Regional power grid prediction scheduling method and system based on topology balance

Abstract

The invention discloses a regional power grid prediction scheduling method and system based on topological balance, and relates to the technical field of power grid prediction scheduling, and the method comprises the steps: building a monitoring model reflecting the real-time power flow of a whole network; based on the monitoring model, combining a probability statistical method to construct a new energy output scene library; obtaining the critical operation state of the power grid in each probability scene based on the scene library, and forming a mapping relation between the power transmission capability and the scene; according to the real-time operation state matching mapping relation, correction and joint optimization are carried out to obtain the self-adaptive power transmission limit; through the dynamic scheduling method based on real-time monitoring, scene analysis and joint optimization, the problem that the traditional static power transmission limit cannot accurately reflect high-proportion new energy fluctuation and power grid operation state change is solved.

Description

Technical Field

[0001] This invention relates to the field of power grid predictive scheduling technology, and more specifically, to a regional power grid predictive scheduling method and system based on topology balance. Background Technology

[0002] With the continuous increase in the proportion of new energy power generation and the large-scale integration of intermittent power sources such as wind power and photovoltaics into regional power grids, the operation of the power grid faces increasing uncertainties and dynamic challenges. Traditional power grid dispatching and forecasting methods largely rely on historical load forecasts and statistical patterns of unit output, optimizing operations through static power flow calculations and empirical dispatching rules. However, these methods have several technical bottlenecks.

[0003] Traditional dispatching methods often employ static or empirical safety margins when calculating transmission channel capacity, neglecting the differences in critical states and instantaneous carrying capacity under different operating scenarios. This makes it difficult to provide precise constraints for adjusting renewable energy output. Furthermore, high-proportion renewable energy output fluctuates significantly and exhibits strong uncertainty, leading to increasingly complex and variable power flow distribution and power transmission characteristics. Maximum transmission capacity is closely related to grid operation modes and renewable energy output adjustment methods, making offline limits based on long-period timescales ill-suited to drastic changes in grid operation. To address these issues, this invention proposes a solution. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a regional power grid predictive scheduling method and system based on topology balance. By using a dynamic scheduling method based on real-time monitoring, scenario-based analysis, and joint optimization, the method solves the problem that traditional static transmission limits cannot accurately reflect high-proportion renewable energy fluctuations and changes in power grid operating status.

[0005] To achieve the above objectives, the present invention provides the following technical solution: Firstly, this application provides a regional power grid predictive scheduling method based on topological balance. The method includes: establishing a monitoring model reflecting real-time power flow across the entire network; constructing a renewable energy output scenario library based on the monitoring model and combining it with probabilistic statistical methods; obtaining the critical operating states of the power grid under each probabilistic scenario based on the scenario library, forming a mapping relationship between transmission capacity and scenarios; generating a scenario-capacity mapping table by weighted averaging of the transmission capacity matrix with corrected probability weights; and obtaining an adaptive transmission limit by matching the mapping relationship according to the real-time operating state, correcting and jointly optimizing the limit, wherein the correction is performed by calculating the variance to correct the confidence interval, and the optimization is performed by constructing a multi-objective joint optimization function and constraints based on the confidence interval correction.

[0006] In one embodiment, a monitoring model reflecting the real-time power flow of the entire network is established, specifically by: preprocessing real-time power grid data to form a standardized input dataset; constructing a node-branch network model based on the power grid topology and dynamically assigning values ​​to it to form a power flow monitoring framework; performing parameter matching and parameter mapping on the standardized input dataset and inputting it into the power flow monitoring framework; and using an improved Newton-Raphson iterative method to calculate the power flow; and generating a real-time power flow monitoring model of the entire network that includes node voltage, branch power flow, and system power status based on the calculation results.

[0007] In one embodiment, parameter matching and mapping are performed on the standardized input data, specifically: the standardized input data is initialized as a set of particles and initial weights are assigned; the predicted state of each particle is calculated based on the power grid topology; real-time measurement data is acquired and compared with the predicted state of the particles, the updated weights of each particle are calculated and the set of particles is resampled; the resampled set of particles is weighted and averaged to obtain the optimal particle state and mapped to the node-branch network model.

[0008] In one embodiment, an improved Newton-Raphson iterative method is used for power flow calculation. Specifically, an initial power flow state vector is constructed based on the matched standardized input dataset; a nonlinear equation for the entire power flow is established based on the power grid topology and the whole-network power flow monitoring framework; a Jacobian matrix is ​​constructed through a sparse matrix and the initial power flow state vector is iteratively solved to output the power flow calculation results including node voltage, line power flow, and system power.

[0009] In one embodiment, based on the monitoring model, a new energy power output scenario is constructed using probabilistic statistical methods. Specifically, the following steps are taken: a first dataset is extracted based on the real-time monitoring model, and a probability distribution is fitted to the first dataset; an initial random scenario set is generated from the probability distribution using a random sampling method; the initial random scenario set is subjected to dimensionality reduction and clustering processing, representative scenarios are selected, and an initial probability weight is assigned to each scenario; multi-scenario power flow calculation and verification are performed based on the representative scenarios to generate a new energy power output scenario library containing node voltage, branch power flow, and system power balance status.

[0010] In one embodiment, the critical operating state of the power grid under each probability scenario is obtained based on the scenario library. Specifically, this involves: performing time-series power flow calculations on each representative scenario in the new energy output scenario library to obtain dynamic operating data; calculating the dynamic margin index for each scenario based on the dynamic operating data; and identifying the critical operating state under each scenario based on the dynamic margin index.

[0011] In one embodiment, the mapping relationship is obtained by weighting the transmission capacity matrix with corrected probability weights to generate a scenario-capacity mapping table. Specifically, this involves: analyzing the limit transmission capacity of the transmission channel under each scenario to form a scenario transmission capacity matrix; constructing a conditional probability correction model based on the initial probability weights of each scenario, and outputting corrected probability weights through the probability correction model; combining the corrected probability weights with the transmission capacity matrix corresponding to each scenario, and using a weighted averaging method to obtain the corrected weighted transmission capacity matrix; and generating a scenario-capacity mapping table based on the corrected weighted transmission capacity matrix.

[0012] In one embodiment, the adaptive transmission limit is obtained by correcting and jointly optimizing the mapping relationship based on the real-time operating status. The correction step specifically includes: constructing a high-dimensional state vector based on the real-time operating status and extracting the state vectors corresponding to historical scenarios; performing preliminary screening between the current high-dimensional state vector and the corresponding historical state vectors, and selecting several scenarios with the smallest distance as a subset based on the Mahalanobis distance calculation method; using a multiple linear regression model to predict the transmission capacity data in the subset to obtain the preliminary transmission capacity; inputting the preliminary transmission capacity into a pre-trained random forest regression model, and obtaining the final transmission capacity prediction value through multi-decision tree ensemble prediction; calculating the variance based on the final transmission capacity prediction value, correcting the confidence interval based on the variance, and outputting the corrected actual available transmission capacity.

[0013] In one embodiment, the adaptive transmission limit is obtained by matching and jointly optimizing the mapping relationship based on the real-time operating status. The optimization steps are as follows: constructing a multi-objective joint optimization function that includes the total power loss of the system, the output and consumption of renewable energy, and the dynamic margin index of grid security, and setting constraints, and using the confidence interval of the corrected actual available transmission capacity to constrain the instantaneous maximum transmission capacity of each transmission channel; solving the optimization function using a rolling time-domain optimization algorithm to obtain the optimal adjustment scheme of the new energy output strategy and the grid operation mode; calculating the adaptive limit of each transmission channel and issuing it to the dispatching system, and establishing a closed-loop update mechanism to realize the dynamic adjustment of transmission capacity.

[0014] Secondly, this application provides a regional power grid predictive dispatch system based on topological balance. This system includes a model building module for establishing a monitoring model reflecting real-time power flow across the entire network; a scenario library building module for constructing a renewable energy output scenario library based on the monitoring model and using probabilistic statistical methods; a mapping table generation module for obtaining the critical operating states of the power grid under various probabilistic scenarios based on the scenario library, forming a mapping relationship between transmission capacity and scenarios. This mapping relationship is obtained by weighted averaging of the transmission capacity matrix with corrected probability weights to generate a scenario-capacity mapping table; and a transmission limit generation module for matching the mapping relationship according to the real-time operating state, correcting and jointly optimizing to obtain an adaptive transmission limit. The correction is achieved by calculating the variance to correct the confidence interval, and the optimization is performed by constructing a multi-objective joint optimization function and constraints based on the confidence interval correction.

[0015] This invention enables real-time reflection of grid status, scientific description of renewable energy fluctuation characteristics, reduced computational complexity, and enhanced reliability and flexibility of dispatch decisions under conditions of high renewable energy ratios. It provides a solid data foundation and technical support for dynamic transmission capacity calculation, dispatch optimization, and risk assessment. The overall solution not only improves the security, stability, and flexibility of the power grid under high renewable energy ratios but also enhances renewable energy absorption capacity and significantly optimizes the real-time performance and accuracy of dispatch decisions, providing scientific and operable technical support for modern power grid operation. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a regional power grid predictive scheduling method based on topology balance, provided as an embodiment of this application.

[0017] Figure 2 This is a schematic diagram of a regional power grid predictive dispatch system based on topology balancing, provided as an embodiment of this application.

[0018] Figure 3 This is a mapping diagram of the instantaneous maximum power transmission capacity of the power transmission channel under the new energy output scenario provided in the embodiments of this application. Detailed Implementation

[0019] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Reference Figure 1 As shown in the diagram, this invention provides a flowchart of a regional power grid predictive scheduling method based on topology balancing, which includes the following steps: S1. Establish a monitoring model that reflects the real-time power flow of the entire network. Specifically, this includes acquiring real-time power grid data and establishing a real-time monitoring model of the power flow distribution status of the entire network based on the power grid topology and operating mode. The real-time power grid data includes new energy output data, conventional unit operating data, and key operating parameters.

[0021] Specifically, the system collects real-time power output data from renewable energy units through the power dispatch automation system and the renewable energy power station monitoring system. This data includes active power output, reactive power output, and available installed capacity. It also acquires conventional unit operating data from thermal power, hydropower, and nuclear power units using status monitoring devices and an energy management system. This data includes real-time output levels, reserve capacity, and start / stop status. Furthermore, it synchronously acquires key operating parameters, including voltage, current, frequency, and power flow direction, through the power grid SCADA system (monitoring and data acquisition system) to ensure consistency and comparability between renewable energy output data and conventional unit operating data. Finally, it performs time synchronization and data cleaning on the renewable energy output data, conventional unit operating data, and key operating parameters to form a standardized real-time operating database for subsequent establishment of a dynamic monitoring model of the power flow distribution status of the entire network.

[0022] In this embodiment, real-time power grid data is acquired, and a real-time monitoring model of the power flow distribution status of the entire network is established based on the power grid topology and operating mode. The real-time power grid data includes renewable energy output data, conventional unit operating data, and key operating parameters, specifically: Real-time power grid data is preprocessed, including time alignment and data cleaning, outlier removal and missing value filling, to form a standardized input dataset; Based on the power grid topology information, the busbars, substations and important load points in the power grid are modeled as nodes in the figure, and the transmission lines, transformers and controllable branches are modeled as branches in the figure. Each node and branch contains electrical parameter information, including impedance, rated capacity, rated voltage and power flow control equipment parameters, forming a complete node-branch network model. Based on the actual operation mode of the power grid, the node and branch parameters are dynamically assigned to the node-branch network model to construct a power flow monitoring framework that can reflect the current operating status of the power grid. For example, dynamic assignment can be understood as updating the parameters of nodes and branches to the current values ​​in real time or by time step according to the actual operation of the power grid, rather than using fixed, pre-set parameters. The node and branch parameters include the unit start-up status and output power, load distribution and power factor, and power flow control strategy parameters, such as reactive power regulation, phase angle control, and power shunting device settings. The purpose of this is to allow the power flow monitoring framework to truly reflect the current state of the power grid and to update the power flow calculation results in real time as the units start and stop, load changes, or control strategies are adjusted, rather than performing static calculations. Perform parameter matching and parameter mapping on the standardized input dataset to make the standardized input dataset correspond to the corresponding nodes and branches in the node-branch network model; The standardized input dataset after matching is input into the whole network power flow monitoring framework. The improved Newton-Raphson iterative method is used to perform power flow calculation, and solve the voltage magnitude and phase angle of each node, the active power flow and reactive power flow of each line, and the power balance state of the whole system. Based on the power flow calculation results, a real-time monitoring model of the power flow distribution status of the entire network is constructed, including the voltage of each node, the power flow of each branch, and the system power status.

[0023] It should be noted that power grid topology information refers to a complete set of information describing the structure and connections of the power grid, including all nodes, branches, and their attributes. Nodes mainly refer to the locations and capacities of buses, substations, load points, and renewable energy or energy storage devices that can be connected. Each node includes parameters such as voltage level and power demand. Branches mainly refer to transmission lines, transformers, and other power transmission channels, and their attributes include impedance, rated capacity, turns ratio, and switch status. It also includes the connection relationships between nodes and branches, reflecting the topology of the power grid.

[0024] Furthermore, parameter matching and mapping are performed on the standardized input dataset, specifically as follows: The standardized input dataset is initialized as a set of particles, where each particle represents a state in the node-branch network model. Initial weights are set for each particle based on the initial uncertainty. Can be taken as Where N is the total number of particles; Based on the power grid topology and node-branch network model, the predicted state of each particle is calculated using the power flow approximation method. The predicted state includes the injected power and voltage of each node, the power flow magnitude and direction of each branch, forming a set of particle predicted states. The actual real-time measurement data is compared with the predicted state of the particles, and the updated weight of each particle is calculated. The weight update can be in the form of a Gaussian likelihood function. The updated weights are calculated using the following formula:

[0025] In the formula, Let be the updated weight of the i-th particle at time step k. Let i be the weight of the i-th particle at the previous time step k-1. It is an exponential function used to map observation errors to weight decay factors. This is the standard form of the Gaussian likelihood function, which measures the difference between the predicted state of a particle and real-time measurement data. This is real-time measurement data at time step k, including all measurable grid parameters such as node voltage magnitude, voltage phase angle, branch power flow, and active / reactive power. Let R be the predicted state of the i-th particle at time step k, and let R be the measurement noise covariance matrix, which reflects the uncertainty and noise intensity of each real-time measurement data.

[0026] The particle set is resampled, particles whose updated weights are lower than the preset weight threshold are eliminated, particles whose weights are higher than the preset weight threshold are retained, and new particles are generated to ensure that the number of particles remains at N, so as to enhance the robustness to abnormal data and measurement noise. The optimal particle state is obtained by weighted averaging the resampled particle set and mapped onto the node-branch network model as the optimal estimate of the voltage amplitude, phase angle and power flow of each branch, thus achieving the matching of standardized input data with the node-branch network model.

[0027] It should be noted that the advantage of using particle filtering for parameter matching and mapping is that it can accurately correspond multi-source, real-time operating data of new energy and conventional units, which may contain missing or noisy data, to the grid node-branch model. Through particle set prediction, weight update, and weighted averaging, it can achieve optimal estimation of node voltage, phase angle, and branch power flow, thereby ensuring that the input data is completely matched with the network model in terms of structure and value. This improves the accuracy, robustness, and real-time performance of power flow calculation. At the same time, it can dynamically respond to fluctuations in new energy output, unit start-up and shutdown, and topology changes, providing a reliable foundation for real-time monitoring, dynamic transmission capacity calculation, and scheduling optimization.

[0028] Furthermore, the standardized input dataset after matching is input into the network-wide power flow monitoring framework, and the improved Newton-Raphson iterative method is used to perform power flow calculations, specifically: The matched and standardized input dataset is used to construct an initial power flow state vector, which includes the initial voltage magnitude, initial phase angle, injected active power, injected reactive power, impedance parameters and rated capacity of each branch for each node. Based on the power grid topology and the whole-network power flow monitoring framework, a whole-network power flow nonlinear equation is established, namely, the node power imbalance function f(x) = 0, where:

[0029] In the formula, Let i be the injected active power of the i-th node. The reactive power injected into the i-th node, Let be the voltage amplitude of the i-th node. Let be the voltage amplitude at the j-th node. Let be the difference between the voltage phase angles at node i and node j. Let be the real part of the admittance matrix, representing the active power transmission capacity between node i and node j. Let be the imaginary part of the admittance matrix, representing the reactive power transfer capacity between node i and node j, and let x be the initial power flow state vector, including the voltage magnitude and phase angle of all nodes. This represents the total number of nodes in the power grid.

[0030] A Jacobian matrix J(x) is constructed using a sparse matrix, storing only non-zero elements. In each iteration, if the change in the system state is less than the preset average change, the previous Jacobian matrix is ​​reused; otherwise, only a portion of the submatrices are updated to reduce matrix construction overhead. The initial power flow state vector is iteratively updated based on the improved Newton-Raphson iterative method until the preset maximum number of iterations is reached, at which point convergence is achieved. The improved Newton-Raphson iteration method has the following specific calculation formula:

[0031] In the formula, Let be the initial power flow state vector at the nth iteration. This is the initial power flow state vector at the (n+1)th iteration. Let n be the node power imbalance function in the nth iteration. For the nth iteration, denoted as the Jacobian matrix of the nodal power imbalance function with respect to the initial power flow state variables.

[0032] Once the iteration converges, the voltage magnitude and phase angle of each node, the active and reactive power flow of each line, and the power balance state of the entire system are obtained, forming a complete power flow calculation result.

[0033] It should be noted that in each iteration, the power imbalance in the current state is calculated. Using the Jacobian matrix Describe the linear approximation of the power equation and obtain the correction amount. Then use Update the state to gradually approach the solution that makes the power imbalance function zero.

[0034] By improving the Newton-Raphson iterative method for power flow calculation, computational efficiency and convergence stability can be significantly improved while maintaining computational accuracy. Its function lies in precisely combining the mapped, standardized input data with the whole-network power flow monitoring framework to quickly solve for the voltage magnitude and phase angle of each node, the active and reactive power flow of each branch, and the power balance state of the entire system. Its advantages include: reducing the computational load of large-scale power grid calculations by utilizing sparse matrix techniques and a partial Jacobian update mechanism, improving algorithm efficiency, and meeting real-time monitoring requirements.

[0035] S2, Based on the monitoring model, a new energy output scenario library is constructed using probabilistic statistical methods. Specifically, this includes quantifying the uncertainty of new energy output based on the real-time monitoring model and constructing a variety of new energy output scenario libraries using probabilistic statistical methods.

[0036] In this embodiment, the uncertainty of new energy output is quantified based on a real-time monitoring model, and a database of various new energy output scenarios is constructed using probabilistic statistical methods, specifically: Based on the real-time monitoring model, a first dataset related to new energy is extracted. The first dataset includes the historical power curve and real-time output data of each new energy unit. The probability distribution of the new energy output is obtained by fitting the probability distribution of the first dataset. The probability distribution includes the marginal probability distribution and the joint probability distribution, which are used to quantify the randomness and fluctuation range of the output. For the output of a single unit, a Gaussian mixture model or a non-parametric kernel density estimation method can be used to obtain its marginal probability distribution. For the combined output of multiple units, a Copula function can be used to construct the combined probability distribution. An initial set of random scenarios is generated from the probability distribution using a random sampling method, with each scenario corresponding to the power output status of each new energy unit; The random sampling method may be Monte Carlo sampling, Latin hypercube sampling, or importance sampling. The initial random scene set is subjected to dimensionality reduction and clustering to select representative scenes, and each scene is assigned an initial probability weight to represent its probability of occurrence. Among them, the PCA principal component analysis method is used to extract the most important power fluctuation patterns (the core patterns that describe the overall trend of new energy output and can explain most of the fluctuations, which are used to reduce complexity and retain key system fluctuation information). K-means clustering, spectral clustering or hierarchical clustering algorithms are applied to the dimensionality-reduced data to select representative scenarios, and the original power curves corresponding to each cluster center are retained as representative scenarios. Based on the aforementioned representative scenarios, and combined with a real-time monitoring model, multi-scenario power flow calculations and verifications are performed, and node voltages, branch power flow, and system power balance status are output. For each scenario, the improved Newton-Raphson method is used to calculate the power flow of the entire network, outputting the voltage amplitude and phase angle of each node, the active and reactive power flow of each branch, and the power balance status of the entire system. The feasibility of the scenario is verified, including voltage over-limit, line overload, and power balance constraints, and scenarios that do not meet the safety constraints are eliminated. Based on the power flow calculation results of multiple scenarios, a library of multiple new energy power output scenarios is generated. The scenario library includes new energy power output and corresponding power flow states at different time scales.

[0037] It should be noted that by quantifying the uncertainty of renewable energy output and generating multiple representative output scenarios, the grid dispatch and operation can fully consider the impact of renewable energy power fluctuations. Specifically, this approach has the following advantages: First, probability distribution fitting accurately describes the randomness of each unit's output; second, the representative scenarios generated through random sampling and dimensionality reduction clustering retain the main fluctuation trends of the system while significantly reducing computational complexity; third, combined with real-time power flow calculations, power balance and safety constraint verification are performed on each scenario to ensure that the generated scenarios are both feasible and representative; finally, the resulting library of multiple renewable energy output scenarios provides real-time and reliable data support for grid dispatch optimization, risk assessment, and transmission capacity calculation, significantly improving the safety, reliability, and flexibility of the grid under conditions of high renewable energy output, while reducing reliance on offline forecasting and manual experience.

[0038] S3. Based on the scenario library, obtain the critical operating state of the power grid under various probability scenarios, and form a mapping relationship between transmission capacity and scenario. Specifically, this includes obtaining the critical operating state of the power grid under various scenarios based on a database of multiple new energy output scenarios, determining the instantaneous maximum transmission capacity of each transmission channel, and forming a scenario-capacity mapping table.

[0039] In this embodiment, the critical operating states of the power grid under various scenarios are obtained based on a database of multiple new energy power output scenarios, specifically as follows: Power flow calculations are performed on each representative scenario in the database of multiple new energy power output scenarios to obtain dynamic operating data of each scenario in continuous time series. The dynamic operating data includes time series data of node voltage amplitude, phase angle and active and reactive power flow of each branch. The power flow calculation is performed by the improved Newton-Raphson iterative method. For each scenario, dynamic margin indicators are calculated based on dynamic operating data. These dynamic margin indicators include node voltage dynamic margin, branch power flow dynamic margin, and system power balance deviation dynamic margin. Among them, the node voltage dynamic margin is the difference between the node voltage and the allowable upper and lower limits at each time step, and the minimum margin is recorded; the branch power flow dynamic margin is the difference between the branch power and the rated capacity at each time step, and the maximum occupancy ratio and extreme values ​​are recorded; the system power balance deviation dynamic margin is the extreme value and rate of change of the difference between the total system generation, load and loss at each time step. Identify the critical operating state in each scenario based on the dynamic margin index; The dynamic margin indicators for each scenario are statistically analyzed according to the time series, and key indicators are extracted. The key indicators include minimum margin, maximum occupancy ratio, rate of change, and extreme fluctuation time points. Based on key indicators, the critical operating states in each scenario are identified. These critical operating states include the moment when a node or branch has the minimum margin or maximum occupancy ratio, the time when the system power balance deviation reaches the preset critical threshold, and the instantaneous over-limit of voltage or power flow that may be caused by short-term rapid fluctuations.

[0040] It should be noted that the purpose of identifying critical operating states is to discover in advance the extreme operating conditions that the power grid may reach under conditions of high fluctuations in new energy sources. Dynamic margin indicators can accurately quantify the time and magnitude at which node voltages, branch power flows, and system power balance approach the critical point. This not only helps to identify the weakest link and potential risks in the system, but also provides reliable data for subsequent dispatch optimization and transmission capacity adjustment. This enables the power grid to take timely measures in the face of rapid output fluctuations or extreme scenarios, improving safety, stability, and operational flexibility, while reducing the risk of accidents caused by overload or voltage exceeding limits.

[0041] Furthermore, the instantaneous maximum transmission capacity of each transmission channel is determined, forming a scenario-capacity mapping table, specifically: Based on each scenario and its corresponding critical operating state, the transmission channels are analyzed one by one. The transmission power of each channel is gradually increased or decreased until a preset limit condition is reached, thus obtaining the instantaneous maximum transmission capacity of each channel in the current scenario. Figure 3 As shown; The preset limit condition is to gradually increase the transmission power until any node or branch reaches its safety limit, while keeping the active and reactive power of the system within the preset allowable deviation range and keeping the voltage of each node within the allowable safety range, with the branch power flow margin as the constraint.

[0042] The analysis operation is repeated for all transmission channels in each scenario to form the transmission capacity matrix corresponding to that scenario. The matrix records the instantaneous maximum transmission capacity of each transmission channel in that scenario. Based on the initial probability weights for each scenario, a conditional probability correction model is constructed, and the corrected probability weights are output. The correction model adopts the Bayesian conditional probability formula. The specific calculation formula for the conditional probability correction model is as follows:

[0043] In the formula, Scenario Y under the current power grid operating state Corrected probability weights, In the scene The conditional probability of the running state Y occurring. These are the initial probability weights. This is for normalization processing of all scenarios.

[0044] The power grid operating status includes the voltage amplitude and phase angle of each node, the power flow of each branch, active / reactive power, load demand, and the output level of each new energy unit.

[0045] The corrected probability weights are replaced with the initial probability weights and combined with the transmission capacity matrix corresponding to each scenario. The weighted average method is used to obtain the corrected weighted transmission capacity matrix. The specific calculation formula for the modified weighted transmission capacity matrix is ​​as follows:

[0046] In the formula, This is the weighted transmission capacity matrix after probability correction. For the scene The power transmission capacity matrix below.

[0047] A scenario-capacity mapping table is generated based on the modified weighted transmission capacity matrix.

[0048] It should be noted that generating a scenario-capacity mapping table can associate the instantaneous maximum capacity of transmission channels under each scenario with the probability of scenario occurrence and operating status. This provides a quantitative basis for adjusting renewable energy output strategies and grid operation modes, enabling the dispatching system to accurately determine the grid's transmission limits under different operating conditions, thereby achieving joint optimization. Based on the mapping table, the system can dynamically adjust transmission limits, adaptively updating them according to changes in renewable energy output and grid status. This ensures both the safe and stable operation of the grid and maximizes the absorption capacity of renewable energy, improving overall operational efficiency and flexibility.

[0049] S4, based on the real-time operating status matching mapping relationship, corrects and jointly optimizes to derive adaptive transmission limits, specifically including: S41: Obtain the real-time power grid operating status and match and correct it with the scenario-capacity mapping table to obtain the corrected actual available transmission capacity under the current power grid operating conditions. S42, and jointly optimizes the new energy output adjustment strategy and grid operation mode to generate a transmission limit that is adaptive to the grid status.

[0050] Specifically, S41 involves acquiring the real-time power grid operating status and matching and correcting it with the scenario-capacity mapping table to obtain the corrected actual available transmission capacity under the current power grid operating conditions. The real-time power grid operating status is obtained and a high-dimensional state vector is constructed. The real-time power grid operating status includes the voltage amplitude and phase angle of each node, the power flow of each branch, active / reactive power, load demand, and the output level of each new energy unit. Extract the corresponding historical state vector for each historical scenario in the scenario-capability mapping table; The current high-dimensional state vector and the historical state vector are initially screened. The multidimensional similarity measure is obtained by calculating Mahalanobis distance. The multidimensional similarity measures are sorted from smallest to largest, and the scenes with the smallest distance are selected as a subset. For the transmission capacity data in the aforementioned subset, a preliminary transmission capacity is predicted using a multiple linear regression model, specifically as follows: The power grid operation status of each scenario in the subset is paired with the corresponding transmission capacity matrix to form a training dataset; A multiple linear regression model was constructed, with transmission capacity as the dependent variable and various elements of the power grid operation status as independent variables, to establish the regression model. The regression model is specifically calculated using the following formula:

[0051] In the formula, For the initial transmission capacity forecast, The intercept is... For regression coefficients, Let j be the characteristic value of the power grid operation state in the i-th scenario. For residuals, The dimension is the state vector.

[0052] Use the least squares method to analyze the model parameters. , , ..., Training is performed to minimize the error between the predicted values ​​and the actual transmission capacity, thereby obtaining the optimal model parameters; By inputting the trained regression model with the current real-time power grid operating status, the preliminary power transmission capacity can be obtained.

[0053] The initial power transmission capacity is input into a pre-trained random forest regression model, and the corresponding power transmission capacity is predicted independently by each regression tree to obtain a set of prediction values ​​for a single tree. Among them, the pre-trained random forest regression model uses the state vectors of a subset of historical scenarios and the corresponding power transmission capacity as training data to train multiple regression decision trees, and generates different training samples through bootstrap to ensure the diversity and robustness of the model. The final power transmission capacity prediction value is obtained by averaging the prediction results of all decision trees and outputting the random forest regression model. The specific calculation formula for the predicted final transmission capacity is as follows:

[0054] In the formula, This represents the final predicted transmission capacity value from the random forest regression model. This represents the total number of decision regression trees included in the random forest. This is the predicted output of the j-th regression tree for the current input state vector.

[0055] Based on the variance of the predicted final transmission capacity of each tree The confidence interval of the random forest output is corrected to obtain the corrected actual usable transmission capacity with uncertainty boundary.

[0056] The variance is calculated using the following formula:

[0057] The specific calculation formula for the corrected actual available transmission capacity is as follows:

[0058] In the formula, To correct for the actual available transmission capacity, Given a preset confidence coefficient, find the corresponding quantile in the standard normal distribution table according to the required confidence level. denoted as the standard deviation of the predicted values ​​for each tree in the random forest.

[0059] It should be noted that, taking into account both the current real-time operating status of the power grid and the high-dimensional characteristics of historical scenarios, a preliminary prediction was made using a multiple linear regression model, which was further optimized using a random forest regression model. Simultaneously, the prediction uncertainty was quantified and corrected using confidence intervals, resulting in a more robust and reliable transmission capacity estimate. This method effectively reflects the dynamic operating characteristics of the power grid under different renewable energy output conditions, ensuring more scientific and reasonable constraints on transmission channels under high volatility and uncertainty conditions. It provides a reliable basis for the joint optimization of subsequent renewable energy output adjustment strategies and power grid operation modes, improving the power grid's safety margin and renewable energy absorption capacity, while enhancing the real-time performance and accuracy of dispatching decisions.

[0060] Among them, S42, and jointly optimize the new energy output adjustment strategy and grid operation mode to generate a transmission limit that is adaptive to the grid state, specifically: A joint optimization objective function is constructed, which includes the total power loss of each system, the renewable energy output and consumption, and the dynamic margin index of grid security. Constraints are set, and the confidence interval of the modified actual available transmission capacity is used to constrain the instantaneous maximum transmission capacity of each transmission channel. The joint optimization objective function is calculated using the following formula:

[0061] In the formula, To minimize the joint optimization objective function, , , These are the weighting coefficients, For the total power loss of the system, To contribute to the consumption of renewable energy, It serves as a dynamic margin indicator for power grid security.

[0062] The constraints include node voltage, branch power flow, and power balance constraints, specifically:

[0063]

[0064]

[0065] In the formula, Voltage amplitude, , These are the minimum and maximum allowable values ​​for the node voltage, respectively. Let j be the active power flow of the j-th transmission line (branch) in the power grid. Let j be the corrected actual available transmission capacity of the j-th transmission line, and let j be the actual available transmission capacity of the j-th transmission line. The total active power output of all generators (conventional units and new energy sources), This represents the sum of active power at all load nodes.

[0066] Based on the rolling time-domain optimization algorithm, the current power grid operating state and constraints are input, and the joint optimization objective function is solved to obtain the optimal power output adjustment strategy for each new energy unit and the optimal adjustment scheme for the power grid operating mode. The adaptive limit for each transmission channel is calculated based on the joint optimization results. Its value is dynamically updated according to the fluctuation of new energy output and the changes in grid status, reflecting the instantaneous carrying capacity of the grid under the current operating conditions. The process involves obtaining the solution results of the joint optimization model, including the optimal output adjustment of each new energy unit and the grid operation mode adjustment scheme. The optimization results are then used to perform power flow calculations using an improved Newton-Raphson iterative method. The power flow distribution and node voltage of each branch are recalculated to obtain the instantaneous carrying capacity of each transmission channel under the current grid conditions. The aforementioned instantaneous carrying capacity is combined with the confidence interval of the corrected actual available transmission capacity to obtain an adaptive limit. The adaptive limit is recorded as the transmission channel constraint value for reference by the dispatch control system. When the output of new energy units or the grid conditions change, the limit is dynamically updated to ensure that the limit adapts to the grid operating conditions. The adaptive transmission limit is sent to the dispatch and control center in real time, and a closed-loop update mechanism is established: based on the new real-time operating status, it is updated periodically or by event triggering to ensure that the limit is always consistent with the grid status, thereby realizing dynamic adjustment and safe operation of transmission capacity.

[0067] This invention constructs a high-precision, dynamically updated real-time monitoring model of the power flow distribution status of the entire network by collecting real-time operating data of new energy sources and conventional generating units, combined with the power grid topology and operating mode. Particle filtering is used to achieve precise matching between standardized input data and the node-branch network model, ensuring the accuracy and robustness of power flow calculations. An improved Newton-Raphson iterative method is used for efficient power flow solution, improving the computational speed and convergence stability of large-scale power grids. Furthermore, the uncertainty of new energy output is quantified by generating a representative scenario library through probability distribution fitting, random sampling, and dimensionality reduction clustering. Power flow verification is performed on each scenario to ensure its feasibility and representativeness.

[0068] By employing a high-precision, dynamically updatable power flow monitoring model and particle filter data matching, the system achieves a precise correspondence between input data and the node-branch network model, ensuring the accuracy and robustness of power flow calculations. It quantifies the uncertainties of new energy sources based on various output scenarios and identifies critical operating states under each scenario, forming a scenario-capacity mapping table to provide a scientific basis for the instantaneous carrying capacity of transmission channels. Furthermore, by combining multiple linear regression and random forest regression models with confidence interval correction, it obtains robust and reliable practically available transmission capacity. Finally, through joint optimization and rolling time-domain algorithms, it generates transmission limits that adapt to the grid conditions, achieving dynamic adjustment and safe operation of transmission capacity.

[0069] Reference Figure 2 As shown in the diagram, this invention provides a schematic diagram of a regional power grid predictive dispatch system based on topology balance, including a model building module, a scenario library building module, a mapping table generation module, and a transmission limit generation module. These modules are interconnected. The model building module is used to build a monitoring model that reflects the real-time power flow of the entire network. The scenario library construction module is used to construct a new energy output scenario library based on the monitoring model and combined with probabilistic statistical methods; The mapping table generation module is used to obtain the critical operating state of the power grid under each probability scenario based on the scenario library, form a mapping relationship between transmission capacity and scenario, and generate a scenario-capacity mapping table by performing weighted averaging on the transmission capacity matrix after correcting the probability weights. The power transmission limit generation module is used to match the mapping relationship based on the real-time operating status, correct and jointly optimize to obtain an adaptive power transmission limit. The correction is performed by calculating the variance to correct the confidence interval, and the optimization is performed by constructing a multi-objective joint optimization function and constraints by combining the confidence interval correction.

[0070] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0071] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0072] Those skilled in the art will recognize that the modules 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 implementation should not be considered beyond the scope of this application.

[0073] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0074] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0075] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A predictive dispatch method for regional power grids based on topology balance, characterized in that, include: Establish a monitoring model that reflects the real-time power flow of the entire network; Based on the monitoring model, a new energy output scenario library is constructed using probabilistic statistical methods; Based on the scenario library, the critical operating state of the power grid under each probability scenario is obtained, forming a mapping relationship between transmission capacity and scenario. The mapping relationship is obtained by performing a weighted average of the transmission capacity matrix by correcting the probability weights, and generating a scenario-capacity mapping table. Based on the real-time operating status matching mapping relationship, the adaptive transmission limit is obtained by correction and joint optimization. The correction is achieved by calculating the variance to correct the confidence interval, and the optimization is achieved by constructing a multi-objective joint optimization function and constraints by combining the confidence interval correction.

2. The regional power grid predictive dispatch method based on topology balancing according to claim 1, characterized in that, The establishment of a monitoring model reflecting the real-time power flow of the entire network specifically includes: Real-time power grid data is preprocessed to form a standardized input dataset; A node-branch network model is constructed based on the power grid topology and dynamically assigned values ​​to form a power flow monitoring framework. The standardized input dataset is matched and mapped with parameters, and then fed into the power flow monitoring framework. The improved Newton-Raphson iterative method is used to calculate the power flow. Based on the calculation results, a real-time power flow monitoring model for the entire network is generated, which includes node voltage, branch power flow, and system power status.

3. The regional power grid predictive dispatch method based on topology balancing according to claim 2, characterized in that, The parameter matching and parameter mapping of the standardized input data specifically involves: Initialize the normalized input data into a set of particles and assign initial weights; Predicted states of each particle are calculated based on the power grid topology. Acquire real-time measurement data and compare it with the predicted state of the particles, calculate the updated weight of each particle, and resample the particle set; The optimal particle state is obtained by weighted averaging the resampled particle set and then mapped to the node-branch network model.

4. The regional power grid predictive dispatch method based on topology balancing according to claim 3, characterized in that, The improved Newton-Raphson iterative method is used for power flow calculation, specifically as follows: Construct an initial power flow state vector based on the matched standardized input dataset; Based on the power grid topology and the whole-network power flow monitoring framework, establish the whole-network power flow nonlinear equation; The Jacobian matrix is ​​constructed using sparse matrices, and the initial power flow state vector is solved iteratively to output power flow calculation results including node voltages, line power flow, and system power.

5. The regional power grid predictive dispatch method based on topology balancing according to claim 4, characterized in that, The construction of a new energy output scenario library based on the monitoring model and combining probabilistic statistical methods specifically includes: The first dataset is extracted based on the real-time monitoring model, and a probability distribution is fitted to the first dataset. An initial set of random scenes is generated from a probability distribution using a random sampling method. The initial random scene set is subjected to dimensionality reduction and clustering to select representative scenes and assign initial probability weights to each scene; Based on representative scenarios, multi-scenario power flow calculations and verifications are performed to generate a new energy output scenario library that includes node voltage, branch power flow, and system power balance status.

6. The regional power grid predictive dispatch method based on topology balancing according to claim 5, characterized in that, The process of obtaining the critical operating state of the power grid under each probability scenario based on the scenario library specifically includes: Time-series power flow calculations are performed on representative scenarios in the new energy power output scenario library to obtain dynamic operation data; The dynamic margin index for each scenario is calculated based on the dynamic operating data. Identify critical operating states in various scenarios based on dynamic margin indicators.

7. The regional power grid predictive dispatch method based on topology balancing according to claim 1, characterized in that, The mapping relationship is obtained by weighting the transmission capacity matrix with corrected probability weights to generate a scenario-capacity mapping table. The specific steps are as follows: Analyze the maximum power transmission capacity of power transmission channels under various scenarios to form a scenario power transmission capacity matrix; A conditional probability correction model is constructed based on the initial probability weights for each scenario, and the corrected probability weights are output through the probability correction model. The corrected probability weights are combined with the transmission capacity matrix corresponding to each scenario, and the weighted average method is used to obtain the corrected weighted transmission capacity matrix. A scenario-capacity mapping table is generated based on the modified weighted transmission capacity matrix.

8. The regional power grid predictive dispatch method based on topology balancing according to claim 7, characterized in that, The adaptive transmission limit is derived by matching the mapping relationship based on the real-time operating status, correcting and jointly optimizing the result. The correction step specifically includes: A high-dimensional state vector is constructed based on the real-time running status, and the state vector corresponding to the historical scene is extracted. The current high-dimensional state vector is initially screened by comparing it with the corresponding historical state vectors. Based on the Mahalanobis distance calculation method, several scenarios with the smallest distance are selected as a subset. The transmission capacity data in the subset is predicted using a multiple linear regression model to obtain the preliminary transmission capacity; The initial power transmission capacity is input into a pre-trained random forest regression model, and the final power transmission capacity prediction value is obtained through multi-decision tree ensemble prediction. Variance is calculated based on the final predicted transmission capacity, and confidence intervals are corrected based on the variance to output the corrected actual available transmission capacity.

9. The regional power grid predictive dispatch method based on topology balancing according to claim 8, characterized in that, The optimization steps for obtaining the adaptive transmission limit by matching the mapping relationship based on the real-time operating status and then modifying and jointly optimizing it are as follows: A multi-objective joint optimization function is constructed, which includes dynamic margin indicators of total system power loss, renewable energy output and consumption, and grid security. Constraints are set, and the confidence interval of the modified actual available transmission capacity is used to constrain the instantaneous maximum transmission capacity of each transmission channel. The optimization function is solved using a rolling time-domain optimization algorithm to obtain the optimal adjustment scheme for new energy output strategy and power grid operation mode; The adaptive limit for each transmission channel is calculated and sent to the dispatching system, and a closed-loop update mechanism is established to realize the dynamic adjustment of transmission capacity.

10. A system using the topology-balanced regional power grid predictive dispatch method as described in any one of claims 1-9, characterized in that, include: The model building module is used to build a monitoring model that reflects the real-time power flow of the entire network. The scenario library construction module is used to construct a new energy output scenario library based on the monitoring model and combined with probabilistic statistical methods; The mapping table generation module is used to obtain the critical operating state of the power grid under each probability scenario based on the scenario library, form a mapping relationship between transmission capacity and scenario, and generate a scenario-capacity mapping table by performing weighted averaging on the transmission capacity matrix after correcting the probability weights. The power transmission limit generation module is used to match the mapping relationship based on the real-time operating status, correct and jointly optimize to obtain an adaptive power transmission limit. The correction is performed by calculating the variance to correct the confidence interval, and the optimization is performed by constructing a multi-objective joint optimization function and constraints by combining the confidence interval correction.

Images