A virtual power plant optimization scheduling method, device and medium

Abstract

The application discloses a kind of virtual power plant optimization scheduling method, equipment and medium, it is related to virtual power plant optimization technical field, including, according to power resource operation preprocessing data, establish dynamic correlation graph, with dynamic correlation graph is carried out space-time alignment and coupling analysis to power grid state preprocessing data, obtain dynamic credible capacity;Dynamic credible capacity is used as power operation constraint, and with the lowest virtual power plant total operation cost as optimization goal, constructs MILP optimization model and carries out solution, generates power scheduling scheme;According to power grid state preprocessing data, the reactive power demand intensity of each power node is calculated, and reactive power support demand thermodynamic atlas is generated, and the active and reactive power decision variables of power scheduling scheme and reactive power support demand thermodynamic atlas are solved by particle swarm optimization algorithm, and the power resource scheduling instruction set is output;The application improves the scheduling efficiency, reduces the operation cost, enhances the overall economy and flexibility of virtual power plant.

Description

Technical Field

[0001] This invention relates to the field of virtual power plant optimization technology, and in particular to a virtual power plant optimization scheduling method, equipment and medium. Background Technology

[0002] In current power systems, with the widespread application of distributed energy resources, Virtual Power Plants (VPPs), as a management platform integrating multiple distributed power resources, have gradually become one of the key technologies in modern power dispatching. The core objective of VPPs is to achieve flexible resource dispatching by integrating different types of distributed energy resources (such as wind power, solar power, and battery storage), thereby improving the operational efficiency and economy of the power system. In recent years, with the development of information and communication technologies, VPPs have made significant progress in real-time monitoring, data analysis, and dispatch optimization. Many optimization algorithms have been applied to power dispatching problems, especially mathematical optimization models such as linear programming (LP) and mixed-integer linear programming (MILP). These methods have effectively improved the quality of power dispatching schemes.

[0003] Existing virtual power plant dispatching methods still face certain challenges. When processing operational data from different power resources, current technologies often neglect dynamic correlation and spatiotemporal coupling analysis, which may lead to low resource dispatching efficiency or high operating costs. To address this, our invention establishes a dynamic correlation graph and combines it with spatiotemporal alignment and coupling analysis to more accurately calculate the dynamic reliable capacity of the power grid. This capacity serves as a constraint on power operation, and dispatching schemes are optimized based on this, thereby improving the accuracy and efficiency of dispatching. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a virtual power plant optimization scheduling method to solve the problem that the prior art has failed to fully consider dynamic correlation and spatiotemporal coupling analysis.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a virtual power plant optimized scheduling method, which includes collecting distributed power resource operation data and power grid operation parameters, performing preprocessing, and outputting power resource operation preprocessing data and power grid status preprocessing data;

[0008] Based on the preprocessed data of power resource operation, a dynamic correlation map is established. The preprocessed data of power grid status and the dynamic correlation map are spatiotemporally aligned and coupled to obtain dynamic reliable capacity.

[0009] Dynamic reliable capacity is used as a power operation constraint, and the minimum total operating cost of virtual power plants is used as the optimization objective. A MILP optimization model is constructed and solved to generate a power dispatch scheme.

[0010] The reactive power demand intensity of each power node is calculated based on the preprocessed data of the power grid status, and a reactive power support demand heat map is generated. The active and reactive power decision variables of the power dispatch scheme and the reactive power support demand heat map are solved collaboratively by the particle swarm optimization algorithm, and the power resource dispatch instruction set is output.

[0011] Through a secure and encrypted communication channel, the power resource dispatch instruction set is sent to the local controller of each distributed resource, and the execution of the instructions is monitored in real time.

[0012] As a preferred embodiment of the virtual power plant optimization scheduling method of the present invention, the distributed power resource operation data includes new energy power operation data, energy storage power status data and load power demand data;

[0013] The power grid operating parameters include node voltage, current, active power, reactive power, and switch status.

[0014] As a preferred embodiment of the virtual power plant optimized scheduling method of the present invention, the specific steps for outputting power resource operation preprocessing data and grid status preprocessing data are as follows:

[0015] The distributed power resource operation data is denoised, cleaned, synchronized in time, and normalized to output preprocessed power resource operation data.

[0016] The system performs time-series correction, filtering and smoothing, normalization and baseline adjustment on the power grid operating parameters, and outputs preprocessed power grid status data.

[0017] As a preferred embodiment of the virtual power plant optimization scheduling method described in this invention, the following steps are taken: Based on the preprocessed power resource operation data, a dynamic correlation graph is established; the preprocessed power grid status data and the dynamic correlation graph are then subjected to spatiotemporal alignment and coupling analysis to obtain dynamic reliable capacity.

[0018] By using the dynamic time warping method, time series similarity analysis is performed on the preprocessed data of power resource operation to extract the dynamic correlation between different preprocessed data of power resource operation.

[0019] Using preprocessed power resource operation data as graph nodes, edges between graph nodes are constructed based on dynamic relationships to form a dynamic relationship graph;

[0020] The power grid state preprocessing data and the dynamic correlation map are aligned in time and space, and the spatiotemporal coupling relationship is extracted by calculating the correlation between the power grid state preprocessing data and the dynamic correlation map.

[0021] Based on the spatiotemporal coupling relationship, the mutual influence between the power grid status and the operation data of distributed power resources is analyzed, and the dynamic reliable capacity is calculated.

[0022] As a preferred embodiment of the virtual power plant optimization scheduling method described in this invention, the following steps are taken: Dynamic reliable capacity is used as a power operation constraint, and the optimization objective is to minimize the total operating cost of the virtual power plant. A MILP optimization model is constructed and solved to generate a power scheduling scheme.

[0023] Dynamic reliable capacity is used as a power operation constraint to define the upper limit constraint of the grid connection point power of the virtual power plant;

[0024] With the goal of minimizing the total operating cost of the virtual power plant, an objective function for optimizing electricity costs is defined.

[0025] Based on the upper limit constraint of grid connection power and the objective function of electricity cost optimization, a MILP optimization model is constructed;

[0026] A mathematical programming solver is used to numerically solve the MILP optimization model to obtain the active power setpoints for each distributed resource; the active power setpoints are then arranged in a time series to generate a power dispatching scheme.

[0027] As a preferred embodiment of the virtual power plant optimized scheduling method of the present invention, the specific steps for calculating the reactive power demand intensity of each power node based on the preprocessed power grid state data and generating a reactive power support demand heat map are as follows.

[0028] Reactive demand intensity is calculated from the preprocessed power grid status data to obtain the reactive demand intensity of each power node.

[0029] The reactive power demand intensity of each power node is mapped to the spatial location of the power grid and converted into a continuous power grid area through Kriging interpolation, forming an initial reactive power support demand heat map.

[0030] The reactive power demand intensity at different locations in the initial reactive power support demand heatmap is marked by color intensity and then smoothed to generate a reactive power support demand heatmap.

[0031] As a preferred embodiment of the virtual power plant optimization scheduling method described in this invention, the step of collaboratively solving the active and reactive power decision variables of the power scheduling scheme and the reactive power support demand heatmap using the particle swarm optimization algorithm to output a power resource scheduling instruction set is as follows:

[0032] The active and reactive power decision variables of the power dispatch scheme and the reactive power support demand heat map are solved by the particle swarm optimization algorithm. At the same time, the active power allocation in the power dispatch scheme and the reactive power allocation in the reactive power support demand heat map are optimized to obtain the optimal active and reactive power decision variables.

[0033] The optimal active and reactive power decision variables are arranged in time series and the constraints are verified to generate a set of power resource dispatch instructions.

[0034] As a preferred embodiment of the virtual power plant optimized scheduling method of the present invention, the step of distributing the power resource scheduling instruction set to the local controller of each distributed resource through a secure encrypted communication channel and monitoring the execution of the instructions in real time includes the following specific steps.

[0035] The power resource dispatch instruction set is encrypted and then sent to the local controller of each distributed power resource through a secure encrypted communication channel.

[0036] After receiving the power resource dispatch instruction set, the local controller decrypts and verifies the contents of the power resource dispatch instruction set, executes the power dispatch task according to the power resource dispatch instruction set, and monitors the execution status of power resources in real time.

[0037] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the virtual power plant optimization scheduling method as described in the first aspect of the present invention.

[0038] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the virtual power plant optimization scheduling method as described in the first aspect of the present invention.

[0039] The beneficial effects of this invention are as follows: By establishing a dynamic correlation graph and performing spatiotemporal alignment analysis, the dynamic changes of power resources and grid status are accurately captured, realizing the dynamic and reliable capacity calculation of the grid, providing reliable power operation constraints, and improving the resource management capability and grid stability of the dispatching system; at the same time, the particle swarm optimization algorithm is used to collaboratively optimize the active and reactive power decision variables, optimize the power dispatching scheme, ensure the rational allocation of active and reactive power, improve dispatching efficiency, reduce operating costs, and enhance the overall economy and flexibility of the virtual power plant. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the 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.

[0041] Figure 1 A flowchart for optimizing the scheduling method for virtual power plants.

[0042] Figure 2 A flowchart for constructing a dynamic association graph and aligning it with the spatiotemporal context.

[0043] Figure 3 A flowchart for solving the MILP optimization model and generating the scheduling scheme.

[0044] Figure 4 A flowchart for active-reactive power collaborative optimization and command issuance monitoring. Detailed Implementation

[0045] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0046] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0048] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a virtual power plant optimized scheduling method, including the following steps:

[0049] S1. Collect distributed power resource operation data and power grid operation parameters, perform preprocessing, and output power resource operation preprocessing data and power grid status preprocessing data.

[0050] It should be noted that distributed power resource operation data includes new energy power operation data, energy storage power status data, and load power demand data.

[0051] Power grid operating parameters include node voltage, current, active power, reactive power, and switch status.

[0052] S1.1 Perform noise reduction, data cleaning, time synchronization and normalization on the distributed power resource operation data, and output preprocessed power resource operation data.

[0053] Furthermore, methods such as filtering or wavelet decomposition are used to eliminate measurement noise in the distributed power resource operation data. Data cleaning is completed through outlier removal, missing value imputation, and consistency verification. The distributed power resource operation data is synchronized with time according to a unified time base to ensure that various types of distributed power resource operation data are aligned on the same time axis. Distributed power resource operation data with different dimensions and value ranges are normalized to obtain preprocessed power resource operation data.

[0054] S1.2 Perform time-series correction, filtering and smoothing, normalization and baseline adjustment on the power grid operating parameters, and output preprocessed power grid status data.

[0055] Furthermore, based on the clock deviation and frame loss, the power grid operating parameters are time-series corrected to ensure that data such as voltage, current, and power are aligned on a unified time axis; low-pass filtering or moving average methods are used to remove instantaneous fluctuations in distributed power resource operating data to achieve data smoothing; distributed power resource operating data of different dimensions are normalized to ensure comparability between parameters; baseline adjustment is performed based on historical statistical values ​​to correct long-term offsets and obtain preprocessed power grid status data.

[0056] S2. Based on the preprocessed data of power resource operation, establish a dynamic correlation map, and perform spatiotemporal alignment and coupling analysis on the preprocessed data of power grid status and the dynamic correlation map to obtain dynamic reliable capacity.

[0057] It should be noted that existing methods typically analyze distributed power resource operation data and grid status data separately, relying on static statistical indicators or simple correlation analysis to describe the relationship between resources and the grid operation status. They lack consideration for the dynamic changes in time series and do not systematically integrate the interactive features of time and space dimensions into quantifiable dynamic capacity indicators. Therefore, they cannot accurately reflect the nonlinear time-series dependence between resources and the comprehensive impact of grid status on the adjustable capacity of distributed resources.

[0058] Our invention uses preprocessed power resource operation data as graph nodes and establishes a dynamic correlation graph based on dynamic relationships. It aligns the preprocessed power grid status data with the dynamic correlation graph in both time and space. By extracting continuous connections across time dimensions and interaction patterns across spatial regions through stable and statistically verified correlation sequences, it quantifies the dynamic reliable capacity of each distributed power resource operation preprocessed data within each time window. This invention achieves precise quantification of power resource operation constraints, improves the reliability and security of virtual power plant scheduling, and simultaneously considers the volatility of renewable energy and the flexibility of energy storage.

[0059] S2.1. Using the dynamic time warping method, time series similarity analysis is performed on the preprocessed data of power resource operation to extract the dynamic correlation between different preprocessed data of power resource operation.

[0060] Furthermore, statistical analysis is performed on the preprocessed data of power resource operation to extract average trend curves or sequences with high fluctuation stability. Combined with representative periods in the operation scenario (such as peak load periods or periods with significant fluctuations in renewable energy output), the sequence that best reflects the overall operation mode is determined as a reference sequence. Based on the parameter sequence, the time-series trajectories of different preprocessed power resource operation data are compared using the dynamic time warping method, allowing nonlinear time stretching and compression to overcome the time-series differences in load, energy storage, and renewable energy output. The cumulative distance or similarity index under the alignment path is calculated using cosine similarity to identify the synchronicity and differences between resources. Based on the similarity threshold, strong or weak correlations between preprocessed power resource operation data are screened and quantified, and dynamic correlation relationships are output.

[0061] It should be noted that the similarity threshold is determined by analyzing the similarity distribution of a large number of samples to determine a reasonable range. Combined with the sensitivity requirements of power grid dispatch to the correlation strength (the stringent requirements for voltage stability and power balance), the boundary point is selected by methods such as clustering and cross-validation.

[0062] Based on the statistical distribution of similarity of historical data samples (e.g., dividing intervals through cluster analysis), and closely combined with the actual sensitivity and reliability requirements of power grid dispatch to correlation strength (e.g., optimizing the boundary point through cross-validation), a comprehensive similarity threshold range (e.g., 0 to 1) is determined.

[0063] S2.2 Using preprocessed power resource operation data as graph nodes, edges are constructed between graph nodes based on dynamic association relationships to form a dynamic association graph.

[0064] Furthermore, each type of power resource operation preprocessing data is identified as an independent graph node, such as new energy power operation preprocessing data, energy storage power status preprocessing data, and load power demand preprocessing data. Based on dynamic association relationships, strong and weak associations between power resource operation preprocessing data are mapped one-to-one, and edges between graph nodes are established according to the quantification results of similarity indicators in dynamic association relationships. For power resource operation preprocessing data that meet the similarity threshold, edges with edge weights are generated between the corresponding graph nodes, where the edge weights represent the dynamic similarity strength between power resource operation preprocessing data. After all graph nodes have completed edge construction, the graph nodes and the edges between them are combined as a whole to form a dynamic association graph that reflects the temporal coupling characteristics between power resource operation preprocessing data.

[0065] S2.3. Align the preprocessed power grid status data with the dynamic correlation map in terms of time and space, and extract the spatiotemporal coupling relationship by calculating the correlation between the preprocessed power grid status data and the dynamic correlation map.

[0066] Furthermore, to unify the time base for preprocessed power grid status data and dynamic correlation graphs, resampling and interpolation are used to map all time series to a common time grid. Graph nodes in the dynamic correlation graph are mapped one-to-one with the corresponding power grid nodes for power grid operating parameters using spatial coordinates. When spatial coordinates are unavailable, nearest neighbor topology mapping or node adjacency mapping is used to analyze the connection relationships or distance similarities of nodes in the power grid topology, performing topology matching to determine the correspondence between graph nodes and power grid nodes in the dynamic correlation graph, completing spatial alignment and obtaining spatial mapping results between nodes. After completing time and spatial alignment, each pair of mapped graph nodes and power grid nodes is aligned... The subsequent time series correlation indices are calculated, including the Pearson correlation coefficient to measure linear correlation, the Granger causality test to identify directional influences between time series, and mutual information to reveal nonlinear dependencies. Statistical tests are used to evaluate whether the correlation indices are statistically significant. Then, unstable or highly random sequences are eliminated by combining the fluctuation characteristics and trend consistency of the time series. The correlation sequences that have been verified in the statistical tests and remain stable in multiple time periods are retained. The correlation sequences between different time periods and spatial nodes are mapped to identify continuous connections across time dimensions and interaction patterns across spatial regions, thus obtaining spatiotemporal coupling relationships.

[0067] It should be noted that the fluctuation characteristics and trend consistency of the time series are obtained by extracting statistical features (such as standard deviation, variance, moving average trend, correlation coefficient, etc.) from the time series of power resource operation data and grid status data. Among them, the fluctuation characteristics reflect the short-term change intensity of the series, and the trend consistency is obtained by comparing the long-term trend direction between series (such as moving average or polynomial fitting trend).

[0068] S2.4. Based on the spatiotemporal coupling relationship, analyze the mutual influence between the power grid status and the operation data of distributed power resources, and calculate the dynamic reliable capacity.

[0069] Furthermore, the voltage, current, active power, and reactive power of each power node in the preprocessed power grid state data are mapped to the preprocessed power resource operation data of the corresponding graph nodes, identifying the response relationships across time periods and spatial nodes. Based on the response relationships, the response characteristics of the preprocessed distributed power resource operation data to changes in the power grid state are analyzed. By using correlation analysis (such as Pearson correlation coefficient) and sensitivity analysis (such as partial derivative sensitivity method or Sobol global sensitivity analysis), the correlation and sensitivity between parameters such as voltage, current, active power, and reactive power and resource output are calculated. The adjustable range and stable output capability of distributed power resources under different power grid states are evaluated. The continuous connection across time dimensions and the interaction mode across spatial regions are accumulated as constraint factors to comprehensively reflect the spatiotemporal coupling characteristics of resources, forming the dynamic reliable capacity of each distributed power resource operation data. The dynamic reliable capacity includes active capacity and reactive capacity.

[0070] The expression for calculating dynamic trust capacity is:

[0071] ;

[0072] In the formula, Distributed power resource operation data In the time window Dynamic Trusted Capacity (Unit and Base Capacity) Consistent, for example, megawatts (MW). Distributed power resource operation data Basic capacity (units such as megawatts, MW); It is a time window Number of time sampling points within; Distributed power resource operation data At the point of time The strength of continuous connection across time dimensions (dimensionless). Distributed power resource operation data At the point of time The intensity of cross-spatial region interaction patterns (dimensionless). It is a time window The Middle A point in time; It is an index variable at a specific point in time; It is a time window; It is an index of distributed power resource operation data; It is the weighting coefficient of the spatial region interaction mode, used to measure the contribution of cross-spatial region interaction mode to dynamic trust capacity. It is dimensionless and can take values ​​from 0 to 1. It is a weighting coefficient for continuous connections across time dimensions, used to measure the contribution of continuous connections across time dimensions to dynamic trust capacity. It is dimensionless and can take values ​​from 0 to 1.

[0073] It should be noted that the weighting coefficients of the spatial region interaction mode are based on the correlation, mutual information or power flow coupling degree between nodes in historical operation data to assess the importance of spatial interaction. Then, combined with the grid operation stability index and dispatch sensitivity analysis results, the relative weights of each spatial region's contribution to dynamic reliable capacity are determined through normalization or multi-objective weighted optimization.

[0074] S3. Using dynamic reliable capacity as a power operation constraint and taking the minimum total operating cost of the virtual power plant as the optimization objective, construct and solve the MILP optimization model to generate a power dispatching scheme.

[0075] S3.1. Using dynamic reliable capacity as a power operation constraint, define the upper limit constraint of the grid connection point power of the virtual power plant.

[0076] Furthermore, the dynamic reliable capacity of each distributed power resource's preprocessed operation data within each time window is used as the adjustable power upper limit of the power operation constraint. All grid connection points of distributed power resources in the virtual power plant are aggregated, and the power of each grid connection point is constrained and mapped according to the time series to form the upper limit constraint of the grid connection point power of the virtual power plant within each time window, ensuring that the power of the grid connection point does not exceed the dynamic reliable capacity of each distributed power resource.

[0077] S3.2. Define the objective function for optimizing electricity costs, with the goal of minimizing the total operating cost of the virtual power plant.

[0078] Furthermore, the active power output of each distributed power resource in each time window is associated with the corresponding generation cost or operating cost in the preprocessed data of each distributed power resource. The operating cost of each distributed power resource is accumulated according to the time series. Taking into account factors such as power purchase cost, energy storage charging and discharging cost, and renewable energy output cost, the operating costs of all distributed power resources in the virtual power plant are summed to form a total cost expression covering the entire scheduling period. The output is a power cost optimization objective function with the goal of minimizing the total operating cost.

[0079] S3.3. Based on the upper limit constraint of grid connection point power and the objective function of power cost optimization, construct the MILP optimization model.

[0080] Furthermore, a mixed-integer linear programming (MILP) method is adopted. The dynamic reliable capacity of each distributed power resource's preprocessed operating data within each time window is used as the upper limit constraint of the grid connection point power. At the same time, the expression of the total operating cost of the virtual power plant is used as the objective function for power cost optimization. By expressing the objective function of power cost optimization as a linear weighted sum of the operating costs of all distributed power resources in each time window, and transforming the upper limit constraint of the grid connection point power into a linear inequality constraint of the output power of each distributed resource, the two are unified into the same system of linear equations or matrix expressions to form a linear programming relationship between the objective function and the constraints. The range of active power values ​​and the type of decision variables (integer or continuous variables) of each distributed power resource are clarified. A complete MILP optimization model is formed through mathematical programming methods.

[0081] It should be noted that Mixed-Integer Linear Programming (MILP) is an optimization method where both the constraints and the objective function are linear, with some decision variables being integers and others being continuous variables. This method is often used to handle optimization problems that simultaneously handle discrete decisions (such as equipment start-up and shutdown states) and continuous control quantities (such as power output). It can obtain the globally optimal or near-optimal solution by minimizing or maximizing the objective function while satisfying linear constraints.

[0082] Training the MILP optimization model involves inputting dynamic credibility capacity, cost parameters, and constraints, defining decision variables (integer and continuous variables), and then using a mathematical programming solver (such as CPLEX, Gurobi, or GLPK) to parametrically solve the MILP optimization model. Branch and bound, cutting plane, or heuristic algorithms are used to iteratively update variable values ​​to narrow the feasible region. In each iteration, the objective function value is calculated and constraint satisfaction is checked until the convergence criterion is met (meaning the improvement in the objective function during the MILP solution process reaches the optimality tolerance (e.g., relative difference less than 10⁻)). 6 Complete the training at the designated time.

[0083] S3.4. The mathematical programming solver is used to numerically solve the MILP optimization model to obtain the active power setpoint of each distributed resource.

[0084] Furthermore, the decision variables (integer or continuous variables), grid connection point power upper limit constraints, and electricity cost optimization objective function of the MILP optimization model are exported and input into the mathematical programming solver in a format acceptable to the solver. In the solver, the accuracy of the solution is controlled by setting the optimality tolerance, time constraints are configured to balance the solution speed and computational load, and strategies such as branch and bound, cutting plane, or interior point methods are selected to adapt to the model characteristics. Simultaneously, the solver parameters and strategies can be configured according to the sparsity of the problem or the constraint structure, enabling relaxation-repair heuristics, preprocessing, or parallel computing modes. For example, setting optimality tolerance, time constraints, branch and bound, cutting plane, and internal point methods are all possible. A bounding strategy or relaxation-repair heuristic is used to balance solution timeliness and solution quality. The mathematical programming solver is then started to perform numerical solutions. The mathematical programming solver seeks the optimal or near-optimal solution that minimizes the total operating cost of the virtual power plant while satisfying the upper limit constraint of the grid connection point power through numerical methods such as branch and bound or interior point methods. After the mathematical programming solver outputs the optimal or near-optimal solution, the feasibility of the decision variable values ​​output by the mathematical programming solver is verified to check whether the upper limit constraint of the grid connection point power is satisfied. After the verification is passed, the active power sequence corresponding to the distributed power resource operation preprocessing data output by the mathematical programming solver is organized to form the active power setpoint of the distributed resource.

[0085] It should be noted that mathematical programming solvers are software tools used to automatically solve optimization problems (such as linear programming, integer programming, and nonlinear programming). They can iteratively search for the optimal solution using numerical algorithms (such as the simplex method, interior point method, and branch and bound method) based on the input objective function and constraints. Common mathematical programming solvers include Gurobi, CPLEX, GLPK, and SCIP, and are widely used in fields such as power dispatching, operations research, and resource allocation.

[0086] S3.5 Arrange the active power setpoints according to the time series to generate a power dispatching scheme.

[0087] Furthermore, the active power setpoints for each time period are indexed and marked to ensure that the setpoints of all distributed resources correspond one-to-one with the scheduling cycle. Based on the continuity of the scheduling cycle and the operating characteristics of the resources, the active power setpoints within the same time period are combined to form the overall scheduling instruction for the distributed resources in that time period. The combined results of each time period are then sequentially spliced ​​together to construct an active power time series covering the complete scheduling cycle. The compiled active power time series is then organized into a power dispatching scheme.

[0088] S4. Calculate the reactive power demand intensity of each power node based on the preprocessed power grid status data, and generate a reactive power support demand heat map. Use the particle swarm optimization algorithm to collaboratively solve the active and reactive power decision variables of the power dispatch scheme and the reactive power support demand heat map, and output the power resource dispatch instruction set.

[0089] It should be noted that existing methods typically determine the reactive power demand of power nodes through static load estimation or historical experience data. Reactive power support demand is usually expressed in terms of fixed node distribution or simplified regional distribution. The active and reactive power decisions in power dispatching schemes often adopt step-by-step optimization, that is, optimizing active power allocation first and then reactive power allocation, which lacks coordinated optimization of active and reactive power. When outputting dispatching instructions, the spatial continuity and real-time changes of reactive power allocation and power node demand are not fully considered, making it difficult to form a comprehensive dispatching scheme for dynamic operating conditions.

[0090] Our invention extracts operating parameters such as voltage amplitude, current phase, and active power flow direction of power nodes from preprocessed power grid data, calculates the reactive power demand intensity of each power node, maps the reactive power demand to the spatial location of the power grid, and forms a continuous reactive power support demand heat map using Kriging interpolation. We then use a particle swarm optimization algorithm to collaboratively solve the active power allocation in the power dispatch scheme and the reactive power allocation in the reactive power support demand heat map, outputting a power resource dispatch instruction set. This achieves joint optimization of active and reactive power while balancing spatial continuity constraints, improving power grid operation safety, load matching accuracy, and dispatch response efficiency.

[0091] S4.1 Calculate the reactive power demand intensity of the preprocessed power grid status data to obtain the reactive power demand intensity of each power node.

[0092] Furthermore, operating parameters such as voltage amplitude, current phase, and active power flow direction of power nodes are extracted from the preprocessed power grid status data. Feature matching is performed on the voltage amplitude and current phase in the preprocessed data to obtain the voltage-current relationship of the power nodes. Combined with the active power flow direction, the power components of the power nodes are decomposed to extract reactive power demand within each time window. The reactive power demand is accumulated and balanced according to the time series to extract feature quantities that reflect the voltage support pressure and reactive power consumption level of the power nodes. These feature quantities are then combined and mapped according to the power component decomposition principle to obtain the reactive power components of the power nodes within each time window. The time series of the reactive power components is then normalized to reactive power demand intensity, which is used to reflect the voltage support pressure and reactive power consumption level of the power nodes.

[0093] S4.2 Map the reactive power demand intensity of each power node to the spatial location of the power grid, and convert it into a continuous power grid area using the Kriging interpolation method to form an initial reactive power support demand heat map.

[0094] Furthermore, based on the physical coordinates of power nodes in the power grid, the reactive power demand intensity corresponding to each power node is mapped to the spatial location of the power grid, resulting in a discrete set of reactive power demand intensity points. Kriging interpolation is then used on the reactive power demand intensity point set to integrate the correlation between spatial location and reactive power demand intensity, expanding the discrete point set into a continuous distribution covering the entire power grid area. Based on the continuous distribution, an initial reactive power support demand heat map is generated by mapping the reactive power demand intensity of each spatial location obtained by Kriging interpolation into a color or numerical matrix, so that the reactive power support demand level of different areas of the power grid can be intuitively presented in space.

[0095] It should be noted that the spatial location of the power grid is obtained by parsing the geographical coordinates of the power nodes in the power grid topology and the line connection information.

[0096] S4.3. Mark the reactive power demand intensity at different locations in the initial reactive power support demand heat map by color depth, and perform smoothing to generate a reactive power support demand heat map.

[0097] Furthermore, a mapping relationship from low to high intensity is defined on the color scale. For example, a normalized reactive power demand intensity example value of 0 is mapped to a light color, and a normalized reactive power demand intensity example value of 1 is mapped to a dark color. The reactive power demand intensity located at each grid spatial location in the initial reactive power support demand heat map is normalized, and a color depth is assigned to each grid spatial location according to the normalization result, forming an initial heat map distribution with color coding. Spatial smoothing is applied to the initial heat map distribution with color coding. For example, Gaussian filtering or multi-scale smoothing methods are used to remove local noise and enhance regional continuity to obtain a continuous and smooth spatial color block distribution. The smoothed spatial color block distribution is saved as a reactive power support demand heat map.

[0098] S4.4 Solve the active and reactive power decision variables of the power dispatch scheme and reactive power support demand heat map by using the particle swarm optimization algorithm. At the same time, optimize the active power allocation in the power dispatch scheme and the reactive power allocation in the reactive power support demand heat map to obtain the optimal active and reactive power decision variables.

[0099] Furthermore, the active and reactive power decision variables are represented as position vectors using particle encoding in the particle swarm optimization algorithm. Each position vector contains the active power setpoint for each distributed power resource in each time window and the reactive power allocation value for each spatial location. A fitness function is established for each position vector, which is a weighted combination of the total operating cost index of the virtual power plant and the voltage deviation constraint and reactive power deficiency penalty term from the reactive power support demand heatmap. This fitness function is used to evaluate the merits of particle positions. During the iteration process, the particle swarm optimization algorithm adjusts the position vectors using velocity and position update rules, and performs a feasibility check on the position vectors after each update. The feasibility check covers the adjustable power upper limit, the grid connection point power upper limit constraint, and the power equipment operating boundary constraint. If the constraints are not met, the position is corrected to a feasible position using the constraint satisfaction adjustment method. At the convergence of the iteration, the particle position with the optimal fitness function value is selected as the optimal active and reactive power decision variable.

[0100] It should be noted that the boundary constraints of power equipment operation are derived from the physical operating limits and safety operating standards of the power equipment, and are boundary conditions determined by operating parameters such as voltage allowable range, reactive power adjustment range and line transmission capacity.

[0101] Voltage deviation constraint refers to the requirement, during the optimization process, that the voltage values ​​of each power node in the power grid be maintained within a specified safe operating range to prevent excessively high or low voltages caused by uneven reactive power distribution. Voltage deviation constraint limits the deviation to an allowable range by comparing the actual voltage of a node with its rated or target voltage.

[0102] Velocity and position update rules are fundamental mechanisms in particle swarm optimization (PSO) that guide particle search direction and position changes. Particles dynamically adjust their movement speed based on the difference between their current velocity, their historical best position, and the global best position; subsequently, they correct their current position based on the updated velocity, gradually bringing the particle closer to the target optimal region.

[0103] Constraint satisfaction adjustment methods are correction mechanisms used in particle swarm optimization iterations to ensure the feasibility of solutions. When a particle's updated position violates conditions such as power limits, voltage constraints, or device operating boundaries, the particle's variable values ​​are adjusted or corrected to bring them back into the allowable range. Common methods include boundary truncation (forcibly restricting variables exceeding the boundary to their boundary values), penalty function methods (imposing penalties on violating solutions in the fitness function to guide them back to the feasible region), and projection repair methods (projecting solutions that do not satisfy constraints back to the feasible region along the gradient direction).

[0104] S4.5 Perform time series arrangement and constraint verification on the optimal active and reactive power decision variables to generate a power resource dispatch instruction set.

[0105] Furthermore, the optimal active and reactive power decision variables are arranged according to time series, and the active power setpoints and reactive power allocation values ​​of each distributed power resource in each time window and each spatial location are organized into a continuous time series table. At the same time, each continuous time series table is verified to ensure that each active and reactive power setting meets the adjustable power limit, the virtual power plant grid connection point power limit constraint, and the power equipment operation boundary constraint. After the constraint conditions are verified, the continuous time series table is converted into a power resource dispatch instruction set.

[0106] S5. Through a secure and encrypted communication channel, the power resource dispatch instruction set is sent to the local controller of each distributed resource, and the execution of the instructions is monitored in real time.

[0107] S5.1. Encrypt the power resource dispatch instruction set and send the encrypted power resource dispatch instruction set to the local controller of each distributed power resource through a secure encrypted communication channel.

[0108] Furthermore, each power resource dispatch instruction is serialized into a digital signal according to a unified format. The serialized power resource dispatch instruction set is then encrypted using a symmetric or asymmetric encryption algorithm to generate an encrypted power resource dispatch instruction set. This encrypted power resource dispatch instruction set is then sent one by one to the local controller of each distributed power resource through a secure encrypted communication channel. Integrity verification and identity authentication are performed during transmission to ensure the confidentiality, integrity, and availability of the power resource dispatch instruction set during transmission.

[0109] It should be noted that a secure encrypted communication channel is a pathway that protects information transmission through encryption algorithms and security authentication mechanisms during the communication process. It is used to ensure that the power resource dispatch instruction set maintains confidentiality, integrity, and verifiability during transmission, and to prevent unauthorized access or tampering.

[0110] Secure encrypted communication channels are obtained by configuring encryption protocols and security authentication mechanisms in the communication link.

[0111] S5.2 After receiving the power resource dispatch instruction set, the local controller decrypts and verifies the contents of the power resource dispatch instruction set, executes the power dispatch task according to the power resource dispatch instruction set, and monitors the power resource execution status in real time.

[0112] Furthermore, after receiving the power resource dispatch instruction set, the local controller uses a pre-distributed or negotiated key (such as a symmetric key or private key) and a corresponding encryption algorithm (such as AES or RSA) to decrypt the encrypted data, remove the encryption layer to restore the original instruction content, and recover the original instruction content. The decrypted power resource dispatch instruction set is then verified for integrity and origin to ensure that the instructions have not been tampered with and that their source is reliable. After successful verification, the power dispatch task is executed according to the time sequence and power setpoints in the power resource dispatch instruction set, including the active power output and reactive power support allocation of each distributed power resource. During execution, the local controller collects power resource operating parameters in real time, monitors the power resource response, records deviations and anomalies, and can provide feedback for subsequent dispatch adjustments or optimizations, ultimately completing the execution of the power dispatch task.

[0113] This embodiment also provides a computer device applicable to the virtual power plant optimization scheduling method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the virtual power plant optimization scheduling method proposed in the above embodiment.

[0114] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0115] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the virtual power plant optimized scheduling method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0116] In summary, this invention achieves dynamic and reliable capacity calculation of the power grid by establishing a dynamic correlation graph and performing spatiotemporal alignment analysis, accurately capturing the dynamic changes in power resources and grid status, providing reliable power operation constraints, and improving the resource management capabilities and grid stability of the dispatching system. Simultaneously, it employs a particle swarm optimization algorithm to collaboratively optimize active and reactive power decision variables, thereby optimizing the power dispatching scheme, ensuring the rational allocation of active and reactive power, improving dispatching efficiency, reducing operating costs, and enhancing the overall economy and flexibility of the virtual power plant.

[0117] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A virtual power plant optimized scheduling method, characterized in that: include, Collect distributed power resource operation data and power grid operation parameters, perform preprocessing, and output preprocessed power resource operation data and preprocessed power grid status data; Based on the preprocessed data of power resource operation, a dynamic correlation graph is established. The preprocessed power grid status data and the dynamic correlation graph are then subjected to spatiotemporal alignment and coupling analysis to obtain the dynamic reliable capacity. The specific steps are as follows: By using the dynamic time warping method, time series similarity analysis is performed on the preprocessed data of power resource operation to extract the dynamic correlation between different preprocessed data of power resource operation. Using preprocessed power resource operation data as graph nodes, edges between graph nodes are constructed based on dynamic association relationships to form a dynamic association graph; The power grid state preprocessing data and the dynamic correlation map are aligned in time and space, and the spatiotemporal coupling relationship is extracted by calculating the correlation between the power grid state preprocessing data and the dynamic correlation map. Based on the spatiotemporal coupling relationship, the voltage, current, active power and reactive power of each power node are associated with the power resource operation preprocessing data corresponding to the graph nodes mapped to the same power node, so as to identify the response relationship between changes in grid status and changes in distributed power resource operation at different time periods and different spatial nodes. Based on response relationships, the response characteristics of preprocessed data of distributed power resources under changes in grid conditions are analyzed. Correlation analysis and sensitivity analysis are used to determine the correlation strength and sensitivity between voltage, current, active power, reactive power and resource output, thereby assessing the adjustable range and stable output capability of distributed power resources under different grid conditions. The continuous connection across time and the interaction pattern across spatial regions are accumulated as constraint factors to form dynamic reliable capacity. Dynamic reliable capacity is used as a power operation constraint, and the optimization objective is to minimize the total operating cost of the virtual power plant. A MILP optimization model is constructed and solved to generate a power dispatch scheme. Reactive demand intensity is calculated from the preprocessed power grid status data to obtain the reactive demand intensity of each power node. Based on the physical coordinates of power nodes in the power grid, the reactive power demand intensity corresponding to each power node is mapped to the spatial location of the power grid, resulting in a discrete set of reactive power demand intensity points. The Kriging interpolation method is used on the set of reactive power demand intensity points to integrate the correlation between spatial location and reactive power demand intensity, expanding the discrete set of reactive power demand intensity points into a continuous reactive power demand intensity distribution covering the power grid area. Based on the continuous reactive power demand intensity distribution, an initial reactive power support demand heat map is formed. The reactive power demand intensity at different locations in the initial reactive power support demand heatmap is marked by color intensity and smoothed to generate a reactive power support demand heatmap. The active and reactive power decision variables of the power dispatch scheme and the reactive power support demand heatmap are solved collaboratively by the particle swarm optimization algorithm, and the power resource dispatch instruction set is output. The process involves using particle swarm optimization to collaboratively solve for the active and reactive power decision variables in the power dispatch scheme and reactive power support demand heatmap, outputting a power resource dispatch instruction set. The specific steps are as follows: The active and reactive power decision variables of the power dispatch scheme and the reactive power support demand heat map are solved by the particle swarm optimization algorithm. At the same time, the active power allocation in the power dispatch scheme and the reactive power allocation in the reactive power support demand heat map are optimized to obtain the optimal active and reactive power decision variables. Time series arrangement and constraint verification are performed on the optimal active and reactive power decision variables to generate a power resource dispatch instruction set. Through a secure and encrypted communication channel, the power resource dispatch instruction set is sent to the local controller of each distributed resource, and the execution of the instructions is monitored in real time.

2. The virtual power plant optimization scheduling method as described in claim 1, characterized in that: The distributed power resource operation data includes new energy power operation data, energy storage power status data, and load power demand data. The power grid operating parameters include node voltage, current, active power, reactive power, and switch status.

3. The virtual power plant optimization scheduling method as described in claim 1, characterized in that: The specific steps for outputting preprocessed power resource operation data and preprocessed power grid status data are as follows. The distributed power resource operation data is denoised, cleaned, synchronized in time, and normalized to output preprocessed power resource operation data. The system performs time-series correction, filtering and smoothing, normalization and baseline adjustment on the power grid operating parameters, and outputs preprocessed power grid status data.

4. The virtual power plant optimization scheduling method as described in claim 1, characterized in that: The process involves using dynamic reliable capacity as a power operation constraint and minimizing the total operating cost of the virtual power plant as the optimization objective. A MILP optimization model is constructed and solved to generate a power dispatching scheme. The specific steps are as follows: Dynamic reliable capacity is used as a power operation constraint to define the upper limit constraint of the grid connection point power of the virtual power plant; With the goal of minimizing the total operating cost of the virtual power plant, an objective function for optimizing electricity costs is defined. Based on the upper limit constraint of grid connection power and the objective function of electricity cost optimization, a MILP optimization model is constructed; A mathematical programming solver is used to numerically solve the MILP optimization model to obtain the active power setpoints for each distributed resource; the active power setpoints are then arranged in a time series to generate a power dispatching scheme.

5. The virtual power plant optimization scheduling method as described in claim 1, characterized in that: The process involves distributing power resource dispatching instruction sets to the local controllers of each distributed resource through a secure and encrypted communication channel, and monitoring the execution of these instructions in real time. The specific steps are as follows: The power resource dispatch instruction set is encrypted and then sent to the local controller of each distributed power resource through a secure encrypted communication channel. After receiving the power resource dispatch instruction set, the local controller decrypts and verifies the contents of the power resource dispatch instruction set, executes the power dispatch task according to the power resource dispatch instruction set, and monitors the execution status of power resources in real time.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the virtual power plant optimization scheduling method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the virtual power plant optimization scheduling method according to any one of claims 1 to 5.

Images