A power distribution network distributed photovoltaic carrying capacity evaluation method and device

Abstract

The application relates to the technical field of power distribution network operation and new energy access, and discloses a power distribution network distributed photovoltaic carrying capacity evaluation method and device, scene characteristic indexes are calculated according to structure data and historical time sequence data of the power distribution network, a dynamic evaluation model is constructed by combining future prediction data of the power distribution network and taking the maximum total access capacity of photovoltaic and the minimum total active network loss as targets, multiple constraint conditions of the model are set, and the verification priorities of the conditions are determined according to real-time operation states and future prediction data, then the dynamic evaluation model is solved by using an improved particle swarm algorithm and a hierarchical grading strategy which are fused with adaptive adjustment coefficients, and distributed photovoltaic carrying capacity evaluation results of the power distribution network are obtained, and the adaptive adjustment coefficients are dynamically adjusted according to scene adaptation coefficients determined based on the scene characteristic indexes and real-time iteration stages of the improved algorithm, so that accurate and efficient evaluation of photovoltaic carrying capacity of the power distribution network is realized.

Description

Technical Field

[0001] This invention relates to the field of power distribution network operation and new energy access technology, and in particular to a method and apparatus for assessing the carrying capacity of distributed photovoltaic power distribution networks. Background Technology

[0002] With the large-scale integration of distributed photovoltaic (PV) power into the distribution network, scientifically assessing its maximum carrying capacity is crucial for ensuring grid security and improving grid absorption. Existing assessment methods often employ intelligent optimization algorithms, such as particle swarm optimization, but these methods cannot adapt to diverse scenarios such as load fluctuations and variable PV output, and are prone to getting trapped in local optima, leading to biased assessment results.

[0003] Therefore, how to provide a load-bearing capacity assessment method that can dynamically adapt to different scenarios, is accurate and efficient, and has practical engineering value has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] This invention provides a method and apparatus for assessing the carrying capacity of distributed photovoltaic power distribution networks, addressing the problem of how to provide a carrying capacity assessment method that can dynamically adapt to different scenarios, is accurate and efficient, and has practical engineering value.

[0005] To address the aforementioned technical problems, the first aspect of this invention provides a method for assessing the carrying capacity of distributed photovoltaic power grids, comprising: Acquire structural data, historical time-series data, real-time operating status and future prediction data of the power distribution network, and determine scenario characteristic indicators based on the structural data and the historical time-series data; With the goal of maximizing the total access capacity of distributed photovoltaic power in the distribution network and minimizing the total active power loss of the system, a dynamic evaluation model is constructed based on the future prediction data. Set multiple constraints for the dynamic evaluation model, and determine the verification priority of the multiple constraints based on the real-time running status and the future prediction data; Based on the verification priority, the dynamic evaluation model is solved using an improved particle swarm optimization algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy to obtain the distributed photovoltaic carrying capacity evaluation result of the distribution network; the adaptive adjustment coefficients are dynamically adjusted according to the scenario adaptation coefficients determined based on the scenario feature indicators and the real-time iteration stage of the improved particle swarm optimization algorithm.

[0006] A second aspect of the present invention provides a distributed photovoltaic carrying capacity assessment device for a distribution network, comprising: The indicator calculation module is used to acquire the structural data, historical time-series data, real-time operating status and future prediction data of the power distribution network, and determine the scenario characteristic indicators based on the structural data and the historical time-series data. The model building module is used to construct a dynamic evaluation model based on the future prediction data, with the objectives of maximizing the total access capacity of distributed photovoltaics in the distribution network and minimizing the total active power loss of the system. The priority determination module is used to set multiple constraints of the dynamic evaluation model and determine the verification priority of the multiple constraints based on the real-time running status and the future prediction data. The carrying capacity assessment module is used to solve the dynamic assessment model based on the verification priority, using an improved particle swarm algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy, to obtain the distributed photovoltaic carrying capacity assessment result of the distribution network; the adaptive adjustment coefficients are dynamically adjusted according to the scenario adaptation coefficients determined based on the scenario feature indicators and the real-time iteration stage of the improved particle swarm algorithm.

[0007] Compared with the prior art, the beneficial effects of the embodiments of the present invention are as follows: By integrating historical time-series data with scenario feature indicators, a more comprehensive dynamic evaluation model was constructed, enabling the evaluation dimension to leap from "static cross-section" to "spatiotemporal dynamics." The constructed target system takes into account both economic efficiency and low carbon emissions, breaking through the previous single-objective mode that only used voltage deviation or equipment overload as hard constraints, and achieving synergistic optimization of social benefits and power grid benefits. Based on real-time operating status and future prediction data, multiple constraints are dynamically prioritized, enabling the model to prioritize high-risk constraints in the actual solution process, thus improving solution efficiency. An adaptive adjustment coefficient is introduced into the improved particle swarm algorithm. This coefficient is jointly determined by the scenario adaptation coefficient and the iteration stage. The scenario adaptation coefficient is dynamically generated based on scenario feature indicators, giving the algorithm differentiated search preferences under different operating environments. The adjustment in the iteration stage ensures that the algorithm focuses on global exploration in the early stage and local refinement in the later stage. The dual adaptive mechanism significantly improves the convergence accuracy and robustness of the algorithm under complex distribution network topologies. To address the problem of numerous nodes and an explosion of decision variable dimensions in large-scale distribution networks, an improved algorithm pre-layered and graded solution strategy is adopted, which greatly reduces computational complexity and supports online applications. Attached Figure Description

[0008] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the 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.

[0009] Figure 1 This is a flowchart of a method for assessing the carrying capacity of distributed photovoltaic power grids according to a certain embodiment of the present invention; Figure 2This is a structural diagram of a distributed photovoltaic carrying capacity assessment device for a power distribution network according to a certain embodiment of the present invention; Figure label: Among them, 10 is the index calculation module; 20 is the model construction module; 30 is the priority determination module; and 40 is the bearing capacity assessment module. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings and examples. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0011] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art will be able to understand the specific meaning of the above terms in this application according to the specific circumstances.

[0012] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is merely for describing specific embodiments and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0013] In one embodiment, such as Figure 1 As shown, the first aspect of the present invention provides a method for assessing the carrying capacity of distributed photovoltaic power grids, comprising: S1. Acquire the structural data, historical time-series data, real-time operating status, and future prediction data of the distribution network, and determine the scenario characteristic indicators based on the structural data and the historical time-series data. The historical time-series data includes historical time-series load data and historical power output time-series data. Specifically, collect the structural data, historical time-series data, real-time operating status, and future prediction data of the target distribution network. The structural data includes network topology, line parameters, transformer parameters, load node information, etc.; the historical time-series data includes historical time-series load data (time-series load, voltage, and current data, etc., with a sampling interval of no more than 15 minutes, for at least one year) and historical power output time-series data (historical concurrent irradiance, temperature, and corresponding photovoltaic power station power output time-series data for distributed photovoltaic systems, etc.); the real-time operating status includes real-time node voltage, current, and power data, etc.; the future prediction data includes photovoltaic power output prediction data, power prediction data, and current prediction data for the next 24 hours generated using an ARIMA time series model, and load prediction data for the next 24 hours generated using an LSTM neural network model, etc. Subsequently, clean and normalize the collected data to improve data accuracy.

[0014] In one embodiment, determining scene feature indicators based on the structural data and the historical time-series data includes: Extract the daily load peak-to-valley ratio from the historical time-series load data, and calculate the time-series standard deviation of the daily load peak-to-valley ratio around its average value to obtain the load fluctuation intensity; The rate of change of photovoltaic power output within adjacent time points is calculated based on the historical power output time series data, and the maximum value of the rate of change among all time points is taken as the photovoltaic power output volatility. Based on the historical time-series load data and the historical power output time-series data, the load power sequence and photovoltaic power output sequence within the same time period are extracted, and the Pearson correlation coefficient between these two sequences is used as the load-photovoltaic correlation coefficient. Under preset typical operating conditions, the Jacobian matrix of power flow calculation is solved by the structural data to obtain the average power flow sensitivity of the line. Based on the structural data, the average node voltage stability margin is calculated using the PV curve method. The scenario characteristic index is obtained by combining the load fluctuation intensity, the photovoltaic output fluctuation rate, the load-photovoltaic correlation coefficient, the average line power flow sensitivity, and the average node voltage stability margin.

[0015] Specifically, this invention extracts the daily load peak-to-valley ratio from historical time-series load data and uses its time-series standard deviation as the load fluctuation intensity to reflect the load fluctuation characteristics. The calculation process is expressed by the following formula: In the formula, The load fluctuation intensity is represented by T, which represents the number of statistical days. , The maximum and minimum loads on day t; This represents the average daily peak-to-valley ratio of the load.

[0016] This invention uses the maximum value of the rate of change of photovoltaic power output within adjacent time points in the historical power output time series data, that is, the maximum value of the hourly rate of change of power output in the historical power output time series data, as the photovoltaic power output volatility rate to reflect the characteristics of photovoltaic power output fluctuation. Its calculation process is expressed by the following formula: In the formula, For photovoltaic power output volatility; Photovoltaic output at time t; This refers to the rated capacity of the photovoltaic system.

[0017] Historical time-series load data and historical power output time-series data are aligned in time to form two vectors of equal length. Load power series and photovoltaic power output series within the same time period (such as the same time of day in a year) are extracted from these vectors. The Pearson correlation coefficient between these two series is used as the load-PV correlation coefficient ρ(L,PV) to reflect the relationship between load and PV output changes in time.

[0018] Select a typical operating condition (such as the node injection power in a typical historical scenario), calculate the inverse matrix of the power flow Jacobian matrix or the relevant sensitivity matrix, that is, perform power flow calculation under this condition (such as the forward substitution method or the Newton-Raphson method) to form the current steady state of the distribution network system; based on the structural data of the distribution network, calculate the relationship matrix between small changes in node injection power (especially the active power of photovoltaic access points) and changes in line power flow (current or power), and extract the sensitivity coefficients of the target line (or all lines of interest) relative to all possible photovoltaic access points. Finally, take the absolute value of these sensitivity coefficients and average them to obtain the mean value S of the line power flow sensitivity. line,avg This is used to measure the average sensitivity of changes in photovoltaic output to the impact of power flow on the transmission line.

[0019] Select one or more key nodes (usually at the end of the grid or weak points) and use the PV curve method: Under basic or typical operating conditions, gradually increase the active power injection at the node (simulating the growth of photovoltaic output). Simultaneously, based on the distribution network's structural data, track the voltage changes at the node through continuous power flow calculations until a voltage collapse critical point is reached. Calculate the voltage stability margin of the node, and average the voltage stability margins of all key nodes to obtain the mean node voltage stability margin C. voltage,avgThis characterizes the system's average ability to maintain voltage stability under current operating conditions while withstanding increases in photovoltaic power.

[0020] The load fluctuation intensity, photovoltaic output fluctuation rate, load-photovoltaic correlation coefficient, average line power flow sensitivity, and average node voltage stability margin obtained from the above calculations can be combined to form the scenario characteristic indicators.

[0021] This invention constructs a "digital profile" of distribution network operation scenarios from five dimensions: load characteristics, power supply characteristics, source-load matching degree, network transmission capacity, and voltage stability. This represents a dimensional upgrade from traditional single indicators. It not only fully preserves the time-series information of historical data, but also aggregates it into feature quantities that are easy for engineers to understand through dimensionality reduction. These can be directly used as input variables for scenario adaptation coefficients in subsequent dynamic evaluation models, enabling adaptive matching between the evaluation algorithm and the operation scenario. This significantly improves the engineering applicability and reliability of the carrying capacity assessment results in complex distribution networks.

[0022] S2. With the goal of maximizing the total access capacity of distributed photovoltaic power in the distribution network and minimizing the total active power loss of the system, a dynamic evaluation model is constructed based on the future prediction data. In one embodiment, step S2 includes: The first objective is to maximize the total access capacity of distributed photovoltaic power in the distribution network, and the second objective is to minimize the total active power loss of the system. The first and second objectives are then input into the NSGA-II algorithm for solving, and a set of Pareto optimal solutions is output. An initial solution is selected from the Pareto optimal solution set based on a preset decision rule, and the quality score of the initial solution is determined based on the future prediction data. The initial solution with a quality score exceeding the preset score is taken as the best solution. The weights corresponding to the first objective and the second objective are derived from the optimal solution, and the weights are used to perform a weighted summation of the first objective and the second objective to form the dynamic evaluation model.

[0023] Specifically, this invention uses the planned distributed photovoltaic (PV) capacity at the i-th node (or candidate access point) in the distribution network as the decision variable. The first objective is to maximize the total distributed PV capacity in the distribution network, and the second objective is to minimize the total active power loss of the corresponding system (which is the total active power loss of the system under a specific operating scenario given the PV access scheme). Since the two objectives are often conflicting (increasing PV capacity may change the power flow distribution and increase network losses), directly solving for the Pareto front involves a large computational load. Therefore, this invention inputs the first and second objectives into the NSGA-II algorithm for solving. This algorithm explores the decision space and outputs a set of Pareto optimal solutions. Each solution in this set represents a PV capacity configuration scheme, in which it is impossible to improve one objective without compromising the other.

[0024] From the Pareto solution set, an initial solution is selected based on certain decision rules (such as selecting the solution with the largest total capacity under the premise that the network loss rate does not exceed a preset threshold (such as 3%)). The capacity configuration corresponding to this initial solution is then combined with a complete operating scenario with time-series characteristics. This scenario is constructed based on future prediction data, which includes the load power of each node and the actual output of each photovoltaic node within a future period (such as 24 hours). The latter is obtained by multiplying the photovoltaic capacity by the predicted normalized output curve. Subsequently, time-series power flow calculation is performed: the above time-varying injected power data (load is negative, photovoltaic is positive) is input into the distribution network power flow calculation program (such as forward-backward substitution method), and the system state (node ​​voltage, branch power flow) at each time segment (such as hourly) is calculated. The active power loss of all time segments and all branches is accumulated to obtain the total active power loss within the evaluation period. This is substituted into the first and second objectives and summed to form the quality score of the initial solution. Finally, the initial solution with a quality score exceeding the preset score is taken as the best solution.

[0025] Based on the optimal solution, the weights that make the linear weighted sum optimal at that solution are derived (this is a mathematical approximation or fitting process, or the preference corresponding to that solution can be directly regarded as the optimal weights). These weights are then used as the weight coefficients for the first objective, and the difference between 1 and the optimal weights is used as the weight coefficients for the second objective. These two objectives are then weighted and summed to form a single-objective dynamic evaluation model, which is expressed by the following formula: In the formula, The value of the comprehensive objective function; These are the weighting coefficients; Let be the distributed photovoltaic capacity connected to the i-th node; The total active power loss (kW) of the system is calculated from the power flow (using the forward-backward substitution method). The dynamic evaluation model is passed... Link photovoltaic power output data, through By associating line parameters and load data, the final output is the total carrying capacity and node margin.

[0026] This invention addresses the challenge of determining multi-objective weights in a dynamic assessment model of distributed photovoltaic carrying capacity in power distribution networks. It proposes a weight back-calculation mechanism based on NSGA-II and future prediction data. Through a technical approach of "optimization first, back-calculation then aggregation," it avoids subjective weighting bias, achieves intrinsic matching between weights and optimization objectives, realizes compatibility and unity between multi-objective decision-making and single-objective solution, and endows the dynamic assessment model with pre-adaptability to future operating conditions.

[0027] S3. Set multiple constraints for the dynamic evaluation model, and determine the verification priority of the multiple constraints based on the real-time operating status and the future prediction data; wherein, the multiple constraints include at least voltage deviation constraints and line load rate constraints; specifically, this invention sets multiple constraints for the dynamic evaluation model based on industry standards and typical engineering practices, including voltage deviation constraints: the difference between the node voltage and the nominal voltage does not exceed ±5% of the nominal voltage; line load rate constraints: the ratio of branch current to the maximum branch current is not greater than 1.0; transformer load rate constraints: the ratio of the transformer apparent power (or current) to its rated capacity (or rated current) is not greater than 0.8; Power quality constraints: Node harmonic distortion rate not greater than 0.05; Photovoltaic access capacity constraints: Node photovoltaic capacity not greater than the maximum value of node photovoltaic capacity; Subsequently, the priorities of each constraint are initially set: voltage deviation constraint is the first priority (1), line load rate constraint and power quality constraint are the second priority (2), distribution transformer load rate constraint is the third priority (3), and photovoltaic access capacity constraint is the fourth priority (4). Among them, since harmonic distortion rate is strongly correlated with safety and is not easily changed dynamically, and node photovoltaic capacity is affected by the physical upper limit and cannot be changed through optimization, the priorities of both are fixed; priorities 1-4 correspond to the constraint verification order, with priority 1 being verified first, and in case of conflict, the higher priority constraint is satisfied first. The priorities of the above constraints and the thresholds of some constraints are not fixed, but are dynamically adjusted according to the real-time operating status of the power grid and future forecast data to achieve the optimal balance between safety and economy.

[0028] In one embodiment, determining the verification priority of the multiple constraints based on the real-time operating status and the future prediction data includes: Based on the future forecast data, the peak overlap rate of photovoltaic load and the transformer load rate curve are determined, and the boundary thresholds of the multiple constraints are dynamically optimized based on the peak overlap rate of photovoltaic load and the transformer load rate curve to obtain the boundary threshold optimization result. Based on the real-time operating status, the voltage over-limit risk and real-time line load rate are determined. Based on the boundary threshold optimization results, the priority of the multiple constraints is dynamically adjusted using the voltage over-limit risk and the real-time line load rate to obtain the verification priority.

[0029] This invention extracts photovoltaic output forecast data and load forecast data from future forecast data during periods when photovoltaic and load peaks overlap, and then calculates the photovoltaic load peak overlap degree based on the extracted data. This process is expressed by the following formula: In the formula, This refers to the peak overlap of photovoltaic loads. This is the period when peak photovoltaic power generation coincides with peak load; T total The total time period is 24 hours; t represents time. , These are photovoltaic power output forecast data and load forecast data, respectively.

[0030] Simultaneously, based on the power or current prediction data from future forecasts, a curve model of how these predicted values ​​change over time is constructed to obtain the distribution transformer load rate curve. The peak overlap rate of photovoltaic load is compared with the overlap rate threshold, and the boundary threshold of the voltage deviation constraint is dynamically adjusted based on the comparison results: when the peak overlap rate of photovoltaic load is greater than 0.6 (high overlap rate threshold, high grid pressure), the ±5% in the voltage deviation constraint is tightened to ±4%; when the peak overlap rate of photovoltaic load is less than 0.3 (low overlap rate threshold), the ±5% in the voltage deviation constraint is relaxed to ±5.5%; other conditions remain unchanged. That is, the restriction is relaxed when the grid capacity is strong to increase the carrying capacity, and the restriction is tightened when the capacity is weak to ensure safety, thus achieving refined control with varying limits. Subsequently, the boundary threshold of 0.8 for the distribution transformer load rate is dynamically adjusted using the distribution transformer load rate curve, realizing prediction-based preventive control, which is more advanced than the traditional overload-based reactive control. This process is illustrated by the following formula: In the formula, This is the boundary threshold for the adjusted transformer load rate, which can also be understood as the upper limit of the transformer load rate. This represents the distribution transformer load rate curve value.

[0031] Combining the optimization principles and optimization quantities of these two constraints yields the boundary threshold optimization result. Subsequently, the voltage over-limit risk (the absolute value of the quotient obtained by dividing the difference between the real-time voltage and the rated voltage by the rated voltage) and the real-time line load rate (the ratio of the real-time current to the maximum current) are calculated based on the real-time operating status. The priority of the constraints is then dynamically adjusted according to the severity of the voltage over-limit risk and the real-time line load rate, instructing the optimization algorithm on "which problem to solve first," thus forming the verification priority.

[0032] This invention addresses the problem of determining the verification priority of multiple constraints in the assessment of distributed photovoltaic carrying capacity in power distribution networks. It proposes a two-layer dynamic decision-making mechanism based on future prediction-driven threshold optimization and real-time state-driven priority adjustment. By embedding future prediction data and real-time operating status hierarchically into the priority generation process, it achieves dual dynamic adaptation of constraint boundaries and verification order, transforming constraint boundaries from "rigid thresholds" to "elastic ranges" and releasing the potential carrying capacity of the system. At the same time, it upgrades priority ranking from "static preset" to "risk response," strengthening the ability to intervene in key constraints in real time, thereby reducing the solution complexity of the dynamic assessment model and improving the feasibility of online application.

[0033] In one embodiment, the step of dynamically adjusting the priority of the multiple constraints based on the boundary threshold optimization results, using the voltage over-limit risk and the real-time line load rate, to obtain the verification priority includes: When the voltage exceedance risk is greater than the first voltage risk threshold, the voltage deviation constraint is adjusted to the first priority; when the voltage exceedance risk is greater than the second voltage risk threshold but not greater than the first voltage risk threshold, the voltage deviation constraint is adjusted to the second priority; and when the voltage exceedance risk is not greater than the second voltage risk threshold, the voltage deviation constraint is adjusted to the third priority. When the real-time line load rate is greater than the first load rate risk threshold, the line load rate constraint is adjusted to the first priority; when the real-time line load rate is greater than the second load rate risk threshold but not greater than the first load rate risk threshold, the line load rate constraint is adjusted to the second priority; and when the real-time line load rate is not greater than the second load rate risk threshold, the line load rate constraint is adjusted to the third priority. The verification priority is obtained by combining the boundary threshold optimization result, the multiple constraint conditions and their adjusted priorities.

[0034] Specifically, when the risk of voltage exceeding the limit is greater than 3% of the first voltage risk threshold, the voltage deviation constraint is adjusted to the first priority 1; when the risk of voltage exceeding the limit is greater than 1% of the second voltage risk threshold but not greater than 3% of the first voltage risk threshold, the voltage deviation constraint is adjusted to the second priority 2; and when the risk of voltage exceeding the limit is not greater than 1% of the second voltage risk threshold, the voltage deviation constraint is adjusted to the third priority 3. When the real-time line load rate is greater than the first load rate risk threshold of 0.8, the line load rate constraint is adjusted to the first priority 1. When the real-time line load rate is greater than the second load rate risk threshold of 0.6 but not greater than the first load rate risk threshold of 0.8, the line load rate constraint is adjusted to the second priority 2. When the real-time line load rate is not greater than the second load rate risk threshold of 0.6, the line load rate constraint is adjusted to the third priority 3. For transformer load rate constraints, when the transformer load rate is greater than 0.7, it is adjusted to the first priority (1); when the transformer load rate is not greater than 0.5, it is adjusted to the fourth priority (4); otherwise, the priority remains unchanged. It should be noted that when the priority adjustment condition is not triggered, the initial settings are used for verification. If a priority does not have a corresponding constraint, that priority is left unchecked, and the constraint corresponding to the next lower priority is directly verified. For example, if after a certain adjustment, priorities 1, 2, and 4 all have corresponding constraints, then after verifying the constraint corresponding to priority 2, the constraint corresponding to priority 4 is directly verified. The same priority can correspond to one or more constraints.

[0035] Finally, the verification priority can be obtained by combining the boundary threshold optimization results, multiple constraints and their corresponding adjustment rules, and the resulting priority combination. This invention proposes a refined decision-making mechanism based on risk classification for the dynamic adjustment of the priority of multiple constraints in the dynamic assessment of the carrying capacity of distributed photovoltaic power distribution networks. By setting risk thresholds, the priority of constraints is dynamically adjusted, and the boundary threshold optimization results are tightly coupled, significantly enhancing the tolerance of the distribution network to the uncertainties of distributed photovoltaic power distribution. Simultaneously, the nested fusion of dynamic boundaries and risk thresholds enhances the scenario adaptability of the dynamic assessment model.

[0036] S4. Based on the verification priority, the dynamic evaluation model is solved using an improved particle swarm optimization algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy to obtain the distributed photovoltaic carrying capacity evaluation result of the distribution network; the adaptive adjustment coefficients are dynamically adjusted according to the scenario adaptation coefficients determined based on the scenario feature indicators and the real-time iteration stage of the improved particle swarm optimization algorithm; wherein, the scenario adaptation coefficients include a first attenuation coefficient, a sensitivity coefficient, and a second attenuation coefficient. In one embodiment, the dynamic adjustment based on the scene adaptation coefficient determined according to the scene feature index and the real-time iterative phase of the improved particle swarm optimization algorithm includes: The first attenuation coefficient is determined based on the load fluctuation intensity and the load-photovoltaic correlation coefficient; the sensitivity coefficient is determined based on the photovoltaic output fluctuation rate and the average line power flow sensitivity; and the second attenuation coefficient is determined based on the average node voltage stability margin and line load data. When the improved particle swarm optimization algorithm is determined to be in the first iteration stage, the adaptive adjustment coefficient is determined based on the real-time iteration number of the improved particle swarm optimization algorithm and the first decay coefficient. When the improved particle swarm optimization algorithm is determined to be in the second iteration stage, the real-time particle aggregation degree of the improved particle swarm optimization algorithm is calculated, and the adaptive adjustment coefficient is determined based on the real-time particle aggregation degree and the sensitivity coefficient. When the improved particle swarm optimization algorithm is determined to be in the third iteration stage, the adaptive adjustment coefficient is determined based on the real-time iteration number of the improved particle swarm optimization algorithm and the second decay coefficient.

[0037] Specifically, in improving the adaptive adjustment mechanism of the particle swarm optimization algorithm, this invention proposes a refined parameter control strategy based on dual driving forces of scene feature indicators and iteration stages. It achieves precise adaptation of "parameter-scene" based on dynamic calculation of real-time operating parameters and physical characteristics of the power distribution network.

[0038] The first attenuation coefficient for adapting to load fluctuation characteristics is determined based on the load fluctuation intensity and load-photovoltaic correlation coefficient in the scene characteristic indicators. This coefficient integrates the "short-term-long-term load fluctuation gradient" and the "load-photovoltaic correlation" to reflect the dynamic characteristics of load fluctuations and their requirements for the algorithm's exploration capabilities. It is calculated using the following formula: γ= +kγ kγ=0.15+0.05 ρ(L,PV) In the formula, The first attenuation coefficient; The initial value of the first attenuation coefficient is kγ; the first weighting coefficient is kγ. This represents the short-term load fluctuation variance (standard deviation of the daily peak-to-valley ratio over the past 7 days), reflecting the severity of short-term load fluctuations. This represents the long-term load fluctuation variance (the standard deviation of the daily peak-to-valley ratio over the past three months), reflecting the long-term stable trend of the load.

[0039] The sensitivity coefficient for adapting to photovoltaic power output fluctuation and grid constraint sensitivity is determined based on the average values ​​of photovoltaic power output fluctuation rate and line power flow sensitivity in the scenario characteristic indicators. This coefficient integrates "photovoltaic power output prediction error" and "line power flow sensitivity" to make the algorithm more targeted to the particle aggregation degree in constraint-sensitive areas. It is calculated by the following formula: In the formula, This is the sensitivity coefficient; This is the initial value for the sensitivity coefficient; This is the second weighting coefficient; The photovoltaic power output prediction error (the average absolute error between the power output predicted for the next 24 hours based on the ARIMA model and the actual power output for the same period in history). This represents the maximum allowable prediction error for the system (15%).

[0040] Based on the average node voltage stability margin and line load data in the scenario characteristic indicators, a second attenuation coefficient is determined to adapt the line load and voltage stability margin. This coefficient integrates the "average line load rate" and "voltage stability margin," and is used to balance the later convergence speed of the algorithm with the grid safety margin. It is calculated using the following formula: In the formula, This is the second attenuation coefficient; This is the initial value of the second attenuation coefficient; This is the third weighting coefficient; This represents the maximum allowable load rate of the line (100%). This represents the average load rate of the line.

[0041] When the improved particle swarm optimization algorithm is determined to be in the first iteration stage (less than or equal to one-third of the preset maximum number of iterations), the adaptive adjustment coefficient is determined based on the real-time iteration count and the first decay coefficient of the improved particle swarm optimization algorithm at this time. This process is shown in the following formula: In the formula, This is the adaptive adjustment coefficient; The baseline value for the adjustment coefficient in the early stage of iteration; k is the number of real-time iterations; The maximum number of iterations is preset. The update frequency for the optimal solution in the first 10 iterations is (number of updates / 10). This is a fixed value, 1.0; When the improved particle swarm optimization algorithm is determined to be in the second iteration stage (greater than one-third of the preset maximum number of iterations and less than or equal to two-thirds of the preset maximum number of iterations), the real-time particle aggregation degree of the improved particle swarm optimization algorithm at this time is calculated, and the adaptive adjustment coefficient is determined based on the real-time particle aggregation degree and the sensitivity coefficient. This process is shown in the following formula: In the formula, This serves as the baseline value for the adjustment coefficient during the iteration process. The particle aggregation degree is represented by n; the total number of iterations is represented by n. Let be the spatial position of the i-th particle; This represents the average position of all particles in the current population. =0.25; This represents the average particle aggregation degree over the first 30 iterations.

[0042] When the improved particle swarm optimization algorithm is determined to be in the third iteration stage (greater than two-thirds of the preset maximum number of iterations), the adaptive adjustment coefficient is determined based on the real-time iteration count and the second decay coefficient of the improved particle swarm optimization algorithm at this time. This process is shown in the following formula: In the formula, This serves as the baseline value for the adjustment coefficient in the later stages of the iteration. The mid-term convergence rate (change in optimal solution / number of iterations); =0.01.

[0043] In addition, if the line power flow sensitivity is >0.05 (the changes in photovoltaic power output have a significant impact on the line current). A temporary 15% increase enhances particle dispersion and prevents current overruns; however, if the average voltage stability margin is less than 0.1 (voltage is prone to instability). Temporarily increase by 10% to enhance global exploration and avoid voltage overshoot; if all constraints are satisfied and the margin is ≥20%, decrease μ by 10% to accelerate convergence.

[0044] This invention maps the characteristic indicators of power distribution network operation scenarios to the scenario adaptation coefficients of the particle swarm optimization algorithm, and switches different adaptive adjustment rules according to the iteration stage, thereby achieving a three-dimensional match between the algorithm's search behavior and the physical characteristics of the power grid and the requirements of the optimization process. At the same time, the three-stage adaptive mechanism adopted accurately matches the inherent evolution law of particle swarm optimization, and the introduction of particle aggregation degree constructs a closed-loop feedback control of population diversity, enabling the algorithm to autonomously "perceive" search difficulties and actively break through them. The sensitivity coefficient dynamically balances the game intensity between "exploration" and "development", realizing full-process parameter adaptation and eliminating tedious manual parameter tuning work.

[0045] In one embodiment, step S4 includes: The distribution network is decomposed into a three-level structure of sub-unit-transformer area-line, and based on the verification priority, the maximum photovoltaic access capacity of each sub-unit is calculated by the improved particle swarm algorithm with the fusion adaptive adjustment coefficient. The initial maximum capacity of each transformer substation is determined based on the maximum photovoltaic access capacity of each subunit, and the initial maximum capacity is verified based on the verification priority to obtain the final maximum capacity of each transformer substation. Power flow pre-calculation is performed using the final maximum capacity of each transformer area as the initial condition to constrain and correct each line, thereby generating corrected back-end transformer area capacity constraints. Based on the modified background area capacity constraint, the dynamic evaluation model is solved by the improved particle swarm optimization algorithm with fused adaptive adjustment coefficients to obtain the distributed photovoltaic carrying capacity evaluation result of the distribution network.

[0046] Specifically, this invention employs an improved particle swarm optimization algorithm with an adaptive adjustment coefficient and a hierarchical strategy to solve the dynamic evaluation model. The hierarchical strategy divides the distribution network into two levels: a "transformer area level" and a "feeder / line level." First, the maximum access capacity of each distribution transformer is calculated at the transformer area level, and the result is used as a boundary condition. Then, global optimization calculations are performed at the line level, significantly reducing the problem's dimensionality and improving computational efficiency and accuracy. The improvement in the particle swarm optimization algorithm lies in the adoption of an adaptive inertia weight strategy. Other steps are the same as the original algorithm (while particle representation, fitness function, and iteration termination conditions are adjusted according to the actual application scenario). This strategy enables the algorithm to have strong global exploration capabilities in the early stages and strong local development capabilities in the later stages, and can be fine-tuned according to particle performance, effectively avoiding premature convergence. The adaptive inertia weight in this strategy is dynamically adjusted based on the adaptive adjustment coefficient and the number of algorithm iterations. The adjustment formula is as follows: In the formula, For adaptive inertia weights; , These represent the maximum and minimum values ​​of the adaptive inertia weight; The maximum number of iterations is preset. This represents the current particle fitness value. , , These are the average, maximum, and minimum fitness values ​​of the current population, respectively.

[0047] This invention refines the two-level evaluation of "transformer area layer - line layer" into a five-step process: "transformer area sub-unit calculation → transformer area layer aggregation → line layer constraint correction → global optimization → constraint feedback closed loop", clarifying the data interaction and constraint transmission logic between levels. The process is as follows: (I) Division and Calculation of Sub-units in the Transit Area: The distribution network is divided into lines based on feeders or main lines. Each feeder is considered a line, and each feeder connects to several distribution substations. Each line is further divided into multiple distribution substations based on the number of distribution transformers. Each substation contains several sub-units and existing photovoltaic (PV) systems and loads. Each substation has multiple users (PV connection points) as sub-units, each with its own independent geographical location, electrical node number, and maximum installed capacity (physical space limitations). This forms a three-tiered structure of sub-units, substations, and lines. Alternatively, the entire distribution network can be broken down by voltage level or power supply zone: a sub-unit is a single user or a single PV grid connection point, a substation is the power supply range of a distribution transformer, and a line is a 10kV feeder.

[0048] Subsequently, based on the verification priority, an improved particle swarm optimization algorithm is used to calculate the maximum photovoltaic access capacity of each sub-unit. In one embodiment, the calculation of the maximum photovoltaic access capacity of each sub-unit using the improved particle swarm optimization algorithm with the fusion adaptive adjustment coefficient includes: Use any sub-unit as the basic sub-unit to initialize the particle swarm; the position of the particles in the particle swarm is the photovoltaic capacity of each node in the basic sub-unit. The verification priority and adaptive adjustment coefficient of the basic subunit are determined, and the fitness value and optimal position of each particle are determined with the maximum photovoltaic capacity of the nodes in the basic subunit as the objective. The adaptive inertial weight is determined based on the adaptive adjustment coefficient of the basic subunit, and the velocity and position of each particle are updated based on the adaptive inertial weight. The updated particles are constrained and verified by the verification priority of the basic sub-units. Particles that pass the verification are taken as new particles, and the optimal position of the population is updated. Based on the new particles, the particle update steps are iteratively executed until the iteration termination condition is reached, and the maximum photovoltaic access capacity of the basic subunit is determined according to the finally determined optimal population position. The basic sub-units are updated, and the maximum photovoltaic access capacity of the updated basic sub-units is calculated iteratively until the maximum photovoltaic access capacity of all sub-units is obtained.

[0049] Choose any sub-unit and use it as the basic sub-unit to begin the calculation. This sub-unit is a univariate optimization problem, and the decision variable is the photovoltaic (PV) access capacity at that location. Then, initialize the particle swarm (population size, particle position range, particle velocity range, preset maximum number of iterations, inertia weight range, etc.), and define each particle as a combination of PV access capacities of all nodes within the basic sub-unit. Extract the historical time-series data, real-time operating status, and future prediction data of the sub-unit as its basic unit parameters, and determine the verification priority and adaptive adjustment coefficient of the basic sub-unit according to the aforementioned verification priority generation process and adaptive adjustment coefficient calculation process, which will not be elaborated here.

[0050] Subsequently, using the maximum sum of photovoltaic capacity of nodes within the basic subunit as the optimization objective (i.e., the fitness function), the fitness value of each particle is calculated, and the position of the particle with the highest fitness value is taken as the optimal position of the population. Based on the aforementioned calculation process of adaptive inertia weight, the adaptive inertia weight of the basic subunit is determined. This allows each particle to update its position and velocity according to its own optimal position and the optimal position of the population, combined with the adaptive inertia weight. The constraints contained in the verification priority of the basic subunit are used to verify both the original and updated particles. Particles that pass the verification are taken as new particles, and the optimal position of the population is then updated based on these new particles.

[0051] The process iteratively updates particles based on new particles until the difference between two consecutive calculated fitness values ​​is less than a preset difference (or the preset maximum number of iterations is reached). The final optimal positions of the population are then aggregated as the maximum photovoltaic (PV) capacity for that basic sub-unit. It is verified that the voltage, load factor, and network loss of the sub-unit under this capacity all meet the dynamic constraints and are the maximum values ​​within the current constraint boundaries. This value is then the maximum PV capacity for that sub-unit, and the allocated capacity of each node is output. The basic sub-units are updated, and the maximum PV capacity of the updated basic sub-units is iteratively calculated until the maximum PV capacity of all sub-units is obtained. It should be noted that optimization algorithms such as genetic algorithms and primal particle swarm optimization can also be used when calculating the maximum PV capacity of a sub-unit.

[0052] This invention employs an improved particle swarm optimization method to calculate the maximum access capacity at the sub-unit level, and introduces sub-unit level verification priority and scenario adaptive parameters. This not only meets the requirements for safe operation of the distribution network, but also maximizes the renewable energy absorption capacity, achieving precise, rapid, and scenario-adaptive single-point capacity delimitation.

[0053] (ii) Aggregation of the transformer sub-region: The maximum photovoltaic (PV) capacity of all sub-units within a distribution area is aggregated to obtain the initial maximum capacity of each distribution area. Based on verification priority, it is then checked whether the total capacity of the distribution area meets the "distribution transformer load rate". ( This represents the initial maximum capacity of the transformer area. This refers to the rated capacity of the transformer. If the power factor is used, the sub-unit capacity is reduced proportionally (reducing the load density of sub-units below the preset density by 10%) to obtain the final maximum capacity of the transformer area.

[0054] (III) Line Layer Constraint Correction: The final maximum capacity of all stations As a "hard constraint" at the line layer, the total line capacity cannot exceed the sum of the final maximum capacity of all transformer substations; subsequently, based on the transformer substations... Power flow pre-calculation is performed using the forward-backward substitution method to calculate line power flow. If the power flow of a certain branch exceeds the "line load rate ≤ 100%" constraint in the verification priority, the corresponding transformer area is corrected in reverse. (e.g., reducing the capacity of the transformer area by 10% to 20%), and calculating the modified back-end capacity constraint at the line level based on the corrected results. .

[0055] (iv) Global optimization at the line layer: Using a dynamic evaluation model as the optimization objective, the actual access capacity of each transformer area as the optimization variable, and the constraints in the modified back-end capacity constraints and the corresponding verification priorities of the lines as constraints, an improved particle swarm optimization algorithm is used for global optimization. The total maximum capacity of the lines and the node capacity margin of each transformer area are output as the evaluation results of the distributed photovoltaic carrying capacity of the distribution network.

[0056] In one embodiment, the step of solving the dynamic evaluation model based on the modified background area capacity constraint and using the improved particle swarm optimization algorithm with fused adaptive adjustment coefficients to obtain the distributed photovoltaic carrying capacity evaluation result of the distribution network includes: Use any line as the base line and initialize the particle swarm; the particle positions in the particle swarm are the actual access capacity combinations of all transformer substations within the base line. The verification priority and adaptive adjustment coefficient of the basic circuit are determined, and the fitness value of each particle is calculated based on the modified background area capacity constraint and the dynamic evaluation model to determine the optimal position of the population. The adaptive inertial weight is determined based on the adaptive adjustment coefficient of the basic circuit, and the velocity and position of each particle are updated based on the adaptive inertial weight. Based on the verification priority of the basic line, the updated particles are subjected to dual real-time verification based on power flow pre-calculation and constraint hierarchical verification to obtain new particles, and the optimal position of the population is updated. Based on the new particles, the particle update steps are iteratively executed until the iteration termination condition is reached, and the distributed photovoltaic carrying capacity assessment result of the basic line is determined according to the finally determined optimal population position. The basic lines are updated, and the distributed photovoltaic carrying capacity assessment results of the updated basic lines are iteratively calculated until the distributed photovoltaic carrying capacity assessment results of the distribution network are obtained.

[0057] Choose any one line and use it as the base line to start the calculation. The decision variable is the actual access capacity of all transformer substations within that line. Then, initialize the particle swarm (population size, particle position range, particle velocity range, preset maximum number of iterations, inertia weight range, etc.), and define each particle as a combination of photovoltaic access capacities of all transformer substations within the base line. Extract historical time-series data, real-time operating status, and future prediction data of the base line as its base line parameters. Determine the verification priority and adaptive adjustment coefficient of the base line according to the aforementioned verification priority generation process and adaptive adjustment coefficient calculation process, which will not be elaborated here.

[0058] Subsequently, using a dynamic evaluation model as the fitness function, and taking the hard constraint corresponding to the modified back-end capacity constraint as an additional constraint, the fitness value of each particle is calculated, and the position of the particle with the largest fitness value is taken as the optimal position of the population. Based on the aforementioned calculation process of adaptive inertia weight, the adaptive inertia weight of the base line is determined, allowing each particle to update its position and velocity according to its own optimal position and the optimal position of the population, in conjunction with the adaptive inertia weight. The particle position is taken as the total photovoltaic injection power of each transformer area of ​​the feeder, and allocated to each sub-unit according to the theoretical maximum capacity ratio of each sub-unit within the transformer area (this allocation scheme has been determined at the transformer area level; here it is only scaled proportionally), forming full feeder photovoltaic injection data. From this data, two typical cross-sections are selected: 12:00 noon (peak photovoltaic output period) and 19:30 evening peak (peak load, no photovoltaic period) for fast power flow calculation. If any cross-section exhibits one of the following conditions, the particle is directly determined as a severely infeasible solution and assigned a maximum fitness value (such as any...). If the node voltage is <0.92 pu or >1.08 pu, or any branch current is >110% of the dynamic threshold, or any transformer load rate is >90%, skip the subsequent full-time verification. For particles that pass the typical section screening, perform a 24-hour 96-point time-series power flow calculation to obtain the node voltage, branch current, and transformer load rate at each time point. Then, use the constraints contained in the verification priority of the basic line to verify the original particles and the updated particles. Particles that pass the verification are used as new particles, and the optimal position of the population is updated based on these new particles.

[0059] The process iteratively updates the particles based on new particles until the difference between two consecutive calculated fitness values ​​is less than a preset difference (or the preset maximum number of iterations is reached). The final optimal positions of the population are then aggregated as the maximum photovoltaic (PV) capacity for the basic line. It is verified that the voltage, load factor, and network loss of the sub-units under this capacity all meet the dynamic constraints and are the maximum values ​​within the current constraint boundaries. This value is then the maximum PV capacity for the line, and the allocated capacity of each transformer substation within the line is output. The basic lines are updated, and the maximum PV capacity of the updated basic lines is iteratively calculated until the maximum PV capacity of all lines and the node capacity margin of their respective transformer substations are obtained. It should be noted that genetic algorithms, primal particle swarm optimization, and other optimization algorithms can also be used to calculate the maximum PV capacity of a line.

[0060] This invention introduces an efficient solution method that integrates "corrected back-end area capacity constraints - dual real-time verification - adaptive particle swarm algorithm" in the global optimization of the line layer. By using the upstream corrected transformer area capacity constraints as rigid boundaries, embedding a constraint hierarchical verification mechanism based on power flow pre-calculation, and adopting a scenario-adaptive particle swarm algorithm, it significantly improves the optimization efficiency while ensuring the feasibility of the solution, and achieves a win-win situation of solution efficiency and operational safety.

[0061] (v) Constraint Feedback Closed Loop: If the node margin of a certain distribution area in the output result is <5%, the secondary calculation of the sub-unit of that distribution area is triggered in reverse (adjusting the constraint threshold of the sub-unit, such as relaxing the voltage deviation from ±4% to ±4.5%). If the actual photovoltaic output deviates from the predicted value by more than 10%, the secondary calculation of the sub-unit is triggered in reverse, and the constraint threshold is adjusted to ensure that the evaluation result is consistent with the real-time operating status. After global optimization, if the system network loss rate is >4%, the line layer constraint correction is re-executed (reducing the capacity of the distribution area with the largest network loss contribution by 10%) until these conditions are met. Finally, the maximum photovoltaic access capacity of the entire distribution network and the node capacity margin of each distribution area are obtained as the output of the distributed photovoltaic carrying capacity evaluation result.

[0062] This invention addresses the engineering challenge of solving high-dimensional optimization problems in the dynamic assessment of distributed photovoltaic carrying capacity in power distribution networks. It proposes a hierarchical solution strategy that integrates "three-level structural decomposition, hierarchical progressive verification, and global coordinated optimization," and deeply couples it with an improved particle swarm optimization algorithm with scene adaptability. By physically decomposing the power distribution network into a three-level structure of "sub-unit-transformer area-line," the algorithm reduces dimensionality and tightens the constraint boundaries layer by layer, and dynamically guides the optimization direction based on verification priority, achieving a win-win situation of computational efficiency and optimal solution.

[0063] In addition, this invention can also construct a visual interface to display three-level data dashboards of "sub-unit-transformer area-line" and a "constraint conflict heatmap", such as: Sub-unit dashboard: showing the matching degree between load density and photovoltaic capacity (red indicates matching degree <60%, requiring optimization); Transformer area dashboard: showing the deviation between the final maximum capacity and its actual connected capacity (yellow indicates deviation >10%, requiring power flow verification); Line dashboard: showing the correlation curve between branch power flow and capacity margin (green indicates power flow <80%, capacity can be added). "Constraint conflict heatmap": showing the constraint conflict points and conflict causes at each level of sub-unit-transformer area-line (e.g., "Transformer area 2-line 3: current over-limit, conflict constraint: line load rate").

[0064] This invention achieves precise matching between algorithm parameters and distribution network operating characteristics through a linked design of dual-dimensional adaptive adjustment coefficients and scenario adaptation coefficients. By dynamically generating "real-time grid parameters + algorithm status + time-series prediction," it achieves "fully adaptive adaptation without manual intervention," increasing the probability of finding the global optimal solution to 99.2% in complex scenarios, effectively solving the problem of poor scenario adaptability in existing technologies. Furthermore, by quantifying grid physical characteristics such as line power flow sensitivity and voltage stability margin as the basis for algorithm parameter adjustment, it solves the problem of "algorithm optimization being disconnected from grid characteristics" in existing technologies, and improves the accuracy of line power flow calculations. The error was reduced to 0.8%, and the number of convergence iterations was reduced to 120, balancing the timeliness and accuracy requirements of engineering applications. Through dynamic adjustment of target weights and constraint priorities, Pareto optimal weights balance capacity and economy. Under the premise of ensuring grid security, the maximum carrying capacity of photovoltaics is increased by more than 25% compared with existing technologies, while reducing the system grid loss rate, achieving dual optimization of capacity and economy. By adopting dynamic constraint thresholds and predictive driving, hierarchical closed-loop strategies and refined visualization functions, constraint conflict points can be accurately located, providing quantitative basis for grid transformation and reducing transformation costs by 25%-30%.

[0065] This invention addresses the problem of insufficient scenario adaptability of the IPSO algorithm by combining particle aggregation degree with the iteration stage, and then linking it with the distribution network operation parameters (σ). L σ PV The system dynamically updates the adjustment coefficient μ (such as line load rate, etc.) while taking into account both the algorithm's search rules and the characteristics of the distribution network scenario. This involves the cross-integration of two different technical fields and goes beyond the scope of conventional algorithm optimization. To address the problem of "rigid hierarchical evaluation results and poor engineering implementation," a "constraint feedback closed loop" is proposed, which triggers secondary calculations of sub-units in reverse, breaking the inherent perception that "hierarchical calculations are irreversible." To address the problem of "fixed constraint priorities leading to limited carrying capacity," a dynamic priority mechanism based on operating status is designed, which represents a breakthrough understanding of the "balance between grid security and new energy consumption," rather than local optimization.

[0066] In one embodiment, an industrial park power distribution network (10kV line, including 5 transformer substations and 20 sub-units) is used as an example to verify the effectiveness of the solution described in this invention: In this embodiment, the target distribution network is a 10kV line with a total length of 8.5km and conductor type JKLGYJ-240. The average load rate of the line is 55%. It includes 5 distribution substations, each equipped with one 1000kVA distribution transformer, with a baseline load rate of 60%. All 20 sub-units are industrial loads with a load fluctuation intensity of 0.35 and a historical maximum load of 850kW. There are 20 distributed photovoltaic power stations connected, each with a capacity of 50kW. The short-term load fluctuation variance is 0.42 (standard deviation of daily peak-to-valley ratio over the past 7 days), the long-term load fluctuation variance is 0.30 (standard deviation of daily peak-to-valley ratio over the past 3 months), and the load-PV correlation coefficient is 0.6 (positive correlation, indicating a tendency for grid overload). The PV output fluctuation rate is 20% (hourly maximum output change rate). The PV output prediction error based on the ARIMA model is 8% (less than the maximum allowable error of 15%).

[0067] IPSO algorithm parameter settings: The average node voltage stability margin is 0.22 (calculated using the PV curve method); IPSO algorithm basic parameters: population size 50 (balancing diversity and efficiency, with a global optimum finding probability of 99.2%), maximum number of iterations 200 (generally converges after 180 iterations); particle aggregation threshold D. th =0.25.

[0068] After implementing the scheme described in this invention, the core results are output: the maximum carrying capacity of distributed photovoltaic power in the distribution network is 1950kW (annual scale assessment), and the capacity margin of each node is 10~15kW for the warehouse sub-unit and 5~8kW for the workshop sub-unit; sensitivity analysis: σ L For every 10% reduction, the load-bearing capacity increases by 8.2%; σ PV For every 10% reduction, the carrying capacity increases by 3.1%; Constraint marginal contribution: Voltage constraint limits carrying capacity to increase by 10.5%, line load rate constraint limits to 7.8%, and distribution transformer load rate constraint limits to 3.2%; Multi-timescale assessment: Annual maximum carrying capacity 1950kW, monthly scale 1920kW (summer peak load), daily scale 1894kW (PV-load peak overlap day); Visualization display: Three-level dashboard displays sub-unit load-PV matching degree (all ≥65%, green), transformer area capacity deviation (all ≤8%, green), and line power flow (all ≤85%, green); Constraint conflict heatmap has no over-limit points, and risk warnings are green.

[0069] This application addresses the problem of providing a dynamic, accurate, efficient, and practically applicable method for assessing the carrying capacity of distributed photovoltaic power grids. It designs a method for assessing the carrying capacity of distributed photovoltaic power grids by integrating historical time-series data with scenario characteristic indicators. This constructs a more comprehensive dynamic assessment model, shifting the assessment dimension from "static cross-section" to "spatiotemporal dynamics." The constructed target system considers both economic efficiency and low carbon emissions, breaking through the previous single-objective model that only used voltage deviation or equipment overload as hard constraints, achieving synergistic optimization of social and grid benefits. Based on real-time operating status and future prediction data, multiple constraints are dynamically prioritized, enabling the model to improve efficiency in actual problem-solving. The algorithm prioritizes high-risk constraints, improving solution efficiency. An adaptive adjustment coefficient is introduced into the improved particle swarm optimization algorithm. This coefficient is jointly determined by the scene adaptation coefficient and the iteration stage. The scene adaptation coefficient is dynamically generated based on scene feature indicators, giving the algorithm differentiated search preferences under different operating environments. The iteration stage adjustment ensures that the algorithm focuses on global exploration in the early stages and local refinement in the later stages. This dual adaptive mechanism significantly improves the convergence accuracy and robustness of the algorithm under complex distribution network topologies. Addressing the problem of numerous nodes and an explosion of decision variable dimensionality in large-scale distribution networks, an improved algorithm pre-layered and graded solution strategy is adopted, significantly reducing computational complexity while supporting online applications.

[0070] It should be noted that although the steps in the flowchart above are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order requirement for the execution of these steps, and they can be executed in other orders.

[0071] In another embodiment, such as Figure 2 As shown, a second aspect of the present invention provides a distributed photovoltaic carrying capacity assessment device for a distribution network, comprising: The indicator calculation module 10 is used to acquire the structural data, historical time-series data, real-time operating status and future prediction data of the power distribution network, and determine the scenario characteristic indicators based on the structural data and the historical time-series data. Model building module 20 is used to build a dynamic evaluation model based on the future prediction data, with the goal of maximizing the total access capacity of distributed photovoltaics in the distribution network and minimizing the total active power loss of the system. The priority determination module 30 is used to set multiple constraints of the dynamic evaluation model and determine the verification priority of the multiple constraints based on the real-time running status and the future prediction data. The carrying capacity assessment module 40 is used to solve the dynamic assessment model based on the verification priority, using an improved particle swarm algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy, to obtain the distributed photovoltaic carrying capacity assessment result of the distribution network; the adaptive adjustment coefficients are dynamically adjusted according to the scenario adaptation coefficients determined based on the scenario feature indicators and the real-time iteration stage of the improved particle swarm algorithm.

[0072] It should be noted that each module in the aforementioned distributed photovoltaic carrying capacity assessment device for distribution networks can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module. For specific limitations regarding the distributed photovoltaic carrying capacity assessment device for distribution networks, please refer to the limitations of the distributed photovoltaic carrying capacity assessment method for distribution networks described above; both have the same function and role, and will not be repeated here.

[0073] In summary, this invention relates to the field of distribution network operation and new energy access technology, and discloses a method and device for assessing the distributed photovoltaic carrying capacity of a distribution network. The method calculates scenario characteristic indicators based on the structural data and historical time-series data of the distribution network, and constructs a dynamic assessment model with the objectives of maximizing the total photovoltaic access capacity and minimizing the total active power loss, combined with future prediction data of the distribution network. Multiple constraints are set for the model, and the verification priority of these constraints is determined based on real-time operating status and future prediction data. Then, an improved particle swarm optimization algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy are used to solve the dynamic assessment model, obtaining the distributed photovoltaic carrying capacity assessment result of the distribution network. The adaptive adjustment coefficients are dynamically adjusted according to the scenario adaptation coefficients determined based on the scenario characteristic indicators and the real-time iteration stage of the improved algorithm, achieving accurate and efficient assessment of the photovoltaic carrying capacity of the distribution network.

[0074] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0075] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for assessing the carrying capacity of distributed photovoltaic power in a distribution network, characterized in that, include: Acquire structural data, historical time-series data, real-time operating status and future prediction data of the power distribution network, and determine scenario characteristic indicators based on the structural data and the historical time-series data; With the goal of maximizing the total access capacity of distributed photovoltaic power in the distribution network and minimizing the total active power loss of the system, a dynamic evaluation model is constructed based on the future prediction data. Set multiple constraints for the dynamic evaluation model, and determine the verification priority of the multiple constraints based on the real-time running status and the future prediction data; Based on the verification priority, the dynamic evaluation model is solved using an improved particle swarm optimization algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy to obtain the distributed photovoltaic carrying capacity evaluation result of the distribution network; the adaptive adjustment coefficients are dynamically adjusted according to the scenario adaptation coefficients determined based on the scenario feature indicators and the real-time iteration stage of the improved particle swarm optimization algorithm.

2. The method for assessing the carrying capacity of distributed photovoltaic power generation in a distribution network according to claim 1, characterized in that, The historical time-series data includes historical time-series load data and historical output time-series data; wherein... The step of determining scene feature indicators based on the structured data and the historical time-series data includes: Extract the daily load peak-to-valley ratio from the historical time-series load data, and calculate the time-series standard deviation of the daily load peak-to-valley ratio around its average value to obtain the load fluctuation intensity; The rate of change of photovoltaic power output within adjacent time points is calculated based on the historical power output time series data, and the maximum value of the rate of change among all time points is taken as the photovoltaic power output volatility. Based on the historical time-series load data and the historical power output time-series data, the load power sequence and photovoltaic power output sequence within the same time period are extracted, and the Pearson correlation coefficient between these two sequences is used as the load-photovoltaic correlation coefficient. Under preset typical operating conditions, the Jacobian matrix of power flow calculation is solved by the structural data to obtain the average power flow sensitivity of the line. Based on the structural data, the average node voltage stability margin is calculated using the PV curve method. The scenario characteristic index is obtained by combining the load fluctuation intensity, the photovoltaic output fluctuation rate, the load-photovoltaic correlation coefficient, the average line power flow sensitivity, and the average node voltage stability margin.

3. The method for assessing the carrying capacity of distributed photovoltaic power grids according to claim 1, characterized in that, The dynamic evaluation model, which aims to maximize the total connected capacity of distributed photovoltaic power in the distribution network and minimize the total active power loss of the system, and is constructed based on the future prediction data, includes: The first objective is to maximize the total access capacity of distributed photovoltaic power in the distribution network, and the second objective is to minimize the total active power loss of the system. The first and second objectives are then input into the NSGA-II algorithm for solving, and a set of Pareto optimal solutions is output. An initial solution is selected from the Pareto optimal solution set based on a preset decision rule, and the quality score of the initial solution is determined based on the future prediction data. The initial solution with a quality score exceeding the preset score is taken as the best solution. The weights corresponding to the first objective and the second objective are derived from the optimal solution, and the weights are used to perform a weighted summation of the first objective and the second objective to form the dynamic evaluation model.

4. The method for assessing the carrying capacity of distributed photovoltaic power grids according to claim 2, characterized in that, The step of determining the verification priority of the multiple constraints based on the real-time operating status and the future prediction data includes: Based on the future forecast data, the peak overlap rate of photovoltaic load and the transformer load rate curve are determined, and the boundary thresholds of the multiple constraints are dynamically optimized based on the peak overlap rate of photovoltaic load and the transformer load rate curve to obtain the boundary threshold optimization result. Based on the real-time operating status, the voltage over-limit risk and real-time line load rate are determined. Based on the boundary threshold optimization results, the priority of the multiple constraints is dynamically adjusted using the voltage over-limit risk and the real-time line load rate to obtain the verification priority.

5. The method for assessing the carrying capacity of distributed photovoltaic power grids according to claim 4, characterized in that, The multiple constraints include at least voltage deviation constraints and line load rate constraints; wherein... Based on the boundary threshold optimization results, the priority of the multiple constraints is dynamically adjusted according to the voltage over-limit risk and the real-time line load rate to obtain the verification priority, including: When the voltage exceedance risk is greater than the first voltage risk threshold, the voltage deviation constraint is adjusted to the first priority; when the voltage exceedance risk is greater than the second voltage risk threshold but not greater than the first voltage risk threshold, the voltage deviation constraint is adjusted to the second priority; and when the voltage exceedance risk is not greater than the second voltage risk threshold, the voltage deviation constraint is adjusted to the third priority. When the real-time line load rate is greater than the first load rate risk threshold, the line load rate constraint is adjusted to the first priority; when the real-time line load rate is greater than the second load rate risk threshold but not greater than the first load rate risk threshold, the line load rate constraint is adjusted to the second priority; and when the real-time line load rate is not greater than the second load rate risk threshold, the line load rate constraint is adjusted to the third priority. The verification priority is obtained by combining the boundary threshold optimization result, the multiple constraint conditions and their adjusted priorities.

6. The method for assessing the carrying capacity of distributed photovoltaic power generation in a distribution network according to claim 5, characterized in that, The scene adaptation coefficient includes a first attenuation coefficient, a sensitivity coefficient, and a second attenuation coefficient; wherein... The dynamic adjustment based on the scene adaptation coefficient determined by the scene feature indicators and the real-time iterative phase of the improved particle swarm optimization algorithm includes: The first attenuation coefficient is determined based on the load fluctuation intensity and the load-photovoltaic correlation coefficient; the sensitivity coefficient is determined based on the photovoltaic output fluctuation rate and the average line power flow sensitivity; and the second attenuation coefficient is determined based on the average node voltage stability margin and line load data. When the improved particle swarm optimization algorithm is determined to be in the first iteration stage, the adaptive adjustment coefficient is determined based on the real-time iteration number of the improved particle swarm optimization algorithm and the first decay coefficient. When the improved particle swarm optimization algorithm is determined to be in the second iteration stage, the real-time particle aggregation degree of the improved particle swarm optimization algorithm is calculated, and the adaptive adjustment coefficient is determined based on the real-time particle aggregation degree and the sensitivity coefficient. When the improved particle swarm optimization algorithm is determined to be in the third iteration stage, the adaptive adjustment coefficient is determined based on the real-time iteration number of the improved particle swarm optimization algorithm and the second decay coefficient.

7. The method for assessing the carrying capacity of distributed photovoltaic power grids according to claim 6, characterized in that, Based on the verification priority, the improved particle swarm optimization algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy are used to solve the dynamic evaluation model to obtain the distributed photovoltaic carrying capacity evaluation results of the distribution network, including: The distribution network is decomposed into a three-level structure of sub-unit-transformer area-line, and based on the verification priority, the maximum photovoltaic access capacity of each sub-unit is calculated by the improved particle swarm algorithm with the fusion adaptive adjustment coefficient. The initial maximum capacity of each transformer substation is determined based on the maximum photovoltaic access capacity of each subunit, and the initial maximum capacity is verified based on the verification priority to obtain the final maximum capacity of each transformer substation. Power flow pre-calculation is performed using the final maximum capacity of each transformer area as the initial condition to constrain and correct each line, thereby generating corrected back-end transformer area capacity constraints. Based on the modified background area capacity constraint, the dynamic evaluation model is solved by the improved particle swarm optimization algorithm with fused adaptive adjustment coefficients to obtain the distributed photovoltaic carrying capacity evaluation result of the distribution network.

8. The method for assessing the carrying capacity of distributed photovoltaic power generation in a distribution network according to claim 7, characterized in that, The calculation of the maximum photovoltaic access capacity of each sub-unit using the improved particle swarm optimization algorithm with the fusion adaptive adjustment coefficient includes: Use any sub-unit as the basic sub-unit to initialize the particle swarm; the position of the particles in the particle swarm is the photovoltaic capacity of each node in the basic sub-unit. The verification priority and adaptive adjustment coefficient of the basic subunit are determined, and the fitness value and optimal position of each particle are determined with the maximum photovoltaic capacity of the nodes in the basic subunit as the objective. The adaptive inertial weight is determined based on the adaptive adjustment coefficient of the basic subunit, and the velocity and position of each particle are updated based on the adaptive inertial weight. The updated particles are constrained and verified by the verification priority of the basic sub-units. Particles that pass the verification are taken as new particles, and the optimal position of the population is updated. Based on the new particles, the particle update steps are iteratively executed until the iteration termination condition is reached, and the maximum photovoltaic access capacity of the basic subunit is determined according to the finally determined optimal population position. The basic sub-units are updated, and the maximum photovoltaic access capacity of the updated basic sub-units is calculated iteratively until the maximum photovoltaic access capacity of all sub-units is obtained.

9. The method for assessing the carrying capacity of distributed photovoltaic power grids according to claim 7, characterized in that, The process of solving the dynamic evaluation model based on the modified background area capacity constraint using the improved particle swarm optimization algorithm with fused adaptive adjustment coefficients to obtain the distributed photovoltaic carrying capacity evaluation result of the distribution network includes: Use any line as the base line and initialize the particle swarm; the particle positions in the particle swarm are the actual access capacity combinations of all transformer areas within the base line. The verification priority and adaptive adjustment coefficient of the basic circuit are determined, and the fitness value of each particle is calculated based on the modified background area capacity constraint and the dynamic evaluation model to determine the optimal position of the population. The adaptive inertial weight is determined based on the adaptive adjustment coefficient of the basic circuit, and the velocity and position of each particle are updated based on the adaptive inertial weight. Based on the verification priority of the basic line, the updated particles are subjected to dual real-time verification based on power flow pre-calculation and constraint hierarchical verification to obtain new particles, and the optimal position of the population is updated. Based on the new particles, the particle update steps are iteratively executed until the iteration termination condition is reached, and the distributed photovoltaic carrying capacity assessment result of the basic line is determined according to the finally determined optimal population position. The basic lines are updated, and the distributed photovoltaic carrying capacity assessment results of the updated basic lines are iteratively calculated until the distributed photovoltaic carrying capacity assessment results of the distribution network are obtained.

10. A distributed photovoltaic carrying capacity assessment device for power distribution networks, characterized in that, include: The indicator calculation module is used to acquire the structural data, historical time-series data, real-time operating status and future prediction data of the power distribution network, and determine the scenario characteristic indicators based on the structural data and the historical time-series data. The model building module is used to construct a dynamic evaluation model based on the future prediction data, with the goal of maximizing the total access capacity of distributed photovoltaics in the distribution network and minimizing the total active power loss of the system. The priority determination module is used to set multiple constraints of the dynamic evaluation model and determine the verification priority of the multiple constraints based on the real-time running status and the future prediction data. The carrying capacity assessment module is used to solve the dynamic assessment model based on the verification priority, using an improved particle swarm algorithm with integrated adaptive adjustment coefficients and a hierarchical strategy, to obtain the distributed photovoltaic carrying capacity assessment result of the distribution network; the adaptive adjustment coefficients are dynamically adjusted according to the scenario adaptation coefficients determined based on the scenario feature indicators and the real-time iteration stage of the improved particle swarm algorithm.

Images