Network-configuration-type SVG dynamic capacity sizing and site selection method based on load fluctuation

Abstract

The present application relates to the technical field of power grid reactive power compensation, and discloses a network type SVG dynamic capacity and site selection method based on load fluctuation. The method collects real-time load data and historical reactive power demand data in the power grid system, forms a load fluctuation characteristic sequence and a reactive power compensation demand characteristic sequence; performs multi-time scale correlation analysis on the above two sequences, identifies the dynamic coupling mode between load fluctuation and reactive power demand, and generates a reactive power compensation priority evaluation result; according to the load fluctuation characteristic sequence and the reactive power compensation priority evaluation result, a dynamic capacity optimization model of the network type SVG is established, and the model parameters are adaptively adjusted; the real-time collected power grid operation data is matched and analyzed with the dynamic capacity optimization model, the potential installation position and capacity configuration range of the network type SVG are determined, and a preliminary site selection and capacity configuration scheme is generated; based on the preliminary scheme, a comprehensive evaluation of technical economy is performed, and the optimal network type SVG configuration scheme is selected, which can better adapt to the dynamic operation demand of the power grid.

Description

Technical Field

[0001] This invention relates to the field of reactive power compensation technology for power grids, specifically to a dynamic varistor location method for grid-based SVG based on load fluctuations. Background Technology

[0002] As the power system continues to expand, the number and types of electrical equipment in the power grid are constantly increasing, leading to significant fluctuations in the power grid load. Whether it's the start-up and shutdown of large motors in industrial production, seasonal peak electricity consumption changes, or output fluctuations caused by the connection of new energy power generation equipment, all these factors cause the real-time load of the power grid to be in a dynamic state, thereby leading to instability in the reactive power demand of the power grid.

[0003] Traditional reactive power compensation schemes for power grids often employ fixed-capacity reactive power compensation equipment. The capacity and installation location of this equipment are difficult to adjust based on load fluctuations after it is put into operation. When the grid load fluctuates significantly, the fixed-capacity reactive power compensation equipment cannot promptly match the changes in reactive power demand, potentially leading to voltage deviations exceeding permissible limits and impacting power supply quality. Furthermore, traditional site selection methods rely heavily on experience or static data, failing to fully consider the dynamic correlation between load fluctuations and reactive power demand. Determining the installation location of reactive power compensation equipment solely based on historical average data can easily result in a mismatch between the compensation location and actual high-demand nodes, leading to reactive power surplus in some areas and reactive power shortage in others.

[0004] Traditional methods for developing reactive power compensation schemes often fail to conduct in-depth multi-timescale analysis of load fluctuation characteristics and reactive power compensation demand characteristics. This makes it difficult to accurately identify the coupling relationship between load fluctuations and reactive power demand across different time dimensions, resulting in compensation schemes lacking specificity. In the capacity determination stage, traditional models often use fixed parameters, lacking the ability to adaptively adjust based on real-time grid operation data. When the grid's operating state changes, the model's output capacity configuration deviates significantly from actual demand, failing to meet the needs of dynamic reactive power compensation. Furthermore, after forming a preliminary site selection and capacity determination scheme, traditional methods do not comprehensively consider technical and economic factors, focusing only on a single indicator and failing to coordinate equipment costs, operation and maintenance conditions, compensation effects, response performance, and other factors. This makes it difficult for the final compensation scheme to balance technical rationality and economic feasibility in practical applications, failing to fully leverage the supporting role of reactive power compensation equipment in grid stability. Summary of the Invention

[0005] The purpose of this invention is to provide a dynamic dimensionality-based addressing method for network-type SVG based on load fluctuations, so as to solve the problems mentioned in the background art.

[0006] To achieve the above object, the application provides a network configuration type SVG dynamic capacity and site selection method based on load fluctuation, which comprises the following steps:

[0007] Real-time load data and historical reactive power demand data in the power grid system are collected to form load fluctuation characteristic sequences and reactive power compensation demand characteristic sequences;

[0008] Multi-time scale correlation analysis is performed on the load fluctuation characteristic sequences and the reactive power compensation demand characteristic sequences to identify the dynamic coupling mode between the load fluctuation and the reactive power demand, and to generate a reactive power compensation priority evaluation result;

[0009] According to the load fluctuation characteristic sequences and the reactive power compensation priority evaluation result, a dynamic capacity optimization model of the network configuration type SVG is established, and the model parameters are adaptively adjusted;

[0010] Real-time collected power grid operation data are matched and analyzed with the dynamic capacity optimization model to determine the potential installation location and capacity configuration range of the network configuration type SVG, and a preliminary site selection and capacity configuration scheme is generated;

[0011] Based on the preliminary site selection and capacity configuration scheme, a comprehensive technical and economic evaluation is performed to select the optimal network configuration type SVG configuration scheme.

[0012] Preferably, the real-time load data and historical reactive power demand data in the power grid system are collected to form load fluctuation characteristic sequences and reactive power compensation demand characteristic sequences, which comprises the following steps:

[0013] The load change rate and reactive power fluctuation of the power grid node are monitored to extract load time sequences and reactive time sequences;

[0014] The load time sequences and the reactive time sequences are standardized to eliminate dimensional differences, and standardized load fluctuation characteristic sequences and reactive power compensation demand characteristic sequences are generated;

[0015] The standardized sequences are integrated to form a multi-dimensional data set for subsequent analysis.

[0016] Preferably, the multi-time scale correlation analysis is performed on the load fluctuation characteristic sequences and the reactive power compensation demand characteristic sequences to identify the dynamic coupling mode between the load fluctuation and the reactive power demand, and to generate a reactive power compensation priority evaluation result, which comprises the following steps:

[0017] The load fluctuation characteristic sequences and the reactive power compensation demand characteristic sequences are decomposed into short-term, medium-term and long-term components according to the time scale;

[0018] The correlation coefficient and coupling strength between the load fluctuation and the reactive power demand under different time scales are calculated to identify the significant coupling mode;

[0019] According to the coupling strength and the grid operation state, the reactive power compensation priority is allocated, and a reactive power compensation priority evaluation result is generated.

[0020] Preferably, according to the load fluctuation characteristic sequence and the reactive power compensation priority evaluation result, a dynamic capacity optimization model of the grid-forming SVG is established, and the model parameters are adaptively adjusted, including:

[0021] Using the load fluctuation characteristic sequence as an input variable and the reactive power compensation priority evaluation result as a constraint condition, a capacity optimization objective function is constructed;

[0022] An iterative algorithm is used to solve the objective function to obtain preliminary capacity configuration parameters;

[0023] According to real-time grid data, the objective function and the constraint condition are dynamically updated, and the model parameters are adaptively adjusted.

[0024] Preferably, the real-time collected grid operation data are matched and analyzed with the dynamic capacity optimization model, potential installation positions and capacity configuration ranges of the grid-forming SVG are determined, and a preliminary site selection and capacity determination scheme is generated, including:

[0025] The grid topology structure and node impedance information are extracted, and a reactive power compensation sensitivity matrix is calculated;

[0026] In combination with the output of the dynamic capacity optimization model, the reactive power compensation effect and capacity demand of each node are evaluated;

[0027] The nodes with high reactive power compensation sensitivity and large capacity demand are screened as potential installation positions, and the capacity configuration ranges are determined to generate the preliminary site selection and capacity determination scheme.

[0028] Preferably, based on the preliminary site selection and capacity determination scheme, a comprehensive technical and economic evaluation is performed to select an optimal grid-forming SVG configuration scheme, including:

[0029] The device cost, operation and maintenance cost, and life cycle data of the grid-forming SVG are collected;

[0030] Technical indexes of different configuration schemes are calculated, including compensation accuracy, response time, and reliability;

[0031] In combination with the economic indexes and the technical indexes, a weighted scoring method is used for comprehensive evaluation to select the optimal configuration scheme.

[0032] Preferably, the method further includes:

[0033] The operation data of the grid-forming energy storage device are integrated, and the dynamic capacity optimization model is adjusted to consider the synergy of the energy storage and the SVG;

[0034] According to the energy storage charging and discharging characteristics and the reactive power compensation demand, the capacity configuration and site selection decision are optimized.

[0035] Preferably, the method further comprises:

[0036] Performing technical and economic comparison and analysis of the network construction type equipment and the phase modifier, including technical advancement and economic evaluation;

[0037] Building a comprehensive evaluation index system of technical and economic efficiency for optimizing the configuration scheme selection.

[0038] Preferably, the method further comprises:

[0039] Designing a multi-time scale AVC coordination control strategy of multiple reactive power sources, and analyzing the dynamic reactive power regulation characteristics of different reactive power sources;

[0040] Realizing optimal allocation of reactive power dispatching resources and ensuring voltage stability of the power grid.

[0041] Preferably, the method further comprises:

[0042] Real-time monitoring of power grid load fluctuation and reactive power demand change, dynamic updating of reactive power compensation priority evaluation results and capacity optimization model;

[0043] Adjusting the site selection and capacity determination scheme according to the updated results to ensure adaptation to the power grid operation conditions.

[0044] Compared with the prior art, the method has the advantages that:

[0045] The network construction type SVG dynamic capacity determination and site selection method based on load fluctuation can comprehensively obtain dynamic information of power grid load and reactive power demand by collecting real-time load data and historical reactive power demand data in the power grid system, breaking the limitation of traditional methods which only rely on static data or partial fragment data, providing a more comprehensive and accurate data basis for subsequent analysis, and accurately reflecting the actual change law of power grid load and reactive power demand.

[0046] The multi-time scale correlation analysis of the load fluctuation characteristic sequence and the reactive power compensation demand characteristic sequence can identify the dynamic coupling mode between load fluctuation and reactive power demand and generate reactive power compensation priority evaluation results, which can deeply mine the internal relationship between the two from different time dimensions, clearly master the influence law of load fluctuation on reactive power demand under different time scales such as short-term, medium-term and long-term, and then clearly determine the reactive power compensation priority of different regions and different time periods, so that the subsequent capacity optimization and site selection decision making is more targeted, and the problem of unreasonable allocation of compensation resources caused by lack of understanding of dynamic coupling relationship in traditional methods is avoided.

[0047] A dynamic capacity optimization model of grid-connected SVG is established according to the load fluctuation characteristic sequence and the reactive power compensation priority evaluation result, and the model parameters are adaptively adjusted, so that the model can closely fit the actual operation state of the power grid. The model takes the load fluctuation characteristics as the input and the reactive power compensation priority as the constraint, and the capacity optimization objective function constructed is more in line with the actual demand. At the same time, the initial capacity configuration parameters are solved by an iterative algorithm, and the target function and the constraint condition are dynamically updated according to the real-time power grid data, so that the model parameters always keep synchronization with the changes of the power grid operation, avoiding the inadaptation problem of the traditional fixed parameter model when the state of the power grid changes, and improving the accuracy and timeliness of the capacity configuration result.

[0048] The real-time collected power grid operation data are matched and analyzed with the dynamic capacity optimization model to determine the potential installation location and capacity configuration range of the grid-connected SVG and generate a preliminary site selection and capacity determination scheme. The reactive power compensation sensitivity matrix is calculated by extracting the power grid topology and node impedance information, and the compensation effect and capacity demand of each node are evaluated in combination with the model output, so that the nodes with high reactive power compensation sensitivity and large capacity demand can be accurately screened as the potential installation location, ensuring that the installation location is highly consistent with the high demand area of the power grid, and the reasonable capacity configuration range is determined, providing a scientific preliminary basis for subsequent scheme optimization and avoiding the blindness of traditional site selection methods.

[0049] Based on the technical and economic comprehensive evaluation of the preliminary site selection and capacity determination scheme, the optimal grid-connected SVG configuration scheme is selected. By collecting equipment cost, operation and maintenance cost and life cycle data, calculating technical indicators such as compensation accuracy, response time and reliability, and then combining with the weighted scoring method for comprehensive evaluation, the technical performance and economic cost of the scheme can be considered as a whole, ensuring that the scheme meets the demand of power grid reactive power compensation in the technical aspect, and ensuring that the scheme is reasonable in the economic aspect, avoiding the overall imbalance problem caused by the traditional scheme which focuses on only one aspect of technology or economy, so that the finally determined configuration scheme is more feasible and practical in practical application. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 A working principle diagram of the load fluctuation-based grid-connected SVG dynamic capacity selection and site selection method described in the present application;

[0051] Figure 2 A working principle diagram for forming a load fluctuation characteristic sequence and a reactive power compensation demand characteristic sequence;

[0052] Figure 3 A working principle diagram for load fluctuation and reactive power demand multi-time scale correlation analysis and reactive power compensation priority evaluation. DETAILED DESCRIPTION

[0053] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0054] With reference to Figure 1 The present application provides a network configuration type SVG dynamic capacity selection method based on load fluctuation, which comprises collecting real-time load data and historical reactive power demand data in the power grid system, forming load fluctuation characteristic sequence and reactive power compensation demand characteristic sequence, and realizing data standardization by monitoring the operating parameters of the power grid nodes. The load fluctuation characteristic sequence and the reactive power compensation demand characteristic sequence are subjected to multi-time scale correlation analysis to identify the dynamic coupling mode between load fluctuation and reactive power demand, and to generate a reactive power compensation priority evaluation result. The analysis involves time series decomposition and correlation calculation. According to the load fluctuation characteristic sequence and the reactive power compensation priority evaluation result, a dynamic capacity optimization model of the network configuration type SVG is established, and the model parameters are adaptively adjusted. The model construction uses an objective function and an iterative algorithm. Then, the real-time collected power grid operation data are matched and analyzed with the dynamic capacity optimization model to determine the potential installation location and capacity configuration range of the network configuration type SVG, and to generate a preliminary site selection and capacity selection scheme, which is combined with the power grid topology and sensitivity calculation. Based on the preliminary site selection and capacity selection scheme, a comprehensive technical and economic evaluation is performed to select the optimal network configuration type SVG configuration scheme, and the evaluation covers equipment cost and technical indicators.

[0055] Embodiment 1: With reference to Figure 2 The collection and processing of real-time load data and historical reactive power demand data in the power grid system rely on monitoring equipment deployed at each node of the power grid, such as smart meters, power sensors, and data acquisition and monitoring systems. These devices continuously record the active power, reactive power, voltage, and current of the node at a preset time interval. The construction of the load time sequence focuses on capturing the change trend and instantaneous fluctuation of the load. The load change rate is quantified by calculating the difference between the load values of adjacent time points, thereby forming an original sequence reflecting the dynamic characteristics of the load. The reactive time sequence is directly derived from the monitored reactive power readings, and is combined with fault records, planned maintenance events, and typical daily operation curves in the historical database to form a data set that can represent the change law of the system reactive power demand.

[0056] The standardization of the original load time series and reactive power time series is a crucial step, which aims to eliminate the dimensional differences caused by different measurement units, so that the subsequent correlation analysis can be carried out on the same basis. The standardization process usually adopts the minimum-maximum scaling method to linearly transform the original data into the [0, 1] interval, or adopts the Z-score transformation method to convert the data into a distribution with a mean of 0 and a standard deviation of 1. In the processing process, it is also necessary to integrate the outlier detection algorithm, such as the quartile range based method, to identify and eliminate outliers caused by measurement noise or transmission errors; for the inevitable missing values, linear interpolation method or time series trend based prediction method is used for filling, to ensure the continuity and integrity of the data sequence. After these steps, the original data is transformed into standardized load fluctuation feature sequence and reactive power compensation demand feature sequence, each data point is associated with the corresponding time stamp and node identifier.

[0057] Integrating the standardized sequences to form a multi-dimensional data set requires designing a structured data model. The data model usually includes time dimension, space dimension and feature dimension, the time dimension is accurate to milliseconds or seconds to match the dynamic process of the power system, the space dimension clearly records the transformer station, line or specific node number to which the data belongs, and the feature dimension stores the standardized load fluctuation value and reactive power demand value. This multi-dimensional data set can be organized into table form in time series database, or stored using data cube-like structure, to facilitate efficient multi-dimensional query and aggregation analysis. The data integration process also includes aligning and fusing information from different data sources, ensuring that the time stamps are synchronized and the spatial positions are consistent.

[0058] The refinement of the load fluctuation feature sequence is not limited to basic standardization processing, but also needs to further extract features that can deeply reflect the dynamic behavior of the load. In addition to the load change rate, the load fluctuation amplitude, the fluctuation duration, the daily average load curve shape factor and other characteristics can be calculated, which together constitute a multi-angle description of the load fluctuation characteristics. Calculating these features usually needs to be done in a sliding time window, and the size of the window is dynamically adjusted according to the time scale of interest, for example, a short window is used for second-level fluctuation analysis, and a long window is used for trend analysis in hours. The feature extraction process may involve digital signal processing techniques, such as filtering to separate fundamental and harmonic components, or Fourier transform to analyze the frequency domain characteristics of load fluctuations, thereby more fully revealing the nature of load behavior. The generation of the reactive power compensation demand feature sequence also requires in-depth analysis, which cannot simply be a record of historical reactive power. The construction of this sequence needs to consider the actual operating state of the power grid, such as events such as node voltage deviation exceeding the allowed range, line reactive power flow exceeding the limit, etc. as important markers of high reactive power demand. By analyzing the change pattern of reactive power before and after these events in historical data, a correlation model between reactive power demand and system faults or operating constraints can be established. Power factor is also an important reference index, nodes that run at low power factor for a long time usually have stronger reactive power compensation demand, and this information is integrated into the reactive power compensation demand feature sequence, making it more accurately reflect the true demand of the system.

[0059] The multi-dimensional data set formed ultimately needs to be converted into a format suitable for subsequent machine learning or optimization algorithms to read. The common practice is to export the data as a structured text file, such as CSV format, or use a binary format such as HDF5 to support faster read-write speed and larger data volume. Before exporting, data partitioning is usually needed, such as dividing by time range or power grid region, to facilitate parallel processing by distributed computing frameworks. Metadata is also included in the data set to describe the source of the data, processing history, quality identification, and other information to ensure traceability of the data throughout the analysis process. This carefully prepared multi-dimensional data set.

[0060] Embodiment 2: see Figure 3The core work of this stage is the multi-time scale correlation analysis based on the dynamic relationship between load fluctuation and reactive power demand. The standardized load fluctuation characteristic sequence and reactive power compensation demand characteristic sequence are decomposed according to different time granularities. The division of time scales is based on the periodic characteristics and control requirements of power grid operation. Generally, short-term components are defined as minute to hour level, medium-term components are defined as hour to day level, and long-term components are defined as day to month level. The decomposition process is realized by digital signal processing technology, such as wavelet transform. By selecting appropriate wavelet basis function and decomposition level, the high-frequency details, low-frequency profiles and trend components in the original signal are separated, which correspond to short-term, medium-term and long-term fluctuation components respectively. This decomposition makes the different rhythm change rules hidden in the original data visible.

[0061] To quantify the degree of cooperative change between sequences at different time scales, Pearson correlation coefficient is used to measure the strength of linear relationship between short-term components, while mutual information and other indicators that can capture nonlinear correlation are introduced to calculate the coupling strength of medium-term and long-term components. For each time scale, the calculation is recursively performed within a sliding time window, resulting in a sequence of correlation coefficients that changes over time, rather than a single static value. The determination of coupling strength requires setting a threshold, for example, defining the relationship with absolute correlation coefficient value continuously higher than 0.7 as strong coupling mode, and lower than 0.3 as weak coupling mode. The significant coupling modes identified are recorded and marked with the time period they occur and the grid nodes they belong to.

[0062] The allocation of reactive power compensation priority is a comprehensive decision-making process, which not only depends on the calculated coupling strength values, but also needs to integrate the actual operating state information of the current power grid. The operating state information includes whether the node voltage is close to the limit value, the size of line load rate, and whether the system has special situations such as faults or maintenance. A typical priority allocation rule is that for those regions with strong positive coupling between load fluctuation and reactive power demand and node voltage near the lower limit, the highest compensation priority is given; for regions with strong coupling but normal voltage level, medium priority is given; for regions with weak coupling or high voltage level, lower priority is temporarily given. The final reactive power compensation priority evaluation result is usually presented in the form of a list or matrix with weight coefficients, which clearly indicates the urgency of reactive power compensation demand of different grid regions at different time scales.

[0063] The dynamic capacity optimization model of networked SVG is the next step to utilize the aforementioned analysis results. The model takes the load fluctuation characteristic sequence as the main input variable, which reflects the driving source of the system's reactive power demand. The priority evaluation results of reactive power compensation are converted into constraint conditions of the optimization problem, for example, setting more stringent voltage deviation constraints or smaller reactive power shortage allowance range for high-priority nodes. The capacity optimization objective function is usually designed to pursue the optimization of the overall system operation performance, and common goals include minimizing the total active network loss of the entire network, minimizing the sum of voltage deviations, or maximizing the input benefit of the reactive power compensation equipment. The decision variable, i.e., the SVG capacity configuration value of each candidate installation site, is usually included in the objective function. Iterative algorithms are used to solve this objective function due to its possible nonlinearity and complexity. Genetic algorithm is a common choice, which simulates the natural selection process by initializing a set of random capacity configuration schemes as the population, and gradually approaches the optimal solution of the objective function through selection, crossover, mutation, and other operations. Gradient descent method is suitable for the case where the objective function is differentiable, which gradually adjusts the capacity configuration parameters in the direction of the gradient descent of the objective function, and converges to a local optimal solution. The solution process obtains a preliminary capacity configuration parameter set, which indicates the recommended SVG capacity size to be installed at each node.

[0064] Dynamic updating of the model according to real-time grid data is crucial to its effectiveness, as grid load levels, network topology, and generator output states are constantly changing. The adaptive adjustment mechanism continuously receives the latest operation data from the energy management system or the data acquisition and monitoring control system through the establishment of a data interface. When important operating state changes are detected, such as the commissioning or decommissioning of main lines, the start or stop of large loads, or significant fluctuations in distributed power output, the model updating process is triggered. The updating operation may include: retraining or adjusting the coefficients in the objective function using new load fluctuation data; recalculating the power flow distribution according to the current network topology to update the sensitivity parameters in the constraint conditions; or adjusting the priority evaluation results based on the latest voltage quality evaluation results, and then correcting the boundary values of the constraint conditions. This process makes the capacity optimization model a dynamic entity that can track changes in the actual operating state of the system, and its output is more practical and meaningful.

[0065] Example 3: Real-time grid operation data is matched with the dynamic capacity optimization model to determine potential installation locations and capacity configuration ranges for grid-forming SVG, and a preliminary scheme is generated, followed by a comprehensive technical and economic evaluation to select the optimal configuration. The matching analysis process begins with the extraction of grid topology and node impedance information, which is obtained from the grid geographic information system and parameter database, including line resistance, reactance, transformer ratio, and other data. Based on these basic data, a reactive power compensation sensitivity matrix is calculated, which represents the degree of influence on the voltage of each node in the system when a unit of reactive power is injected at each node. The calculation typically uses the inverse of the Jacobian matrix of the power flow equation or direct sensitivity analysis, and is implemented through numerical simulation tools such as the power system analysis package.

[0066] The calculation of the reactive power compensation sensitivity matrix provides a quantitative basis for evaluating the compensation effect of candidate nodes, combined with the preliminary capacity configuration parameters output by the dynamic capacity optimization model, the reactive power compensation demand and effect of each node are carefully evaluated. The evaluation process simulates the system operation scenario after installing different capacity SVG at the candidate nodes, and analyzes key indicators such as voltage deviation improvement, line loss reduction, and static voltage stability margin improvement. When screening potential installation locations, multiple criteria are set, including the sensitivity value of the node being higher than a certain threshold, the historical voltage out-of-limit frequency of the node being higher, and the capacity demand matching the optimization model recommended value. Through this multi-condition filtering, nodes that are sensitive to reactive power compensation and have urgent demand are identified as key candidate objects. When determining the capacity configuration range, not only the static demand is considered, but also dynamic process analysis is introduced, such as investigating the instantaneous change range of reactive power demand of the node under load mutation or fault conditions, to ensure that the SVG capacity can cover extreme scenarios. The preliminary scheme of site selection and capacity determination is presented in the form of a report, containing recommended node list, suggested SVG capacity upper and lower limit for each node, expected compensation effect summary, and preliminary investment cost estimate.

[0067] The comprehensive technical and economic evaluation stage aims to select the optimal solution from multiple preliminary schemes. The cost data of grid-forming SVG equipment is collected, including procurement price, installation cost, civil engineering cost, etc. The operation and maintenance cost involves regular maintenance, spare parts replacement, energy efficiency loss, etc. The life cycle data refers to the design life and reliability indicators provided by the equipment manufacturer. The economic index calculation includes initial investment, annual operation and maintenance cost discount, residual value disposal, etc., forming a full life cycle cost model. The calculation of technical indicators focuses on performance evaluation. The compensation accuracy is measured by the deviation of SVG output reactive power from the command value. The response time refers to the time span from receiving the control command to outputting the stable value. The reliability index is represented by the mean time between failures or availability. These technical indicators are obtained through laboratory test data, field operation records, or simulation, ensuring the objectivity of the evaluation.

[0068] The economic and technical indicators are normalized and combined into a comprehensive score using a weighted scoring method. This method quantifies the pros and cons of each configuration scheme through a specific formula. The formula for calculating the comprehensive score is: Where: represents the comprehensive evaluation score of a certain network configuration scheme of SVG, the higher the score, the better the scheme. represents the total number of indicators participating in the evaluation, which covers economic and technical dimensions. is the weight coefficient assigned to the th evaluation indicator, reflecting the importance of the indicator in the overall decision-making, and the sum of all weight coefficients is 1. is the standardized score value of the th evaluation indicator, which is linearly transformed to the [0, 1] interval by extreme value method or vector normalization method to eliminate the influence of different indicators dimension.

[0069] The determination of the weight coefficient combines objective data and subjective judgment, for example, using the analytic hierarchy process to construct a judgment matrix and calculate the characteristic vector to obtain the relative weight, ensuring that the importance of each indicator is reasonably reflected. The calculation of the standardized score uses different standardization functions for benefit-type indicators (such as compensation accuracy, the larger the better) and cost-type indicators (such as investment cost, the smaller the better), ensuring the consistency of the scoring direction. For each generated site and capacity preliminary scheme, its value is calculated independently, and the scheme with the highest comprehensive score is finally selected as the optimal configuration scheme. The entire evaluation process may be iterative, and if the highest score scheme has short boards in some key indicators, the weights can be adjusted to recalculate until the optimal solution that balances the needs of all parties is obtained. The result of the scheme selection is accompanied by a detailed sensitivity analysis report, explaining the possible impact of weight changes or input data fluctuations on the final ranking, providing sufficient reference for decision-makers. The final output of the optimal configuration scheme includes clear installation location, equipment specifications, investment budget, expected operation and maintenance plan, and technical performance guarantee value, forming an executable engineering project proposal.

[0070] ​Example 4: Coordinated integration of grid-forming energy storage and comparison with traditional compensators, the core of which is to extend the optimization model to consider the comprehensive benefits of hybrid systems. Take a regional power grid as an example, there is a problem of large fluctuation of photovoltaic power output and load center voltage stability in the region, and it is planned to configure reactive power compensation resources at key nodes. The operation data of the existing grid-forming energy storage in the region are integrated, which are obtained through the power station monitoring system, including the real-time charge and discharge power of the battery energy storage system, the DC side voltage, the AC side output reactive power, and the battery state of charge history curve. These data are recorded at a set time interval (such as every 5 minutes) and marked with their corresponding grid node positions. For example, the energy storage power station at node N12 is in a charging state (absorbing active power) when the photovoltaic power is large at noon, but its converter also has the ability to provide capacitive reactive power; while at the evening peak load, the energy storage system discharges to support the grid, and at this time its reactive power output range is restricted by the discharge power. These operating characteristics are quantified and input into the dynamic capacity optimization model.

[0071] Adjusting the dynamic capacity optimization model to adapt to the synergistic effect of energy storage and SVG means that the objective function and the constraint conditions need to be restructured. The objective function is no longer just to minimize network loss or voltage deviation, and new optimization objectives may be introduced, such as maximizing the overall regulation benefit of the energy storage-SVG hybrid system, or minimizing the reactive power shortage of the hybrid system at different times of the day. The constraint conditions are significantly increased, including the state of charge constraint of the energy storage system (which must be maintained within a certain range to prevent overcharging and overdischarging), the apparent power constraint of the energy storage converter, and the power coordination and allocation constraint between energy storage and SVG. For example, in the optimization calculation, when considering both configuring SVG and utilizing the reactive power support capability of existing energy storage at the same node, the model needs to coordinate the two: when the state of charge of the energy storage is high and the active power output is low, prefer to use the energy storage to provide reactive power to save SVG capacity; when the energy storage needs high power charge and discharge, then mainly by SVG to undertake the task of reactive power compensation. This synergistic optimization model is solved by mixed integer programming or model predictive control algorithm, and the output is the optimal capacity configuration and operation strategy of SVG and existing energy storage system working together.

[0072] According to the energy storage charging and discharging characteristics and the optimization of capacity configuration and site selection decision of reactive power compensation, the space-time characteristics of the power grid need to be analyzed in detail. For example, by analyzing the historical data of the region, it is found that node N07 will have a voltage drop in the afternoon of summer due to the surge of air conditioning load and the decline of photovoltaic output, and there is a group of energy storage power stations near the node, which are usually in the discharge terminal in the afternoon, and the reactive power output capability is limited. Based on this, the optimization decision may tend to additionally configure a certain capacity of SVG at node N07 to specifically cope with the reactive power shortage and voltage problem in this specific period, instead of completely relying on energy storage. On the contrary, for node N15, the load fluctuation is relatively flat, and the nearby energy storage capacity is sufficient and the charging and discharging plan is flexible, so the optimization result may show that only the reactive power function of energy storage needs to be fully utilized to meet the requirements, without the need for new SVG equipment. The site selection decision thus changes from simply "where to install SVG" to a comprehensive layout problem of "where to install SVG and how to use the existing energy storage for reactive power support".

[0073] To conduct a technical and economic comparative analysis of network-forming devices and phase modulators, a comprehensive evaluation dimension needs to be established. Technical advancement evaluation covers multiple aspects, such as dynamic response speed: the response time of network-forming SVG and energy storage converters is usually in milliseconds, while the response speed of traditional phase modulators is in seconds; adjustment accuracy: power electronic devices can achieve fine and continuous adjustment of reactive power, while the adjustment of phase modulators has a stepwise nature; fault ride-through capability: network-forming devices can provide active support during grid faults, while the behavior of phase modulators depends more on the excitation system. Economic evaluation runs through the entire life cycle of the device, and the one-time investment cost of phase modulators may be lower than that of large-capacity SVG, but their operation and maintenance costs (including wear and tear of rotating parts, energy consumption of lubrication and cooling systems, etc.) are usually higher than those of static power electronic devices. Land occupation, installation complexity, noise impact, etc. are also factors to be considered in economic evaluation. The construction of a comprehensive evaluation index system of technical and economic efficiency is to quantify the above comparisons, and this system adopts a hierarchical structure. Referring to Table 1, some index comparisons are shown when considering two different technical schemes (Scheme A: configure a new network-forming SVG; Scheme B: expand a traditional phase modulator) for the same node (such as N07).

[0074] Table 1: Technical and economic index comparison of reactive power compensation schemes for node N07

[0075]

[0076] The indicators listed in the table need to be further processed for comprehensive evaluation. Numerical indicators need to be standardized to eliminate dimensional effects, such as using different standardization functions for cost indicators (the smaller the better) and performance indicators (the larger the better). Text indicators such as "technology maturity" need to be converted to grade scores. Weight distribution reflects the preferences of decision-makers. If the power grid pays special attention to fast response and flexible control, the weight of indicators such as dynamic response time will be set higher. If the project investment budget is tight, the investment cost weight may dominate. The comprehensive score of each scheme is calculated by the weighted scoring method.

[0077] In Example 5, a multi-reactive power source automatic voltage control coordination strategy under multiple time scales is designed, and a real-time monitoring and dynamic updating mechanism is established. Taking a regional power grid with a large number of wind power and photovoltaic power as an example, the reactive power sources of the power grid include distributed network-type SVG, centralized synchronous compensator, and traditional shunt capacitor banks and reactors. Designing a coordinated control strategy requires analyzing the dynamic reactive voltage regulation characteristics of these heterogeneous reactive power sources. Network-type SVG has extremely fast response speed and can achieve smooth bidirectional adjustment of reactive power within tens of milliseconds, but its sustained overload capacity is poor and the capacity is limited. Synchronous compensator has huge short-time overload capacity and can provide strong voltage support, but its response needs to go through excitation system adjustment, which takes hundreds of milliseconds to seconds. Shunt capacitor banks and reactors can only be switched in steps, and the response speed depends on the mechanical switch action time, but their cost is low and the capacity can be large.

[0078] Based on these differences in characteristics, a multi-time scale AVC coordinated control strategy is designed as a hierarchical structure. The ultra-short-term control of seconds or even milliseconds mainly relies on network-type SVG. The AVC master system or local controller sends precise reactive power or voltage instructions to the SVG through a high-speed communication network to suppress rapid voltage flicker or transient voltage drop caused by wind power and photovoltaic power fluctuations. The short-term control task from seconds to minutes is undertaken by the synchronous compensator. The AVC system sends voltage set values or reactive power output targets to the compensator according to the voltage change trend of the power grid, uses its good overload characteristics to cope with the gradual changes of load or the switching of units, and avoids the SVG from exiting operation due to long-time overload. The medium and long-term control above the minute level is executed by shunt capacitor banks and reactors. The AVC system formulates an economic and efficient switching plan according to the predicted load level and new energy output, and performs rough adjustment of regional reactive power balance, thereby leaving the capacity of valuable dynamic reactive power resources (SVG and compensator) to cope with unpredictable rapid fluctuations.

[0079] The core objective of this strategy is to achieve optimal configuration of reactive power dispatch resources, ensuring voltage stability and economic operation of the power grid. In the regional power grid, optimal configuration requires solving a constrained optimization problem. The objective function usually considers minimization of system loss and optimization of voltage quality, and the constraints include that the voltage of each node must be within a safe range, the output of each reactive power source must not exceed the limit, and the static voltage stability margin must be ensured. Optimization calculation will result in, for example, during periods of high wind power generation and light load, preferentially activating line reactors to absorb excess reactive power, and keeping SVGs in standby mode; while during the evening load peak and wind power output decline, preferentially activating capacitor banks, and instructing phase-modulating machines and SVGs to generate reactive power to support the hub node voltage. This optimal configuration enables each type of reactive power source to function at the time scale most suitable for its technical characteristics, achieving synergy between static and dynamic reactive power compensation, and ensuring that the system can maintain voltage stability under various operating modes. Real-time monitoring of power grid load fluctuations and changes in reactive power demand is a prerequisite for dynamic updating, and this system collects full-network power flow data, node voltage values, and real-time output of each reactive power source at a high sampling frequency through the deployment of synchronized phasor measurement units, smart meters, and SCADA systems at key nodes. The monitoring system has early warning thresholds, and when it detects that the load change rate of a certain area has continuously exceeded the set value, or that the reactive power flow of the main transmission channel has undergone significant reversal, or that the voltage of the hub node has deviated from the normal range, it is determined that an important change has occurred in the operating state of the power grid.

[0080] The dynamic updating of reactive power compensation priority evaluation results and capacity optimization model is the embodiment of the adaptive ability of the system. Once the monitoring system detects the above changes, the updating process will be triggered. Updating the reactive power compensation priority evaluation results means recalculating the urgency of reactive power demand at each node. For example, if a new impact load is detected in an industrial area that was originally load-stable, the reactive power compensation priority of that node needs to be immediately increased. Correspondingly, the parameters of the dynamic capacity optimization model also need to be adjusted, including updating the weight coefficients in the objective function (such as focusing more on the areas where stability problems are prominent), modifying the voltage safety limit in the constraint condition, and refreshing the network topology parameters. This process can be achieved through online rolling optimization or event-triggered model parameter identification. According to the updating results, the siting and sizing scheme is adjusted to ensure that the strategy can always adapt to the grid operating conditions. Event-triggered model parameter identification refers to automatically starting the re-identification of key model parameters when a specific event (such as large load switching, main generator start-stop, network topology change) occurs. For example, when an important line is detected to be put into operation, the system needs to update the parameters corresponding to the line in the network admittance matrix. The identification process uses high-precision synchrophasor measurement unit measurement data within a certain period of time after the event occurs, and uses parameter estimation techniques such as least squares to estimate the new, more accurate model parameters. Compared with periodic online rolling optimization, the event-triggered mechanism is an on-demand updating efficient way that ensures that the model is only modified when the system structure changes. For example, the initial siting and sizing scheme may plan to install a large-capacity SVG at node A, but real-time monitoring finds that due to the commissioning of a new transmission line, the voltage stability problem of node A has been alleviated, while the reactive power demand and voltage sensitivity of node B have increased sharply due to the construction of a large data center. Based on this, the system will automatically start the recalculation, and the updated optimization results may suggest adjusting the installation location of the SVG from node A to node B, and making corresponding modifications to the capacity configuration recommendations.

[0081] It should be noted that, in this document, the 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. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device.

[0082] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A network-constructing type SVG dynamic capacity sizing and site selection method based on load fluctuation, characterized in that, The method comprises: Collecting real-time load data and historical reactive power demand data in the power grid system to form load fluctuation characteristic sequences and reactive power compensation demand characteristic sequences; Performing multi-time scale correlation analysis on the load fluctuation characteristic sequences and the reactive power compensation demand characteristic sequences, identifying the dynamic coupling mode between load fluctuation and reactive power demand, and generating a reactive power compensation priority evaluation result; According to the load fluctuation characteristic sequences and the reactive power compensation priority evaluation result, a dynamic capacity optimization model of the grid-forming SVG is established, and the model parameters are adaptively adjusted; Matching and analyzing the real-time collected power grid operation data with the dynamic capacity optimization model to determine the potential installation location and capacity configuration range of the grid-forming SVG, and generating a preliminary site selection and capacity determination scheme; Based on the preliminary site selection and capacity determination scheme, a comprehensive technical and economic evaluation is performed to select the optimal grid-forming SVG configuration scheme; The multi-time scale correlation analysis on the load fluctuation characteristic sequences and the reactive power compensation demand characteristic sequences, the identification of the dynamic coupling mode between load fluctuation and reactive power demand, and the generation of the reactive power compensation priority evaluation result comprise: Decomposing the load fluctuation characteristic sequences and the reactive power compensation demand characteristic sequences into short-term, medium-term and long-term components according to time scales; Calculating the correlation coefficient and coupling strength between load fluctuation and reactive power demand under different time scales to identify significant coupling modes; According to the coupling strength and the power grid operation state, the reactive power compensation priority is allocated to generate a reactive power compensation priority evaluation result; According to the load fluctuation characteristic sequences and the reactive power compensation priority evaluation result, a dynamic capacity optimization model of the grid-forming SVG is established, and the model parameters are adaptively adjusted, which comprises: Using the load fluctuation characteristic sequences as input variables and the reactive power compensation priority evaluation result as constraint conditions, a capacity optimization objective function is constructed; An iterative algorithm is used to solve the objective function to obtain preliminary capacity configuration parameters; The objective function and the constraint conditions are dynamically updated according to real-time power grid data to adaptively adjust the model parameters; The matching and analysis of the real-time collected power grid operation data with the dynamic capacity optimization model to determine the potential installation location and capacity configuration range of the grid-forming SVG, and the generation of the preliminary site selection and capacity determination scheme comprise: Extracting power grid topology and node impedance information to calculate a reactive power compensation sensitivity matrix; Combining the output of the dynamic capacity optimization model to evaluate the reactive power compensation effect and capacity demand of each node; Filtering nodes with high reactive power compensation sensitivity and large capacity demand as potential installation locations and determining the capacity configuration range to generate a preliminary site selection and capacity determination scheme.

2. The load fluctuation based network configuration type SVG dynamic capacity sizing and siting method according to claim 1, characterized in that, The collection of real-time load data and historical reactive power demand data in the power grid system to form load fluctuation characteristic sequences and reactive power compensation demand characteristic sequences comprises: Monitoring the load change rate and reactive power fluctuation of power grid nodes to extract load time series and reactive power time series; Standardizing the load time series and the reactive power time series to eliminate dimensional differences and generate standardized load fluctuation characteristic sequences and reactive power compensation demand characteristic sequences; Integrating the standardized sequences to form a multi-dimensional data set for subsequent analysis.

3. The load fluctuation based network configuration type SVG dynamic capacity sizing and siting method according to claim 1, characterized in that, The preliminary scheme based on site selection and capacity determination is subjected to technical and economic comprehensive evaluation to select the optimal network configuration SVG configuration scheme, including: Collecting the equipment cost, operation and maintenance cost, and life cycle data of network configuration SVG; Calculating the technical indicators of different configuration schemes, including compensation accuracy, response time, and reliability; Combining economic indicators and technical indicators, using weighted scoring method for comprehensive evaluation, and selecting the optimal configuration scheme.

4. The network configuration type SVG dynamic capacity sizing site selection method based on load fluctuation according to claim 3, characterized in that, The method further comprises: Integrating the operation data of network configuration energy storage equipment, adjusting the dynamic capacity optimization model to consider the synergy of energy storage and SVG; According to the charging and discharging characteristics of energy storage and the reactive power compensation demand, the capacity configuration and site selection decision are optimized.

5. The load fluctuation based network configuration type SVG dynamic capacity sizing and siting method according to claim 4, characterized in that, The method further comprises: Performing technical and economic comparative analysis of network configuration equipment and phase modifier, including technical advancement and economic evaluation; Building a comprehensive evaluation index system of technical and economic indicators for optimal configuration scheme selection.

6. The load fluctuation based network configuration type SVG dynamic capacity sizing and siting method according to claim 5, characterized in that, The method further comprises: Designing a multi-time scale AVC coordinated control strategy for multiple reactive power sources, analyzing the dynamic reactive power regulation characteristics of different reactive power sources; Optimizing the configuration of reactive power dispatch resources to ensure grid voltage stability.

7. The load fluctuation based network configuration type SVG dynamic capacity sizing and siting method according to claim 6, characterized in that, The method further comprises: Real-time monitoring of grid load fluctuation and reactive power demand change, dynamic updating of reactive power compensation priority evaluation results and capacity optimization model; Adjusting the site selection and capacity determination scheme according to the updated results to ensure adaptation to grid operating conditions.

Images