Reactive power planning optimization method and system for high-voltage power distribution network

Abstract

The invention relates to the field of power distribution network reactive power optimization, and discloses a reactive power planning optimization method and system for a high-voltage power distribution network, and the method comprises the steps: obtaining real-time power data collected by a sensor, calculating a reactive power demand, and obtaining reactive power demand distribution; when the variance of the reactive power demand distribution exceeds a preset threshold value, the compensation capacity is adjusted, and a preliminary planning scheme is obtained; simulating an operation state according to the preliminary planning scheme, and obtaining a voltage deviation; when the voltage deviation value is larger than a preset threshold value, parameters are dynamically adjusted, and refining scheme output is obtained; reactive power compensation capacity is adjusted according to the refining scheme output, a first reactive power compensation scheme is obtained, reactive power distribution is optimized according to the first reactive power compensation scheme, a final reactive power optimization scheme is obtained, reactive power compensation parameters are adjusted according to the final reactive power optimization scheme, and final reactive power optimization configuration is obtained. According to the method, a reactive power planning scheme adapting to dynamic characteristics and prediction deviation can be constructed in the face of load random change.

Description

Technical Field

[0001] This invention relates to the field of reactive power optimization in power distribution networks, and more particularly to a reactive power planning and optimization method and system for high-voltage power distribution networks. Background Technology

[0002] Reactive power optimization planning in high-voltage distribution networks is crucial in power systems, directly impacting grid stability, energy efficiency, and operating costs. By rationally configuring reactive power compensation equipment, power loss can be effectively reduced, voltage quality improved, and power supply reliability ensured. However, current reactive power planning methods have significant limitations in practical applications. Many methods rely too heavily on static load models or single-scenario assumptions, making it difficult to adapt to dynamic load changes and uncertainties. Furthermore, some schemes, when dealing with complex grid structures, ignore the differences in load characteristics between nodes, leading to significant deviations between planning results and actual needs. These limitations make it difficult for distribution networks to achieve efficient reactive power allocation when facing load fluctuations or sudden emergencies.

[0003] For example, the electricity demand in industrial parks may fluctuate dramatically during peak hours due to adjustments in production plans, while the load in residential areas varies depending on weather or holidays. This dynamic characteristic makes traditional fixed planning schemes insufficient to meet real-time demands, leading to either insufficient or excessive reactive power compensation equipment, thus affecting grid operating efficiency. A deeper challenge lies in the fact that the prediction bias caused by this dynamic characteristic further exacerbates the planning difficulty. For instance, during the summer peak electricity consumption period in a certain city's distribution network, the surge in air conditioning load can cause prediction bias that leads to voltage drops at some nodes, and in severe cases, even localized power outages.

[0004] It is evident that existing technologies struggle to construct reactive power planning schemes that adapt to dynamic characteristics and prediction biases when faced with random load changes, which is detrimental to the stable operation of the power grid. Summary of the Invention

[0005] This invention provides a reactive power planning optimization method for high-voltage distribution networks to solve the problem that existing methods are difficult to construct reactive power planning schemes that adapt to dynamic characteristics and prediction deviations when faced with random load changes.

[0006] Firstly, in order to solve the above-mentioned technical problems, the present invention provides a reactive power planning and optimization method for high-voltage distribution networks, comprising:

[0007] Real-time power data collected from each load node of the high-voltage distribution network by a sensor network is acquired, and uncertainty features are extracted based on the real-time power data to obtain an uncertainty feature set;

[0008] The reactive power demand is calculated based on the uncertainty feature set to obtain the reactive power demand distribution.

[0009] When the variance of the reactive power demand distribution exceeds a preset variance threshold, the particle swarm optimization algorithm is used to initialize the particle positions of nodes with high uncertainty, and the initial particles are iteratively calculated to obtain the updated particle velocity and position.

[0010] Based on the updated particle velocity and position, the compensation capacity of nodes with high uncertainty is adjusted to obtain a preliminary planning scheme;

[0011] Based on the preliminary planning scheme, the operation status is simulated to obtain voltage deviation and energy consumption data; when the voltage deviation value is greater than the preset deviation threshold, the operation parameters are adjusted to obtain the adjusted operation parameters;

[0012] When the voltage deviation in the adjusted operating parameters exceeds the preset deviation threshold, the operating parameters are updated to obtain the first parameter adjustment scheme.

[0013] The parameters are dynamically adjusted according to the first parameter adjustment scheme to obtain the final refining scheme output;

[0014] The reactive power compensation capacity is adjusted according to the output of the refining scheme to obtain the first reactive power compensation scheme, and the performance indicators are quantified according to the first reactive power compensation scheme to obtain the final performance indicators.

[0015] The reactive power allocation is optimized based on the final performance indicators to obtain the final reactive power optimization scheme. The reactive power compensation parameters are then adjusted according to the final reactive power optimization scheme to obtain the final reactive power optimization configuration.

[0016] In one optional implementation, real-time power data collected by a sensor network from each load node of the high-voltage distribution network is acquired; uncertainty feature extraction is performed on the real-time power data to obtain an uncertainty feature set; reactive power demand is calculated based on the uncertainty feature set to obtain the reactive power demand distribution, including:

[0017] The real-time power data is compressed to obtain a compressed power data stream.

[0018] The power data stream is separated to obtain trend data and fluctuation data.

[0019] When the variance of the trend data and fluctuation data exceeds a preset variance threshold, an uncertainty feature extraction operation is performed to obtain an uncertainty feature set.

[0020] Based on the uncertainty feature set, simulated scenarios are generated, resulting in multiple random scenarios;

[0021] The reactive power demand is calculated based on the multiple random scenarios to obtain the reactive power demand distribution.

[0022] In one optional implementation, when the variance of the reactive power demand distribution exceeds a preset variance threshold, the particle positions of the high-uncertainty nodes are initialized, and the initial particles are iteratively calculated using a particle swarm optimization algorithm to obtain updated particle velocities and positions. Based on the updated particle velocities and positions, the compensation capacity of the high-uncertainty nodes is adjusted to obtain a preliminary planning scheme, including:

[0023] When the variance of the reactive power demand distribution exceeds a preset variance threshold, the real-time power data of the high uncertainty node is standardized to obtain standardized power data.

[0024] Based on the standardized power data, the particle swarm optimization algorithm is used to initialize the particle positions of nodes with high uncertainty to obtain an initial particle set.

[0025] Based on the initial particle set, iterative calculations are performed to obtain the updated particle velocity and position;

[0026] When the updated particle velocity and position satisfy the iterative convergence condition, the compensation capacity of the high-uncertainty node is adjusted to obtain the preliminary planning scheme.

[0027] In one optional implementation, an operational state simulation is performed based on the preliminary planning scheme to obtain voltage deviation and energy consumption data; when the voltage deviation value is greater than a preset deviation threshold, the operating parameters are adjusted to obtain the adjusted operating parameters, including:

[0028] Based on the preliminary planning scheme, an operational state simulation was performed to obtain the first dataset;

[0029] The first dataset contains voltage values ​​and energy consumption data;

[0030] The voltage deviation is calculated based on the first dataset to obtain the second dataset;

[0031] The second dataset contains voltage deviations;

[0032] When the voltage deviation value in the second dataset is greater than the preset deviation threshold, the relationship between voltage deviation and energy consumption is fitted to obtain a correlation model;

[0033] The operating parameters are adjusted according to the correlation model to obtain the adjusted operating parameters.

[0034] In one optional implementation, when the voltage deviation in the adjusted operating parameters exceeds a preset deviation threshold, the operating parameters are updated to obtain a first parameter adjustment scheme; the parameters are dynamically adjusted according to the first parameter adjustment scheme to obtain the final refined scheme output, including:

[0035] When the voltage deviation in the adjusted operating parameters exceeds the preset deviation threshold, random variable sampling is performed to obtain the first probability distribution data.

[0036] Based on the first probability distribution data, the operating parameters are optimized to obtain the first set of operating parameters;

[0037] When the first set of operating parameters does not meet the preset convergence criterion, parameter fine-tuning is performed until the preset convergence criterion is met, and the first parameter adjustment scheme is obtained.

[0038] The parameters are dynamically adjusted according to the first parameter adjustment scheme to obtain the final refining scheme output.

[0039] In one optional implementation, the reactive power compensation capacity is adjusted according to the output of the refining scheme to obtain a first reactive power compensation scheme, and the performance indicators are quantified according to the first reactive power compensation scheme to obtain the final performance indicators, including:

[0040] The initial reactive power demand distribution is calculated based on the output of the refining scheme to obtain the first power factor and the first voltage distribution data;

[0041] The reactive power compensation capacity is adjusted based on the first power factor and the first voltage distribution data to obtain the first reactive power compensation scheme.

[0042] Based on the first reactive power compensation scheme, a global optimal search is performed using the particle swarm optimization algorithm to obtain the optimal reactive power compensation configuration.

[0043] The performance indicators are quantified based on the optimal reactive power compensation configuration to obtain the final performance indicators.

[0044] The final performance indicators include voltage stability, second power factor, and load balance.

[0045] In one optional implementation, reactive power allocation is optimized based on the final performance index to obtain a final reactive power optimization scheme, including:

[0046] Calculate the probability distribution of the performance indicators based on the final performance indicators to obtain the first set of performance indicators;

[0047] If the voltage stability or load balance in the first set of performance indicators is lower than the preset probability threshold, the process returns to the particle position initialization step, iteratively calculates the initial particles, and obtains the second reactive power compensation configuration.

[0048] Based on the second reactive power compensation configuration quantification performance index, a second set of performance indexes is obtained;

[0049] If the voltage stability or load balance in the second set of performance indicators is lower than the preset probability threshold, the process returns to the particle position initialization step and iteratively calculates the initial particles until the voltage stability or load balance is higher than the preset probability threshold, thus obtaining the final reactive power optimization scheme.

[0050] In one optional implementation, adjusting the reactive power compensation parameters according to the final reactive power optimization scheme to obtain the final reactive power optimization configuration includes:

[0051] Based on the final reactive power optimization scheme, power flow calculation is performed to obtain the first reactive power demand distribution data;

[0052] Calculate the voltage fluctuation probability distribution based on the first reactive power demand distribution data to obtain the first voltage fluctuation feature set;

[0053] When the voltage fluctuation amplitude in the first voltage fluctuation feature set exceeds the preset amplitude threshold, the particle swarm optimization algorithm is used to perform local optimal tracking, adjust the reactive power compensation parameters, and obtain the third reactive power compensation configuration.

[0054] Based on the quantitative performance indicators of the third reactive power compensation configuration, a third set of performance indicators is obtained;

[0055] If the voltage fluctuation amplitude in the third performance index set is higher than the preset amplitude threshold, the process returns to the local optimal tracking step of the particle, adjusts the reactive power compensation parameters, and continues until the voltage fluctuation amplitude is lower than the preset amplitude threshold to obtain the final reactive power optimization configuration.

[0056] Secondly, the present invention provides a reactive power planning and optimization system for a high-voltage distribution network, comprising:

[0057] The feature extraction module is used to acquire real-time power data collected by the sensor network from each load node of the high-voltage distribution network, extract uncertain features based on the real-time power data to obtain an uncertain feature set, and calculate reactive power demand based on the uncertain feature set to obtain the reactive power demand distribution.

[0058] The intelligent algorithm optimization module is used to initialize the particle positions of high uncertainty nodes using the particle swarm optimization algorithm when the variance of the reactive power demand distribution exceeds a preset variance threshold, and to perform iterative calculations on the initial particles to obtain updated particle velocities and positions; based on the updated particle velocities and positions, the compensation capacity of high uncertainty nodes is adjusted to obtain a preliminary planning scheme.

[0059] The simulation evaluation module is used to simulate the operating state according to the preliminary planning scheme to obtain voltage deviation and energy consumption data; when the voltage deviation value is greater than the preset deviation threshold, the operating parameters are adjusted to obtain the adjusted operating parameters.

[0060] The operating parameter refinement module is used to update the operating parameters when the voltage deviation in the adjusted operating parameters exceeds a preset deviation threshold, thereby obtaining a first parameter adjustment scheme; and to dynamically adjust the parameters according to the first parameter adjustment scheme to obtain the final refined scheme output.

[0061] The reactive power compensation adjustment module is used to adjust the reactive power compensation capacity according to the output of the refining scheme to obtain the first reactive power compensation scheme, and to quantify the performance indicators according to the first reactive power compensation scheme to obtain the final performance indicators.

[0062] The cyclic verification decision module is used to optimize reactive power allocation based on the final performance index to obtain the final reactive power optimization scheme.

[0063] The local fine-tuning module is used to adjust the reactive power compensation parameters according to the final reactive power optimization scheme to obtain the final reactive power optimization configuration.

[0064] Thirdly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the reactive power planning and optimization method for high-voltage distribution networks described in any one of the above-mentioned methods.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] (1) This invention perceives load dynamics through real-time data, and accurately separates trend and fluctuation components through data compression and decomposition; it uses variance threshold to trigger analysis to extract core uncertainty features; and finally generates a large number of random scenarios through Monte Carlo simulation, outputting the probability distribution of reactive power demand to adapt to various fluctuation scenarios, thus solving the problem of existing technologies having "single static models that are difficult to cope with uncertainty".

[0067] (2) This invention intelligently identifies nodes with high uncertainty by using variance threshold, eliminates dimensional differences by data standardization, simulates swarm intelligence by using particle swarm algorithm, and dynamically optimizes by using speed update formula. Finally, through statistical analysis, it verifies that the invention solves the problems of ignoring node differences and difficulty in adapting to dynamic changes in the prior art.

[0068] (3) This invention obtains real-time operating data through power simulation, quickly identifies abnormal nodes through voltage deviation calculation, establishes a quantitative correlation model using linear regression, and finally adjusts parameters precisely based on the model. This solves the problem of ignoring node differences and prediction deviations in the background technology, ensuring voltage stability and improving energy efficiency.

[0069] (4) This invention quantifies the probability distribution of voltage deviation through random sampling; performs small-step parameter iterative optimization based on probability data; and ensures adjustment stability through strict convergence criteria. It solves the problem of difficulty in adapting to dynamic load changes in the prior art, which can not only reduce voltage deviation, but also reduce energy consumption and system losses, and achieve a dual improvement in voltage quality and operating economy.

[0070] (5) Based on topology and load data, this invention accurately locates problem nodes through Newton-Raphson power flow calculation; dynamically adjusts the compensation capacity in the simulation environment to initially improve voltage; then uses particle swarm optimization to perform a global optimal search, simultaneously optimizing power factor and load balance; finally, the robustness of the scheme is verified through peak scenarios. This solves the problems of ignoring node differences and insufficient ability to cope with sudden situations in existing technologies, ensuring that remote nodes remain stable under extreme conditions. Attached Figure Description

[0071] Figure 1 This is a schematic diagram of the reactive power planning and optimization method for high-voltage distribution networks provided by the present invention;

[0072] Figure 2 This is a schematic diagram of the reactive power planning and optimization system structure for high-voltage distribution networks provided by the present invention. Detailed Implementation

[0073] 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.

[0074] Reference Figure 1 The first embodiment of the present invention provides a reactive power planning optimization method for high-voltage distribution networks, including steps S1 to S7, as follows:

[0075] S1. Acquire real-time power data collected from each load node of the high-voltage distribution network by the sensor network; extract uncertainty features based on the real-time power data to obtain an uncertainty feature set; calculate reactive power demand based on the uncertainty feature set to obtain the reactive power demand distribution.

[0076] S2, when the variance of the reactive power demand distribution exceeds a preset variance threshold, the particle swarm optimization algorithm is used to initialize the particle positions of the high uncertainty nodes, and the initial particles are iteratively calculated to obtain the updated particle velocity and position; based on the updated particle velocity and position, the compensation capacity of the high uncertainty nodes is adjusted to obtain a preliminary planning scheme.

[0077] S3, simulate the operation status according to the preliminary planning scheme to obtain voltage deviation and energy consumption data; when the voltage deviation value is greater than the preset deviation threshold, adjust the operation parameters to obtain the adjusted operation parameters;

[0078] S4, when the voltage deviation in the adjusted operating parameters exceeds the preset deviation threshold, the operating parameters are updated to obtain a first parameter adjustment scheme; the parameters are dynamically adjusted according to the first parameter adjustment scheme to obtain the final refining scheme output;

[0079] S5, adjust the reactive power compensation capacity according to the refined scheme output to obtain the first reactive power compensation scheme, and quantify the performance indicators according to the first reactive power compensation scheme to obtain the final performance indicators.

[0080] S6. Optimize reactive power allocation based on the final performance index to obtain the final reactive power optimization scheme.

[0081] S7. Adjust the reactive power compensation parameters according to the final reactive power optimization scheme to obtain the final reactive power optimization configuration.

[0082] In step S1, real-time power data collected by a sensor network from each load node of the high-voltage distribution network is acquired. Uncertainty feature extraction is performed on the real-time power data to obtain an uncertainty feature set. Reactive power demand is calculated based on the uncertainty feature set to obtain the reactive power demand distribution, including:

[0083] The real-time power data is compressed to obtain a compressed power data stream.

[0084] The power data stream is separated to obtain trend data and fluctuation data.

[0085] When the variance of the trend data and fluctuation data exceeds a preset variance threshold, an uncertainty feature extraction operation is performed to obtain an uncertainty feature set.

[0086] Based on the uncertainty feature set, simulated scenarios are generated, resulting in multiple random scenarios;

[0087] The reactive power demand is calculated based on the multiple random scenarios to obtain the reactive power demand distribution.

[0088] In this embodiment, firstly, in the real-time power data acquisition and processing of the high-voltage distribution network, a sensor network is deployed at each load node to collect power data in real time. For example, a distribution network has 100 load nodes, and active and reactive power data are collected once per second, resulting in a huge amount of data. To reduce the transmission burden, a data compression algorithm is used to process this real-time power data to obtain a compressed power data stream.

[0089] In one possible implementation, a wavelet transform-based compression algorithm is used to decompose the original power data into different frequency components, retaining only the key feature data, achieving a compression ratio of up to 5:1. For example, taking the active power sequence [100,102,101,99,98,105,150,95,100,98] collected within 1 second at a load node as an example, the compression process based on wavelet transform is as follows: First, the original data is decomposed into low-frequency approximation coefficients (such as [101,122.5], reflecting the steady-state trend of power) and high-frequency detail coefficients (such as [1,-0.5,-27.5,-2.5], capturing transient changes) using Haar wavelets; then, the detail coefficients are threshold quantized (coefficients with absolute values ​​greater than 5 are retained, such as only -27.5 is retained, and the rest are set to zero), while all approximation coefficients are retained to ensure that the overall trend is not lost; finally, the compressed data only needs to store 2 approximation coefficients and 1 non-zero detail coefficient and its position information, and the data volume is compressed from the original 40 bytes to 13 bytes, with a compression ratio of about 2.8:1 (in actual multi-level decomposition and optimization thresholds, it can be further improved to 5:1). During decompression, the data is reconstructed using the retained coefficients through inverse wavelet transform. This significantly reduces the transmission burden while still preserving key features such as power mutations, thus meeting the needs of real-time monitoring.

[0090] Next, the compressed power data stream is processed by a time series decomposition algorithm to separate trend and fluctuation components. The STL decomposition method is used to divide the data into long-term trends (such as daily load variation patterns) and short-term fluctuations (such as sudden power spikes). For example, the power data for a given node may show a morning and evening peak trend, while the fluctuation component reflects instantaneous changes caused by equipment start-up and shutdown. If the variance of the trend data is 1000 and the variance of the fluctuation data is 500, both exceeding the preset variance threshold of 200, it indicates significant uncertainty in the data, requiring further analysis. This decomposition helps to accurately identify the regularity and abnormal fluctuations in power grid operation, providing a basis for subsequent optimized scheduling.

[0091] It should be noted that the preset variance threshold is usually not a fixed value. The basis for setting it mainly includes: determining a statistical boundary that can distinguish between normal and abnormal fluctuations by analyzing long-term operating data of the high-voltage distribution network (such as reactive power fluctuation range, voltage deviation, etc.); setting an upper limit of variance that can ensure stable operation of the system according to power grid operation standards (such as the requirements for voltage fluctuation and reactive power balance in national and industry standards); setting a fluctuation threshold within its controllable range considering the response speed and capacity limitations of reactive power compensation equipment (such as capacitors, SVG, etc.); and evaluating the risk probability of the system (such as voltage over-limit probability) under different variance levels by combining Monte Carlo simulation calculations, thereby determining an acceptable variance threshold.

[0092] Then, if the variance of the trend data and the fluctuation data exceeds a preset variance threshold, principal component analysis (PCA) is used to extract the uncertainty feature set. For example, the input data contains a two-dimensional feature matrix of trend components (such as a smoothed daily load curve) and fluctuation components (such as the hourly power standard deviation). PCA first centers X (subtracts the mean), and after calculating the covariance matrix, obtains eigenvalues ​​λ1 = 1.8 and λ2 = 0.2 (corresponding eigenvectors representing the directions of "amplitude change" and "frequency fluctuation," respectively). By retaining the principal components with a cumulative contribution rate exceeding 80% (here, λ1 contributes 90%), the original data is projected onto the direction of the first principal component, generating a one-dimensional uncertainty feature set, the magnitude of which directly reflects the main driving factors of power change (such as the intensity of load abrupt changes). For example, a feature value of 2.5 for a certain period indicates that more than 80% of the power fluctuations during that period are caused by high-amplitude load switching or seasonal temperature changes, rather than random noise. This process significantly improves analytical and computational efficiency by discarding minor components with small variance (such as measurement errors) while retaining 90% of the original information and compressing the data dimension from two dimensions to one dimension.

[0093] Finally, based on the uncertainty feature set, Monte Carlo simulation is used to generate multiple random scenarios to calculate reactive power demand. For example, the uncertainty feature set indicates that power fluctuations are mainly affected by industrial load and weather, allowing for the generation of 1000 scenarios to simulate reactive power demand under different load combinations and weather conditions. For instance, in a high-temperature weather scenario, increased air conditioning load may lead to a 20% increase in reactive power demand; while in a low-load scenario, reactive power demand may only be 70% of the baseline value. By statistically analyzing the reactive power demand distribution of these scenarios, a demand probability distribution curve can be obtained, with a mean of 50 MVar and a standard deviation of 10 MVar. This distribution provides a basis for configuring reactive power compensation equipment and optimizing grid stability.

[0094] In step S2, when the variance of the reactive power demand distribution exceeds a preset variance threshold, the particle positions of the high-uncertainty nodes are initialized, and the particle swarm optimization algorithm is used to iteratively calculate the initial particles to obtain updated particle velocities and positions. Based on the updated particle velocities and positions, the compensation capacity of the high-uncertainty nodes is adjusted to obtain a preliminary planning scheme, including:

[0095] When the variance of the reactive power demand distribution exceeds a preset variance threshold, the real-time power data of the high uncertainty node is standardized to obtain standardized power data.

[0096] Based on the standardized power data, the particle swarm optimization algorithm is used to initialize the particle positions of nodes with high uncertainty to obtain an initial particle set.

[0097] Based on the initial particle set, iterative calculations are performed to obtain the updated particle velocity and position;

[0098] When the updated particle velocity and position satisfy the iterative convergence condition, the compensation capacity of the high-uncertainty node is adjusted to obtain the preliminary planning scheme.

[0099] In this embodiment, firstly, when the variance of the reactive power demand distribution exceeds a preset variance threshold, it indicates that the reactive power fluctuations at some load nodes are large, requiring further processing to optimize grid operation. For real-time power data acquisition at nodes with high uncertainty, active and reactive power data can be obtained every minute through a sensor network. For example, a distribution network may contain 50 nodes with high uncertainty; each node collects data once per minute, generating 3000 sets of data per hour, each set containing active power, reactive power, and a timestamp. Data standardization is a crucial step, which can be achieved using a min-max normalization method to map the power data to a range of 0 to 1. For example, if the reactive power range of a node is 10 MVar to 50 MVar, the normalized data range is 0 to 1, facilitating subsequent algorithm processing. It should be noted that standardization eliminates dimensional differences, making the power data from different nodes comparable and facilitating the optimization algorithm's identification of key features.

[0100] Next, based on standardized power data, each particle represents a reactive power value Q, with its initial position randomly generated within the range of [20MVar, 40MVar], and its initial velocity v randomly assigned within the range of [-1MVar / s, 1MVar / s]. Simultaneously, the individual optimal position (initially its own position) and the global optimal position (initially the position with the best fitness in the swarm) of each particle are recorded. In each iteration, the particle velocity is determined using the formula: Dynamically updated, where i represents the particle number, t represents the iteration number, and ν i The initial velocity of each particle is randomly generated within the range of [-1MVar / s, 1MVar / s] (e.g., ν1 = 0.5MVar / s), Q i This means that each particle represents a reactive power value, pbest i For the optimal position of an individual, gbest i To determine the global optimal position, w is the inertia weight, set to w = 0.7; learning factors c1 and c2 are set to c1 = c2 = 1.5; random numbers r1, r2 ∈ [0, 1]; subsequently, particle positions are determined according to... Adjust and calculate the fitness of the new position (e.g., the absolute deviation of reactive power from the ideal value of 30 MVar). If it is better than the current individual optimum, update pbest. i If it is better than the global best, then update gbest. iDuring the iteration process, the particle velocity gradually decreases (e.g., from the initial 0.5 MVar / s to 0.2 MVar / s), and the position moves closer to the optimal solution. When the maximum number of iterations (e.g., 50 times) is met or the global optimal position changes less than 0.01 MVar / s for 10 consecutive times, the algorithm converges and outputs the optimal reactive power value, such as 30 MVar.

[0101] Finally, the compensation capacity of nodes with high uncertainty is adjusted based on the globally optimal location. For example, the adjusted reactive power compensation capacity of a node is 35 MVar, which can be achieved by adjusting the capacitor bank. Statistical analysis methods are used to verify the adjustment effect. For instance, power data for one week after the adjustment is collected, and the mean and variance of reactive power demand are calculated. If the mean is 34 MVar and the variance drops to 50, which is below the preset variance threshold of 100, it indicates that the adjustment is effective, and the adjustment scheme is confirmed as a preliminary planning scheme. It should be noted that statistical analysis can further verify the stability of reactive power demand through probability density functions, such as generating a normal distribution curve to confirm that 90% of the reactive power demand is concentrated between 30 MVar and 38 MVar. This verification ensures that the compensation capacity adjustment meets actual operational needs.

[0102] In step S3, an operational state simulation is performed based on the preliminary planning scheme to obtain voltage deviation and energy consumption data; when the voltage deviation value is greater than a preset deviation threshold, the operating parameters are adjusted to obtain the adjusted operating parameters, including:

[0103] Based on the preliminary planning scheme, an operational state simulation was performed to obtain the first dataset;

[0104] The first dataset contains voltage values ​​and energy consumption data;

[0105] The voltage deviation is calculated based on the first dataset to obtain the second dataset;

[0106] The second dataset contains voltage deviations;

[0107] When the voltage deviation value in the second dataset is greater than the preset deviation threshold, the relationship between voltage deviation and energy consumption is fitted to obtain a correlation model;

[0108] The operating parameters are adjusted according to the correlation model to obtain the adjusted operating parameters.

[0109] In this embodiment, firstly, the real-time operating state is simulated according to the preliminary planning scheme. For example, a city's power distribution network contains 10 nodes. By inputting operating parameters such as load power and line impedance, a first dataset is generated. This first dataset includes the voltage values ​​and energy consumption data of each node. For example, node 1 has a voltage value of 0.98 pu and an energy consumption of 50 kW; node 2 has a voltage value of 0.95 pu and an energy consumption of 60 kW. These data reflect the operating characteristics of the power distribution network under specific conditions, providing a foundation for subsequent analysis.

[0110] Next, after extracting voltage values ​​from the first dataset, the voltage deviation of each node is calculated. Assuming the standard voltage is 1.0 pu, the voltage deviation of node 1 is |1.0 - 0.98| = 0.02 pu, and the deviation of node 2 is 0.05 pu, thus forming the second dataset. If the preset deviation threshold is 0.03 pu, then the deviation of node 2 exceeds the threshold and requires further analysis. This deviation calculation method is simple and intuitive, and can quickly identify nodes with abnormal voltage, providing a basis for optimization. In one embodiment, for nodes with voltage deviations exceeding the threshold, a linear regression algorithm is used to analyze the relationship between voltage deviation and energy consumption. For example, linear regression fitting is performed on the data of node 2, inputting multiple sets of voltage deviation and energy consumption data, such as (0.05 pu, 60 kW), (0.04 pu, 58 kW), etc., to obtain a correlation model, the form of which is P. loss = a·ΔV+b, where a is the slope (unit: kW / pu), representing the change in energy consumption for every 1 p.u. increase in voltage deviation; b is the intercept (reference energy consumption). This indicates that for every 0.01 pu increase in voltage deviation, energy consumption increases by 2 kW.

[0111] Finally, based on the correlation model, the particle swarm optimization algorithm is used to adjust the operating parameters. It is assumed that by adjusting the capacitor switching at node 2, the voltage deviation is reduced from 0.05 pu to 0.02 pu, and the energy consumption is reduced from 60 kW to 55 kW. The particle swarm optimization algorithm performs iterative calculations; the optimization process is as follows: each particle represents a capacitor capacity C. i , set C i The range of values ​​(e.g., C) i ∈[0,50]kVar). Initial positions for generating N particles (C i ) and velocity (v) i ), v i Randomly generated within the range of [-1kVar / s, 1kVar / s] (e.g., v1 = 0.5kVar / s), for each particle C i Calculate the corresponding objective function value J(C) i )=α·P loss (C i)+β·(ΔV(C i )-ΔV target )2, where P loss (C i ) is the energy consumption (e.g., P) predicted through a correlation model. loss =200·(ΔV(C) i )+50); ΔV(C i ) represents the voltage deviation after capacitor adjustment, ΔV target For the target voltage deviation, α and β are weighting coefficients (e.g., α = 1, β = 10 emphasizes voltage stability). In each iteration, the particle velocity is determined by the formula: Dynamically updated, where i represents the particle number, t represents the iteration number, and ν i The initial velocity of each particle is randomly generated within the range of [-1 kVar / s, 1 kVar / s] (e.g., ν1 = 0.5 kVar / s), Q i Each particle represents a reactive power value, pbest is the individual optimal position, gbest is the global optimal position, w is the inertia weight set to w = 0.7; c1 and c2 are learning factors set to c1 = c2 = 1.5; random numbers r1, r2 ∈ [0, 1]; subsequently, particle positions are determined according to... The algorithm is adjusted and the fitness of the new position is calculated. When the maximum number of iterations (e.g., 50 times) or the global optimal position changes less than 0.01 kVar / s for 10 consecutive times, the algorithm converges and outputs the capacitor capacity (e.g., 15 kVar).

[0112] Determining the optimal capacitor capacitance ensures voltage stability within a reasonable range. This adjustment method, guided by fitted parameters, balances the dual objectives of voltage stability and energy reduction.

[0113] In step S4, when the voltage deviation in the adjusted operating parameters exceeds a preset deviation threshold, the operating parameters are updated to obtain a first parameter adjustment scheme; the parameters are dynamically adjusted according to the first parameter adjustment scheme to obtain the final refined scheme output, including:

[0114] When the voltage deviation in the adjusted operating parameters exceeds the preset deviation threshold, random variable sampling is performed to obtain the first probability distribution data.

[0115] Based on the first probability distribution data, the operating parameters are optimized to obtain the first set of operating parameters;

[0116] When the first set of operating parameters does not meet the preset convergence criterion, parameter fine-tuning is performed until the preset convergence criterion is met, and the first parameter adjustment scheme is obtained.

[0117] The parameters are dynamically adjusted according to the first parameter adjustment scheme to obtain the final refining scheme output.

[0118] In this embodiment, firstly, when the voltage deviation in the adjusted operating parameters exceeds a preset deviation threshold, the Monte Carlo method is used to randomly sample and simulate the uncertainty of voltage deviation in the distribution network, generating first probability distribution data. For example, a city's distribution network has 8 nodes, and the deviation of a certain node follows a normal distribution (e.g., node 1's deviation is N(0.03, 0.005)). 2 ), Node 2 is N(0.06, 0.01) 2 If there is a correlation between nodes (e.g., the correlation coefficient between nodes 1 and 2 is 0.7), then the independent standard normal samples are modeled by covariance matrix and linearly transformed by Cholesky decomposition to generate 1000 sets of related samples jointly from 8 nodes. Then, extreme values ​​that exceed the physical constraint range (e.g., ±0.1 pu) are truncated to finally form the first probability distribution data containing the deviation probability characteristics of each node.

[0119] Next, based on the first probability distribution data, the gradient descent method is used to optimize the operating parameters, resulting in the first set of operating parameters. Gradient descent calculates the gradient of the objective function and iteratively updates the parameters to approximate the optimal solution. For example, for the voltage deviation at node 2, the operating parameters are initialized, setting the initial value of the capacitor capacity C0 = 10MVar, the step size α = 0.01, and the convergence threshold ∈ = 0.001. Then, the iterative loop begins: based on the current parameter X... k Perform power flow calculations to obtain the node voltage V. 2k and system loss P lossk The objective function is calculated as a weighted sum of deviation and energy consumption: J(X) k )=w1×|1.0-V 2k ∣+w2·P lossk (Where w1 and w2 are weighting coefficients). Subsequently, the gradient is calculated using the finite difference method. (δ = 0.01MVar represents a small perturbation) to determine the optimization direction. Then, the parameters are updated along the negative gradient direction: To adjust the capacitor's capacitance. During the iteration process, if the change in the objective function ΔJ=|J(X) k+1 )-J(X k If ) | <∈, then the loop terminates and the optimal parameter X is finally output. * (e.g. C) * =15.2MVar) and the minimum objective function value J(X) * This method ensures the stability of parameter adjustment through small-step iterations, while reducing energy consumption.

[0120] Then, if the first set of operating parameters does not meet the preset convergence criteria, parameter fine-tuning is performed. For example, the convergence threshold is set to a voltage deviation change of less than 0.001 pu, or the number of iterations reaches 100. Assuming that after 10 iterations, the voltage deviation of node 2 decreases from 0.06 pu to 0.025 pu, satisfying the convergence condition, the first parameter adjustment scheme is obtained. This scheme avoids over-optimization and ensures computational efficiency by clearly defining the termination condition.

[0121] Finally, based on the first parameter adjustment scheme, the quality indicators and probability distribution data of the distribution network are calculated. For example, after adjusting the capacitor capacity of node 2, the voltage deviation stabilizes at 0.025 pu, the system loss decreases by 3%, and the probability distribution data shows that the probability of deviation exceeding the threshold decreases from 20% to 5%. These indicators are quantitatively analyzed using the following formulas and steps, combined with Monte Carlo simulation, probabilistic statistics, and dynamic optimization methods, to comprehensively verify the effectiveness of the scheme: N sets of input scenarios (such as load power P) are generated through Monte Carlo sampling. L,i Renewable energy output P R,i In each scenario, power flow calculations are performed to obtain the node 2 voltage deviation ΔV. 2,i =∣V 2,i -V ref |, where V 2,i Let V be the node 2 voltage under i samples. ref For the target voltage value, such as V ref =0.03pu, the average voltage deviation is:

[0122]

[0123] For example, before adjustment After adjustment This reduces the efficiency by 37.5%. Furthermore, an extended approach can be introduced to predict peak-hour deviations and dynamically adjust parameters, ultimately yielding a refined solution output. This method, through forward-looking analysis, further enhances the stability of the distribution network.

[0124] In step S5, the reactive power compensation capacity is adjusted according to the output of the refining scheme to obtain a first reactive power compensation scheme. Then, the performance indicators are quantified based on the first reactive power compensation scheme to obtain the final performance indicators, including:

[0125] The initial reactive power demand distribution is calculated based on the output of the refining scheme to obtain the first power factor and the first voltage distribution data;

[0126] The reactive power compensation capacity is adjusted based on the first power factor and the first voltage distribution data to obtain the first reactive power compensation scheme.

[0127] Based on the first reactive power compensation scheme, a global optimal search is performed using the particle swarm optimization algorithm to obtain the optimal reactive power compensation configuration.

[0128] The performance indicators are quantified based on the optimal reactive power compensation configuration to obtain the final performance indicators.

[0129] The final performance indicators include voltage stability, second power factor, and load balance.

[0130] In this embodiment, firstly, based on the refined scheme output, a power flow calculation method is used to calculate the initial reactive power demand by analyzing line impedance and load distribution. Assume a remote distribution network contains 10 nodes, with node 8 being a remote node, and load data including active power of 100kW and reactive power of 50kvar. In one embodiment, based on the Newton-Raphson method, the voltage and power factor of each node are calculated, specifically: for each node i of the distribution network, the active power P... i and reactive power Q i The equilibrium equation is:

[0131]

[0132] Among them, V i V j θ represents the node voltage amplitude. ij =θ i -θ j G represents the phase difference of the node voltage. ij B ij Let represent the real and imaginary parts of the line admittance matrix; n is the total number of nodes (n = 10 in this example). Linearize the power flow equations to construct the Jacobian matrix J:

[0133]

[0134] The elements of each submatrix are:

[0135]

[0136] The initial voltage amplitude is set to V. i (0) =1.0pu, phase angle θ I (0) =0, updated iteratively:

[0137]

[0138] Until the power imbalance max(|ΔP|,|ΔQ|) < ∈ (e.g., ∈ = 10) -6 ).

[0139] For example, the voltage at node 8 is V8 = 0.95 pu after the initial iteration; the power factor is cosφ = 0.85. The first power factor is 0.85, and the voltage at node 8 is 0.95 pu. This method accurately reflects the power flow of the distribution network by iteratively solving nonlinear equations, providing fundamental data for subsequent optimization.

[0140] Next, based on the first power factor and voltage distribution data, a real-time simulation scenario of node 8 is constructed in the simulation environment. For example, the load fluctuations of node 8 are simulated, and the node voltage should satisfy: V min ≤V i ≤V max (usually V) min =0.95pu,V max 1 If the voltage at node 8 exceeds this range (0.05 p.u.), the reactive power compensation capacity is adjusted, such as by adding a 20 kvar capacitor. Iterative calculations show that the voltage increases from 0.95 p.u. to 0.98 p.u., forming the first reactive power compensation scheme. This dynamic adjustment, through real-time feedback, quickly responds to load changes, ensuring voltage stability.

[0141] Then, for the first reactive power compensation scheme, a particle swarm optimization algorithm is used for global search to optimize the load balance and the second power factor. The particle swarm optimization algorithm simulates swarm behavior and iteratively updates the configuration of the reactive power compensation equipment to obtain the optimal reactive power compensation configuration. Assuming the initial particle swarm size is 50, with each particle representing a reactive power compensation configuration scheme, a multi-objective fitness function is designed, combining the load balance and power factor objectives: f(x... i )=w1·LB(x i )+w2·cosφ(x i ), where position x i To randomly generate the compensation capacity combination for each particle, the load balance (LB) can be defined as the reciprocal of the voltage deviation of each node or the degree of balance of the line load rate. The power factor (cosφ) is taken as the power factor of the key node (e.g., node 8), with a weighting coefficient of w1 + w2 = 1, adjusted according to priority (e.g., w1 = 0.6, w2 = 0.4). After 20 iterations, each iteration includes the following steps: updating the individual optimum and the global optimum, and updating the particle velocity and position. When the maximum number of iterations is reached or the global optimum fitness has not significantly improved after k consecutive iterations, the iteration ends, and the particle swarm optimization algorithm finds the optimal compensation configuration (e.g., increasing node 8 by 30 kvar, and adjusting adjacent nodes by ±10 kvar). The load balance of node 8 improves from 0.7 to 0.9, and the second power factor reaches 0.92. This method avoids getting trapped in local optima through global search, thus optimizing the overall distribution network performance.

[0142] Finally, the adaptability of node 8 was verified in a simulation environment based on the optimal reactive power compensation configuration. For example, a peak load scenario of 150kW was simulated, and the performance indicators were quantified as follows: voltage stability remained at 0.98 pu, the second power factor was 0.92, and the load balance was 0.9. These indicators, through comprehensive evaluation, verified the high adaptability of the solution in remote nodes and improved the reliability of the distribution network operation.

[0143] In step S6, reactive power allocation is optimized based on the final performance index to obtain the final reactive power optimization scheme, including:

[0144] Calculate the probability distribution of the performance indicators based on the final performance indicators to obtain the first set of performance indicators;

[0145] If the voltage stability or load balance in the first set of performance indicators is lower than the preset probability threshold, the process returns to the particle position initialization step, iteratively calculates the initial particles, and obtains the second reactive power compensation configuration.

[0146] Based on the second reactive power compensation configuration quantification performance index, a second set of performance indexes is obtained;

[0147] If the voltage stability or load balance in the second set of performance indicators is lower than the preset probability threshold, the process returns to the particle position initialization step and iteratively calculates the initial particles until the voltage stability or load balance is higher than the preset probability threshold, thus obtaining the final reactive power optimization scheme.

[0148] In this embodiment, firstly, the probability distributions of voltage stability and load balance are calculated using the Monte Carlo simulation method based on the final performance indicators, resulting in a first set of performance indicators. Preferably, a normal distribution model is used to describe load fluctuations, where the reactive power has a mean of 60 kvar and a standard deviation of 5 kvar. In the Monte Carlo simulation, it is initially assumed that the reactive power load Q follows a normal distribution Q ~ N(60 kvar, 5 kvar). 2 ), generate 10,000 random load samples; for each sample Q(k), obtain the voltage stability index Vs through power flow calculation. (k) (such as node voltage amplitude) and load balance indicators Where k represents the current scenario number or time step in the calculation, n is the number of nodes participating in the calculation in the distribution network, and Vs (k) Let V be the actual voltage value (in pu, per unit) of the s-th node in a specific scenario (such as the k-th Monte Carlo simulation). ref The reference voltage was used. The results show that the probability of voltage stability being between 0.94 and 0.96 pu is 80%, and the probability of the average load balance being greater than 0.7 is 75%. This probabilistic analysis can quantify uncertainty and provide a basis for optimization decisions.

[0149] Next, when the voltage stability or load balance in the first performance index set is lower than a preset probability threshold—for example, the preset probability threshold for voltage stability between 0.94 and 0.96 pu is set to P(0.94≤Vs≤0.96) = 90%, and the preset probability threshold for a load balance mean greater than or equal to 0.7 is set to P(LB≥0.7) = 80%, where P is calculated by dividing the number of samples meeting the conditions (e.g., a load balance mean greater than or equal to 0.7) by the total number of random samples—then the process returns to the particle position initialization step. For example, the particle swarm size is set to 40, and the configuration of the reactive power compensation equipment is adjusted using a velocity update formula. After 15 iterations, a second reactive power compensation configuration is obtained, such as adding a 25 kvar capacitor at node 10. This method optimizes reactive power distribution and improves the operating efficiency of the distribution network through swarm intelligence search.

[0150] Then, based on the second reactive power compensation configuration, the performance of node 10 was verified in a simulation environment. A peak load scenario of 160kW was simulated to quantify voltage stability and load balance. The results showed that the voltage reached 0.98 pu and the load balance improved to 0.88. The probability distributions of voltage stability and load balance were then calculated using Monte Carlo simulation to obtain the second set of performance indicators.

[0151] Finally, if the voltage stability or load balance in the second set of performance indicators is lower than the preset threshold, the process returns to the particle position initialization step to perform iterative calculations on the initial particles until the voltage stability or load balance is higher than the preset probability threshold, thus obtaining the final reactive power optimization scheme.

[0152] In step S7, the reactive power compensation parameters are adjusted according to the final reactive power optimization scheme to obtain the final reactive power optimization configuration, including:

[0153] Based on the final reactive power optimization scheme, power flow calculation is performed to obtain the first reactive power demand distribution data;

[0154] Calculate the voltage fluctuation probability distribution based on the first reactive power demand distribution data to obtain the first voltage fluctuation feature set;

[0155] When the voltage fluctuation amplitude in the first voltage fluctuation feature set exceeds the preset amplitude threshold, the particle swarm optimization algorithm is used to perform local optimal tracking, adjust the reactive power compensation parameters, and obtain the third reactive power compensation configuration.

[0156] Based on the quantitative performance indicators of the third reactive power compensation configuration, a third set of performance indicators is obtained;

[0157] If the voltage fluctuation amplitude in the third performance index set is higher than the preset amplitude threshold, the process returns to the local optimal tracking step of the particle, adjusts the reactive power compensation parameters, and continues until the voltage fluctuation amplitude is lower than the preset threshold to obtain the final reactive power optimization configuration.

[0158] In this embodiment, firstly, based on the final reactive power optimization scheme and the topology data of the high-voltage distribution network, power flow calculation is performed on the load characteristics of remote nodes to obtain the first reactive power demand distribution data. For example, a distribution network contains 15 nodes, with node 12 being a remote node. The analysis of the connection between the main line and branch lines focuses on topology identification and electrical parameter correlation. Topology identification clarifies the hierarchical relationship between the main line and branch lines using a single-line diagram or geographical wiring diagram of the distribution network. For example, in a 15-node distribution network, the main line may connect nodes 1-5, while the branch line extends from node 5 to the remote node 12, forming a "main line-branch" radial structure. Electrical parameter correlation is performed through line impedance calculation. Based on the conductor type, length, and arrangement of the branch lines, the resistance R and reactance X are calculated. For example, the branch line from node 5 to node 12 is 3km long, using LGJ-95 wire. Looking up the table, the unit length impedance Z0 = 0.35 + j0.42Ω / km, so the total impedance is Z. 5-12

[0159] =3×(0.35+j0.42)=1.05+j1.26Ω.

[0160] Determine the load characteristics of node 12, such as active power 150kW and reactive power 80kvar. Using the Newton-Raphson method, combined with line impedance and load distribution, calculate the initial voltage of node 12 as 0.93pu and the power factor as 0.82.

[0161] Next, based on the first reactive power demand distribution data, a probability distribution model for remote nodes is constructed. Reactive power fluctuations are affected by load changes, equipment operating status, etc., and a normal distribution Q is adopted based on the central limit theorem. 12 ~N(80kvar,6 2 A probabilistic model was constructed, where the mean of 80 kvar reflects typical load demand, and the standard deviation of 6 kvar quantifies the impact of random factors such as load abrupt changes or equipment switching. For example, the reactive power mean at node 12 is 80 kvar, and the standard deviation is 6 kvar. Monte Carlo simulation was used to generate 8000 random load samples, and the probability distribution of voltage fluctuations was calculated to obtain the first voltage fluctuation characteristic set. The results show that the probability of voltage being between 0.92 and 0.95 pU is 75%. This probabilistic analysis quantifies the characteristics of voltage fluctuations, providing a basis for optimization decisions.

[0162] Then, if the first voltage fluctuation feature set shows that the voltage fluctuation amplitude exceeds the preset amplitude threshold, the voltage fluctuation amplitude is defined as the percentage of the instantaneous change in the effective voltage value relative to the rated voltage, and its standard calculation formula is: ΔV p.u. =(V max -V min ) / V N ), where V max V represents the maximum effective voltage value within the observation period. min V is the minimum effective value of the voltage within the observation period. N The system rated voltage, ΔV p.u. Voltage fluctuation amplitude, for example, according to distribution network requirements, a voltage fluctuation amplitude greater than 3% triggers flicker risk; therefore, a preset amplitude threshold of 0.03 pu represents a voltage fluctuation of ±3%. Reactive power compensation parameters are adjusted. The particle swarm optimization algorithm searches for local optima by simulating swarm intelligence to adjust the configuration of reactive power compensation equipment. Preferably, the particle swarm size is set to 50, and after 12 iterations, a 30 kvar capacitor is added at node 12 to form a third reactive power compensation configuration. This method optimizes reactive power distribution by dynamically adjusting particle positions.

[0163] Finally, the voltage fluctuation characteristics of node 12 were verified in a simulation environment using a third reactive power compensation configuration. For example, a peak load scenario of 180kW was simulated, and the voltage fluctuation amplitude was quantified. For instance, the simulation duration was set to 1 minute (6000 time steps, step size 0.01s), the voltage amplitude sequence of node 12 was recorded, and the voltage extreme values ​​were extracted. 12,max =max{V 12 (t)},V 12,min =min{V 12 (t)}, where V 12,max and

[0164] V 12,min These represent the maximum and minimum voltage values ​​of node 12 during the observation period, respectively, and are used to quantify the voltage fluctuation range of this node.

[0165] Formula for calculating voltage fluctuation amplitude: ΔV 12 =V 12,max -V 12,min To further evaluate the probabilistic characteristics of voltage fluctuations under uncertainty conditions, the Monte Carlo method is used to generate stochastic scenarios, such as load fluctuation P. 12 ~N(180kw,8 2 ), Q 12 ~N(80kvar,8 2N = 8000 sets of random samples are generated, and the voltage fluctuation amplitude distribution is calculated. For example, the probability P(ΔV12≤0.02pu) of voltage fluctuation amplitude being less than 0.02pu is calculated by dividing the number of samples that meet the condition of voltage fluctuation amplitude being less than 0.02pu by the total number of random samples. Then, the probability distribution of voltage stability and load balance is calculated using the Monte Carlo simulation method to further evaluate the performance indicators and generate a third set of performance indicators.

[0166] If the voltage fluctuation amplitude in the third performance index set is higher than the preset amplitude threshold, the process returns to the local optimal tracking step of the particle, adjusts the reactive power compensation parameters, and continues until the voltage fluctuation amplitude is lower than the preset threshold to obtain the final reactive power optimization configuration.

[0167] Reference Figure 2 The second embodiment of the present invention provides a reactive power planning and optimization system for a high-voltage distribution network, comprising:

[0168] The feature extraction module is used to acquire real-time power data collected by the sensor network from each load node of the high-voltage distribution network, extract uncertain features based on the real-time power data to obtain an uncertain feature set, and calculate reactive power demand based on the uncertain feature set to obtain the reactive power demand distribution.

[0169] The particle swarm optimization module is used to initialize the particle positions of high-uncertainty nodes using the particle swarm algorithm when the variance of the reactive power demand distribution exceeds a preset variance threshold, and to perform iterative calculations on the initial particles to obtain updated particle velocities and positions; based on the updated particle velocities and positions, the compensation capacity of the high-uncertainty nodes is adjusted to obtain a preliminary planning scheme.

[0170] The optimization and evaluation module is used to simulate the operating status according to the preliminary planning scheme to obtain voltage deviation and energy consumption data; when the voltage deviation value is greater than the preset deviation threshold, the operating parameters are adjusted to obtain the adjusted operating parameters.

[0171] The parameter adjustment module is used to update the operating parameters when the voltage deviation in the adjusted operating parameters exceeds a preset deviation threshold, thereby obtaining a first parameter adjustment scheme; and to dynamically adjust the parameters according to the first parameter adjustment scheme to obtain the final refined scheme output.

[0172] The reactive power compensation adjustment module is used to adjust the reactive power compensation capacity according to the output of the refining scheme to obtain the first reactive power compensation scheme, and to quantify the performance indicators according to the first reactive power compensation scheme to obtain the final performance indicators.

[0173] The reactive power allocation module is used to optimize the reactive power allocation based on the final performance indicators to obtain the final reactive power optimization scheme.

[0174] The reactive power compensation parameter adjustment module is used to adjust the reactive power compensation parameters according to the final reactive power optimization scheme to obtain the final reactive power optimization configuration.

[0175] It should be noted that the reactive power planning and optimization system for high-voltage distribution networks provided in this embodiment of the invention is used to execute all the process steps of the reactive power planning and optimization method for high-voltage distribution networks described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0176] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a reactive power planning and optimization program for high-voltage distribution networks. When the processor executes the computer program, it implements the steps described in the various embodiments of the reactive power planning and optimization methods for high-voltage distribution networks, for example... Figure 1 The step S1 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the feature extraction module.

[0177] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0178] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0179] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting various parts of the electronic device through various interfaces and lines.

[0180] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0181] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0182] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0183] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A reactive power planning optimization method for a high-voltage distribution network, characterized in that, The application relates to a power distribution network uncertainty feature extraction method and device. The application comprises the following steps: Uncertainty feature extraction is performed according to real-time power data collected by a sensor network from each load node of a high-voltage power distribution network, and an uncertainty feature set is obtained; Reactive power demand is calculated according to the uncertainty feature set, and a reactive demand distribution is obtained; When the variance of the reactive demand distribution exceeds a preset variance threshold, particle position initialization is performed on high-uncertainty nodes by using a particle swarm algorithm, and initial particles are iteratively calculated to obtain updated particle speed and position; The compensation capacity of the high-uncertainty nodes is adjusted according to the updated particle speed and position, and a preliminary planning scheme is obtained; Operation state simulation is performed according to the preliminary planning scheme, and voltage deviation and energy consumption data are obtained; When the voltage deviation value exceeds a preset deviation threshold, the operation parameters are adjusted, and adjusted operation parameters are obtained; When the voltage deviation of the adjusted operation parameters exceeds the preset deviation threshold, the operation parameters are updated, and a first parameter adjustment scheme is obtained; Parameter dynamic adjustment is performed according to the first parameter adjustment scheme, and a final refined scheme output is obtained; The reactive power compensation capacity is adjusted according to the refined scheme output, a first reactive power compensation scheme is obtained, and performance index quantification is performed according to the first reactive power compensation scheme, and a final performance index is obtained; 2. The method for reactive power planning optimization of high voltage distribution network according to claim 1, characterized in that, Optimized reactive power distribution is performed according to the final performance index, a final reactive power optimization scheme is obtained, and reactive power compensation parameters are adjusted according to the final reactive power optimization scheme, and a final reactive power optimization configuration is obtained. Real-time power data collected by a sensor network from each load node of a high-voltage power distribution network is obtained, and uncertainty feature extraction is performed according to the real-time power data, and an uncertainty feature set is obtained; Reactive power demand is calculated according to the uncertainty feature set, and a reactive demand distribution is obtained, including: Data compression is performed according to the real-time power data, and compressed power data streams are obtained; Separation is performed according to the power data streams, and trend data and fluctuation data are obtained; When the variance of the trend data and the fluctuation data exceeds a preset variance threshold, uncertainty feature extraction is performed, and an uncertainty feature set is obtained; A simulation scenario is generated according to the uncertainty feature set, and a plurality of random scenarios are obtained; 3. The method for reactive power planning optimization of high voltage distribution network according to claim 1, characterized in that, Reactive power demand calculation is performed according to the plurality of random scenarios, and a reactive demand distribution is obtained. When the variance of the reactive demand distribution exceeds a preset variance threshold, particle position initialization is performed on high-uncertainty nodes, and initial particles are iteratively calculated by using a particle swarm algorithm, and updated particle speed and position are obtained; the compensation capacity of the high-uncertainty nodes is adjusted according to the updated particle speed and position, and a preliminary planning scheme is obtained, including: When the variance of the reactive demand distribution exceeds a preset variance threshold, the real-time power data of the high-uncertainty nodes is standardized, and standardized power data is obtained; Particle position initialization is performed on the high-uncertainty nodes according to the standardized power data by using a particle swarm algorithm, and an initial particle set is obtained; Iterative calculation is performed according to the initial particle set, and updated particle speed and position are obtained; When the updated particle velocity and position satisfy an iteration convergence condition, a compensation capacity of a high-uncertainty node is adjusted to obtain a preliminary planning scheme.

4. The method for reactive power planning optimization of high voltage distribution network according to claim 1, characterized in that, An operating state simulation is performed according to the preliminary planning scheme to obtain voltage deviation and energy consumption data. When the voltage deviation value is greater than a preset deviation threshold, operating parameters are adjusted to obtain adjusted operating parameters, including: An operating state simulation is performed according to the preliminary planning scheme to obtain a first data set. The first data set includes voltage values and energy consumption data. A voltage deviation is calculated according to the first data set to obtain a second data set. The second data set includes voltage deviations. When the voltage deviation value in the second data set is greater than a preset deviation threshold, a relationship between the voltage deviation and the energy consumption is fitted to obtain a correlation model. The operating parameters are adjusted according to the correlation model to obtain adjusted operating parameters.

5. The method for reactive power planning optimization of high voltage distribution network according to claim 1, characterized in that, When the voltage deviation in the adjusted operating parameters exceeds the preset deviation threshold, operating parameter updating is performed to obtain a first parameter adjustment scheme. Parameter dynamic adjustment is performed according to the first parameter adjustment scheme to obtain a final refined scheme output, including: When the voltage deviation in the adjusted operating parameters exceeds the preset deviation threshold, random variable sampling is performed to obtain first probability distribution data. Operating parameter optimization is performed according to the first probability distribution data to obtain a first operating parameter set. When the first operating parameter set does not satisfy a preset convergence criterion, parameter fine-tuning is performed until the preset convergence criterion is satisfied to obtain a first parameter adjustment scheme. Parameter dynamic adjustment is performed according to the first parameter adjustment scheme to obtain a final refined scheme output.

6. The method for reactive power planning optimization of high voltage distribution network according to claim 1, characterized in that, The reactive power compensation capacity is adjusted according to the refined scheme output to obtain a first reactive power compensation scheme, and performance index quantification is performed according to the first reactive power compensation scheme to obtain a final performance index, including: Initial reactive power demand distribution calculation is performed according to the refined scheme output to obtain first power factor and first voltage distribution data. The reactive power compensation capacity is adjusted according to the first power factor and first voltage distribution data to obtain a first reactive power compensation scheme. Global optimal search is performed according to the first reactive power compensation scheme using a particle swarm optimization algorithm to obtain an optimal reactive power compensation configuration. Performance index quantification is performed according to the optimal reactive power compensation configuration to obtain a final performance index. The final performance index includes voltage stability, second power factor, and load balance degree.

7. The method for reactive power planning optimization of high voltage distribution network according to claim 1, characterized in that, Optimized reactive power distribution is performed according to the final performance index to obtain a final reactive power optimization scheme, including: Performance index probability distribution is calculated according to the final performance index to obtain a first performance index set. When the voltage stability or load balance degree in the first performance index set is lower than a preset probability threshold, the particle position initialization step is returned to for iterative calculation of initial particles to obtain a second reactive power compensation configuration. Performance index quantification is performed according to the second reactive power compensation configuration to obtain a second performance index set. When the voltage stability or load balance degree in the second performance index set is lower than a preset probability threshold, the particle position initialization step is returned to again, and the initial particle is iteratively calculated until the voltage stability or load balance degree is higher than the preset probability threshold, and a final reactive power optimization scheme is obtained.

8. The method for reactive power planning optimization of high voltage distribution network according to claim 1, characterized in that, According to the final reactive power optimization scheme, the reactive power compensation parameter is adjusted to obtain a final reactive power optimization configuration, including: According to the final reactive power optimization scheme, power flow calculation is performed to obtain first reactive power demand distribution data; According to the first reactive power demand distribution data, voltage fluctuation probability distribution is calculated to obtain a first voltage fluctuation feature set; When the voltage fluctuation amplitude in the first voltage fluctuation feature set exceeds a preset amplitude threshold, a local optimal tracking is performed by using a particle swarm optimization algorithm to adjust the reactive power compensation parameter to obtain a third reactive power compensation configuration; According to the third reactive power compensation configuration, performance indexes are quantified to obtain a third performance index set; When the voltage fluctuation amplitude in the third performance index set is higher than the preset amplitude threshold, the particle local optimal tracking step is returned to again to adjust the reactive power compensation parameter until the voltage fluctuation amplitude is lower than the preset amplitude threshold, and a final reactive power optimization configuration is obtained.

9. A reactive power planning optimization system for a high voltage distribution network, characterized in that, Including: The feature extraction module is configured to acquire real-time power data collected by the sensor network from each load node of the high-voltage power distribution network, perform uncertainty feature extraction according to the real-time power data, and obtain an uncertainty feature set; and calculate reactive power demand according to the uncertainty feature set to obtain a reactive power demand distribution. The intelligent algorithm optimization module is configured to, when a variance of the reactive power demand distribution exceeds a preset variance threshold, perform particle position initialization on a high-uncertainty node by using a particle swarm algorithm, and iteratively calculate an initial particle to obtain updated particle speed and position; and adjust compensation capacity of the high-uncertainty node according to the updated particle speed and position to obtain a preliminary planning scheme. The simulation evaluation module is configured to perform operating state simulation according to the preliminary planning scheme to obtain voltage deviation and energy consumption data. When the voltage deviation value is greater than a preset deviation threshold, the operating parameter is adjusted to obtain an adjusted operating parameter. The operating parameter refining module is configured to, when the voltage deviation in the adjusted operating parameter exceeds the preset deviation threshold, perform operating parameter updating to obtain a first parameter adjustment scheme; and perform parameter dynamic adjustment according to the first parameter adjustment scheme to obtain a final refined scheme output. The reactive power compensation adjustment module is configured to adjust reactive power compensation capacity according to the refined scheme output to obtain a first reactive power compensation scheme, and perform performance index quantification according to the first reactive power compensation scheme to obtain final performance indexes. The cyclic verification decision module is configured to perform optimized reactive power distribution according to the final performance indexes to obtain a final reactive power optimization scheme. The local fine adjustment module is configured to adjust the reactive power compensation parameter according to the final reactive power optimization scheme to obtain a final reactive power optimization configuration.

Images