Method and system for analyzing and calling uncertainty demand response potential of virtual power plant

Abstract

The invention relates to the technical field of demand response analysis, and discloses an uncertain demand response potential analysis and calling method and system for a virtual power plant, and the method comprises the steps: building a physical constraint model according to a resource operation state set of a target virtual power plant, carrying out the causal analysis of the resource operation state set, generating an initial causal graph, and carrying out the calculation of the initial causal graph; identifying a latent variable with an implicit dependency relationship in the resource operation state set, embedding the initial causal graph to obtain an updated causal graph, performing proxy observation construction on the updated causal graph by using a direct influence variable of the latent variable, and performing demand response potential analysis on a target virtual power plant by using the target causal graph to obtain a demand response potential analysis result; and obtaining a resource response potential table and a hierarchical scheduling strategy, and carrying out rolling optimization calling on the target virtual power plant to obtain a target response calling result. According to the method, the cause and effect structure definition and the response potential evaluation accuracy are effectively improved, and the prediction deviation and uncertainty influence are effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of demand response analysis technology, and in particular to a method and system for analyzing and invoking the uncertainty demand response potential of a virtual power plant. Background Technology

[0002] In modern power systems, virtual power plants achieve flexible regulation of system power balance by aggregating distributed energy resources and controllable loads. However, due to the diverse resource types and complex coupling relationships within virtual power plants, as well as the uncertainties in renewable energy output, load response, and energy storage status, accurately assessing demand response potential has become a challenging technical task.

[0003] Existing methods typically rely on probability distribution assumptions to model and predict the uncertainties in responses to various resources. However, these methods are highly dependent on probability distributions, making it difficult to fully characterize potential nonlinearities and multi-source coupling characteristics. In practical applications, prediction biases can easily amplify scheduling risks, leading to unreliable resource allocation at critical moments and impacting system security and response reliability.

[0004] Furthermore, the causal structure in virtual power plants is often difficult to define precisely, with numerous latent variables, such as user comfort preferences, local control strategies, and equipment aging status. These latent variables have a significant impact on resource response but cannot be directly observed, leading to uncertainty in causal relationship inferences and further increasing the difficulty of assessing demand response potential. The unclear causal structure may cause misjudgments in response prediction, making it difficult to refine hierarchical dispatch strategies, leading to frequent adjustments in the rolling optimization process, decreased scheduling efficiency, and reduced model interpretability and regulatory acceptability.

[0005] In summary, existing technologies face challenges in assessing and utilizing the demand response potential of virtual power plants. These challenges include the amplification of risks due to reliance on probability distribution assumptions, and the uncertainties arising from complex causal relationships and latent variables. Consequently, it is difficult to achieve high-precision assessment and flexible utilization while ensuring safety, reliability, and interpretability.

[0006] Therefore, existing virtual power plant demand response potential analysis methods rely heavily on probability distribution assumptions and have unclear causal structures, which can easily lead to amplified prediction bias, inaccurate response potential assessment, and increased risk of strategy invocation. Summary of the Invention

[0007] This invention provides a method for analyzing and invoking the uncertain demand response potential of a virtual power plant. Its main purpose is to solve the problems of existing virtual power plant demand response potential analysis methods, which rely heavily on probability distribution assumptions and have unclear causal structures, easily leading to amplified prediction bias, inaccurate response potential assessment, and increased risk of invoking strategies.

[0008] Firstly, to achieve the above objectives, the present invention provides a method for analyzing and invoking the uncertainty demand response potential of a virtual power plant, comprising: Collect the set of resource operation statuses of the target virtual power plant, and construct a physical constraint model based on the set of resource operation statuses; The physical constraint model is used to perform causal analysis on the set of resource operating states, and an initial causal graph is generated based on the analysis results; Identify latent variables with implicit dependencies within the resource runtime state set, and embed these latent variables into the initial causal graph to obtain an updated causal graph; Obtain the direct influencing variables of the latent variables, and use the direct influencing variables to construct the updated causal graph by proxy observation, thereby obtaining the target causal graph; The target virtual power plant is analyzed for demand response potential using the target cause-effect graph to obtain a resource response potential table and a hierarchical scheduling strategy. The target virtual power plant is subjected to rolling optimization calls using the resource response potential table and the hierarchical scheduling strategy to obtain the target response call result.

[0009] Secondly, the present invention also provides a system for analyzing and invoking the uncertainty demand response potential of a virtual power plant, the system comprising: The constraint model construction module is used to collect the resource operation status set of the target virtual power plant and construct a physical constraint model based on the resource operation status set; The cause-effect graph generation module is used to perform causal analysis on the set of resource operating states using the physical constraint model, and generate an initial cause-effect graph based on the analysis results; The latent variable embedding module is used to identify latent variables with implicit dependencies in the resource running state set, and embed the latent variables into the initial causal graph to obtain the updated causal graph; The surrogate observation construction module is used to obtain the direct influencing variables of the latent variables, and to construct the updated causal graph using the direct influencing variables to obtain the target causal graph. The response potential analysis module is used to perform demand response potential analysis on the target virtual power plant using the target cause-effect graph, and obtain a resource response potential table and a hierarchical scheduling strategy. The response call module is used to perform rolling optimization calls on the target virtual power plant using the resource response potential table and the hierarchical scheduling strategy to obtain the target response call result.

[0010] This invention constructs a target causal graph containing latent variables to perform deep causal analysis on the multi-source resource operation status of a virtual power plant, achieving accurate identification of implicit dependencies and surrogate observation completion. Compared to traditional methods based solely on statistical correlation, this invention effectively reduces prediction bias and uncertainty, improving the accuracy and robustness of demand response potential assessment. By combining a resource response potential table with rolling optimization of hierarchical scheduling strategies, dynamic coordination and flexible scheduling of virtual power plant resources can be achieved, significantly improving system response efficiency, operational stability, and overall economy. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. 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.

[0012] Figure 1 This is a flowchart illustrating a method for analyzing and invoking the uncertainty demand response potential of a virtual power plant, as provided in an embodiment of the present invention. Figure 2 A functional block diagram of a virtual power plant uncertainty demand response potential analysis and invocation system provided in an embodiment of the present invention; The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0013] To enable those skilled in the art to better understand the technical solutions of this disclosure, and to fully understand and implement the process of how this disclosure applies technical means to solve technical problems and achieve corresponding technical effects, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. The embodiments of this disclosure and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort should fall within the protection scope of this disclosure.

[0014] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0015] This application provides a method for analyzing and invoking the uncertainty demand response potential of a virtual power plant. This method can be executed by software or hardware installed on a terminal device or server-side device. The server-side includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be a standalone server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and demand forecasting platforms.

[0016] Reference Figure 1 The diagram shown is a flowchart illustrating a method for analyzing and invoking the uncertainty demand response potential of a virtual power plant according to an embodiment of the present invention. In this embodiment, the method includes: S1. Collect the set of resource operation statuses of the target virtual power plant, and construct a physical constraint model based on the set of resource operation statuses.

[0017] In this embodiment of the invention, the resource operation status set is a set of status data of various dispatchable resources (such as photovoltaic, battery energy storage, controllable load, distributed generator sets, etc.) in the virtual power plant during operation. The resource operation status includes: power output (active and reactive), voltage, current, frequency, energy storage state of charge (SOC), equipment start-up and shutdown status, output constraints, temperature and other operating conditions.

[0018] The physical constraint model is a mathematical model based on the operating laws of the power system and the physical characteristics of the equipment. It is used to describe the inherent constraint relationship between the operating states of resources, including: power balance constraints, energy conservation constraints, safe operating boundaries of voltage, current and frequency, and operating limit constraints of each device.

[0019] Specifically, the step of constructing a physical constraint model based on the set of resource operating states includes: The resource operation status set is timestamped to obtain an aligned resource operation status set; The set of aligned resource operating states is normalized to obtain a set of standard resource operating states; Extract the key operational characteristics of the standard resource operational status set; Obtain the constraint range of each key operational feature, and construct the parameter constraint function of each key operational feature based on the constraint range; The physical constraint model of the target virtual power plant is generated based on all the parameter constraint functions.

[0020] In this embodiment of the invention, the collected resource operating status set is timestamped according to a unified time base to eliminate time deviations caused by inconsistent sampling periods between different devices or monitoring channels, ensuring that the resource statuses are comparable and correlated at the same time. Furthermore, the aligned data is normalized to convert operating parameters of different dimensions and ranges (such as power, voltage, SOC, etc.) into a unified standard scale, thereby obtaining a standardized set of resource operating statuses.

[0021] In this embodiment of the invention, key operating features that reflect the dynamics and operating characteristics of the system are extracted from a set of standard resource operating states, such as power change rate, voltage deviation, energy storage SOC fluctuation, and load response amplitude. Combining the physical characteristics of the equipment with the operational safety boundaries, the constraint range corresponding to each key operating feature (such as maximum / minimum output, voltage upper and lower limits, and SOC safety range) is obtained.

[0022] Based on these constraints, parametric constraint functions are further constructed to quantitatively describe the feasible domain and variation patterns of various operational characteristics under physical constraints. These parametric constraint functions are then summarized and coupled according to resource type and physical relationship (e.g., power balance, energy conservation, equipment limits, time-series charge / discharge constraints, and network electrical constraints). Equality constraints and inequality constraints are uniformly expressed in mathematical form (e.g., time-series constraints described by linear / nonlinear equations, convex / non-convex inequalities, and difference equations).

[0023] Furthermore, cross-resource constraints (such as the mutual exclusion of power output between converters and energy storage, and load controllability boundaries) are incorporated into the same model framework. The model undergoes numerical processing (discretization of the time domain, linearization, or retention of nonlinear terms and selection of a solution strategy), feasibility testing, and historical data calibration. Finally, a physical constraint model that can be used for causal analysis and optimal scheduling is obtained. The calculation formula is shown below:

[0024]

[0025] in, Represents the coefficient matrix of the equality constraints. Indicates the standard resource's operating status. Represents the equality constraint constant. This represents the coefficient matrix of the inequality constraints. The constants represent inequality constraints. Specifically, the equality constraint coefficient matrix represents all constraints that "must be strictly balanced," such as the power balance equation and energy conservation. The inequality constraint coefficient matrix represents operating boundaries, such as power limits, energy storage SOC range, and line capacity limits.

[0026] This invention collects the resource operation status set of the target virtual power plant and constructs a physical constraint model, which can realistically reflect the physical coupling relationship and operation boundary between various distributed energy sources, energy storage and controllable loads, thereby ensuring the physical feasibility and accuracy of subsequent modeling and analysis results.

[0027] S2. Perform causal analysis on the set of resource operating states using the physical constraint model, and generate an initial causal graph based on the analysis results.

[0028] In this embodiment of the invention, the true causal relationships between various resources are revealed under the premise of satisfying physical laws, rather than relying solely on statistical correlation. By introducing physical constraints, spurious correlations caused by noisy data and external disturbances can be effectively eliminated, making causal inference more consistent with objective laws such as energy conservation and power balance. The initial causal graph generated based on the analysis results can intuitively reflect the mutual influence paths and control logic of resources within the virtual power plant.

[0029] In detail, the step of performing causal analysis on the set of resource operating states using the physical constraint model and generating an initial causal graph based on the analysis results includes: Obtain the causal structure edge constraints, and construct an initial causal search space based on the causal structure edge constraints and the physical constraint model; Independence tests and conditional correlation analyses are performed on the standard resource operating state set to obtain the causal direction of each resource variable within the standard resource operating state set; The initial causal search space is corrected using the causal direction of the variables to obtain an updated causal search space; Identify the target causal link in the updated causal search space; Each resource variable in the standard resource operating state set is used as a node in the cause-effect graph. Generate causal graph edges based on the target causal link and the updated causal search space; An initial causal graph is generated based on the causal graph nodes and the causal graph edges.

[0030] In this embodiment of the invention, constraints on the causal structure edges are obtained from system analysis or empirical knowledge. For example, certain resources cannot have a direct causal relationship or must follow a unidirectional dependency. These constraints on the causal structure edges are combined with constraints in the physical constraint model, such as power balance, energy conservation, and equipment operating limits, to form a feasible causal graph search space that conforms to physical laws and is constrained by edge constraints. This provides clear boundaries for subsequent causal inference, narrowing the search range while ensuring that the generated causal structure has physical interpretability and computational feasibility.

[0031] Furthermore, independence tests are conducted on each resource variable in the standard resource operation status set to determine which variables do not have a direct dependency relationship; at the same time, conditional correlation analysis is performed to examine whether the correlation between each variable is significant under the condition of controlling for other variables, thereby identifying potential causal relationships and making a preliminary inference on the causal direction of each resource variable—that is, which variables may affect other variables as causes and which variables may be affected as results.

[0032] Furthermore, this causal direction information is used to modify the initial causal search space by adding reasonable causal edges and removing edges that do not conform to statistical and physical laws, resulting in an updated causal search space that conforms to both data-driven characteristics and physical constraints.

[0033] In this embodiment of the invention, target causal links with significant causal relationships are identified from the updated causal search space, determining which variables may have direct causal influences. Each resource variable in the standard resource operating state set is mapped to a node in the causal graph, and corresponding causal graph edges are generated based on the target causal links and the information from the updated causal search space, representing the causal direction and connection relationship between variables. Finally, all nodes and edges are combined to form an initial causal graph, which reflects the potential causal structure between resource variables and provides a structured foundation for subsequent latent variable embedding and target causal graph optimization.

[0034] By using physical constraint models to perform causal analysis on the set of resource operating states, the true causal relationships between resources can be accurately identified under the premise of following physical laws such as power balance, energy conservation and equipment operating limits. At the same time, the initial causal graph can intuitively present the mutual influence paths of resources within the virtual power plant. Compared with analysis methods based solely on statistical correlation, it can effectively reduce false associations caused by noise and external disturbances.

[0035] S3. Identify latent variables with implicit dependencies within the resource running state set, and embed the latent variables into the initial causal graph to obtain the updated causal graph.

[0036] In this embodiment of the invention, implicit dependencies are the mutual influences or constraints between variables in the resource operating state set that are not directly observed but actually exist. Latent variables are variables in the resource operating state set that cannot be directly measured or observed but affect system behavior, and are usually used to explain implicit dependencies between resource variables. Examples include: control strategies within a virtual power plant, external environmental disturbances (weather, temperature, and light changes), and cluster load behavior patterns.

[0037] Specifically, the latent variables that identify implicit dependencies within the set of resource operating states include: Identify groups of potentially related variables with implicit dependencies within the set of resource runtime states; Calculate the covariance matrix of the potential associated variable group; Principal component analysis is performed on the covariance matrix to generate potential common factor structures; Extract the principal component loading vectors of the potential common factor structure to obtain the potential factor structure; Convert the latent factor structure into a latent factor vector; The latent factor vector is normalized and made interpretable to obtain latent variables.

[0038] In this embodiment of the invention, by analyzing the statistical dependencies between resource variables in the resource operation status set, combinations of variables with potential covariance characteristics or implicit dependencies are identified to form potential related variable groups; covariance matrix calculation is performed on each potential related variable group to quantify the degree of linear correlation and consistency of change among the variables.

[0039] Furthermore, principal component analysis (PCA) is performed on the covariance matrix of the potential associated variable groups to extract the main linear variation directions in the data, thereby identifying potential common factor structures—that is, the potential influence patterns that can explain most of the variance of the resource variables. The loading vectors of each principal component are extracted to reflect the contribution weight of each resource variable in the potential common factor structure, ultimately yielding a quantifiable potential factor structure.

[0040] In this embodiment of the invention, the extracted latent factor structure is converted into a numerical latent factor vector, so that each latent factor has a clear representation in the resource variable space. These latent factor vectors are standardized in terms of dimensions, scaled to a fixed range, and subjected to interpretability processing (such as rotation, sparsification, or association with resource variable features) to obtain latent variables that are both quantitative and easy to understand. These latent variables can be directly embedded into a causal graph to reveal the implicit dependencies between resource variables.

[0041] In this embodiment of the invention, embedding the latent variables into the initial causal graph to obtain an updated causal graph includes: Embed the latent variables into the initial causal graph; The initial causal graph after embedding is modified by applying directional constraints to obtain the modified causal graph. The modified causal graph is then subjected to edge weight correction using a Bayesian structure learning method to obtain an updated causal graph.

[0042] In this embodiment of the invention, latent variables are treated as new nodes and connected to resource variables influenced by the latent variables to supplement unobserved implicit dependencies in the initial causal graph. The causal graph with embedded latent variables is then corrected by directional constraints. The direction of causal edges is adjusted based on physical laws, edge constraints, and statistical inference results to ensure that each causal edge in the graph conforms to system logic and is interpretable. This generates a complete and corrected causal graph that more accurately reflects the true causal structure between variables within the virtual power plant.

[0043] In this embodiment of the invention, a modified causal graph is used as the initial structural input Bayesian structure learning method. By learning the joint probability distribution of resource variables and latent variables, the edge weight of each causal edge is calculated. Based on the Bayesian inference results, the edge weights in the causal graph are modified to enhance the weights of credible edges and weaken or delete insignificant edges, thereby obtaining an updated causal graph with a more robust structure and more accurate edge weights.

[0044] This invention identifies implicit dependencies in a set of resource operating states and embeds the corresponding latent variables into an initial causal graph. This can compensate for the influence of factors in resource variables that are not directly measured on the system, thereby making the causal graph more complete and accurate. It not only reveals the potential dependencies between resource variables, but also enhances the interpretability and physical rationality of the causal structure.

[0045] S4. Obtain the direct influencing variables of the latent variables, and use the direct influencing variables to construct the surrogate observation of the updated causal graph to obtain the target causal graph.

[0046] In this embodiment of the invention, the directly influencing variable is the resource variable that the latent variable directly affects in the updated causal graph, and is also the node in the updated causal graph that is directly affected by the latent variable.

[0047] Proxy observation construction is used because latent variables cannot be directly observed. Therefore, the behavior or effect of latent variables is indirectly represented by variables that directly influence latent variables. Usually, an observable "proxy variable" is generated through statistical modeling, linear combination, or feature extraction, which is used to replace the latent variable in the analysis and inference in the updated causal graph.

[0048] In detail, the step of constructing the target causal graph by using the directly influencing variables to perform surrogate observations on the updated causal graph includes: Correlation analysis was performed on the latent variables and the directly influencing variables to obtain the time series correlation. Calculate the lagged correlation between the latent variables and the directly influencing variables; Conditional correlation analysis was performed on the latent variables and the directly influencing variables to obtain the conditional independence of the directly influencing variables; Candidate proxy variables for the latent variables are selected from the directly influencing variables based on the time series correlation, the lagged correlation, and the conditional independence. The candidate proxy variables are linearly mapped to obtain the proxy observations of the latent variables; By establishing a mapping relationship between the proxy observations and the latent variables, the proxy resource nodes of the latent variables are obtained; The proxy resource node is used to replace the latent variable in the updated causal graph to obtain the extended causal graph; The extended causal graph is dynamically intervened and verified to obtain the target causal graph.

[0049] In this embodiment of the invention, time series correlation analysis is performed on latent variables and directly influencing variables. By calculating Pearson correlation coefficient, mutual information or dynamic time warping, the degree of synchronous change of latent variables and directly influencing variables at the same time is quantified, and observation signals that may reflect the effect of latent variables are identified.

[0050] Furthermore, by analyzing the influence of latent variables on directly influencing variables at the previous time or multiple lag times through autocorrelation, crosscorrelation, or Granger causality tests, we can capture the delayed effect of latent variables on directly influencing variables, so as to ensure that the dynamic effect of latent variables on directly influencing variables is fully reflected.

[0051] Furthermore, while controlling for other resource variables or potential factors, the independence of directly influencing variables from latent variables is determined. Conditional correlation coefficients or partial correlation analysis can eliminate variables that appear related but are actually indirect or redundant, thus avoiding the misidentification of non-causal variables as latent variables.

[0052] In this embodiment of the invention, by combining the results of time series correlation, lagged correlation, and conditional independence analyses, candidate proxy variables that best represent the effect of latent variables are selected from the directly influencing variables. These candidate proxy variables can capture both the immediate impact of latent variables and reflect their lagged dynamics, while excluding irrelevant or redundant resource variables.

[0053] In this embodiment of the invention, multiple candidate surrogate variables are typically combined into a continuous numerical sequence through weighted linear combination or regression modeling to obtain surrogate observations of latent variables. This allows the candidate surrogate variables to quantify the changes of latent variables at different time steps. Further, the surrogate observations are considered as observable substitutes for latent variables in the updated causal graph, thereby generating surrogate resource nodes for the latent variables. Specifically, in the updated causal graph, surrogate resource nodes replace the original unobservable latent variables while maintaining the causal connection between latent variables and resource variables, forming an extended causal graph. This preserves the influence of latent variables on the system while simultaneously making latent variables observable.

[0054] In this embodiment of the invention, the step of dynamically intervening to verify the extended causal graph to obtain the target causal graph includes: Obtain the adjustable resources of the target virtual power plant, and construct a small-scale intervention experimental scheme based on the extended causal graph and the adjustable resources; According to the small-scale intervention experimental scheme, a preset intervention signal is sent to the adjustable resources of the target virtual power plant; Collect multidimensional response data after receiving the intervention signal; The intervention signal and the multidimensional response data are time-aligned to generate an intervention response time series. The causal association strength of each resource variable in the extended causal graph is calculated based on the intervention response time series. A causal-weighted regression was performed on the strength of the causal association to obtain the significance of the causal relationship; The extended causal graph is modified based on the significance of the causal relationship to obtain the target causal graph.

[0055] In this embodiment of the invention, adjustable resources in the target virtual power plant, such as energy storage systems, controllable loads, or flexible generation units, are identified. Combined with the causal relationships of each node in the extended causal graph, a small-amplitude intervention experiment scheme is designed to ensure that the intervention amplitude is small enough to avoid disturbing the normal operation of the system, while effectively observing the causal responses between each resource.

[0056] Furthermore, according to the small-amplitude intervention experimental protocol, preset intervention signals (such as slight adjustments to power output or load levels) are sent to adjustable resources. By recording response data, the sensitivity and causal effects of resource variables on latent variables and other extended causal graph nodes are obtained.

[0057] In this embodiment of the invention, the various resource responses of the target virtual power plant after receiving an intervention signal are collected in real time to obtain multi-dimensional response data, including indicators such as power change, energy status, and load adjustment. The intervention signal and the corresponding multi-dimensional response data are precisely aligned according to timestamps to ensure that each intervention action corresponds one-to-one with its resulting system response, thereby generating a complete intervention response time series.

[0058] In this embodiment of the invention, intervention response time series are used to calculate the causal association strength of each resource variable with other resource variables based on an extended causal graph, quantifying the magnitude of the causal impact transmitted by each resource in the system. Furthermore, causal weighted regression analysis is performed on these causal effects to assess the significance of each causal edge, distinguishing between edges with genuine strong causal relationships and those with weak or coincidental correlations. Based on the significance of the causal relationships obtained from the regression, the existence and direction of edges in the extended causal graph are corrected, removing insignificant edges and strengthening reliable causal connections, thereby generating a final target causal graph with a clear structure and sufficient data support.

[0059] This invention constructs proxy observations by utilizing the direct influence variables of latent variables, which can effectively map unmeasurable latent factors into observable nodes. This allows the role of latent variables to be fully reflected in the causal graph, making up for the lack of resource variables and enhancing the interpretability and physical rationality of the causal structure.

[0060] S5. Analyze the demand response potential of the target virtual power plant using the target cause-effect graph to obtain a resource response potential table and a hierarchical scheduling strategy.

[0061] In this embodiment of the invention, the target causal graph is the final causal structure graph obtained by correcting the direction and weight of causal edges after updating the causal graph through latent variable proxy observation and intervention verification. Nodes represent resource variables or latent variable proxy nodes in the virtual power plant, and edges represent the causal influence and intensity between resources.

[0062] Demand response potential analysis is an analysis of the ability of each adjustable resource to participate in demand response under different conditions, given the known causal structure and resource operating characteristics. This includes the adjustable power range, flexibility, response speed, and available adjustment amount under system constraints.

[0063] The resource response potential table is a tabular result derived from demand response potential analysis, recording the response potential indicators of each adjustable resource in the virtual power plant, such as maximum adjustable power, regulation flexibility, and response delay. The hierarchical scheduling strategy is a scheduling strategy designed based on resource response potential and causal relationships, managing the adjustable resources of the virtual power plant hierarchically according to priority, regulation capacity, or response type.

[0064] In detail, the step of using the target causal graph to perform demand response potential analysis on the target virtual power plant to obtain a resource response potential table and a hierarchical scheduling strategy includes: Obtain the historical operation data and external scene information of the target virtual power plant, and construct multiple typical operation scenarios based on the historical operation data and external scene information; Generate a set of response intervention actions based on the adjustable resources of the target virtual power plant; Using the external scene information and the set of response intervention actions, conditional sampling is performed on each resource variable in the target causal graph to obtain the response output distribution for each typical operating scenario; Response characteristic data and cost characteristic data for each resource variable are generated based on the response output distribution; Based on the response characteristic data and the cost characteristic data, the response capabilities of each resource variable in the target virtual power plant under the typical operating scenario are summarized to generate a resource response potential table; Based on the resource response potential table, each resource variable is classified by response reliability to obtain high-reliability resources, medium-reliability resources, and low-reliability resources. Based on the resource response potential table, a hierarchical scheduling strategy is generated for the high-reliability resources, the medium-reliability resources, and the low-reliability resources.

[0065] In this embodiment of the invention, historical operating data includes, but is not limited to: power generation, energy storage status, load levels, and historical control actions of various resources. External scenario information includes, but is not limited to: meteorological conditions (such as temperature, sunlight, and wind speed), market signals, or user behavior patterns. Multiple typical operating scenarios that can represent the main operating states are constructed through clustering or typical day / week extraction methods, covering different load patterns and changes in the external environment.

[0066] In this embodiment of the invention, the set of response intervention actions includes, but is not limited to: fine-tuning power output, adjusting energy storage charging and discharging strategies or load reduction operations, covering the possible adjustment range of resources, while ensuring that the intervention magnitude does not cause system instability.

[0067] In this embodiment of the invention, a target causal graph is used, with external scene information and response intervention actions as conditional inputs, to conditionally sample each resource variable. Under each typical operating scenario, the possible output changes of resource variables under different intervention conditions and external environments are simulated by sampling, and the response output distribution of each resource is obtained, which can quantify the resource's adjustment potential, uncertainty, and response characteristics.

[0068] In this embodiment of the invention, the response output distribution under each typical operating scenario is used to calculate the response characteristic data (such as maximum adjustable power, adjustment flexibility, and response speed) and cost characteristic data (such as adjustment energy consumption, operating costs, and equipment wear and tear) of each resource variable. Furthermore, these response characteristic data and cost characteristic data are combined to summarize the comprehensive adjustment capability of each resource under different typical operating scenarios, forming a systematic resource response potential table. This not only quantifies the demand response potential of each resource but also provides economic and feasibility references.

[0069] In this embodiment of the invention, each resource variable is classified according to its response reliability based on indicators such as the adjustment capability, response speed, and uncertainty of each resource in the resource response potential table. All resource variables are divided into high-reliability resources, medium-reliability resources, and low-reliability resources to distinguish the stability and controllability of resources in demand response.

[0070] In this embodiment of the invention, a corresponding hierarchical scheduling strategy is formulated by combining the partitioning results of all resource variables and the resource response potential table: high-reliability resources are given priority to participate in the global control of the system, medium-reliability resources are used for auxiliary balancing or local adjustment, and low-reliability resources provide supplementary support or backup solutions when necessary.

[0071] This invention utilizes target cause-effect graphs for demand response potential analysis. Based on a comprehensive consideration of causal relationships and the influence of latent variables among resources, it can accurately quantify the adjustment capacity and response characteristics of each adjustable resource, generate a resource response potential table, and formulate a scientific hierarchical scheduling strategy.

[0072] S6. Using the resource response potential table and the hierarchical scheduling strategy, perform rolling optimization calls on the target virtual power plant to obtain the target response call result.

[0073] In this embodiment of the invention, rolling optimization calls are performed by dividing the call problem into consecutive time windows, optimizing and solving within each time window, and then rolling the call results to the next time period.

[0074] The target response call result is the actual resource response result and overall load adjustment result generated during the rolling optimization scheduling process, based on the resource response potential table and hierarchical scheduling strategy.

[0075] In detail, the step of using the resource response potential table and the hierarchical scheduling strategy to perform rolling optimization calls on the target virtual power plant to obtain the target response call result includes: Obtain the real-time operating cycle of the target virtual power plant, and divide the real-time operating cycle into multiple continuously rolling time windows; The set of callable resources within each time window is identified based on the resource response potential table and the hierarchical scheduling strategy. The availability range of the callable resource set is confirmed using the physical constraint model. A response allocation scheme for the callable resource set is generated based on the availability range; Within the continuously scrolling time window, the resource response is invoked step by step using the response allocation scheme to obtain the target response invocation result of the target virtual power plant.

[0076] In this embodiment of the invention, the real-time operating cycle of the target virtual power plant is the continuous operating time period of the current system. This continuous operating time period is divided into several continuously rolling time windows, allowing for independent and dynamic scheduling optimization within each window. Furthermore, by combining the adjustment capabilities, response speeds, and reliability information of each resource in the resource response potential table, as well as the resource priorities and hierarchical management rules in the hierarchical scheduling strategy, the set of callable resources that can participate in regulation within each time window is identified, providing basic input for rolling optimization scheduling and real-time response invocation.

[0077] In this embodiment of the invention, a physical constraint model is used to analyze the operating characteristics, equipment capacity, power limits, and dynamic response constraints of each resource to determine the availability range of each resource within the current scheduling cycle. Furthermore, based on the available power range, response rate, and constraints of different resources, and considering both scheduling objectives and system requirements, a corresponding response allocation scheme is generated. This response allocation scheme can achieve reasonable allocation and coordination among resources while ensuring system security and physical feasibility.

[0078] In this embodiment of the invention, specific adjustment operations are performed on the set of callable resources within each time window according to the response allocation scheme, including: power output adjustment, energy storage charging and discharging control, or controllable load reduction, etc. By gradually applying the response allocation scheme in continuously rolling time windows, rolling call of resource responses is achieved, ensuring that each resource participates in regulation according to priority and hierarchical strategies within its availability range. The target response call results of each resource throughout the entire real-time operation cycle are obtained, reflecting the actual execution effect of the virtual power plant in achieving the demand response target during continuous scheduling.

[0079] This invention utilizes a resource response potential table and a hierarchical scheduling strategy for rolling optimization, enabling dynamic allocation of various adjustable resources throughout continuous operation cycles. This fully leverages the core role of highly reliable resources while rationally utilizing medium- and low-reliability resources for auxiliary adjustment. This method not only ensures the demand response capability and system stability of the virtual power plant within different time windows but also improves scheduling efficiency and economy, achieving optimal resource utilization and continuous controllable response calls. It provides a reliable guarantee for the safe, efficient, and intelligent operation of the virtual power plant.

[0080] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0081] like Figure 2 The diagram shown is a functional block diagram of a virtual power plant uncertainty demand response potential analysis and invocation system provided in an embodiment of the present invention.

[0082] This disclosure provides a system for analyzing and invoking the uncertainty demand response potential of a virtual power plant. This system corresponds one-to-one with the method for analyzing and invoking the uncertainty demand response potential of a virtual power plant described in the above embodiments. Figure 2 As shown, the virtual power plant uncertainty demand response potential analysis and invocation system 100 includes a constraint model construction module 101, a cause-effect graph generation module 102, a latent variable embedding module 103, a surrogate observation construction module 104, a response potential analysis module 105, and a response invocation module 106. Detailed descriptions of each functional module are as follows: The constraint model construction module 101 is used to collect the resource operation status set of the target virtual power plant and construct a physical constraint model based on the resource operation status set; The cause-effect graph generation module 102 is used to perform causal analysis on the set of resource operating states using the physical constraint model, and generate an initial cause-effect graph based on the analysis results; The latent variable embedding module 103 is used to identify latent variables with implicit dependencies in the resource running state set, and embed the latent variables into the initial causal graph to obtain the updated causal graph; The surrogate observation construction module 104 is used to obtain the direct influencing variables of the latent variables, and to construct the updated causal graph using the direct influencing variables to obtain the target causal graph. The response potential analysis module 105 is used to perform demand response potential analysis on the target virtual power plant using the target cause-effect graph, and obtain a resource response potential table and a hierarchical scheduling strategy. The response call module 106 is used to perform rolling optimization calls on the target virtual power plant using the resource response potential table and the hierarchical scheduling strategy to obtain the target response call result.

[0083] In one embodiment, the constraint model construction module 101, when performing the construction of a physical constraint model based on the set of resource operating states, is used for: The resource operation status set is timestamped to obtain an aligned resource operation status set; The set of aligned resource operating states is normalized to obtain a set of standard resource operating states; Extract the key operational characteristics of the standard resource operational status set; Obtain the constraint range of each key operational feature, and construct the parameter constraint function of each key operational feature based on the constraint range; The physical constraint model of the target virtual power plant is generated based on all the parameter constraint functions.

[0084] In one embodiment, the causal graph generation module 102 performs causal analysis on the set of resource operating states using the physical constraint model, and generates an initial causal graph based on the analysis results, for the purpose of: Obtain the causal structure edge constraints, and construct an initial causal search space based on the causal structure edge constraints and the physical constraint model; Independence tests and conditional correlation analyses are performed on the standard resource operating state set to obtain the causal direction of each resource variable within the standard resource operating state set; The initial causal search space is corrected using the causal direction of the variables to obtain an updated causal search space; Identify the target causal link in the updated causal search space; Each resource variable in the standard resource operating state set is used as a node in the cause-effect graph. Generate causal graph edges based on the target causal link and the updated causal search space; An initial causal graph is generated based on the causal graph nodes and the causal graph edges.

[0085] In one embodiment, the latent variable embedding module 103 identifies latent variables with implicit dependencies within the set of resource operating states, and uses them to: Identify groups of potentially related variables with implicit dependencies within the set of resource runtime states; Calculate the covariance matrix of the potential associated variable group; Principal component analysis is performed on the covariance matrix to generate potential common factor structures; Extract the principal component loading vectors of the potential common factor structure to obtain the potential factor structure; Convert the latent factor structure into a latent factor vector; The latent factor vector is normalized and made interpretable to obtain latent variables.

[0086] In one embodiment, the latent variable embedding module 103, when performing the embedding of the latent variables into the initial causal graph to obtain an updated causal graph, is used for: Embed the latent variables into the initial causal graph; The initial causal graph after embedding is modified by applying directional constraints to obtain the modified causal graph. The modified causal graph is then subjected to edge weight correction using a Bayesian structure learning method to obtain an updated causal graph.

[0087] In one embodiment, the surrogate observation construction module 104 performs surrogate observation construction on the updated causal graph using the directly influencing variables to obtain a target causal graph, for the following purposes: Correlation analysis was performed on the latent variables and the directly influencing variables to obtain the time series correlation. Calculate the lagged correlation between the latent variables and the directly influencing variables; Conditional correlation analysis was performed on the latent variables and the directly influencing variables to obtain the conditional independence of the directly influencing variables; Candidate proxy variables for the latent variables are selected from the directly influencing variables based on the time series correlation, the lagged correlation, and the conditional independence. The candidate proxy variables are linearly mapped to obtain the proxy observations of the latent variables; By establishing a mapping relationship between the proxy observations and the latent variables, the proxy resource nodes of the latent variables are obtained; The proxy resource node is used to replace the latent variable in the updated causal graph to obtain the extended causal graph; The extended causal graph is dynamically intervened and verified to obtain the target causal graph.

[0088] In one embodiment, the proxy observation construction module 104 performs dynamic intervention verification on the extended causal graph to obtain a target causal graph, which is used for: Obtain the adjustable resources of the target virtual power plant, and construct a small-scale intervention experimental scheme based on the extended causal graph and the adjustable resources; According to the small-scale intervention experimental scheme, a preset intervention signal is sent to the adjustable resources of the target virtual power plant; Collect multidimensional response data after receiving the intervention signal; The intervention signal and the multidimensional response data are time-aligned to generate an intervention response time series. The causal association strength of each resource variable in the extended causal graph is calculated based on the intervention response time series. A causal-weighted regression was performed on the strength of the causal association to obtain the significance of the causal relationship; The extended causal graph is modified based on the significance of the causal relationship to obtain the target causal graph.

[0089] In one embodiment, the response potential analysis module 105 performs demand response potential analysis on the target virtual power plant using the target causal graph to obtain a resource response potential table and a hierarchical scheduling strategy, for the following purposes: Obtain the historical operation data and external scene information of the target virtual power plant, and construct multiple typical operation scenarios based on the historical operation data and external scene information; Generate a set of response intervention actions based on the adjustable resources of the target virtual power plant; Using the external scene information and the set of response intervention actions, conditional sampling is performed on each resource variable in the target causal graph to obtain the response output distribution for each typical operating scenario; Response characteristic data and cost characteristic data for each resource variable are generated based on the response output distribution; Based on the response characteristic data and the cost characteristic data, the response capabilities of each resource variable in the target virtual power plant under the typical operating scenario are summarized to generate a resource response potential table; Based on the resource response potential table, each resource variable is classified by response reliability to obtain high-reliability resources, medium-reliability resources, and low-reliability resources. Based on the resource response potential table, a hierarchical scheduling strategy is generated for the high-reliability resources, the medium-reliability resources, and the low-reliability resources.

[0090] In one embodiment, the response invocation module 106 performs rolling optimization invocation on the target virtual power plant using the resource response potential table and the hierarchical scheduling strategy to obtain the target response invocation result, for the following purposes: Obtain the real-time operating cycle of the target virtual power plant, and divide the real-time operating cycle into multiple continuously rolling time windows; The set of callable resources within each time window is identified based on the resource response potential table and the hierarchical scheduling strategy. The availability range of the callable resource set is confirmed using the physical constraint model. A response allocation scheme for the callable resource set is generated based on the availability range; Within the continuously scrolling time window, the resource response is invoked step by step using the response allocation scheme to obtain the target response invocation result of the target virtual power plant.

[0091] In this invention, the specific limitations of the uncertainty demand response potential analysis and invocation system for a virtual power plant can be found in the above-described limitations of the uncertainty demand response potential analysis and invocation method for a virtual power plant, and will not be repeated here. Each module in the aforementioned uncertainty demand response potential analysis and invocation system for a virtual power plant can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can invoke and execute the operations corresponding to each module.

[0092] In the embodiments provided by this invention, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0093] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0094] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0095] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0096] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0097] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0098] In the embodiments provided in this disclosure, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0099] It should be noted that, in this disclosure, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element limited by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0100] The above-described 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for analyzing and invoking uncertainty demand response potential of a virtual power plant, characterized in that, The method comprises: collecting a resource running state set of a target virtual power plant, and constructing a physical constraint model based on the resource running state set; performing causal analysis on the resource running state set by using the physical constraint model, and generating an initial causal graph; identifying a latent variable between resources, embedding the latent variable into the initial causal graph, and forming an updated causal graph; obtaining a direct influence variable of the latent variable, constructing optimization of the updated causal graph through proxy observation, and obtaining a target causal graph; performing demand response potential analysis based on the target causal graph, and outputting a resource response potential table and a hierarchical scheduling strategy; performing rolling optimization calling on the virtual power plant by using the resource response potential table and the hierarchical scheduling strategy, and obtaining a final response result.

2. The method of claim 1, wherein, The collecting a resource running state set of a target virtual power plant, and constructing a physical constraint model based on the resource running state set, comprises: timestamp alignment on the resource running state set to obtain an aligned resource running state set; data processing on the resource running state set to obtain a standard resource running state set; extraction of key running features from the standard resource running state set; obtaining constraint ranges of each key running feature, and constructing a parameter constraint function of each key running feature; integration of all the parameter constraint functions to generate the physical constraint model of the target virtual power plant. 3.The virtual power plant's uncertainty demand response potential analysis and invocation method of claim 2, wherein, The performing causal analysis on the resource running state set by using the physical constraint model, and generating an initial causal graph based on the analysis result, comprises: obtaining a causal structure edge limit condition, combining the physical constraint model to construct an initial causal search space; independence test and conditional correlation analysis on the standard resource running state set to determine the causal direction of each resource variable; modifying the initial causal search space by using the variable causal direction to obtain an updated causal search space; identifying a target causal link in the updated causal search space; regarding each resource variable as a causal graph node, determining a causal graph edge according to the target causal link and the updated causal search space; generating the initial causal graph based on the causal graph node and the causal graph edge.

4. The method of claim 1, wherein, The identifying a latent variable between resources comprises: identifying a potential associated variable group with implicit dependency relationship in the resource running state set; calculating a covariance matrix of the potential associated variable group; performing principal component analysis on the covariance matrix to generate a potential common cause structure; extracting a principal component load vector of the potential common cause structure to obtain a potential factor structure; converting the potential factor structure into a potential factor vector; performing data processing on the potential factor vector to obtain a latent variable for causal analysis.

5. The method for analyzing and invoking the uncertainty demand response potential of the virtual power plant according to claim 1, wherein, The identifying a latent variable between resources, embedding the latent variable into the initial causal graph, and forming an updated causal graph, comprises: embedding the latent variable into the initial causal graph; performing directional constraint modification on the embedded initial causal graph to obtain a modified causal graph; performing edge weight modification on the modified causal graph by using a Bayesian structure learning method to obtain an updated causal graph.

6. The method for analyzing and invoking uncertainty demand response potential of virtual power plant according to claim 1, wherein, The obtaining a direct influence variable of the latent variable, constructing optimization of the updated causal graph through proxy observation, and obtaining a target causal graph, comprises: performing correlation, lag correlation and conditional independence analysis on the latent variable and the direct influence variable thereof; Based on the analysis results, candidate proxy variables are selected from the direct influence variables; Linear mapping is performed on the candidate proxy variables to generate proxy observation values of latent variables; The proxy observation values are constructed as proxy resource nodes to replace the latent variables in the updated causal graph, forming an extended causal graph; Dynamic intervention verification is performed on the extended causal graph to obtain a target causal graph.

7. The method for analyzing and invoking uncertainty demand response potential of virtual power plant according to claim 6, wherein, Dynamic intervention verification is performed on the extended causal graph to obtain a target causal graph, including: Obtain the adjustable resources of the target virtual power plant, and construct a small-amplitude intervention experiment scheme based on the extended causal graph; According to the scheme, an intervention signal is applied to the adjustable resources, and subsequent multi-dimensional response data is collected; The intervention signal and the response data are time-aligned to generate an intervention-response time series; According to the time series, the causal correlation strength of each resource variable in the extended causal graph is calculated; Causal weighted regression is performed on the causal correlation strength to obtain the causal relationship significance; According to the causal relationship significance, the extended causal graph is modified to obtain the target causal graph.

8. The method for analyzing and invoking uncertainty demand response potential of virtual power plant according to claim 7, wherein, Based on the target causal graph, demand response potential analysis is performed to output a resource response potential table and a hierarchical scheduling strategy, including: Obtain the operation history data and external scenario information of the target virtual power plant to construct multiple typical operation scenarios; According to the adjustable resources of the target virtual power plant, a set of response intervention actions is generated; Using the external scenario information and the set of response intervention actions, condition sampling is performed on the resource variables in the target causal graph to obtain the response output distribution under each typical operation scenario; According to the response output distribution, response feature data and cost feature data of each resource variable are generated; The response capabilities of each resource variable under different typical operation scenarios are summarized to generate a resource response potential table; According to the resource response potential table, the resource variables are divided into reliability classes to obtain reliable resource categories; Based on the resource response potential table, a hierarchical scheduling strategy is generated for different reliable resource categories. 9.The virtual power plant's uncertainty demand response potential analysis and invocation method of claim 1, wherein, The resource response potential table and the hierarchical scheduling strategy are used to perform rolling optimization calling on the virtual power plant to obtain the final response result, including: Obtain the real-time operation cycle of the target virtual power plant and divide the operation cycle into multiple continuous rolling time windows; According to the resource response potential table and the hierarchical scheduling strategy, identify the callable resource set in each time window; Use the physical constraint model to confirm the availability range of the callable resource set; According to the availability range, generate a response allocation scheme for the callable resource set; In the continuously rolling time windows, use the response allocation scheme to gradually call the resource response of the callable resource set to obtain the target response calling result of the target virtual power plant.

10. An uncertainty demand response potential analysis and invocation system for a virtual power plant, characterized in that, The system includes: A constraint model construction module for collecting a set of resource operation states of a target virtual power plant and constructing a physical constraint model based on the set of resource operation states; A causal graph generation module for performing causal analysis on the resource operation states using the physical constraint model to generate an initial causal graph; A latent variable embedding module for identifying latent variables between resources and embedding them into the initial causal graph to form an updated causal graph; An agent observation construction module is configured to acquire a direct influence variable of the latent variable, optimize the update causal graph through agent observation construction, and obtain a target causal graph; A response potential analysis module is configured to perform demand response potential analysis based on the target causal graph, and output a resource response potential table and a hierarchical scheduling strategy; A response calling module is configured to perform rolling optimization calling on a virtual power plant by using the resource response potential table and the hierarchical scheduling strategy, and obtain a final response result.

Images