Method for analyzing influence of distributed photovoltaic on voltage of power distribution network

Abstract

The invention relates to the technical field of power system analysis, and discloses a method for analyzing the influence of distributed photovoltaic on the voltage of a power distribution network. The method comprises the following steps: collecting geographic position coordinates of distributed photovoltaic access points, real-time output data, power distribution network node load curve data and network impedance parameters, and calculating a dynamic coupling coefficient to generate a photovoltaic load coupling characteristic data set; then, data is extracted based on the data set and grouped according to a topological hierarchical structure, and evaluation sections are marked to form a hierarchical evaluation framework set; then selecting data for each evaluation section and combining a time sequence production simulation method to analyze the influence of photovoltaic output fluctuation on node voltage, and evaluating a voltage stable state; identifying voltage out-of-limit risk points to form a set; and finally, analyzing the node relationship to generate and store voltage safety early warning information of the power distribution network. The method can comprehensively analyze the influence, accurately identify the risk, generate effective early warning, and guarantee the safe and stable operation of the power distribution network.

Description

Technical Field

[0001] This invention relates to the field of power system analysis technology, specifically a method for analyzing the impact of distributed photovoltaic power on distribution network voltage. Background Technology

[0002] With the rapid development of the global economy and the continuous rise in energy demand, traditional non-renewable energy sources such as coal, oil, and natural gas are facing numerous problems, including dwindling reserves, severe environmental pollution, and volatile prices. These non-renewable energy sources were formed in nature over hundreds of millions of years and cannot be replenished in the short term. With large-scale development and utilization, their reserves are decreasing and will eventually be depleted. Taking oil as an example, global oil reserves are gradually decreasing, and the difficulty and cost of extraction are constantly increasing. At the same time, the combustion of oil produces large amounts of greenhouse gases and pollutants, causing serious environmental damage.

[0003] Against this backdrop, energy transition is urgently needed, and renewable energy, due to its cleanliness and sustainability, has become an important direction for global energy development. Distributed photovoltaic (PV) power generation, as a crucial component of renewable energy, has experienced rapid growth in recent years. It utilizes solar energy to generate electricity, converting the heat generated by the melting reaction of hydrogen inside the sun into electrical energy through the photovoltaic effect. Distributed PV power generation boasts economic advantages such as relatively small installed capacity, low initial investment, short construction period, and low investment risk. Furthermore, it produces almost no pollutants during power generation, minimizing environmental impact and making it a green and environmentally friendly power generation method. In addition, distributed PV is typically constructed using building-integrated or building-attached construction methods, making full use of building rooftops and open spaces, greatly contributing to land resource conservation.

[0004] According to data from the National Energy Administration, the newly installed capacity of industrial and commercial photovoltaic (PV) power generation in China reached 88.6 GW in 2024, a year-on-year increase of 68%, accounting for 32% of the total newly installed PV capacity, becoming a key engine driving industry growth. In the first three quarters of 2024, the total newly connected capacity nationwide reached 160.88 million kilowatts, a year-on-year increase of 24.8%, of which distributed PV installations reached 85.22 million kilowatts, fully demonstrating the vigorous vitality and huge potential of the distributed PV market.

[0005] Traditional power distribution networks are primarily power transmission and distribution systems from substations to end users. Their structures are often radial, multi-segmented with single interconnections, or multi-segmented with multiple interconnections. Voltage levels are generally low, and their main responsibility is to step down high-voltage power before distributing it to users. They are characterized by large load variations, low construction costs but high operation and maintenance costs, and direct service to users. However, with the large-scale integration of distributed photovoltaic (PV) systems, the structure and operating characteristics of power distribution networks have changed significantly, gradually transforming into a multi-source structure.

[0006] Distributed photovoltaic (PV) systems are inherently intermittent and random, with their power generation significantly affected by factors such as sunlight intensity and weather. For example, on cloudy days or at night, insufficient sunlight can reduce distributed PV power generation to zero; conversely, during periods of ample sunlight, power generation increases dramatically. Simultaneously, the load on the distribution network is also uncertain, as different users have varying electricity consumption habits and demands, leading to substantial load fluctuations. The combined uncertainty of these two factors presents numerous challenges to the operation of the distribution network.

[0007] When distributed photovoltaic (PV) power generation is high, it may cause the voltage at the connection point to rise, even exceeding the rated voltage range; conversely, when the power generation is low, it may cause a voltage drop, affecting users' normal electricity consumption. This voltage fluctuation not only reduces power quality but may also damage grid equipment. Distributed PV typically uses power electronic converters for grid connection. These devices generate harmonic currents during operation, which, when injected into the grid, cause voltage and current waveform distortion, i.e., harmonic pollution. Harmonic pollution not only reduces power quality but may also cause overheating, vibration, and noise problems in grid equipment, and in severe cases, even damage the equipment. Furthermore, harmonics may cause malfunctions in grid protection devices, affecting the safe and stable operation of the grid. The integration of distributed PV may also lead to increased grid losses, and the uncertainties inherent in distributed PV make the selection of transformers and other equipment more difficult in grid planning. Summary of the Invention

[0008] The purpose of this invention is to provide a method for analyzing the impact of distributed photovoltaic power on distribution network voltage, so as to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, this invention provides a method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage, the method comprising:

[0010] Collect the geographical coordinates and real-time power output data of distributed photovoltaic access points, simultaneously acquire the load curve data and network impedance parameters of distribution network nodes, associate the photovoltaic access points and calculate the dynamic coupling coefficient between photovoltaic power output and load demand, and generate a photovoltaic load coupling characteristic dataset.

[0011] Based on the photovoltaic load coupling feature dataset, dynamic coupling coefficients and node voltage measurements are extracted, and the data are grouped according to the topological hierarchy of the distribution network. The evaluation sections at the transformer substation level and the evaluation sections at the medium-voltage line level are marked to form a hierarchical evaluation framework set.

[0012] For each evaluation segment in the hierarchical evaluation framework set, historical operating data sequences and extreme operating condition scenario data are selected. The impact of photovoltaic power output fluctuations on node voltage is analyzed by combining time-series production simulation methods, the voltage deviation amplitude is calculated, the voltage stability is evaluated, and hierarchical voltage impact analysis results are generated.

[0013] From the results of the hierarchical voltage impact analysis, identify nodes whose voltage exceedance risk values ​​exceed a preset risk threshold and are located in key hierarchical sections, and collect these nodes to form a set of voltage exceedance risk points;

[0014] Based on the set of voltage over-limit risk points, the electrical distance relationship between nodes and the voltage impact propagation path are analyzed, nodes with high propagation risk are marked, and distribution network voltage safety early warning information is generated.

[0015] The voltage safety early warning information of the power distribution network is stored for later retrieval.

[0016] Preferably, the process of associating photovoltaic access points and calculating the dynamic coupling coefficient between photovoltaic output and load demand includes:

[0017] Collect photovoltaic power output time series data and load demand time series data, and obtain the initial coupling coefficient by calculating the matching degree of the changing trends of the two series within the same time period;

[0018] Real-time meteorological data, including cloud cover and ambient temperature, are used to dynamically correct the initial coupling coefficient. At the same time, the weighted adjustment is made by combining the composition ratio of load types, including residential load, commercial load and industrial load, to obtain the dynamic coupling coefficient value.

[0019] Preferably, the formation of the hierarchical evaluation framework set includes:

[0020] Based on the distribution network topology, the nodes are divided into transformer substation level and line level. The transformer substation level corresponds to the power supply area of ​​the low-voltage distribution transformer, and the line level corresponds to the medium-voltage distribution line section.

[0021] Define evaluation parameters for each level, including short-circuit capacity and impedance ratio, and generate a level evaluation parameter table;

[0022] Based on the hierarchical evaluation parameter table, each evaluation segment is labeled to distinguish between normal evaluation segments and key evaluation segments.

[0023] Preferably, the analysis of the impact of photovoltaic power output fluctuations on node voltage using the time-series production simulation method includes:

[0024] By calling up the photovoltaic output and load change sequences from historical operating data, a typical daily operating scenario is simulated;

[0025] Extreme value theory is used to extract the extreme values ​​of photovoltaic power output and load under extreme conditions, and the fluctuation range of node voltage under the extreme values ​​of photovoltaic power output and load is calculated.

[0026] The risk of voltage deviation caused by photovoltaic fluctuations is assessed by comparing the voltage fluctuation range with the safety limit.

[0027] Preferably, the calculation of the voltage deviation amplitude includes:

[0028] Select node voltage data within the evaluation section and calculate the absolute value of the deviation between the measured voltage value and the rated voltage value;

[0029] The distribution of the absolute value of the statistical deviation over time series is used to determine the duration and frequency of voltage deviation.

[0030] By combining the absolute value of the deviation, duration, and frequency, a voltage deviation amplitude index is synthesized.

[0031] Preferably, the nodes identified from the hierarchical voltage impact analysis results that have a voltage exceedance risk value exceeding a preset risk threshold and are located in a critical hierarchical segment include:

[0032] Obtain voltage limit exceedance probability data from the hierarchical voltage impact analysis results;

[0033] The probability of voltage exceeding the limit is compared with a preset risk threshold to filter out nodes that exceed the threshold;

[0034] Check whether nodes exceeding the threshold are located in critical level sections, including important nodes at the transformer substation level or hub nodes at the line level. Once confirmed, include them in the voltage over-limit risk point set.

[0035] Preferably, the electrical distance relationship and voltage influence propagation path between the analysis nodes include:

[0036] Calculate the electrical distance between nodes based on the distribution network impedance parameters;

[0037] Construct an electrical distance matrix to reflect the electrical coupling strength between nodes;

[0038] By combining the voltage influence propagation path, node pairs with short electrical distances are identified as potential influence propagation channels.

[0039] Preferably, the nodes marked with a higher risk of spread include:

[0040] Extract adjacent nodes from the electrical distance matrix whose electrical distance to nodes in the voltage over-limit risk point set is less than a set threshold;

[0041] Evaluate the voltage sensitivity values ​​of the adjacent nodes and select nodes with high sensitivity;

[0042] The selected nodes are marked as locations with a high risk of spread.

[0043] Preferably, the generation of distribution network voltage safety early warning information includes:

[0044] Integrate the set of voltage over-limit risk points and the information on the locations of marked diffusion risk points;

[0045] A detailed description is generated for each risk point, including location identifier, risk level, and scope of impact;

[0046] A structured early warning report containing a set of formatted voltage over-limit risk points, information on marked propagation risk points, and a detailed description of each risk point.

[0047] Preferably, storing the power distribution network voltage safety early warning information for subsequent retrieval includes:

[0048] The early warning information is stored in a database system and an index is created for easy retrieval;

[0049] Set up an information update mechanism to automatically refresh the warning content when new data is input;

[0050] Provides an interface for the power distribution management system to access early warning information in real time.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] The method for analyzing the impact of distributed photovoltaic (PV) power on distribution network voltage in this invention has several significant advantages. In terms of data acquisition and analysis, its comprehensiveness and accuracy are greatly improved. Traditional analysis methods suffer from incomplete and non-real-time data acquisition, failing to fully consider the dynamic coupling relationship between PV output and load demand. This method, however, utilizes high-precision geographic information acquisition equipment and professional power monitoring equipment to accurately acquire the geographical coordinates of distributed PV access points, real-time output data, and load curve data and network impedance parameters of distribution network nodes. Through a specific algorithm, it comprehensively considers multiple factors to calculate the dynamic coupling coefficient, forming a comprehensive and accurate dataset of PV load coupling characteristics, providing a solid and reliable data foundation for subsequent analysis.

[0053] This method stands out in its analytical framework construction. Existing methods lack in-depth analysis of the distribution network topology hierarchy, making it difficult to specifically assess the voltage impact at different levels. This method, however, divides the typical tree-like topology of the distribution network into levels such as transformer substations and medium-voltage lines, marking the assessment sections for each and constructing a hierarchical assessment framework. This framework enables in-depth analysis of the voltage impact at different levels, making the analysis more targeted and systematic, and accurately identifying the key aspects of voltage problems.

[0054] This method also demonstrates significant advantages in tiered voltage impact analysis. Traditional methods struggle to comprehensively and accurately assess the impact of photovoltaic output fluctuations on node voltages when facing complex real-world operating scenarios and extreme conditions. This method, by collecting historical operating data sequences and extreme condition scenario data from the past year, employs time-series production simulation methods combined with mathematical models to calculate voltage deviation amplitudes, assesses voltage stability according to standard specifications, and generates detailed and accurate tiered voltage impact analysis results.

[0055] In terms of risk point identification and early warning, this method far surpasses traditional methods. Traditional methods struggle to effectively identify voltage exceedance risk points and analyze node electrical distances and voltage propagation paths. This method sets a preset risk threshold, filters out nodes whose voltage exceedance risk values ​​exceed the threshold and are located in critical hierarchical sections, forming a set of voltage exceedance risk points. Through electrical distance calculation and voltage impact propagation models, it analyzes node relationships and propagation paths, marks nodes with high diffusion risk, and generates comprehensive and accurate distribution network voltage safety early warning information.

[0056] In practical applications, this method provides power operation and maintenance personnel with clear and explicit early warning information, helping them to take timely measures and effectively ensure the stable operation of the distribution network. It can avoid power outages caused by voltage problems, reduce the risk of equipment damage, and improve the safety and reliability of power system operation. Simultaneously, this hierarchical analysis model and early warning mechanism provides strong support for the large-scale, safe integration and efficient utilization of distributed photovoltaic power in the distribution network. It helps promote the transformation of the energy structure towards a green and sustainable direction, reduces dependence on traditional energy sources, and has a positive impact on environmental protection and energy utilization. Attached Figure Description

[0057] Figure 1 This is a schematic diagram illustrating the working principle of the method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage as described in this invention.

[0058] Figure 2 A flowchart for calculating dynamic coupling coefficients;

[0059] Figure 3 A flowchart for the formation of a hierarchical and graded evaluation framework set. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Please see Figure 1This invention provides a method for analyzing the impact of distributed photovoltaic (PV) power on distribution network voltage. The method includes integrated data acquisition, feature extraction, hierarchical assessment, and risk warning modules to achieve comprehensive analysis of distribution network voltage safety. The core of the method lies in utilizing the dynamic coupling relationship between PV output and load demand, combined with the distribution network topology, to conduct multi-level voltage impact assessment. The geographical coordinates and real-time output data of the distributed PV access points are collected, and load curve data and network impedance parameters of distribution network nodes are simultaneously acquired. The PV access points are spatially mapped to distribution network nodes, and the dynamic coupling coefficient between PV output and load demand is calculated to generate a PV load coupling feature dataset. Using the PV load coupling feature dataset as basic input, the dynamic coupling coefficient and node voltage measurements are extracted, and grouped according to the distribution network topology hierarchy. Evaluation sections at the transformer substation level and medium-voltage line level are marked to form a hierarchical assessment framework set. The hierarchical assessment framework selects historical operating data sequences and extreme operating condition scenario data for each assessment section, and combines time-series production simulation methods to analyze the impact of photovoltaic output fluctuations on node voltage, calculate voltage deviation amplitude, assess voltage stability, and generate hierarchical voltage impact analysis results. These results are used to identify nodes whose voltage exceedance risk values ​​exceed preset risk thresholds and are located in critical hierarchical sections, collecting these nodes to form a voltage exceedance risk point set. This set is analyzed based on the electrical distance relationships between nodes and the voltage impact propagation path, marking nodes with high propagation risk and generating distribution network voltage safety early warning information. This information is stored for later retrieval.

[0062] Example 1: See Figure 2The method for analyzing the impact of distributed photovoltaic (PV) power on distribution network voltage involves collecting time-series data of PV power output and load demand when calculating the dynamic coupling coefficient between PV power output and load demand. The PV power output time-series data originates from real-time monitoring units of the distributed PV system, typically installed in PV inverters or data acquisition units, which record PV power generation values ​​at fixed time intervals. The load demand time-series data is obtained from smart meters or load management systems at distribution network nodes. Smart meters record user power consumption at minute or hourly frequencies, and the load data is transmitted to a central server via a communication network. The matching degree of the changing trends of the two time-series data within the same time period is obtained by calculating the Pearson correlation coefficient. The calculation of the Pearson correlation coefficient requires time alignment of the PV power output time series and the load demand time series. The alignment process uses linear interpolation to fill in any possible data gaps. The calculated correlation coefficient ranges from -1 to 1; a value closer to 1 indicates a stronger positive correlation, and a value closer to -1 indicates a stronger negative correlation. This correlation coefficient serves as the initial coupling coefficient. The initial coupling coefficient reflects the basic synchronization characteristics of photovoltaic output and load demand on a time scale, but the impact of external environmental factors and differences in load composition has not yet been considered.

[0063] Real-time meteorological data, including cloud cover and ambient temperature, are incorporated to dynamically correct the initial coupling coefficient. Cloud cover data is obtained from optical imagery of meteorological satellites or cloud cover observation reports from ground meteorological stations, expressed as a percentage. Ambient temperature data comes from standard temperature measurements at meteorological stations, in degrees Celsius. The dynamic correction process establishes a multiple linear regression model. The dependent variable of the model is the initial coupling coefficient, and the independent variables are cloud cover and ambient temperature. The regression coefficients are determined through training with historical data. The training dataset includes photovoltaic output and load data and corresponding meteorological data under different seasons and weather conditions over the past year. During the correction calculation, the current cloud cover and ambient temperature values ​​are substituted into the regression model to obtain a correction factor. The correction factor is multiplied by the initial coupling coefficient to produce a meteorologically adjusted intermediate coupling coefficient. Increased cloud cover typically leads to a decrease in photovoltaic output, and increased ambient temperature may affect the conversion efficiency of photovoltaic panels. These physical relationships are embedded in the regression model.

[0064] The load type composition ratio is adjusted by weighting residential, commercial, and industrial loads. Load type composition ratio data is extracted from the user profile table of the distribution network planning database. The load type ratio associated with each distribution network node is determined by statistically analyzing the electricity consumption characteristics of users under that node. The weighted adjustment uses a weighted average algorithm, assigning different weight factors to residential, commercial, and industrial loads. These weight factors are set based on the typical daily curve characteristics of the load type and its sensitivity to voltage changes. The weight factor for residential loads considers its morning and evening peak characteristics, the weight factor for commercial loads reflects its concentrated daytime electricity consumption pattern, and the weight factor for industrial loads is related to its continuous and stable electricity consumption characteristics. The weighted calculation combines the intermediate coupling coefficient with the load type weight factors. The weight factors are represented in vector form, and the weighting process is a vector dot product operation, ultimately outputting the dynamic coupling coefficient value. The dynamic coupling coefficient value is a dimensionless numerical value that characterizes the strength of the interaction between photovoltaic output and load demand under specific meteorological conditions and load composition. The dynamic coupling coefficient value is recorded in the corresponding field of the photovoltaic load coupling feature dataset. The photovoltaic load coupling feature dataset is stored in a table structure. Each record contains a timestamp, node identifier, photovoltaic output value, load value, initial coupling coefficient, meteorological parameters, load type ratio and final dynamic coupling coefficient value.

[0065] The acquisition of photovoltaic (PV) output time-series data is completed through data acquisition terminals deployed on the PV array. These terminals upload data using power line carrier communication or wireless communication, and the data format conforms to the IEC 61850 standard. Load demand time-series data acquisition is integrated into the existing electricity information collection system, collecting data from smart meters via concentrators. Time alignment processing ensures that the PV output time-series data and load demand time-series data have identical time resolution and starting point. For data with inconsistent frequencies, downsampling or upsampling methods are used. The Pearson correlation coefficient is calculated using a standard function from a statistical software library. Specifically, it receives normalized PV output and load demand time-series data as input parameters, calculates the ratio of the product of the covariance and standard deviation of the two series, and outputs the Pearson correlation coefficient. Data normalization is performed before calculation to eliminate dimensional differences. The training period for the multiple linear regression model is selected to include complete annual data covering all four seasons to ensure the model's generalization ability. The least squares method is used for parameter estimation during training. The load type composition ratio is updated in sync with changes to the distribution network user files, enabling dynamic maintenance. Weighting factors in the weighted average algorithm are determined through expert surveys or historical data fitting and stored in a configuration file for easy adjustment. The calculation process for the dynamic coupling coefficient is encapsulated as an independent software module. This module takes raw time-series data as input and outputs coupling coefficient values. It is integrated into a distributed computing framework, supporting parallel computation on a large number of nodes. The photovoltaic load coupling characteristic dataset is stored in a time-series database, indexed to support fast queries by time range and node identifier. An anomaly handling mechanism is implemented throughout the calculation process to detect and remove outliers in the input data, ensuring the reliability of the dynamic coupling coefficient values. The dynamic coupling coefficient values ​​serve as crucial input parameters for the subsequent hierarchical evaluation framework set, passed to the voltage impact analysis stage. Parameters in the calculation process, such as regression model coefficients and weighting factors, support online configuration and hot updates, adapting to the differentiated characteristics of distribution networks in different regions. Data flow is logged to track calculation status and diagnose problems. The computational efficiency of the dynamic coupling coefficient calculation module is improved through algorithm optimization and caching mechanisms, meeting the time requirements of real-time analysis. The storage structure of the photovoltaic load coupling characteristic dataset is designed with scalability in mind, and reserved fields are used for adding new influencing factors in the future. The accuracy of the dynamic coupling coefficient calculation is periodically verified by comparing it with actual operating data, and the verification results are used to optimize the model parameters.

[0066] Example 2: See Figure 3The method for analyzing the impact of distributed photovoltaic (PV) power on distribution network voltage divides nodes into transformer substation (RTS) and line substation levels based on the distribution network topology when forming a hierarchical evaluation framework. The distribution network topology is obtained from the power grid company's basic data platform, and the platform data follows the common information model defined by the IEC 61968 standard. The RTS level corresponds to the power supply area of ​​the low-voltage distribution transformer. The boundary of each RTS level is defined by the low-voltage outgoing switch of the distribution transformer. The node division process traverses the distribution network using a topology analysis algorithm, aggregating all user nodes connected to the low-voltage busbar of the same transformer into the same RTS level evaluation segment. The line substation level corresponds to the medium-voltage distribution line segment. The line substation level is divided by feeder. Each 10 kV or 20 kV feeder, from the substation outgoing switch to the end of the line, constitutes an independent line level evaluation segment. Nodes crossing different feeders are not merged. Evaluation parameters are defined for each level, including short-circuit capacity and impedance ratio. Short-circuit capacity is calculated using the equivalent voltage source method recommended by the International Electrotechnical Commission (IEC), requiring input of the equivalent system impedance of the power grid and transformer parameters. Impedance ratio is calculated by selecting the ratio of the impedance magnitude between the first and last nodes of the evaluation section; the impedance magnitude is extracted from the node impedance matrix. The level evaluation parameter table is generated in the form of a relational database table, with fields including level identifier, number of nodes, short-circuit capacity value, impedance ratio value, and average load factor. Each evaluation section is tagged based on the level evaluation parameter table. Tagging rules set thresholds for short-circuit capacity and impedance ratio. Sections with a short-circuit capacity below 10 MVA and an impedance ratio above 0.05 are marked as key evaluation sections, while other sections are marked as normal evaluation sections. Tagging information is written into the metadata attributes of the evaluation section for subsequent differentiated processing.

[0067] A hierarchical evaluation framework, combined with time-series production simulation, analyzes the impact of photovoltaic (PV) output fluctuations on node voltage. The time-series simulation utilizes historical PV output and load change sequences from the distribution automation system's historical database, covering the most recent three full years. Typical daily operating scenarios are extracted from historical data using clustering algorithms. Clustering features include daily maximum PV output, daily minimum load, and daily load factor, generating three typical scenarios: sunny, cloudy, and high-load. Extreme value theory employs a generalized Pareto distribution model to extract the once-in-a-century maximum PV output extreme value from historical PV output sequences and the once-in-five-year maximum load extreme value from historical load sequences. The fluctuation range of node voltage under PV output and load extreme values ​​is obtained through probabilistic power flow calculations. These calculations use point estimation to handle the randomness of PV output, with 2000 Monte Carlo simulations performed for each evaluation segment. Voltage fluctuation range is compared with safety limits, which are set according to the national standard "Power Quality - Supply Voltage Deviation". The allowable voltage deviation range for a 10 kV system is ±7% of the rated voltage. The comparison results generate a voltage deviation risk index, which is calculated as the product of the probability of exceeding the safety limit and the average deviation amplitude. The execution platform for time-series production simulation is deployed based on a cloud computing architecture, and the computing nodes use Docker containerization technology to achieve resource isolation. Historical operating data undergoes preprocessing through a data quality inspection program to remove obvious outliers and fill in reasonable missing values. Clustering of typical daily operating scenarios uses the K-means algorithm. Specifically, it starts by extracting feature variables from historical operating data, including daily maximum photovoltaic output, daily minimum load, and daily load factor. These variables are obtained from the historical database of the distribution automation system, and the data time span covers the most recent three full years. In the data preprocessing stage, the feature variables are standardized by using the Z-score method to convert each feature value into a dimensionless value with a mean of zero and a variance of one, eliminating the influence of different dimensions on the clustering results. The number of clusters is determined using the elbow rule. The threshold selection for the generalized Pareto distribution model employs the average excess function method, while parameter estimation utilizes the maximum likelihood method. Specifically, the model begins by extracting excess values ​​exceeding a threshold from historical photovoltaic (PV) output sequences. These sequences are derived from the historical database of the distribution automation system, covering the most recent three full years. The threshold is determined using the average excess function method, which calculates the average excess value corresponding to different thresholds. The average excess value is the average of the differences between observed values ​​exceeding the threshold and the threshold itself. The threshold point where the average excess value begins to stabilize is selected as the threshold for the generalized Pareto distribution model. The forward-backward algorithm for probabilistic power flow calculations has been parallelized, utilizing GPUs to accelerate matrix operations. Voltage deviation risk indicators are stored in floating-point form, with values ​​mapped to five risk levels: low risk, relatively low risk, medium risk, relatively high risk, and high risk.The output of the hierarchical assessment framework set is encapsulated in JSON format, including assessment segment identifier, timestamp, voltage deviation value, and risk level code.

[0068] The number of evaluation sections at the transformer substation level typically reaches several thousand, and the evaluation process employs a batch parallel processing mode, with each transformer substation-level evaluation section allocated an independent computing thread. The number of evaluation sections at the line level is in the hundreds, and the evaluation process utilizes a distributed in-memory computing framework. Short-circuit capacity calculation requires access to equivalent impedance data from the upstream power grid, which is obtained from the transmission network computing system via the enterprise service bus. Impedance ratio calculation involves complex number operations, using double-precision floating-point numbers to ensure calculation accuracy. Labeling rules support online configuration, and administrators can adjust threshold parameters through a web interface. The time step for time-series production simulation is set to 15 minutes, and the simulation cycle covers a full year of 8760 hours. The typical scenario generation module has a built-in scenario reduction function, using a backsliding method to reduce the initial scenario set to 10 representative scenarios. The extreme value theory analysis module includes an extreme value independence test function, using a bootstrapping method to test the independence assumption of extreme value data. The convergence criterion for probabilistic power flow calculation is set to an adjacent iteration voltage difference of less than 0.0001 per unit. The voltage deviation risk index calculation module integrates sensitivity analysis functionality, enabling the assessment of the impact of different safety limit standards on risk levels. The results visualization of the hierarchical assessment framework is achieved through WebGIS technology. Distribution area-level assessment results are overlaid on the distribution network map using color rendering, while line-level assessment results are displayed in a dynamic power flow diagram. Quality control during the assessment process includes data traceability and calculation process log recording. The calculation results for each assessment segment are marked with the data source version and calculation parameter version. The framework's performance monitoring collects CPU utilization and memory usage in real time, automatically triggering computing resource expansion when resource usage exceeds thresholds. Assessment results are stored using a time-series database cluster, with a data retention policy setting to retain original data for three years and aggregated data permanently. The hierarchical assessment framework is integrated with the distribution management system through an enterprise service bus, pushing assessment results to the voltage control application module in real time. The framework's calibration mechanism periodically compares the results with measured voltage data, triggering model parameter recalibration when the deviation exceeds 5%.

[0069] The hierarchical assessment framework's extensibility design supports adding new assessment dimensions, such as power quality levels to assess harmonic impact. The assessment parameter library employs a modular design, allowing new parameters to be added via configuration without modifying program code. The computational workflow engine supports graphical orchestration of the assessment process, allowing users to customize the logical order of assessment steps. Access control is based on role-based access management, ensuring that users from different departments can only view assessment results within their authorized scope. Version management of the hierarchical assessment framework records changes from each algorithm upgrade, supporting historical consistency comparison of assessment results.

[0070] Example 3: Analysis Method for the Impact of Distributed Photovoltaics on Distribution Network Voltage. When calculating the voltage deviation amplitude, node voltage data within the evaluation section is selected. The node voltage data originates from the real-time measurement database of the distribution automation system. Measurement devices include distribution terminal units and smart meters, with a data acquisition frequency set to one sampling point per minute. The absolute value of the deviation between the measured voltage and the rated voltage is obtained through arithmetic difference calculation. The rated voltage is determined according to the nominal voltage level of the distribution line; for a 10 kV system, the rated voltage is set to 10.5 kV, and for a 380 V system, it is set to 0.4 kV. The absolute deviation calculation module employs a sliding time window processing technique with a window width of 15 minutes. Within each window, the absolute value of the difference between the measured voltage and the rated voltage is calculated, generating a continuous time series of absolute deviation values. The distribution of the absolute deviation values ​​over the time series is characterized by statistical analysis methods, including calculating the average, maximum, and 95th percentile of the deviation within each window. The duration of the voltage deviation is extracted from the absolute deviation value series. The duration refers to the length of time during which the absolute deviation value continuously exceeds a threshold, which is set at 7% of the rated voltage according to the national voltage quality standard. The frequency of voltage deviation is calculated by counting the number of deviation events per unit time, using either a day or a week as the time unit. Frequency calculations ensure that the event interval is greater than a minimum duration threshold. The absolute value of the deviation, its duration, and frequency are combined using a weighted composite method to form the voltage deviation amplitude index. The weighted composite formula is as follows:

[0071]

[0072] in: This represents the magnitude of the voltage deviation. This represents the average absolute value of the voltage deviation within the statistical period. Indicates the rated voltage. This represents the average duration of voltage deviation events. This represents the maximum allowed duration threshold of the system. This indicates the average frequency of occurrence of voltage deviation events. This indicates the maximum allowed occurrence frequency threshold of the system. , , These represent the weighting coefficients for the absolute value of the deviation, duration, and frequency, respectively. The weighting coefficients are determined through a combined weighting method, combining qualitative analysis using the analytic hierarchy process (AHP) and objective calculation using the entropy weighting method, to satisfy... Normalization conditions. Voltage deviation amplitude index. It is a dimensionless composite index with a value range between 0 and 1. The larger the value, the more serious the voltage deviation problem.

[0073] The voltage deviation amplitude index is used to identify nodes whose voltage exceedance risk values ​​exceed a preset risk threshold and are located in critical tiers from the results of tiered voltage impact analysis. The tiered voltage impact analysis results are stored in an analysis results database table, whose structure includes node number, timestamp, voltage exceedance probability value, and risk level fields. Voltage exceedance probability data is generated through Monte Carlo simulation with 10,000 simulations. The probability value is calculated as the ratio of the number of simulations in which the voltage exceeds the safety limit to the total number of simulations. The preset risk threshold is set to 0.05 according to the power grid safety operation regulations, indicating that the upper limit of the allowable voltage exceedance probability is 5%. The voltage exceedance probability is compared with the preset risk threshold using a direct numerical comparison to filter out a list of node numbers with a voltage exceedance probability greater than 0.05. The filtered node list needs further verification to determine if it is located in a critical tier, which includes important nodes at the transformer substation level and key nodes at the line level. The criteria for identifying critical nodes are that the users supplied by the node include primary loads or the node is located at the end of the power grid. The criteria for identifying hub nodes are that the node is located in the distribution network topology connecting three or more branch lines. Node location checks are performed by querying the distribution network topology database. The node attribute table in the database contains fields for node type, connectivity, and power supply load level. Nodes that simultaneously meet the conditions of exceeding voltage limit probability and being located in a critical level section are included in the voltage limit risk point set. The set data structure uses a hash table for storage, with the node number as the key and the node details as the value, including voltage limit probability, node type, and geographical coordinates.

[0074] The voltage deviation amplitude calculation module integrates data quality control functions to detect outliers in the input node voltage data. The detection method uses the Raida criterion to remove obviously abnormal data. Specifically, starting with the input node voltage data, which comes from the real-time measurement database of the distribution automation system, the module uses distribution terminal units and smart meters, with a data acquisition frequency of one sampling point per minute. The module preprocesses the input node voltage data, calculating the arithmetic mean, which is obtained by summing all data points and dividing by the total number of data points. It also calculates the standard deviation, which reflects the dispersion of data points relative to the mean. The Raida criterion sets out anomaly detection thresholds based on the arithmetic mean and standard deviation. The threshold range is defined as from the arithmetic mean minus three times the standard deviation to the arithmetic mean plus three times the standard deviation. Each data point is compared to this threshold range; if a data point is below the lower threshold or above the upper threshold, it is marked as an outlier. The sliding time window processing uses an overlapping window technique with a window step size of 5 minutes to ensure data continuity and smoothness. The distribution analysis module outputs statistical reports, including the mean, variance, skewness, and kurtosis of the absolute values ​​of deviations. The duration extraction algorithm uses a state machine model to accurately identify the start and end times of deviation events. The frequency statistics module considers statistical needs at different time scales, supporting multi-granularity frequency analysis at hourly, daily, and weekly levels. The weighting coefficients in the weighted synthesis formula support dynamic adjustment, with the adjustment mechanism modified through the configuration interface based on the actual operation of the power grid. The voltage exceedance probability calculation uses stratified sampling technology to improve Monte Carlo simulation efficiency, employing different sampling strategies for different assessment sections. The preset risk threshold management module supports multiple threshold schemes to adapt to differentiated standards for different voltage levels and regional power grids. The node screening process optimizes the query algorithm, utilizing database indexes to accelerate the processing speed of large-scale node data. The key-level section judgment rules are implemented as a configurable rule engine, with rule conditions supporting logical operator combinations. The maintenance of the voltage exceedance risk point set uses a version control mechanism, retaining historical versions for easy traceability and comparison after each update.

[0075] The time-series data of the voltage deviation amplitude index is stored in a time-series database, which employs a columnar storage structure to improve query efficiency. The computational performance of the identification process is optimized through parallel processing technology, allowing multiple assessment sections to simultaneously identify voltage exceedance risks. The set of voltage exceedance risk points is synchronized in real-time with the distribution management system, and risk point information is immediately pushed to the monitoring screen. A quality verification step is included in the entire identification process, with verification methods including comparison with manual judgment results and cross-system data consistency checks. The voltage deviation amplitude index calculation module is periodically compared and verified with standard sources to ensure the accuracy and reliability of the calculation results. The set of voltage exceedance risk points is visualized using a heatmap, with risk levels represented by color intensity, allowing operators to intuitively understand the risk distribution across the entire network. Redundant backups are set for the node voltage data acquisition channels, automatically switching to backup channels to ensure data continuity in the event of a primary channel failure. The absolute value of the deviation is calculated using double-precision floating-point arithmetic to avoid precision loss during calculation. The selection of statistical analysis methods has undergone rigorous verification to ensure that the statistical results accurately reflect the voltage deviation characteristics. The weighting coefficients of the weighted synthesis formula are periodically recalibrated based on recent actual power grid operating data and expert evaluation opinions. The parameters of the voltage exceedance probability calculation model are updated quarterly, with updated data sourced from the latest power grid structure and equipment parameters. Adjustments to risk thresholds require approval from the power grid dispatching department, and adjustment records are maintained for future reference. The criteria for determining key-level sections are reviewed every six months and revised appropriately based on power grid development. The set of voltage exceedance risk points is stored using a distributed database to ensure data security and scalability.

[0076] The specific methods for obtaining each parameter in the formula need to be explained in detail, including the absolute value and average value of the voltage deviation. The voltage rating is calculated by summing the absolute values ​​of the deviations of all sampling points within the statistical period and dividing by the total number of sampling points. Read directly from the power grid technical parameter database. Average duration. The maximum duration threshold is obtained by calculating the arithmetic mean of the durations of all complete deviation events. According to the power grid operation procedures, it is set to occur continuously for 2 hours. Average occurrence frequency Calculated by averaging the number of daily deviation events. Maximum frequency threshold. The procedure specifies 10 times per day. Weighting coefficient. , , The determination process includes an expert scoring stage, where five experts in the power grid field are invited to score independently, and then a comprehensive weighting is applied based on the entropy value calculation of historical data. The verification of the voltage deviation amplitude index calculation results uses a cross-validation method, comparing the calculated results with actual waveform data to verify the scientific validity and practicality of the index. The update frequency of the voltage exceedance risk point set is set to 15 minutes, consistent with the power grid data acquisition cycle. The computational complexity of the risk identification algorithm has been optimized, enabling risk identification of all network nodes within 5 minutes. The software implementation of the entire embodiment adopts a modular design, including a data acquisition module, an index calculation module, a risk identification module, and a result output module. These modules communicate with each other through a standard data interface. The implementation of this embodiment requires hardware support, including a server cluster, storage array, and network equipment. The server configuration meets the requirements of large memory and high CPU for real-time computing.

[0077] Example 4: Analysis Method for the Impact of Distributed Photovoltaics on Distribution Network Voltage. When analyzing the electrical distance relationships between nodes and the voltage propagation path, this method calculates the electrical distance values ​​between nodes based on distribution network impedance parameters. These parameters are obtained from a network model database that conforms to the Common Information Model (CIM) specification in IEC 61970 and includes line resistance, reactance, transformer turns ratio, and impedance parameters. The electrical distance is calculated using a node impedance matrix method. This matrix is ​​obtained by inverting the network admittance matrix, which is constructed based on the distribution network topology and line parameters. The electrical distance is defined as a measure of the intensity of the mutual influence of voltage changes between nodes. The calculation expression uses a combination of node self-impedance and mutual impedance. Specifically, the electrical distance equals the square of the mutual impedance between node i and node j divided by the product of the self-impedance of node i and node j. The electrical distance reflects the electrical coupling strength between nodes; a smaller electrical distance indicates stronger electrical coupling and more significant voltage mutual influence. The electrical distance matrix is ​​constructed as a square matrix, with rows and columns corresponding to distribution network node numbers, and matrix elements storing the electrical distance values ​​between node pairs. The electrical distance matrix is ​​used to identify node pairs with short electrical distances by combining voltage influence propagation paths. These paths are extracted from power flow calculations, which employ a forward-backward substitution method to calculate the voltage distribution of the distribution network under typical operating conditions. Potential influence propagation channel identification sets electrical distance thresholds, which are determined based on the distribution network's voltage level and network structure characteristics. For 10 kV distribution networks, the electrical distance threshold is typically set to 0.15 per unit. Node pairs with electrical distance values ​​less than the threshold are marked as potential influence propagation channels, and this channel information is stored in a dedicated database table for future reference.

[0078] Electrical distance relationship and propagation path analysis is used to mark nodes with high propagation risk. It extracts adjacent nodes from the electrical distance matrix whose electrical distance to nodes in the voltage exceedance risk point set is less than a set threshold. The set threshold is dynamically adjusted according to voltage safety analysis requirements, taking into account changes in operating mode and risk control levels. Voltage sensitivity values ​​of adjacent nodes are evaluated using sensitivity analysis, which calculates the impact of changes in injected power on node voltage. Sensitivity values ​​are expressed as partial derivatives. Nodes with high sensitivity are selected based on sensitivity threshold comparisons. The sensitivity threshold is determined statistically from historical operating data, and nodes with sensitivity values ​​greater than 0.05 are selected as high-sensitivity nodes. The selected nodes are marked as points with high propagation risk, and this marking information is written to the node attribute field and updated synchronously with the distribution management system. The electrical distance calculation module integrates sparse matrix technology to handle large-scale distribution networks, and the node impedance matrix inversion uses an efficient solver based on the KLU algorithm. The electrical distance matrix is ​​stored using a compressed sparse row format to reduce memory usage, and matrix operations utilize the BLAS library to optimize calculation speed. Voltage impact propagation path analysis combines graph theory algorithms, using Dijkstra's algorithm to find the shortest electrical path between nodes. Potential impact propagation path identification is implemented in parallel, with electrical distance values ​​for multiple node pairs calculated and compared simultaneously. The sensitivity analysis module employs the adjoint network method, constructing an adjoint network to calculate the partial derivatives of node voltage with respect to injected power. Threshold management supports a graphical configuration interface, allowing operators to adjust electrical distance and sensitivity thresholds as needed. The marking process is automated, with node marking information updated in real-time to the distribution geographic information system. The dimension of the electrical distance matrix is ​​consistent with the number of nodes in the distribution network; large distribution networks with tens of thousands of nodes require special processing for matrix storage. Visualization of voltage impact propagation paths uses dynamic coloring technology, displaying paths with different electrical distance values ​​in different colors on the network diagram. Sensitivity calculation considers the special properties of distributed photovoltaic (PV) access points, using a negative load model for PV nodes. The diffusion risk marking is linked to the voltage safety early warning system, with marked nodes triggering the generation of early warning information.

[0079] Referring to Table 1, accurate network parameters are required for calculating electrical distance values; parameter inaccuracies are corrected using state estimation techniques. The electrical distance matrix is ​​updated synchronously with network topology changes, triggering matrix recalculation upon topology changes. Voltage impact propagation path analysis considers multiple operating modes, including maximum and minimum load modes and maximum and minimum photovoltaic output modes. The potential impact propagation channel database is indexed to optimize query efficiency, supporting rapid retrieval by node number and electrical distance range. Sensitivity analysis results are verified by comparing with measured data to ensure the accuracy of sensitivity values. Tagging information management employs version control, recording the operation time and operator for each tagging operation. The electrical distance matrix calculation module is deployed on high-performance computing nodes to ensure the timeliness of large-scale matrix operations. Voltage impact propagation path analysis results are output in a standardized format, supporting data exchange with other analysis software. The potential impact propagation channel identification algorithm is optimized, processing networks with tens of thousands of nodes within 5 minutes. The sensitivity analysis module outputs a node voltage sensitivity matrix, which is used for subsequent voltage control strategy formulation. The diffusion risk tagging system is integrated with the distribution automation system, feeding tagging node information down to field terminal equipment. The electrical distance calculation cycle is set to 15 minutes, consistent with the SCADA system data refresh cycle. The voltage impact propagation path analysis report includes path electrical parameters and an assessment of the degree of impact. The potential impact propagation channel database is backed up regularly, employing a combination of full and incremental backups. Sensitivity analysis model parameters are verified and updated quarterly to adapt to changes in the power grid structure.

[0080] Table 1: Correlation between nodal electrical distance and sensitivity

[0081]

[0082] The node electrical distance and sensitivity correlation table is stored in a relational database system. The table fields include node number, associated risk node, electrical distance value, voltage sensitivity value, risk level, and marking status. Electrical distance values ​​are stored in per-unit format, with a base value of 100 MVA. The voltage sensitivity value represents the change in node voltage when 1 MVA of power is injected. The risk level is comprehensively assessed based on both electrical distance and voltage sensitivity values, using a fuzzy logic method. Specifically, starting with input variable processing, electrical distance and voltage sensitivity values ​​are used as precise input variables. Electrical distance values ​​are stored in per-unit format, with the numerical range calculated based on distribution network impedance parameters. Voltage sensitivity values ​​represent the change in node voltage when 1 MVA of power is injected, with units of per-unit value per MVA. The marking status field records whether a node has been marked as a propagation risk point, and the marking operation is logged. The electrical distance calculation process includes a data verification step to verify the completeness and rationality of network parameters. The electrical distance matrix construction algorithm uses incremental update technology, recalculating only the topology-changed parts. Voltage impact propagation path analysis considers the unique influence of distributed photovoltaic (PV) grid connection points, with PV nodes treated as either PV nodes or PQ nodes depending on their operating status. Potential impact propagation channel identification results are displayed through a visual interface, allowing operators to interactively view detailed channel information. The sensitivity analysis module supports multi-scenario calculations, including normal operating conditions and N-1 fault conditions.

[0083] The marking system implements access control, with different levels of users having different marking operation permissions. The accuracy of electrical distance value calculation is guaranteed through double-precision floating-point arithmetic. Voltage impact propagation path analysis report generation is automated, and the report format conforms to power system analysis standards. A relational index is established in the potential impact propagation channel database to accelerate multi-table join queries. Sensitivity analysis results are verified by comparing simulation data with measured data. Propagation risk marking information is associated with voltage control strategies, and marked nodes trigger automatic voltage control commands. Electrical distance matrix calculation resources are dynamically allocated, automatically expanding calculation nodes when the computational load is high. Voltage impact propagation path analysis considers electromagnetic ring network conditions, and special network topologies are handled using dedicated algorithms. The accuracy of potential impact propagation channel identification is verified by comparing with actual fault waveform data. The linearity test of the sensitivity analysis model ensures the accuracy of small disturbance analysis. The marking information statistics function provides statistical reports on the distribution of risk points. The convergence of electrical distance value calculation is guaranteed through iterative algorithm residual control. Voltage impact propagation path analysis considers the impact of transformer tap changes. The access interface for the potential impact propagation channel database is standardized, supporting multi-system data sharing. The computational efficiency of the sensitivity analysis module is optimized using sparse matrix technology. The electrical distance calculation module has undergone unit testing and integration testing, and the test coverage meets software engineering standards. The voltage impact propagation path analysis function has been verified with actual power grid data, and the analysis results are consistent with operational experience. The potential impact propagation channel identification algorithm has applied for software copyright protection. The sensitivity analysis module's calculation results have been applied to multiple distribution network voltage control projects. The diffusion risk marking system has been put into actual operation, and the marking accuracy has been evaluated through operational data statistics.

[0084] Example 5: The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage integrates a set of voltage exceedance risk points and marked diffusion risk point information when generating distribution network voltage safety early warning information. The set of voltage exceedance risk points is derived from the risk identification results. The marked diffusion risk point information is obtained from the diffusion analysis module, including nodes N101201 and N101203, which have high diffusion risk, along with their electrical distance values ​​of 0.08 pu and 0.11 pu, and voltage sensitivity values ​​of 0.12 pu / MW and 0.09 pu / MW. The integration process employs data fusion technology, using node numbers as the primary key for table join operations. Fields from the voltage exceedance risk point set and the diffusion risk point fields are merged to generate a complete record. The merged data table includes node number, voltage exceedance probability, electrical distance value, voltage sensitivity value, and risk level.

[0085] The generation of detailed descriptions for each risk point is performed by the early warning information generation engine. These descriptions include three core elements: location identifier, risk level, and impact range. The location identifier combines the distribution network node number with geographical coordinates. For example, node N101205 corresponds to latitude and longitude coordinates (30.6585°N, 104.0653°E) and is associated with the specific transformer substation name "Xingfu Community No. 3 Transformer". The risk level is determined by a weighted score of voltage exceedance probability and voltage sensitivity, with five levels: Level 1 (low risk), Level 2 (relatively low risk), Level 3 (medium risk), Level 4 (relatively high risk), and Level 5 (high risk). The impact range is determined using a topology analysis algorithm, searching outwards from the risk point for all adjacent nodes with an electrical distance of less than 0.1 pu to form the impact area. For example, the impact range of node N101205 includes N101201 and N101203. The formatted structured early warning report uses an XML architecture standard, with the report structure consisting of a header, warning body, and footer. The message header includes the warning number, generation time, data version, and metadata information. The warning body displays detailed descriptions of all risk points in a list format. The message footer includes a checksum and signature information. The structured warning report uses JAXB technology to bind and convert objects to XML documents, ensuring the document structure conforms to the IEEE CIM standard. Report files are named with timestamps and stored in a designated directory in the warning database. The directory structure is organized hierarchically by year, month, and day for easy retrieval.

[0086] The power distribution network voltage safety early warning information is stored in a database system using a MySQL cluster architecture. The database tables include a main table for early warning information, a detailed table for risk points, and a table relating the affected areas. The main table stores the early warning number and generation time, the detailed table records the specific parameters of each risk point, and the table relating nodes to their affected areas. A B+ tree structure is used for indexing, with composite indexes created on fields such as node number and generation time to improve query performance. The information update mechanism is implemented through database triggers. When the set of voltage exceedance risk points or the information on spreading risk points changes, the trigger automatically executes a stored procedure to regenerate the early warning information. The interface supports both GET and POST HTTP methods. The GET method is used to query early warning information, and the POST method is used to submit early warning confirmation commands. Interface authentication uses the OAuth 2.0 protocol; the power distribution management system must obtain an access token before calling the interface. Real-time access response time is controlled within 200 milliseconds, and the interface returns data in JSON format, including status codes, message bodies, and pagination information. A specific example demonstrates the early warning information processing flow for node N101205. The voltage over-limit risk point set shows that the probability of voltage over-limit for this node is 0.08. The marked diffusion risk point information shows that the electrical distance value of this node is 0.05 pu and the voltage sensitivity value is 0.15 pu / MW. After integration, a complete record containing all field data is generated. The generation process is described in detail. The geographic information system is called to obtain the node location "Xingfu Community No. 3 Transformer". The risk level calculation score is 72 points, corresponding to a relatively high risk level of level four. The impact range analysis finds N101201. The structured early warning report generates an XML document fragment containing the node number, latitude and longitude coordinates, risk level, and a list of affected nodes. Database storage operations insert a new record into the main early warning information table and insert detailed data of 12 fields into the risk point details table. The interface access demonstration uses the curl command to send a GET request, and the returned JSON data contains all the early warning information for node N101205.

[0087] The early warning information generation engine is deployed with load balancing, using a cluster of three servers to process early warning generation requests. Location identification data in the detailed description is synchronously updated from the equipment management module of the power distribution automation system to ensure the accuracy of location information. Risk level scoring rules support online configuration; operators can adjust scoring thresholds and weight parameters through the management interface. The impact range analysis algorithm is re-executed every 15 minutes to dynamically track network topology changes. The digital signature of the structured early warning report uses the RSA algorithm, and the private key is centrally managed by a security certificate manager. The database system's backup strategy combines full and incremental backups, performing a full backup at 2 AM daily and incremental backups hourly. Index optimization involves periodic rebuilding operations, with maintenance windows set during off-peak business hours. The information update mechanism's trigger conditions support multi-level configuration, allowing updates to be triggered when the voltage exceedance probability changes by more than 0.01 or when the topology changes. Interface access logs record detailed information for each request, and log files are stored for 6 months for audit analysis.

[0088] When the power distribution management system calls the interface, an Authorization field must be added to the request header to carry an access token. The interface response includes HTTP status codes: 200 for success, 400 for request error, and 500 for server error. Real-time access supports filtered queries, which can specify conditions such as time range, risk level, and area range through parameters. The default page size for the interface returned data is 100 records, and custom page settings are supported for the page and size parameters. After the warning information for node N101205 is generated, the database trigger will automatically send a notification to the message middleware of the power distribution management system. The warning information display interface uses the WebSocket protocol to achieve real-time push, and actively refreshes the interface data when a new warning is generated. The confirmation process for risk points requires operators to fill in their handling opinions, and the confirmation information is written back to the confirmation status field of the database. The archiving strategy for historical warning information is set to retain data for 3 years, and expired data is transferred to an offline storage system. The warning information generation module has undergone stress testing and supports processing 1000 warning generation requests per second. The database cluster adopts a master-slave replication architecture, with the master database responsible for write operations and the slave database responsible for query operations, achieving read-write separation. The interface service implements rate limiting and circuit breaking through the gateway to ensure system stability in high-concurrency scenarios. The security mechanism employs multi-layered protection, including a network firewall, application firewall, and encrypted data transmission.

[0089] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0090] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage, characterized in that, The method includes: Collect the geographical coordinates and real-time power output data of distributed photovoltaic access points, simultaneously acquire the load curve data and network impedance parameters of distribution network nodes, associate the photovoltaic access points and calculate the dynamic coupling coefficient between photovoltaic power output and load demand, and generate a photovoltaic load coupling characteristic dataset. Based on the photovoltaic load coupling feature dataset, dynamic coupling coefficients and node voltage measurements are extracted, and the data are grouped according to the topological hierarchy of the distribution network. The evaluation sections at the transformer substation level and the evaluation sections at the medium-voltage line level are marked to form a hierarchical evaluation framework set. For each evaluation segment in the hierarchical evaluation framework set, historical operating data sequences and extreme operating condition scenario data are selected. The impact of photovoltaic power output fluctuations on node voltage is analyzed by combining time-series production simulation methods, the voltage deviation amplitude is calculated, the voltage stability is evaluated, and hierarchical voltage impact analysis results are generated. From the results of the hierarchical voltage impact analysis, identify nodes whose voltage exceedance risk values ​​exceed a preset risk threshold and are located in key hierarchical sections, and collect these nodes to form a set of voltage exceedance risk points; Based on the set of voltage over-limit risk points, the electrical distance relationship between nodes and the voltage impact propagation path are analyzed, nodes with high propagation risk are marked, and distribution network voltage safety early warning information is generated. The voltage safety early warning information of the power distribution network is stored for later retrieval.

2. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 1, characterized in that, The process of associating photovoltaic (PV) access points and calculating the dynamic coupling coefficient between PV output and load demand includes: Collect photovoltaic power output time series data and load demand time series data, and obtain the initial coupling coefficient by calculating the matching degree of the changing trends of the two series within the same time period; Real-time meteorological data, including cloud cover and ambient temperature, are used to dynamically correct the initial coupling coefficient. At the same time, a weighted adjustment is made based on the composition ratio of load types to obtain the dynamic coupling coefficient value. The load types include residential load, commercial load, and industrial load.

3. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 2, characterized in that, The set of hierarchical and graded evaluation frameworks includes: Based on the distribution network topology, the nodes are divided into transformer substation level and line level. The transformer substation level corresponds to the power supply area of ​​the low-voltage distribution transformer, and the line level corresponds to the medium-voltage distribution line section. Define evaluation parameters for each level and generate a level evaluation parameter table, where the evaluation parameters include short-circuit capacity and impedance ratio; Based on the hierarchical evaluation parameter table, each evaluation segment is labeled to distinguish between normal evaluation segments and key evaluation segments.

4. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 3, characterized in that, The analysis of the impact of photovoltaic power output fluctuations on node voltage using the combined time-series production simulation method includes: By calling up the photovoltaic output and load change sequences from historical operating data, a typical daily operating scenario is simulated; Extreme value theory is used to extract the extreme values ​​of photovoltaic power output and load under extreme conditions, and the fluctuation range of node voltage under the extreme values ​​of photovoltaic power output and load is calculated. The risk of voltage deviation caused by photovoltaic fluctuations is assessed by comparing the voltage fluctuation range with the safety limit.

5. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 4, characterized in that, The calculated voltage deviation amplitude includes: Select node voltage data within the evaluation section and calculate the absolute value of the deviation between the measured voltage value and the rated voltage value; The distribution of the absolute value of the statistical deviation over time series is used to determine the duration and frequency of voltage deviation. By combining the absolute value of the deviation, its duration, and its frequency, a voltage deviation amplitude index is synthesized.

6. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 5, characterized in that, The nodes identified from the hierarchical voltage impact analysis results as having voltage exceedance risk values ​​exceeding a preset risk threshold and located in critical hierarchical sections include: Obtain voltage limit exceedance probability data from the hierarchical voltage impact analysis results; The probability of voltage exceeding the limit is compared with a preset risk threshold to filter out nodes that exceed the threshold; Check whether nodes exceeding the threshold are located in critical level sections. Once confirmed, they are included in the voltage over-limit risk point set. Critical level sections include important nodes at the transformer substation level or hub nodes at the line level.

7. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 6, characterized in that, The electrical distance relationships and voltage propagation paths between the analysis nodes include: Calculate the electrical distance between nodes based on the distribution network impedance parameters; Construct an electrical distance matrix to reflect the electrical coupling strength between nodes; By combining the voltage influence propagation path, node pairs with short electrical distances are identified as potential influence propagation channels.

8. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 7, characterized in that, The nodes marked as having a high risk of spread include: Extract adjacent nodes from the electrical distance matrix whose electrical distance to nodes in the voltage over-limit risk point set is less than a set threshold; Evaluate the voltage sensitivity values ​​of the adjacent nodes and select nodes with high sensitivity; The selected nodes are marked as locations with a high risk of spread.

9. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 8, characterized in that, The generated distribution network voltage safety early warning information includes: Integrate the set of voltage over-limit risk points and the information on the locations of marked diffusion risk points; Generate a detailed description for each risk point, including location identifier, risk level, and scope of impact; A structured early warning report containing a set of formatted voltage over-limit risk points, information on marked propagation risk points, and a detailed description of each risk point.

10. The method for analyzing the impact of distributed photovoltaic power generation on distribution network voltage according to claim 9, characterized in that, The storage of the power distribution network voltage safety early warning information for subsequent retrieval includes: The early warning information is stored in a database system and an index is created for easy retrieval; Set up an information update mechanism to automatically refresh the warning content when new data is input; Provides an interface for the power distribution management system to access early warning information in real time.

Images