconstruction optimization result is finally obtained, which maintains the basic characteristics of the optimization scheme while possessing strong adaptability and reliability. It should be noted that in practical applications, the smoothing coefficient can be appropriately adjusted according to the actual system operation to achieve a balance between response speed and stability.

[0123] S4: Based on the reconfiguration optimization results after calibration, the islanding conditions are judged by comprehensively considering power capacity, load demand and network constraints, the reconfiguration operation sequence is optimized, the final islanding and fault reconfiguration scheme is generated, and the islanding and fault reconfiguration of the power distribution system are carried out based on the islanding and fault reconfiguration scheme.

[0124] After error calibration, islanding conditions must be determined and reconfiguration schemes generated. This process begins by assessing the matching degree between distributed generation and loads. Matching degree assessment requires comprehensive consideration of multiple factors: the output capacity of distributed generation, load demand characteristics, geographical location, etc. In particular, for renewable energy sources, the uncertainty and volatility of their output must be considered to ensure that the islanded systems have sufficient power supply capacity.

[0125] Power balance constraints and network connectivity analysis are key steps in islanding. Power balance constraints ensure that the power output within each island region can meet load demand, which requires considering the dynamic characteristics of load changes and the output characteristics of distributed power sources. Network connectivity analysis ensures that the islands maintain topological integrity after partitioning, avoiding isolated nodes or disconnected links.

[0126] The optimization of the reconstruction operation sequence employs an improved dynamic programming algorithm. This algorithm first establishes a state transition network, where nodes represent different states of the system and edges represent transition operations between states. By introducing an operation cost penalty term, and comprehensively considering factors such as the number of switching operations, operation time requirements, and transient impacts, the economic efficiency and reliability of the reconstruction process are ensured.

[0127] The reliability assessment of the scheme employs the Monte Carlo simulation method. By generating a large number of scenario samples to simulate different operating conditions, the system's reliability indices are calculated. These indices include the System Average Outage Frequency Index (SAIFI) and the System Average Outage Duration Index (SAIDI). Statistical analysis of these indices allows for a comprehensive evaluation of the reliability level of the reconfiguration scheme.

[0128] Specifically, this application employs Beta and Weibull distributions to describe the output characteristics of different types of renewable energy sources during scenario generation. This approach more accurately reflects the randomness of the actual system. Simultaneously, confidence interval analysis is used to assess the reliability of the statistical results, ensuring the scientific validity and credibility of the evaluation.

[0129] Finally, based on the evaluation results, if the reliability indicators are found to be unsatisfactory in certain scenarios, it is necessary to return to the previous steps to adjust the solution. This iterative optimization process ensures that the final island partitioning and fault reconfiguration solution not only meets the technical requirements but also has high reliability and practicality.

[0130] Through this systematic processing flow, the embodiments of this application can effectively solve the problems of islanding and fault reconstruction in power distribution systems, and have strong engineering practical value. The innovation of the method is mainly reflected in fuzzy box processing, dynamic optimization strategies, and reliability assessment, providing strong technical support for the safe operation of power distribution systems.

[0131] Among them, such as Figure 5 As shown, S4 specifically includes:

[0132] S4.1: Calculate the matching degree evaluation index between distributed power sources and loads, and perform power balance constraints and network connectivity analysis to obtain the islanding condition judgment results;

[0133] After obtaining the calibrated reconfiguration optimization results, the first step is to determine the islanding conditions (step S4.1). Specifically, it is necessary to calculate the matching evaluation index between distributed generation and load. This index mainly considers three aspects: the output capacity of distributed generation, the electricity demand characteristics of the load, and the geographical distribution relationship between the two. In the output capacity assessment, the type, capacity, and output characteristics of distributed generation need to be considered, especially the random fluctuation characteristics of renewable energy. For load demand characteristics, it is necessary to analyze the electricity consumption patterns and importance of different types of loads. In addition, power balance constraint analysis is required to ensure that the power output in each potential islanding area can meet the load demand. At the same time, through network connectivity analysis, it is ensured that the topological integrity of each islanded area after division is maintained. These analysis results together constitute the islanding condition judgment results, providing a decision-making basis for subsequent reconfiguration operations.

[0134] S4.2: Design the switch operation sequence using an improved dynamic programming algorithm, introduce an operation cost penalty term, and obtain the optimized reconstruction operation sequence;

[0135] After determining the island partitioning conditions, a specific reconfiguration operation sequence needs to be designed. This application employs an improved dynamic programming algorithm to optimize the switching operation sequence. In the design process, a state transition network is first established, where nodes represent different states of the system, and edges represent transition operations between states. Considering the practical constraints of switching operations, an operation cost penalty term is introduced, which includes several aspects: the limit on the number of switching operations, operation time requirements, and transient impacts caused by switching operations. In this way, unnecessary switching operations can be minimized while ensuring the feasibility of the reconfiguration scheme. The algorithm automatically plans the optimal operation path based on the current system state and the target state, obtaining the optimized reconfiguration operation sequence. It should be noted that in practical applications, it may be necessary to consider the operating habits and safety procedures of the operators and make appropriate adjustments to the operation sequence.

[0136] S4.3: Based on Monte Carlo simulation, the reliability of the optimized reconfiguration operation sequence is evaluated, the power supply reliability index is calculated, and the final islanding and fault reconfiguration scheme is obtained.

[0137] Finally, to verify the reliability of the reconfiguration scheme, a detailed evaluation is required. This application employs the Monte Carlo simulation method for reliability assessment. During the simulation, a large number of possible operating scenarios are generated through random sampling, each scenario including different load levels, renewable energy output status, etc. For each scenario, important power supply reliability indicators need to be calculated, such as the System Average Outage Frequency Index (SAIFI) and the System Average Outage Duration Index (SAIDI). These indicators reflect the power supply reliability level of the reconfiguration scheme from different perspectives. By statistically analyzing the simulation results of a large number of scenarios, the performance of the reconfiguration scheme under various operating conditions can be comprehensively evaluated. Based on the evaluation results, if the reliability indicators in some scenarios are found to be unsatisfactory, the previous steps need to be returned to adjust the scheme. This iterative optimization and evaluation process will ultimately yield an islanding partitioning and fault reconfiguration scheme that meets both technical requirements and has high reliability.

[0138] Specifically, S4.3 includes:

[0139] S4.3.1: Generate scenario samples containing renewable energy output and load demand. The scenario samples are generated by random sampling based on Beta distribution and Weibull distribution. Calculate the system average outage frequency index SAIFI and the system average outage duration index SAIDI to obtain the reliability assessment results.

[0140] In the reliability assessment phase, a large number of scenario samples need to be generated for simulation analysis. Different probability distribution models are used to describe the output characteristics of renewable energy: the Beta distribution is used to generate photovoltaic (PV) output scenario samples, which can well reflect the daily variation characteristics of PV power generation; the Weibull distribution is used to generate wind power output scenario samples, as this distribution has been widely validated for describing the stochastic characteristics of wind power output. Regarding load demand, the electricity consumption characteristics of different types of loads need to be considered, including residential load, industrial load, and commercial load, which have different daily load curves and fluctuation characteristics. Based on these scenario samples, two key reliability indicators are calculated: the System Average Outage Frequency Index (SAIFI) and the System Average Outage Duration Index (SAIDI). SAIFI reflects the average number of power outages experienced by users, while SAIDI characterizes the average duration of power outages. These two indicators together constitute the basic data for reliability assessment. It should be noted that when generating scenario samples, the sample size must be large enough to ensure the reliability of the statistical results.

[0141] S4.3.2: Perform confidence interval analysis on the reliability assessment results, evaluate statistical reliability, and obtain the final islanding and fault reconstruction scheme.

[0142] After obtaining the basic data for reliability assessment, further statistical analysis is required. First, confidence interval analysis is performed on the assessment results. This process includes calculating the confidence interval for the sample mean and the confidence interval for extreme values. The width of the confidence interval reflects the degree of uncertainty in the assessment results; a narrower confidence interval indicates a more reliable assessment. Second, statistical reliability needs to be assessed. This includes analyzing the distribution characteristics of the data, verifying the representativeness of the sample, and examining the system performance under extreme conditions. In this process, special attention is paid to extreme scenarios that may significantly degrade system performance, evaluating the system's ability to cope under these conditions. By comprehensively analyzing these statistical results, a comprehensive evaluation of the reliability of the reconfiguration scheme can be obtained. If the assessment results show that the scheme meets all reliability requirements, it can be determined as the final islanding and fault reconfiguration scheme. Otherwise, it is necessary to return to the previous steps based on the problems discovered in the analysis and make corresponding optimizations and adjustments to the scheme. It should be noted that in practical applications, the standards for reliability assessment may vary due to factors such as region and power grid level; reasonable assessment standards should be set according to specific circumstances.

[0143] This application embodiment will use a specific example to illustrate how the solution realizes islanding and fault reconfiguration of the power distribution system.

[0144] Suppose a regional power distribution system comprises one main substation, three distributed photovoltaic power stations, two distributed wind farms, and multiple load nodes. The system is interconnected by 12 feeders and has 15 sectionalizing switches and 10 tie switches. During a lightning-induced fault, a critical feeder in the middle of the system fails, requiring islanding and fault reconfiguration.

[0145] First, the system acquires real-time operational data. This includes output data for each photovoltaic power station (which fluctuates between 45% and 80% of rated capacity due to daily weather conditions), output data for wind farms (which maintains around 60% of rated capacity under current wind speed conditions), and electricity demand data for each load node. Simultaneously, the system automatically collects network topology information and equipment operating status, and assesses the output risk of renewable energy sources using the CVaR (Continuous Value Assurance) index.

[0146] After determining the fault location, the system begins searching for feasible reconfiguration schemes based on a hybrid meta-heuristic algorithm. The algorithm first generates multiple candidate schemes, each containing different combinations of switching operations. Using an adaptive step-size gradient descent method, several potentially high-quality schemes are quickly identified. These schemes can all achieve system reconfiguration, but they have different load transfer strategies and power configurations.

[0147] Next, the system performs fuzzy clustering on these candidate solutions. By analyzing characteristic parameters such as load change rate and voltage deviation, the solutions are divided into three main categories: solutions that minimize the power outage area, solutions that optimize voltage distribution, and solutions that balance power supply reliability and economy. After fuzzy clustering and error analysis, the system identifies an optimal reconfiguration solution.

[0148] In this scheme, the system is divided into three islanded areas. The first island is powered by the main substation, handling approximately 50% of the system load; the second island consists of a combined power supply system of two photovoltaic power plants and one wind farm, supplying approximately 30% of the load; and the third island is powered by the remaining distributed power sources, covering the remaining load area. The system achieves reconfiguration through a five-step switching operation: first, the faulty section is isolated; then, two tie switches are closed to form new power supply channels; next, the states of the three sectionalizing switches are adjusted to establish islanded operating conditions; and finally, one tie switch is fine-tuned to optimize voltage distribution.

[0149] When implementing the reconfiguration scheme, the system strictly follows the optimized operation sequence. First, it is confirmed that the power capacity of each islanded area meets the load demand, and the regulation capability of the distributed power supply is sufficient to cope with load fluctuations. Then, the system executes switching operations in a predetermined sequence through the SCADA system. After each operation, the system status is checked to ensure that key indicators such as voltage and frequency remain within the allowable range.

[0150] The entire reconfiguration process took approximately 3 minutes, successfully containing the impact of the fault to a minimum and ensuring continuous power supply to critical loads. Subsequent Monte Carlo simulation evaluations showed that the solution maintained stable operation under various weather conditions and load levels, with the system's SAIFI and SAIDI indices exceeding preset standards. Particularly noteworthy is the reliability of islanded operation under conditions of significant fluctuations in renewable energy output, thanks to the pre-allocated adjustment margin.

[0151] This example fully demonstrates the practicality and effectiveness of this solution. Through a systematic analysis and optimization process, a reasonable islanding and fault reconfiguration scheme can be quickly generated and implemented, providing strong support for the safe and reliable operation of the power distribution system. This solution not only considers technical feasibility but also takes into account economic efficiency and reliability, making it highly valuable for engineering applications.

[0152] This application also provides a power distribution system islanding and fault reconfiguration system, including:

[0153] The mathematical model building module is used to acquire the topology, load characteristic data and renewable energy output characteristic data of the power distribution system, establish the basic model of the power distribution system based on the node-branch relationship and power flow constraints, and introduce the Conditional Value at Risk (CVaR) index to quantify the impact of renewable energy uncertainty and generate the mathematical model of the power distribution system.

[0154] The fault reconfiguration optimization module is used to perform optimization calculations based on the mathematical model of the power distribution system, using a hybrid meta-heuristic algorithm framework and an adaptive step-size gradient descent method, to obtain the initial reconfiguration optimization results.

[0155] The error estimation module is used to perform fuzzy binning on the initial reconstruction optimization results, dynamically binning the data samples based on the fuzzy clustering algorithm, calculating the error statistical characteristics of each bin interval, and obtaining the calibrated reconstruction optimization results.

[0156] The scheme generation module is used to determine the islanding conditions based on the calibrated reconfiguration optimization results, taking into account power capacity, load demand and network constraints, optimize the reconfiguration operation sequence, generate the final islanding and fault reconfiguration scheme, and perform islanding and fault reconfiguration of the power distribution system based on the islanding and fault reconfiguration scheme.

[0157] This disclosure also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program performs the steps of the power distribution system islanding and fault reconfiguration method described in the above-described method embodiments. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0158] Furthermore, this disclosure also provides a computer program product storing a computer program. When the computer program is run by a processor, it executes the steps of the multi-energy power distribution system optimization planning method for resilience enhancement provided in any of the above embodiments of this disclosure. For details, please refer to the above method embodiments, which will not be repeated here.

[0159] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium, which can be a volatile or non-volatile computer-readable storage medium. In another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0160] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices and apparatuses described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this disclosure, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection may be through some communication interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0161] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0162] In addition, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0163] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0164] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit it. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this disclosure; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the protection scope of the claims.

Claims

1. A method for islanding and fault reconfiguration in a power distribution system, characterized in that, include: The topology, load characteristics, and renewable energy output characteristics of the power distribution system are acquired. A basic model of the power distribution system is established based on node-branch relationships and power flow constraints. The Conditional Value at Risk (CVaR) index is introduced to quantify the impact of renewable energy uncertainties, and a mathematical model of the power distribution system is generated. Using the mathematical model of the power distribution system, optimization calculations are performed based on the hybrid meta-heuristic algorithm framework and the gradient descent method with adaptive step size to obtain the initial reconstruction optimization results; The initial reconstruction optimization results are subjected to fuzzy binning. The data samples are dynamically binned based on the fuzzy clustering algorithm. The error statistical characteristics of each bin interval are calculated to obtain the calibrated reconstruction optimization results. Based on the reconfiguration optimization results after calibration, the islanding conditions are judged by comprehensively considering power capacity, load demand and network constraints, the reconfiguration operation sequence is optimized, the final islanding and fault reconfiguration scheme is generated, and the islanding and fault reconfiguration of the power distribution system are performed based on the islanding and fault reconfiguration scheme.

2. The method according to claim 1, characterized in that, Acquire topology, load characteristic data, and renewable energy output characteristic data of the distribution system. Establish a basic model of the distribution system based on node-branch relationships and power flow constraints. Introduce the Conditional Value at Risk (CVaR) index to quantify the impact of renewable energy uncertainties, generating a mathematical model of the distribution system, including: Based on the topology and load characteristic data of the power distribution system, node power injection balance equations and voltage constraints are established to obtain the power flow model of the power distribution system. For the renewable energy output characteristic data, the Beta distribution is used to describe the probability density function of photovoltaic output, and the Weibull distribution is used to describe the probability distribution of wind power output, so as to obtain the probabilistic characteristic model of renewable energy. By combining the power flow model of the power distribution system and the probabilistic characteristic model of the renewable energy source, and using linearization to transform the conditional value at risk (CVaR) into a linear constraint, the mathematical model of the power distribution system is obtained.

3. The method according to claim 2, characterized in that, Based on the topology and load characteristic data of the power distribution system, node power injection balance equations and voltage constraints are established to obtain the power flow model of the power distribution system, including: Obtain the node admittance matrix elements, node voltage phase angle difference data, and ±5% constraint range of node voltage amplitude. Calculate the active and reactive power injection at the node based on the node power injection balance equation to obtain the power balance equation set. By applying the voltage constraint condition to the power balance equations, the power flow model of the power distribution system is obtained.

4. The method according to claim 1, characterized in that, Using the mathematical model of the power distribution system, optimization calculations are performed based on a hybrid meta-heuristic algorithm framework and an adaptive step-size gradient descent method to obtain initial reconstruction optimization results, including: The optimization population is initialized based on the Latin hypercube sampling method, a penalty function containing the default degree is constructed, and a hybrid metaheuristic optimization framework is established. Iterative optimization is performed using the hybrid meta-heuristic optimization framework and gradient descent method. The search step size is dynamically adjusted according to the changes in the objective function to obtain the initial reconstruction optimization result.

5. The method according to claim 4, characterized in that, Iterative optimization is performed using the aforementioned hybrid meta-heuristic optimization framework and gradient descent method. The search step size is dynamically adjusted according to changes in the objective function to obtain the initial reconstruction optimization results, including: Calculate the continuous change in an individual's position, and determine that local convergence has been achieved when the change is less than a preset threshold, thus obtaining the local convergence result. Calculate the population diversity index. When the population diversity is below the threshold and the optimal solution has not improved for five consecutive generations, global convergence is determined, and the initial reconstruction optimization result is obtained.

6. The method according to claim 1, characterized in that, The initial reconstruction optimization results are subjected to fuzzy binning. Based on a fuzzy clustering algorithm, the data samples are dynamically binned, and the error statistical characteristics of each bin interval are calculated to obtain the calibrated reconstruction optimization results, including: The initial reconstruction optimization results are subjected to z-score standardization to extract feature parameters such as load change rate, voltage deviation and power factor, resulting in a standardized dataset. The standardized dataset is dynamically binned using the fuzzy C-means clustering algorithm. The optimal number of bins is determined based on the silhouette coefficient, and the sample membership degree is calculated to obtain the fuzzy binning result. The error statistics and conditional error correction functions of each bin interval in the fuzzy binning result are calculated to obtain the calibrated reconstruction optimization result.

7. The method according to claim 6, characterized in that, Calculate the error statistics and conditional error correction function for each bin interval in the fuzzy binning result to obtain the calibrated reconstruction optimization result, including: Calculate the mean and standard deviation of the error for each bin interval, construct an error correction model, and obtain the initial correction coefficients; The initial correction coefficients are dynamically updated using an exponential smoothing method to obtain the calibrated reconstruction optimization result.

8. A power distribution system for islanding and fault reconfiguration, characterized in that, include: The mathematical model building module is used to acquire the topology, load characteristic data and renewable energy output characteristic data of the power distribution system, establish the basic model of the power distribution system based on the node-branch relationship and power flow constraints, and introduce the Conditional Value at Risk (CVaR) index to quantify the impact of renewable energy uncertainty and generate the mathematical model of the power distribution system. The fault reconfiguration optimization module is used to perform optimization calculations based on the mathematical model of the power distribution system, using a hybrid meta-heuristic algorithm framework and an adaptive step-size gradient descent method, to obtain the initial reconfiguration optimization results. The error estimation module is used to perform fuzzy binning on the initial reconstruction optimization results, dynamically binning the data samples based on the fuzzy clustering algorithm, calculating the error statistical characteristics of each bin interval, and obtaining the calibrated reconstruction optimization results. The scheme generation module is used to determine the islanding conditions based on the calibrated reconfiguration optimization results, taking into account power capacity, load demand and network constraints, optimize the reconfiguration operation sequence, generate the final islanding and fault reconfiguration scheme, and perform islanding and fault reconfiguration of the power distribution system based on the islanding and fault reconfiguration scheme.

Images