Power distribution network operation parameter optimization identification method based on optimal power flow calculation

Abstract

The invention discloses a power distribution network operation parameter optimization identification method based on optimal power flow calculation, and relates to the technical field of power distribution systems, and the method comprises the steps: collecting historical operation parameters of each node of a power distribution network, screening to obtain a data collection priority of an optimal power flow output index, and reasonably distributing node sampling frequencies according to the data collection priority; measuring and collecting operation parameters of the power distribution network according to the distributed sampling frequency, carrying out pre-detection on the collected data, and meanwhile, carrying out adaptive anomaly judgment and sampling frequency dynamic adjustment and carrying out multi-scale decomposition and sensitivity analysis on node residual errors, so as to realize optimal power flow calculation time window adjustment driven by dynamic parameters; and the load model parameters are verified to ensure the accuracy and reliability of the parameter identification result. Through combination of historical data, real-time measurement and residual sensitivity information, transient disturbance, medium-term fluctuation and long-term trend can be considered in the operation parameter identification of the power distribution network, and the online precision of a load model and line parameters is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution systems, in particular to a power distribution network operation parameter optimization identification method based on optimal power flow calculation. BACKGROUND

[0002] With the access of new types of loads such as distributed power, the power distribution network evolves from a single power radial structure to a multi-power complex network, which puts forward higher standards for the power supply reliability of the power distribution network. The abnormal positioning and alarm technology needs to have a millisecond-level response capability. The line resistance, reactance, transformer ratio and node load model parameters in the power distribution network directly affect the power flow distribution, power loss, node voltage and system safety margin. However, these parameters may deviate or change uncertainly in long-term operation, such as line aging, electrical equipment parameter drift or load behavior change, thereby causing the optimal power flow calculation result to deviate from the actual system state.

[0003] For example, the publication number is: CN114825325A, a power distribution network operation parameter optimization method and device are disclosed, which relates to the technical field of power grid parameter processing. The method comprises: constructing a power distribution network optimization target according to a power distribution network optimization sub-target; the power distribution network optimization sub-target includes a power distribution network single-day node voltage average deviation sub-target, an active network loss sub-target and a single-day carbon emission sub-target; constructing a preset constraint condition; the preset constraint condition includes a power distribution network constraint condition and an electric heating load constraint condition; based on the power distribution network optimization target and the preset constraint condition, the power distribution network single-day node voltage average deviation optimization value, the active network loss optimization value and the single-day carbon emission cost optimization value are obtained.

[0004] For example, the publication number is: CN115378041B, a power distribution network optimization method, system, power distribution network, device and medium are announced, the method comprises: performing injection current source equivalent calculation on the distributed energy and energy storage system in the power distribution network to obtain the first running parameter corresponding to the distributed energy and the second running parameter corresponding to the energy storage system; according to the first running parameter and the second running parameter, and the topology structure of the power distribution network, a plurality of optimization targets corresponding to the power distribution network are set; the target function and / or constraint condition corresponding to the plurality of optimization targets are set, the power distribution network optimization strategy is determined according to the power distribution network optimization algorithm for solving the target function, and the optimization configuration of the energy storage system is adjusted according to the power distribution network optimization strategy, wherein the power distribution network optimization algorithm is used to solve the optimal solution of the target function.

[0005] The above-mentioned technology at least has the following technical problems: The deviation between the power flow calculation and the actual operation state in the power distribution network may be caused by equipment aging, inaccurate parameter nominal value, topology information error or system modification. In addition, the non-stationary, nonlinear and sudden changes of load characteristics caused by the rapid fluctuation of node load over time, the fluctuation of distributed energy output and the adjustment of user-side flexible load may result in large residual error of power flow calculation or load forecasting, which may cause the deviation of node voltage, current and power distribution from the actual operation state, further affecting the accuracy and reliability of operation scheduling, equipment protection setting and real-time control strategy. If not corrected, it may also cause parameter identification error accumulation, reducing the effect of intelligent regulation and control of the power grid and the overall stability of the system. SUMMARY

[0006] In order to solve the technical problems existing in the prior art, the embodiments of the present application provide a power distribution network operation parameter optimization identification method based on optimal power flow calculation. The technical scheme is as follows, comprising: Collecting the historical operation parameters of each node of the power distribution network, screening to obtain the data collection priority of the optimal power flow output index of each node of the power distribution network, and allocating the sampling frequency to each node of the power distribution network.

[0007] According to the sampling frequency of each node of the power distribution network, the measurement collection of the operation parameters of the power distribution network is performed, the pre-position detection of the measurement data of each node of the power distribution network is performed, the preliminary residual error calculation of the operation parameters of the power distribution network is performed, the residual error factor of the load and the model output of each node of the power distribution network is obtained, the adaptive abnormality discrimination of each node of the power distribution network is performed, and the sampling frequency of each node of the power distribution network is adjusted.

[0008] The dynamic parameter adjustment driven by the multi-scale sensitivity of the node residual error of the power distribution network is performed, the optimal power flow calculation time window is adjusted, the operation parameters of the power distribution network are identified, and the load model parameters are verified.

[0009] The technical scheme provided by the embodiments of the present application has at least the following beneficial effects: (1) The application proposes a power distribution network operation parameter optimization identification method based on optimal power flow calculation. First, the historical operation parameters of each node of the power distribution network are collected, the data collection priority of the optimal power flow output index is selected, and the node sampling frequency is reasonably allocated based on the priority to ensure high-precision monitoring of key nodes. Then, the measurement collection of the power distribution network operation parameters is performed according to the allocated sampling frequency, and the collected data is pre-detected. The residual factor of the load and the model output of each node is obtained by combining the preliminary residual calculation, and adaptive abnormality discrimination and dynamic adjustment of the sampling frequency are implemented to improve the data quality and node state identification capability. On this basis, the system performs multi-scale decomposition and sensitivity analysis on the node residual, realizes the optimal power flow calculation time window adjustment driven by dynamic parameters, further performs online identification on the power distribution network operation parameters, and verifies the updated load model parameters to ensure the accuracy and reliability of the parameter identification results. By combining historical data, real-time measurement and residual sensitivity information, the power distribution network operation parameter identification can consider transient disturbance, medium-term fluctuation and long-term trend, improve the online accuracy of the load model and line parameters, and ensure the stability and optimization control effect of the overall system operation, thereby effectively supporting the power distribution network operation parameter optimization and decision-making based on optimal power flow calculation.

[0010] (2) The application helps to determine the collection priority of each node by obtaining the historical sensitivity factor of each node, and dynamically allocates the node sampling frequency based on the priority, thereby ensuring the data collection accuracy and timeliness of key nodes and high-sensitivity nodes. By using historical sensitivity information for priority sorting and sampling strategy optimization, the power distribution network operation parameter optimization identification based on optimal power flow calculation can effectively capture the influence of key nodes on system operation while ensuring data quality, improve the accuracy and stability of parameter identification, and reduce the sampling burden of non-key nodes, thereby realizing efficient use of resources and optimization of overall system operation.

[0011] (3) The application judges whether there is an abnormality in each measurement channel of each node of the power distribution network. When an abnormality in a channel is detected, the weight of the channel in the optimal power flow calculation is automatically reduced, thereby performing pre-detection and preliminary selection on the measurement data of each node of the power distribution network. The abnormal measurement can be eliminated or suppressed before online parameter identification to ensure that residual calculation and parameter updating are based on reliable data, thereby improving the accuracy and stability of the power distribution network operation parameter optimization identification based on optimal power flow calculation, and reducing the negative impact of transient fluctuations or communication abnormalities on the identification results.

[0012] (4) This invention achieves adaptive anomaly detection at the node level. Subsequently, the residual sequence of each node is decomposed into multiple scales according to the set decomposition level to generate real-time sensitive factors of each node at each time scale. By combining node anomaly detection, adaptive adjustment of sampling frequency, multi-scale residual analysis and sensitivity assessment, it can not only accurately identify key sensitive nodes and abnormal nodes, but also provide a reliable basis for subsequent dynamic parameter adjustment and load model parameter identification, thereby effectively improving the accuracy and system stability of online optimization identification of distribution network operation parameters based on optimal power flow calculation. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of the method provided in an embodiment of the present invention; Figure 2 This is a flowchart of a method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation, provided by an embodiment of the present invention. Detailed Implementation

[0015] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0016] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0017] like Figure 1 As shown, this embodiment of the invention provides a method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation, including: Historical operating parameters of each node in the distribution network are collected, and the data collection priority of the optimal power flow output index of each node in the distribution network is obtained by screening. Sampling frequency is then allocated to each node in the distribution network.

[0018] The operating parameters of the distribution network are measured and collected according to the sampling frequency of each node of the distribution network. The measurement data of each node of the distribution network are pre-detected, and the operating parameters of the distribution network are initially residual calculated to obtain the residual factor of the load of each node of the distribution network and the model output. At the same time, adaptive anomaly detection is performed on each node of the distribution network, and the sampling frequency of each node of the distribution network is adjusted.

[0019] Dynamic parameter adjustment driven by multi-scale sensitivity of distribution network node residuals is performed, the optimal power flow calculation time window is adjusted, distribution network operating parameters are identified, and load model parameters are verified.

[0020] It should be noted that each node in the distribution network refers to the physical network node of the distribution network, that is, each bus or key connection point of the distribution network, including but not limited to load center access points (such as user transformer side bus), distributed energy (DER) access points (photovoltaic inverter, wind power access bus), key switches, branch points, and transformer endpoints.

[0021] like Figure 2 As shown, Figure 2 This invention provides a flowchart of a method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation. The method includes collecting historical operating parameters of each node in the distribution network, calculating historical sensitivity factors, and determining the importance of different nodes to the optimal power flow objective function. Subsequently, within a preset time window, the residual factor between the actual observed values ​​and the optimal power flow prediction values ​​is calculated and compared with a threshold to achieve node anomaly detection and dynamic adjustment of the sampling frequency. Based on this, the residual sequence is decomposed into multiple scales, and real-time sensitivity factors are calculated at each time scale to quantify the impact of nodes on the overall operating state. Finally, based on sensitivity and anomaly conditions, these adjustment results are fed back into the load model to perform parameter updates and prediction accuracy verification, thereby ensuring that the identification process can adapt to both instantaneous disturbances and track long-term trends, achieving dynamic optimization and accurate identification of distribution network operating parameters. Determining the importance of different nodes to the optimal power flow objective function involves prioritizing the data collection of optimal power flow output indicators for each node in the distribution network. The specific process is as follows: extracting historical operating parameters of each node in the distribution network, including the standard deviation of power, power factor, frequency, and partial derivative of power of each node; pre-setting a historical time window; and weighting and fusing the standard deviation of power, power factor, frequency, and partial derivative of power of each node within the historical time window to obtain the historical sensitivity factors of each node in the distribution network. The historical sensitivity factors of each node in the distribution network are used to quantitatively evaluate the degree of influence of each node on the optimal power flow objective function.

[0022] It should be noted that the weighted fusion here refers to the unified measurement and combination of multiple characteristic quantities such as power, power factor, standard deviation of frequency, and partial derivative of power within a historical time window. First, different weights are assigned based on the positive and negative correlation between these characteristics and the optimal power flow objective function. For example, a positive weight is assigned when power fluctuation is positively correlated with node power flow deviation, and a negative weight is assigned when power factor deviation has a negative impact on voltage constraints. The comparability of different characteristic quantities in terms of order of magnitude is ensured by normalization. Then, the characteristic quantities are fused by linear weighted summation or weighted mean square method to obtain a single historical sensitivity index, which reflects the comprehensive influence of the node on the overall power flow optimization under historical operating conditions. This allows the generated historical sensitivity factor to scientifically and quantitatively characterize the node's sensitivity to optimal power flow.

[0023] Specifically, sampling frequencies are allocated to each node of the distribution network. The specific process is as follows: the historical sensitivity factors of each node of the distribution network are matched with the data acquisition priorities corresponding to each interval of the sensitivity factors of the distribution network nodes stored in the database to obtain the data acquisition priorities of each node of the distribution network, including the first priority, the second priority and the third priority. The data acquisition priorities of each node of the distribution network are determined according to the data acquisition priorities of each node of the distribution network.

[0024] It should be noted that the specific process of prioritizing data acquisition for each node in the distribution network is as follows: After obtaining the acquisition priority of each node, the system matches corresponding sampling frequency strategies according to different priority ranges: For critical nodes with the highest priority (first priority), high-frequency sampling at the real-time level (e.g., second-level acquisition) is allocated to ensure that abnormal fluctuations are not missed; for sensitive nodes with medium and second priorities, medium-frequency sampling (e.g., tens of seconds to minutes-level) is used to ensure that the main operating trends are captured while also saving computing and communication resources; for stable nodes with low and third priorities, low-frequency sampling (e.g., minutes-level or longer) is used, and automatic downsampling is performed during non-critical periods to reduce redundant data; at the same time, the system introduces an adaptive adjustment mechanism. When the residual factor or anomaly detection result of a node exceeds a threshold, its sampling frequency is temporarily increased to a high-frequency level to track sudden anomalies. In this way, the allocation of sampling frequency reflects the sensitivity of nodes in optimal power flow output while ensuring optimal utilization of communication bandwidth and computing resources.

[0025] It should be noted that among the operating parameters of each node in a distribution network, power typically refers to the active and reactive power of that node, reflecting the actual amount of electrical energy consumed or injected at that node, as well as its support or consumption of the grid voltage. Power factor indicates the degree of closeness between the node's energy utilization efficiency and the phase relationship between voltage and current. Frequency is a core indicator of distribution network operational stability, reflecting the real-time balance between power generation on the power source side and power consumption on the load side; deviations in node frequency often indicate system power supply and demand imbalances or local fluctuations. By jointly monitoring the power, power factor, and frequency of each node, the operating characteristics of the node can be comprehensively characterized from three dimensions: energy flow efficiency, voltage stability, and system dynamic balance, providing a foundation for subsequent residual analysis and sampling priority assessment.

[0026] Specifically, the measurement data of each node in the distribution network is pre-detected. The specific analysis process is as follows: the timestamp, main reference signal, packet loss rate and delay distribution of each measurement channel in each node of the distribution network are obtained, and it is determined whether there is any abnormality in each measurement channel in each node of the distribution network. When there is an abnormality in a certain measurement channel in each node of the distribution network, the weight of that channel in the optimal power flow calculation is reduced. In this way, the measurement data of each node of the distribution network is pre-detected.

[0027] It should be noted that the specific process for determining whether there are anomalies in each measurement channel of each node in the distribution network is as follows: First, the system compares the timestamp and primary reference signal (including but not limited to PMU, GPS, or PTP) of each channel to calculate the time deviation and synchronization error. Simultaneously, it statistically analyzes the data packet loss rate and delay distribution of that channel. If the time deviation exceeds a preset threshold, or the packet loss rate or delay exceeds the corresponding preset threshold, it is preliminarily determined that the channel has a communication or timing anomaly. Furthermore, the system will consider the physical consistency of the node where the channel is located and its upstream and downstream branches. The inspection includes power conservation, current direction and amplitude matching, etc., comparing the data of neighboring sensors at the same node with historical measurements. If a single-point offset, saturation, drift or short-term jump is found, it is judged as a sensor anomaly. Finally, the system combines the above judgment results with historical statistical indicators, residual factors and sensitivity analysis, and uses a rule engine or lightweight Bayesian inference to rank the anomaly probability of each channel. This provides a basis for subsequent node anomaly judgment, weight adjustment and time window expansion, thereby realizing early identification and protection of single-channel anomalies, and ensuring the stability and reliability of subsequent distribution network operation parameter identification and optimal power flow calculation.

[0028] It should be noted that the data flow path collected from a node or device is a complete data acquisition line. A node may have multiple sensors or measurement points, and the data from each sensor is transmitted to the data acquisition system through a communication link (fiber optic, Ethernet, wireless, etc.). This path is a measurement channel.

[0029] Specifically, the residual factors of the load at each node of the distribution network and the model output are obtained. The specific process is as follows: a time window is preset, and the actual observed values ​​of the load at each node of the distribution network and the corresponding predicted values ​​generated by the optimal power flow are processed point by point for difference processing, and normalized to obtain the residual factors of the load at each node of the distribution network and the model output.

[0030] It should be noted that the residual factors of the loads at each node of the distribution network and the model output are analyzed under the following specific conditions: ; In the formula, r i,t R represents the residual factor between the load at the i-th node of the distribution network and the model output. i y1 represents the historical maximum value of the residual between the load of the i-th node in the distribution network and the model output, stored in the database. i,t y2 represents the actual observed value of the i-th node in the distribution network at time t. i,tLet represent the predicted value of the optimal power flow generated at time t for the i-th node in the distribution network, where i represents the node number (i=1,2,3,...n) and n is the total number of nodes. Let t represent the time number within the time window (t=1,2,3,...T) and T is the total number of sampling times within the time window.

[0031] It should be noted that the actual observed values ​​of each node in the distribution network include power, voltage amplitude and phase angle measurements, node frequency measurements, etc. By comparing the differences between the above observed values ​​and the optimal power flow prediction values, the degree of deviation of each node in the distribution network in terms of operating status, power distribution, voltage level and frequency stability can be comprehensively characterized, thereby providing a quantitative basis for subsequent anomaly detection, parameter optimization identification and dynamic adjustment of sampling frequency.

[0032] Specifically, adaptive anomaly detection is performed on each node of the distribution network, and the sampling frequency of each node is adjusted. The specific process is as follows: the residual factor of the load and model output of each node of the distribution network is extracted and compared with the residual factor threshold of the node load and model output stored in the database. If the residual factor of the load and model output of a node of the distribution network is higher than or equal to the residual factor threshold of the node load and model output, the anomaly detection result of the node is recorded as anomaly, and the sampling frequency of the node is set to real-time sampling. If the residual factor of the load and model output of a node of the distribution network is lower than the residual factor threshold of the node load and model output, the anomaly detection result of the node is recorded as normal, and the sampling frequency of the node is set to sampling during non-critical time periods.

[0033] It should be noted that sampling during non-critical time periods refers to periods in the distribution network operation where the load, distributed generation output, or power flow status is relatively stable, with low volatility and uncertainty. During these periods, the node status has a relatively small impact on the overall network's safety and stability. Therefore, communication and computing overhead can be reduced by lowering the sampling frequency. Examples include nighttime periods when residential load is low, periods when some industrial and commercial operations are suspended, stable intervals in renewable energy output curves (such as the midday plateau period for photovoltaics or the low-speed stable period for wind power), and operating periods where network topology and scheduling actions change less frequently. The node's operating status can be maintained with sufficient observability through lower-frequency sampling combined with historical trend inference without significantly affecting the accuracy of anomaly detection and parameter updates.

[0034] Specifically, the dynamic parameter adjustment driven by the multi-scale sensitivity of the residuals of distribution network nodes is carried out as follows: statistically analyze the residual sequence of each node of the distribution network within a time window, decompose the residual sequence of each node of the distribution network according to the set decomposition level, and obtain the residual subsequence of each node of the distribution network at each time scale. Each time scale includes the first time scale, the second time scale, and the third time scale. Calculate the variance of the residual subsequence of each node of the distribution network at each time scale.

[0035] It should be noted that the residual subsequence refers to the components at different time scales obtained after decomposing the residual sequence of the distribution network nodes (i.e., the point-by-point difference between the observed load and the model predicted load) using a preset multi-scale decomposition method (such as wavelet decomposition, empirical mode decomposition, or variational mode decomposition). In this embodiment, the original residual sequence is decomposed into residual signals at the first time scale, the second time scale, and the third time scale, with each time scale being a residual subsequence. Variance is a statistical measure of the point-by-point residual values ​​(i.e., differences) in each residual subsequence, reflecting the strength of residual fluctuations at that time scale. In other words, variance measures the degree of dispersion of the residual values ​​around their mean at a certain time scale; a large variance indicates severe residual fluctuations at that scale, while a small variance indicates relatively stable residuals.

[0036] It should be noted that the residual subsequences of each node in the distribution network at each time scale are obtained by multi-scale decomposition of the residual sequences of each node in the distribution network. Methods such as wavelet decomposition, empirical mode decomposition (EMD), or variational mode decomposition (VMD) are typically used to decompose the original residual sequence into different subsequences according to frequency or periodic characteristics. Each subsequence corresponds to a time scale. The first time scale generally corresponds to short-term, high-frequency components, reflecting transient disturbances or rapid fluctuations; the second time scale corresponds to medium-term components, reflecting intraday or multi-hour periodic changes; and the third time scale corresponds to long-term, low-frequency components, reflecting load trends or long-term deviations. In the specific implementation, the number of decomposition levels and component frequency bandwidths are preset by the system. The time length covered by each time scale is determined according to the load change characteristics of the nodes, thus enabling each residual subsequence to capture its corresponding dynamic characteristics.

[0037] The power, power factor, standard deviation of frequency, and partial derivative of power of the residual subsequences of each node in the distribution network at each time scale are weighted and fused to obtain the real-time sensitivity factors of each node in the distribution network at each time scale. The real-time sensitivity factors of each node in the distribution network at each time scale are used to quantitatively evaluate the degree of influence of each node on the output of the overall optimal power flow model and the load model parameter identification results of the distribution network under the current operating state.

[0038] Specifically, if the anomaly detection result of a node in the distribution network is abnormal and the real-time sensitivity factor of that node in the first time scale is higher than or equal to the real-time sensitivity factor threshold stored in the database, then the time window of the first time scale is expanded.

[0039] If the anomaly detection result of a node in the distribution network is abnormal and the real-time sensitivity factor of that node in the third time scale is higher than or equal to the real-time sensitivity factor threshold stored in the database, then the time window of the third time scale is shortened.

[0040] It should be noted that the real-time sensitivity factors of each node in the distribution network at each time scale have an impact on the identification of the distribution network load model and line parameters under the current operating conditions. This reflects the sensitivity of the node to the optimal power flow output and parameter adjustment at different time scales, such as short-term transient fluctuations, medium-term intraday periodic changes, and long-term trend deviations. This provides a basis for dynamically adjusting the sampling frequency, parameter update step size, and weight allocation, ensuring that the node parameter identification considers both transient disturbances and medium- and long-term trends, thereby achieving online accuracy and stability of the distribution network load model and line parameters.

[0041] It should be noted that a multi-scale decomposition is performed on the initially calculated node residual sequence, that is, the overall residual sequence is decomposed into components reflecting changes at different time scales. In specific operation, the system can use methods such as partitioned wavelet transform or hierarchical empirical mode decomposition (EMD) to classify the node residuals into short-term, high-frequency fluctuation components, medium-term, intraday periodic variation components, and long-term, weekly or longer-term trend components by performing stepwise filtering and signal decomposition on the original residual sequence. The short-term residual components mainly reflect the load fluctuations of the nodes on minute or ten-minute time scales. These fluctuations are usually closely related to distributed energy output, short-term load disturbances, transient scheduling or short-cycle control strategy changes, and can reveal the sensitivity of the nodes to rapidly changing events. The medium-term residual components focus on the load change patterns of the nodes on hourly or intraday scales, such as intraday peak-valley load differences, centralized load scheduling or load dynamic changes in load centers. It reflects the possible deviations and uncertainties of the nodes in medium-term scheduling and load forecasting. Long-term residual components focus on the trend changes of nodes over daily, weekly, or longer periods, such as seasonal load changes, long-term load changes of important users, or small drifts of line / transformer parameters over time. Long-term components can help identify long-term deviations and uncaptured load dynamics in the system model, thus focusing on trend correction in load model updates.

[0042] It should be noted that expanding the time window of the first time scale and shortening the time window of the third time scale are done by proportionally mapping the difference between the real-time sensitivity factors of each node in the distribution network at each time scale and the historical sensitivity factors of the corresponding nodes in the distribution network. For example, for a certain node, the difference between the real-time sensitivity factors of the node at the first and third time scales and the corresponding historical sensitivity factors stored in the database is first calculated. This difference reflects the degree of deviation of the current node's sensitivity at that time scale from the historical level. Then, according to the pre-set mapping relationship, the difference is mapped to a specific time window adjustment amount. The original time window is then added to or subtracted from this adjustment amount: for expanding the time window of the first time scale, the adjustment amount is added to the original first time scale time window to extend the time window and smooth transient fluctuations; for shortening the time window of the third time scale, the adjustment amount is subtracted from the original third time scale time window to shorten the time window and capture trend deviations more quickly. The original time window is dynamically adjusted, and its length changes continuously with the deviation of the node's real-time sensitive factor from the historical level. This helps to ensure that the contribution of the residuals at each time scale in parameter identification balances transient disturbances and responds quickly to long-term trends, thereby achieving online accuracy of node parameter identification and overall system stability.

[0043] It should be noted that if the real-time sensitivity factor of the node in the first time scale distribution network is lower than the real-time sensitivity factor threshold stored in the database, or if the real-time sensitivity factor of the node in the third time scale distribution network is lower than the real-time sensitivity factor threshold stored in the database, the time window of the corresponding time scale remains unchanged.

[0044] Specifically, the operation parameters of the distribution network are identified. The specific process is as follows: the operation parameters of the distribution network are updated online according to the preset iterative optimization algorithm. The iterative process includes: determining the direction and step size of parameter adjustment based on the residual between the node load and the output of the optimal power flow model; continuously correcting the parameters to be identified during the iteration until the residual converges to the preset tolerance range; and recording the parameter update log at the same time.

[0045] It should be noted that the specific process of determining the parameter adjustment direction and step size using residuals is as follows: In each iteration, the actual observations of each node in the distribution network are compared point-by-point with the output of the optimal power flow model to obtain the residual vector. This residual vector is then combined with the parameter sensitivity matrix derived from the power flow equations to obtain the direction and magnitude of the partial derivative of the residuals with respect to each parameter to be identified. The direction of parameter updates is determined based on the product of the residuals and the sensitivity matrix: if the result is positive, it indicates that the current value of the parameter causes the model output to be too large, and the parameter value needs to be decreased in the next iteration; if the result is negative, it indicates that the parameter value causes the model output to be too small, and the parameter value needs to be increased in the next iteration. During updates, a preset iterative optimization algorithm is used, such as gradient descent, which corrects the parameters along the opposite direction of the residual gradient, or Newton's method, which accelerates convergence by introducing a second-order approximation. The step size is adaptively adjusted based on the magnitude of the residuals, the convergence rate, and the stability of the parameter updates to ensure that the iteration can achieve residual convergence to the preset tolerance range within a reasonable number of steps.

[0046] Specifically, the load model parameters are verified as follows: within a preset verification time window, the load model prediction output is compared with the actual observation data to obtain the residual factors of the load at each node of the distribution network and the model output. Within the verification time window, the number of residual factors of the load at each node of the distribution network and the model output that are higher than or equal to the threshold of the residual factor of the node load and the model output is counted and recorded as the number of residual nodes. If the number of residual nodes is higher than or equal to the threshold of the number of residual nodes stored in the database, the built-in parameter identification algorithm is called to re-optimize the parameters to be identified in the power grid model.

[0047] It should be noted that load forecasting models are mathematical or data-driven models used to estimate the electricity demand of the power grid at a certain time period or multiple moments in the future. They take historical load time series, calendar information (hours, weekdays / weekends / holidays), meteorological data (temperature, humidity, irradiance, etc.), controllable load and distributed generation output, socio-economic factors, and event factors as inputs, and output predicted values ​​or probability distributions (including confidence intervals) to meet the needs of different application scenarios in the short term (minutes to days), medium term (days to weeks), and long term (months to years). Traditional time series statistical models (such as ARIMA, exponential smoothing), machine learning methods (such as regression trees, support vector regression), or deep learning methods (such as LSTM, Transformer) are often combined with feature engineering, anomaly detection, and uncertainty quantification (e.g., generating prediction intervals) to improve robustness and interpretability. The prediction results can be used as input for optimal power flow (OPF) and scheduling decisions, as well as for capacity planning, energy markets, and demand response strategies.

[0048] It should be noted that the specific process of re-optimizing the parameters to be identified in the power grid model by calling the built-in parameter identification algorithm can be described as follows: The system first acquires the observation data of the current power grid nodes, including real-time measured values ​​of voltage, current, active power, reactive power, etc., and combines them with historical operating data from the recent period to make initial estimates of the parameters to be identified, such as line resistance, reactance, transformer turns ratio, and node load model coefficients, and calculates the residual sequence of each node, i.e., the difference between the actual observed values ​​and the predicted values ​​of the power grid model; Subsequently, the system inputs these residuals into the built-in parameter identification algorithm module, which can include optimization algorithms such as gradient descent, Newton's method, Gauss-Newton method, recursive least squares, or Kalman filtering. By constructing an objective function of the residuals with respect to the parameters to be identified, the system minimizes the error as the iterative optimization objective; During the iteration process, the system calculates the Jacobian matrix based on the sensitivity of the residuals to each parameter, determines the parameter update direction and step size, and considers the confidence interval of the parameter update. To mitigate safety constraints and prevent over-correction from causing model oscillations or system instability, the system updates the power grid model parameters and recalculates node residuals after each iteration to evaluate the effectiveness of the optimization, such as whether residuals have significantly decreased and whether errors at highly sensitive nodes have been improved. Upon iteration convergence or reaching a preset tolerance range, the system writes the final parameter values ​​into the power grid model and records parameter update logs, including residual changes, parameter updates, and corresponding confidence intervals for each iteration step. Simultaneously, the system maintains redundant estimation or virtual observation processing for nodes with abnormal measurements or those temporarily unavailable to ensure the optimization process is not disturbed by single-point anomalies. After the entire process is complete, the parameters to be identified in the power grid model are re-optimized under the current observation conditions, enabling the model to more accurately reflect the actual operating state of the power grid. This provides a reliable foundation for optimal power flow calculation, load forecasting, and subsequent control, while continuously monitoring node sensitivity and system safety boundaries to ensure that parameter identification is both accurate and robust.

[0049] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0050] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0051] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0052] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0053] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0054] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0055] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0056] In the embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0057] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0058] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0059] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0060] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation, characterized in that, The method includes: Historical operating parameters of each node in the distribution network are collected, and the data collection priority of the optimal power flow output index of each node in the distribution network is obtained by screening. Sampling frequency is then allocated to each node in the distribution network. The operating parameters of the distribution network are measured and collected according to the sampling frequency of each node of the distribution network. The measurement data of each node of the distribution network are pre-detected and the operating parameters of the distribution network are initially calculated to obtain the residual factor of the load of each node of the distribution network and the model output. At the same time, adaptive anomaly detection is performed on each node of the distribution network and the sampling frequency of each node of the distribution network is adjusted. Dynamic parameter adjustment driven by multi-scale sensitivity of distribution network node residuals is performed, the optimal power flow calculation time window is adjusted, distribution network operating parameters are identified, and load model parameters are verified.

2. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 1, characterized in that, The data acquisition priority for obtaining the optimal power flow output index of each node in the distribution network through screening is as follows: Historical operating parameters of each node in the distribution network are extracted, including the power, power factor, standard deviation of frequency, and partial derivative of power of each node. A historical time window is preset, and within the historical time window, the power, power factor, standard deviation of frequency, and partial derivative of power of each node in the distribution network are weighted and fused to obtain the historical sensitivity factor of each node in the distribution network. The historical sensitivity factor of each node in the distribution network is used to quantitatively evaluate the degree of influence of each node in the distribution network on the optimal power flow objective function.

3. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 1, characterized in that, The process of allocating sampling frequencies to each node of the distribution network is as follows: The data acquisition priority of each node in the distribution network is obtained by matching the historical sensitivity factors of each node with the data acquisition priority corresponding to each interval of the sensitivity factors of the distribution network nodes stored in the database. This includes the first priority, the second priority, and the third priority. The data acquisition priority of each node in the distribution network is determined based on the data acquisition priority of each node in the distribution network.

4. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 1, characterized in that, The specific analysis process for performing pre-detection on the measurement data of each node in the distribution network is as follows: The system acquires the timestamps, main reference signals, packet loss rates, and delay distributions of each measurement channel in each node of the distribution network. It then determines whether there are any anomalies in each measurement channel in each node of the distribution network. When an anomaly is detected in a measurement channel in each node of the distribution network, the weight of that channel in the optimal power flow calculation is reduced. This allows for pre-detection of the measurement data of each node of the distribution network.

5. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 1, characterized in that, The specific process for obtaining the residual factors of the load at each node of the distribution network and the model output is as follows: A preset time window is used to perform point-by-point difference processing between the actual observed load values ​​of each node in the distribution network and the corresponding predicted values ​​generated by the optimal power flow, and then normalizes them to obtain the residual factors of the load of each node in the distribution network and the model output.

6. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 1, characterized in that, The process of adaptive anomaly detection for each node in the distribution network and adjusting the sampling frequency of each node is as follows: Extract the residual factors of the load and model output of each node in the distribution network, and compare them with the residual factor thresholds of the node load and model output stored in the database. If the residual factor of the load and model output of a node in the distribution network is higher than or equal to the residual factor threshold, the abnormality judgment result of the node is recorded as abnormal, and the sampling frequency of the node is set to real-time sampling. If the residual factor of the load and model output of a node in the distribution network is lower than the residual factor threshold, the abnormality judgment result of the node is recorded as normal, and the sampling frequency of the node is set to sampling during non-critical time periods.

7. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 4, characterized in that, The specific process of adjusting the dynamic parameters driven by the multi-scale sensitivity of the residuals at distribution network nodes is as follows: The residual sequences of each node in the distribution network are statistically analyzed within a time window. The residual sequences of each node in the distribution network are decomposed according to a set number of decomposition levels to obtain the residual subsequences of each node in the distribution network at each time scale. Each time scale includes a first time scale, a second time scale, and a third time scale. The variance of the residual subsequences of each node in the distribution network at each time scale is statistically analyzed. The power, power factor, standard deviation of frequency, and partial derivative of power of the residual subsequences of each node in the distribution network at each time scale are weighted and fused to obtain the real-time sensitivity factor of each node in the distribution network at each time scale. The real-time sensitivity factor of each node in the distribution network at each time scale is used to quantitatively evaluate the degree of influence of each node on the output of the overall optimal power flow model and the load model parameter identification results of the distribution network under the current operating state.

8. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 7, characterized in that, The specific process for adjusting the optimal power flow calculation time window is as follows: If the anomaly detection result of a node in the distribution network is abnormal and the real-time sensitivity factor of that node in the first time scale is higher than or equal to the real-time sensitivity factor threshold stored in the database, then the time window of the first time scale is expanded. If the anomaly detection result of a node in the distribution network is abnormal and the real-time sensitivity factor of that node in the third time scale is higher than or equal to the real-time sensitivity factor threshold stored in the database, then the time window of the third time scale is shortened.

9. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 1, characterized in that, The specific process for identifying the operating parameters of the distribution network is as follows: The distribution network operation parameters are updated online according to a preset iterative optimization algorithm. The iterative process includes: determining the parameter adjustment direction and step size based on the residual between the node load and the output of the optimal power flow model; continuously correcting the parameters to be identified during the iteration until the residual converges to the preset tolerance range; and recording the parameter update log at the same time.

10. The method for optimizing and identifying distribution network operating parameters based on optimal power flow calculation according to claim 1, characterized in that, The specific process for verifying the load model parameters is as follows: Within the preset verification time window, the load model prediction output is compared with the actual observation data to obtain the residual factors of the load of each node in the distribution network and the model output. Within the verification time window, the number of residual factors of the load of each node in the distribution network and the model output that are higher than or equal to the threshold of the residual factor of the node load and the model output is counted and recorded as the number of residual nodes. If the number of residual nodes is higher than or equal to the threshold of the number of residual nodes stored in the database, the built-in parameter identification algorithm is called to re-optimize the parameters to be identified in the power grid model.

Images