Hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plant oriented to heterogeneous resources

Abstract

According to the heterogeneous resource-oriented virtual power plant hierarchical clustering-cooperative game aggregation and trusted interaction system provided by the invention, refined and structured representation of a strong heterogeneous resource set is realized by constructing a dynamic digital twinborn archive and adaptive clustering of resources, and a hybrid prediction model fusing physical rules and data driving is adopted, so that the reliability of the system is improved, and the reliability of the system is improved. Physical rationality of behavior prediction and quantitative capture of uncertainty are ensured; through a hierarchical distributed optimization architecture of a master-slave-alliance collaborative game, a global complex optimization problem is efficiently decoupled, and real-time collaborative optimal scheduling under large-scale resource access is realized while privacy and autonomy of a resource main body are protected; through an online rolling credibility regulation capability interval evaluation mechanism, a diversified capability vector with a confidence level is output, so that the virtual power plant becomes a reliable transaction object which is transparent to a power grid and has a risk known, and the interaction credibility and credit rating of the virtual power plant in the power market are fundamentally enhanced.

Description

A hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants with heterogeneous resources Technical Field

[0001] This invention relates to the fields of smart energy systems, power cyber-physical systems, and artificial intelligence, and in particular to a hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants with heterogeneous resources. Background Technology

[0002] With the deepening of the construction of new power systems and the acceleration of energy structure transformation, urban virtual power plants, as a key technological carrier for integrating massive, heterogeneous distributed resources, are playing an increasingly important role in enhancing grid resilience, promoting renewable energy consumption, and achieving coordinated interaction between power generation, grid, load, and storage. The resource types aggregated by urban virtual power plants are extremely complex and diverse, including industrial interruptible loads, flexible loads in commercial buildings (such as air conditioning and lighting systems), distributed wind and solar power systems, user-side energy storage equipment, and electric vehicle charging piles with vehicle-to-grid interaction capabilities. These resources differ significantly in physical characteristics, response dynamics, control costs, ownership, and operational objectives, forming a typical complex resource set that is "highly heterogeneous, high-dimensional, and dynamically time-varying." Currently, the industry mainly uses centralized monitoring platforms to collect data and issue commands for various resources, and employs optimization algorithms for unified scheduling, aiming to maximize the overall economic benefits of virtual power plants or specific grid service objectives.

[0003] However, existing mainstream virtual power plant operation technologies have gradually revealed several fundamental bottlenecks in practical applications. First, there is the problem of coarse aggregation: existing systems typically employ simple algebraic summation or simplified aggregation models based on a single type of resource, failing to deeply characterize the physical heterogeneity between different types of resources and the differences in individual behavior within the same type of resource. This coarse aggregation method makes it difficult to accurately predict and model the overall output characteristics, regulation potential, and dynamic response performance of the aggregated virtual power plant, resulting in a significant discrepancy between the declared regulation capacity and the actual deliverable capacity when participating in the electricity market, affecting market credibility and transaction efficiency. Second, there is the dilemma of centralized optimization: facing tens of thousands or even millions of resource objects connected to city-level virtual power plants, traditional centralized optimization methods suffer from a severe "curse of dimensionality," with computational complexity increasing exponentially. It is difficult to find a feasible control scheme that simultaneously satisfies the global optimal objective and all individual operational constraints within the strict time limits of real-time grid dispatch (which typically requires minute-level or even second-level response). Simultaneously, the uploading and centralized processing of massive amounts of data also brings enormous communication bandwidth pressure and privacy leakage risks. Secondly, there is a crisis of trust in the interaction: When participating in the electricity spot market or ancillary services market, virtual power plant operators (VPS) need to declare their available regulation capacity and reliability to the grid dispatching agency. Because the existing system lacks an effective mechanism for quantitatively assessing the uncertainty of their aggregation capabilities (mainly stemming from fluctuations in renewable energy output, randomness in load response, communication delays, and user behavior uncertainty), their declared values ​​are often just "point estimates" lacking probability confidence support. This opaque capability declaration leads to insufficient trust from the grid dispatching agency due to the inability to accurately assess their reliability, making it difficult to use them as reliable backup regulation resources, severely restricting the large-scale commercial application of VPS in critical ancillary services markets. Finally, there is a lack of synergistic incentives: The existing system architecture focuses on technical aggregation and control, lacking a mechanism to automatically, transparently, and verifiably link the technical execution effects with market settlement and economic benefit distribution. Resource subjects (especially the vast number of small and medium-sized users) find it difficult to clearly perceive the correspondence between their contributions and benefits, resulting in low enthusiasm and sustainability for participating in regulation services, hindering the healthy development of the VPS ecosystem.

[0004] Therefore, there is an urgent need to solve core technical challenges in this field, such as how to achieve refined modeling and representation of heterogeneous resources in virtual power plants, how to overcome the computational and communication bottlenecks in the collaborative optimization of massive resources, how to build a transparent and reliable market interaction capability verification system, and how to establish an automatically executable economic incentive closed loop, so as to promote urban virtual power plants from simple "resource aggregation" to a new stage of efficient "reliable collaboration" operation. Summary of the Invention

[0005] The purpose of this invention is to provide a hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants with heterogeneous resources, so as to solve the problems existing in the prior art.

[0006] To achieve the above objectives, the present invention provides the following solution: The present invention provides a hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources, comprising: a resource digital profiling and adaptive dynamic clustering module, used to construct a dynamic digital twin profile for each connected physical resource, and dynamically cluster the resources based on a multi-dimensional feature vector using an adaptive online clustering algorithm; a cluster behavior hybrid prediction module, used to construct a hybrid prediction model for each resource cluster, including a physical engine layer and a data-driven correction layer, to achieve joint prediction of the deterministic trend and uncertain distribution of the cluster aggregation behavior; a hierarchical distributed optimization scheduling module, using a master-slave-alliance hybrid game architecture, to achieve collaborative optimization scheduling between the virtual power plant central controller and each resource cluster alliance through iterative interaction; an online rolling trusted adjustment capability range evaluation module, used to simulate and output the trusted adjustment capability range of the virtual power plant in future periods based on distributed robust optimization or scenario stochastic optimization methods; and an automatic performance verification and revenue distribution module, used to encrypt and store the adjustment process data on the blockchain, and automatically verify the execution effect and complete the revenue distribution through smart contracts.

[0007] Preferably, the dynamic digital twin archive includes static parameters of the resource, real-time operating data, historical response characteristics, reliability statistics, cost function coefficients, and correlation characteristics related to the operating environment.

[0008] Preferably, the adaptive online clustering algorithm is an incremental DBSCAN or an improved spectral clustering algorithm, which can dynamically adjust the resource cluster to which the resource belongs based on the real-time behavioral pattern drift of the resource.

[0009] Preferably, the physics engine layer embeds differential equations or constraints that describe the general physical laws of this type of resource; the data-driven correction layer uses a lightweight neural network, taking real-time state, environmental signals and historical behavior residuals as input, to predict the distribution of group behavior bias and uncertainty.

[0010] Preferably, in the master-slave-alliance hybrid game architecture, the virtual power plant central controller, as the master leader, issues total power regulation demand and price incentive signals to each resource cluster alliance; each resource cluster alliance, as the slave leader, allocates regulation tasks within the alliance through cooperative game and iteratively interacts with the central controller through the projection gradient method or consensus algorithm until a Stackelberg equilibrium is reached.

[0011] Preferably, the credible adjustment capability range includes: a guaranteed output range, a desired output baseline, and a flexibility potential range, along with corresponding confidence levels; the flexibility potential range includes internal risk pricing information.

[0012] Preferably, in the automatic performance verification and profit distribution module, the smart contract has pre-set assessment standards for the adjustment service, automatically compares the execution data stored on the chain with the standards at the agreed verification time, and automatically triggers profit distribution according to the preset profit distribution rules.

[0013] This invention also provides a hierarchical clustering-cooperative game aggregation and trusted interaction method for virtual power plants with heterogeneous resources, comprising: creating multi-dimensional digital profiles of the heterogeneous resources connected to the virtual power plant, constructing dynamic digital twin archives, and performing dynamic clustering using an adaptive online clustering algorithm; for each resource cluster, constructing and updating a hybrid prediction model with enhanced physical information to predict the deterministic trend and uncertainty distribution of its aggregation behavior; based on a master-slave-alliance cooperative game architecture, performing distributed iterative optimization between the virtual power plant central controller and each resource cluster alliance to generate the optimal cooperative scheduling instruction; performing online rolling evaluation of trusted regulation capability ranges and outputting a multi-dimensional trusted capability vector containing the guaranteed output range, the expected output baseline, and the elastic potential range; and using blockchain smart contracts to perform encrypted storage, automatic verification, and revenue distribution throughout the entire regulation service process.

[0014] Preferably, in the dynamic clustering process, the clustering criteria include the physical attributes of the resources, historical response delay, adjustment accuracy, reliability statistics, cost characteristics, and environmental correlation characteristics.

[0015] The present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements a hierarchical clustering-cooperative game aggregation and trusted interaction method for virtual power plants oriented towards heterogeneous resources.

[0016] This invention achieves the following beneficial technical effects compared to existing technologies: The invention provides a hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources, characterized by refined modeling, distributed and efficient optimization, trusted quantification of capabilities, and automatic incentive closure. Specifically, by constructing dynamic digital twin archives of resources and adaptive clustering, it achieves refined and structured representation of highly heterogeneous resource sets, revolutionarily improving the predictability and modeling accuracy of virtual power plant aggregation behavior; it employs a hybrid prediction model that integrates physical laws and data-driven approaches, ensuring the physical rationality of behavior predictions and the quantitative capture of uncertainties; through a hierarchical distributed optimization architecture of master-slave-alliance cooperative game, it efficiently decouples the globally complex optimization problem, protecting the privacy and autonomy of resource subjects while achieving real-time, collaborative optimal scheduling under large-scale resource access, greatly improving the system's scalability and computational efficiency; The online rolling trusted adjustment capability range assessment mechanism outputs diversified capability vectors with confidence levels, making virtual power plants reliable trading objects that are transparent to the power grid and whose risks are known. This fundamentally enhances their interactive credibility and credit rating in the electricity market. Finally, by leveraging blockchain smart contract technology, a fully automated and tamper-proof trust infrastructure for the entire process of "regulation-verification-settlement" is constructed. This seamlessly and transparently links technical execution with economic incentives, creating a fair and credible profit distribution environment. This effectively stimulates the enthusiasm of various resource entities to participate continuously and with high quality, providing key support for building a prosperous and self-driven virtual power plant business ecosystem. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments 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.

[0018] Figure 1 is a system architecture diagram of hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources provided by the present invention; Figure 2 is a flowchart of the method of hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources provided by the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The purpose of this invention is to provide a hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants with heterogeneous resources, in order to solve the problems existing in the prior art.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] Example 1: As shown in Figure 1, the virtual power plant hierarchical clustering-cooperative game aggregation and trusted interaction system for heterogeneous resources provided by this invention adopts a "cloud-edge-device" collaborative overall architecture. A central control and decision-making platform for the virtual power plant is deployed in the cloud as the core of the system, integrating a resource digital profiling and adaptive dynamic clustering module, a physical information-enhanced cluster behavior hybrid prediction module, a hierarchical distributed optimization scheduling module based on alliance cooperative game theory, an online rolling trusted adjustment capability range assessment module, and an automatic performance verification and revenue distribution module based on blockchain smart contracts. Edge aggregation controllers are deployed in locations such as industrial parks and commercial complexes near the resource side, acting as proxy nodes for each resource cluster alliance, responsible for the coordination and management of resources within the cluster and local optimization calculations. The bottom layer consists of various heterogeneous resource intelligent terminals, including data acquisition and execution units installed on air conditioning units, photovoltaic inverters, energy storage equipment, electric vehicle charging piles, and industrial production lines.

[0023] The system workflow is shown in Figure 2, and its specific implementation process is as follows: First, the system performs refined access and dynamic digital profiling of massive heterogeneous resources. Each accessed physical resource (such as a commercial air conditioner, an energy storage cabinet, an adjustable production line, or a photovoltaic array) continuously reports its static parameters (such as rated power, capacity, and physical location), real-time operating data (such as current power, status, and temperature), and historical interaction records through its smart terminal. The resource digital profiling and adaptive dynamic clustering module collects this data and builds and continuously updates a dynamic digital twin profile for each resource. This profile not only contains basic static information, but more importantly, it extracts and updates its "dynamic response feature vector" through machine learning methods, including but not limited to: historical average response delay and variance, adjustment accuracy to commands, reliability statistics within a certain period (such as success rate), cost function coefficients fitted according to changes in market signals, and features strongly correlated with the environment (such as the thermal inertia coefficient of the building envelope of the air conditioner, and the historical cleanliness and efficiency degradation curve of the photovoltaic panel). By aggregating these high-dimensional features, the system gives each resource a digital identity that transcends traditional monitoring and reflects its physical nature and behavioral patterns.

[0024] Based on the formation of rich digital profiles, the system performs adaptive dynamic clustering. The module employs an improved incremental DBSCAN clustering algorithm (or a spectral clustering algorithm based on similarity metrics) to perform online clustering analysis according to the physical type and dynamic response feature vectors of resources. This algorithm can automatically identify high-density regions in the feature space, aggregating resources with similar response characteristics, homogeneous physical constraints, and similar cost functions into the same "resource cluster." For example, a group of energy storage units with fast response speeds, high charging and discharging efficiency, and good reliability records are clustered into a "high-quality fast response cluster"; residential rooftop photovoltaic systems, which are significantly affected by weather, have large output fluctuations, and require significant user intervention, are clustered into a "high-uncertainty elasticity cluster." The clustering process is not instantaneous but dynamic. The system continuously monitors the behavior patterns of each resource. When a significant drift in the response characteristics of a resource is detected (such as equipment performance degradation or changes in user habits), the adaptive algorithm dynamically adjusts its cluster to ensure that the clustering results always reflect the true state of the resources, thus laying a solid foundation for subsequent accurate prediction and optimized scheduling.

[0025] For each stably formed resource cluster, the cluster behavior hybrid prediction module is responsible for constructing a prediction model for its aggregate output. This model is one of the core innovations of this invention and is a physical information neural network architecture. The bottom layer of the model is the "physical engine layer," which is not trained from pure data but pre-embedded with mathematical expressions or constraints describing the general physical laws of this type of resource group. For example, for air conditioning clusters, the physical engine layer embeds building thermodynamic differential equations to describe the relationship between temperature changes and cooling power, outdoor temperature, and building heat capacity; for energy storage clusters, it embeds a battery model that considers charging and discharging efficiency and aging rate. This layer ensures that the predicted trend strictly follows physical laws, avoiding the "model illusion" or absurd outputs that violate fundamental principles such as energy conservation that may occur with purely data-driven models. Above the physical engine layer is the "data-driven correction layer," which typically employs a lightweight gated recurrent unit network. This network takes as input real-time aggregated status of resources within a cluster, short-term external environmental signals (such as electricity price forecasts, temperature forecasts, and irradiance forecasts for the next moment), and historical prediction residuals. It specifically learns and predicts "collective behavior biases" and "uncertainty distributions" that physical models cannot fully cover. This hybrid structure enables the model to output a deterministic output baseline that conforms to physical laws, while simultaneously providing a probability distribution or confidence interval that characterizes uncertainty. This achieves joint prediction of the future behavior of resource clusters based on both "trend" and "uncertainty envelope".

[0026] Based on the accurate prediction models of each resource cluster, the system enters the collaborative optimization scheduling phase. The hierarchical distributed optimization scheduling module adopts a master-slave-alliance hybrid game architecture to overcome the curse of dimensionality in centralized optimization. In this architecture, the virtual power plant central controller acts as the "master leader," aiming to maximize the overall operating revenue of the virtual power plant or minimize the total cost while satisfying the regulation commands issued by the power grid or the market clearing results. It does not directly schedule each specific resource, but rather, based on the prediction models reported by each cluster, it publishes a total power regulation demand signal and a dynamic price incentive signal to each "resource cluster alliance." Each resource cluster forms an autonomous "alliance" at the edge. Within the alliance, each resource agent (i.e., cluster member) engages in internal cooperative game theory on how to fairly and efficiently allocate the regulation tasks to be undertaken by its cluster under the received cluster-level incentive signal. The goal of the game is to minimize the total scheduling cost within the alliance or maximize the overall satisfaction of all members while satisfying each member's own operational constraints (such as the comfort temperature range of air conditioning, the power demand of electric vehicles, and the safe state of charge of energy storage). The central controller and each cluster alliance solve the problem through iterative interaction, such as using the projected gradient method. The central controller adjusts its issued incentive signals based on the marginal costs or scheduling capabilities reported by each alliance; each alliance, based on the new incentive signals, re-engages internally, adjusts its allocation scheme, and reports new cost information. This process iterates rapidly until it converges to a Stackelberg equilibrium, at which point the central controller's objective and the collective interests of each alliance are synergistically optimal, and no one can benefit from unilaterally changing their strategy. This distributed architecture decouples the originally ultra-high-dimensional global optimization problem into a coordination problem between the central controller and multiple cluster alliances, and a local optimization problem within each alliance, greatly reducing computational complexity and communication overhead, and naturally protecting the privacy and autonomy of resource owners.

[0027] To provide the electricity market with a transparent and credible statement of regulation capacity, the online rolling credible regulation capacity range assessment module periodically (e.g., every 15 minutes) performs a rapid forward-looking simulation. Based on the latest resource status, updated forecasting models, and environmental and market signal predictions for multiple future time periods, the module uses distributed robust optimization or scenario-based stochastic optimization methods to simulate the aggregated regulation capacity distribution of a virtual power plant under various typical and extreme uncertainty scenarios (such as sudden drops in photovoltaic power or load response falling short of expectations) within a future time window (e.g., the next 4 hours). The simulation output is not a single numerical value, but a "credible capacity vector" containing multi-dimensional information and accompanying confidence levels. For each future scheduling period, this vector comprises three key components: first, the "guaranteed output range," which refers to the range of upward or downward power adjustments that a virtual power plant can absolutely provide under extremely high confidence (e.g., 99%). This capability is highly reliable and can be submitted to the grid as a "reliability backup product"; second, the "expected output baseline," which is the statistically most likely output value and can serve as the basis for submitting applications for participation in energy market transactions; and third, the "elastic potential range," which refers to the additional regulation capacity that can be provided under higher confidence (e.g., 90%). This capability carries certain risks but can be leveraged to obtain premiums through participation in high-yield markets such as frequency regulation. The system can also automatically calculate an "internal risk pricing" for the elastic potential portion based on historical performance and risk models, providing a quantitative basis for operators to formulate differentiated market pricing strategies.

[0028] Finally, to establish a trustworthy and automated incentive loop, an automated performance verification and revenue distribution module based on blockchain smart contracts is integrated throughout the entire adjustment service. When an adjustment instruction is issued, a smart contract is created and deployed on a permissioned consortium blockchain, pre-defined with the service's performance criteria, such as response time window, allowable error for adjustment accuracy, and duration. During the adjustment service execution, key action instructions from each resource smart terminal, and precise metering data at start and end times (such as meter readings and power curves), are processed using standardized formats and hashed, then uploaded to the blockchain network in real-time or near real-time as immutable transaction records. After the service ends, at the preset verification time, the on-chain smart contract automatically triggers, reads all relevant execution data stored on the chain, and compares it with the pre-defined performance criteria for verification. The verification process is fully automated and transparent, eliminating the possibility of human intervention. Once verification is successful, the smart contract immediately executes the transfer of digital currency or generates legally valid electronic settlement vouchers according to pre-set and agreed-upon profit-sharing rules (e.g., based on the proportion of electricity adjusted or tiered rewards based on response speed), accurately and promptly distributing the profits to the corresponding accounts of each resource entity. This process achieves a seamless and reliable link from "technical execution" to "economic incentives," greatly enhancing the participation and trust of resource entities, especially small and medium-sized entities, and providing crucial infrastructure support for the sustainable development of the virtual power plant ecosystem.

[0029] Through the organic combination and closed-loop operation of the above steps, the system of the present invention realizes the fine modeling, efficient collaborative optimization, transparent and reliable interaction and automated incentive allocation of massive, heterogeneous and dynamic distributed resources in city-level virtual power plants, effectively solving the key bottlenecks in the existing technology.

[0030] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0031] It should be noted that the components mentioned in the above embodiments are all general standard parts or components known to those skilled in the art. Their structures and principles can be learned by those skilled in the art through technical manuals or conventional experimental methods.

[0032] This invention has illustrated its principles and implementation methods using specific examples. The descriptions of these embodiments are merely illustrative of the method and its core ideas; furthermore, those skilled in the art will recognize that modifications may be made to the specific implementation methods and application scope based on the principles of this invention. Therefore, the content of this specification should not be construed as limiting the invention.

Claims

1. A hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants with heterogeneous resources, characterized in that: include: The resource digital profiling and adaptive dynamic clustering module is used to build a dynamic digital twin profile for each connected physical resource and dynamically cluster the resources based on an adaptive online clustering algorithm using multi-dimensional feature vectors; the cluster behavior hybrid prediction module is used to build a hybrid prediction model for each resource cluster, including a physical engine layer and a data-driven correction layer, to achieve joint prediction of the deterministic trend and uncertain distribution of the cluster's aggregation behavior; the hierarchical distributed optimization scheduling module adopts a master-slave-alliance hybrid game architecture, and achieves collaborative optimization scheduling through iterative interaction between the virtual power plant central controller and each resource cluster alliance. The online rolling trusted regulation capability range assessment module is used to simulate and output the trusted regulation capability range of the virtual power plant in the future time period based on distributed robust optimization or scenario stochastic optimization methods; the automatic performance verification and revenue distribution module is used to encrypt and store the regulation process data on the blockchain, and automatically verify the execution effect and complete the revenue distribution through smart contracts.

2. The hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources according to claim 1, characterized in that, The dynamic digital twin archive includes static parameters of the resources, real-time operating data, historical response characteristics, reliability statistics, cost function coefficients, and correlation characteristics related to the operating environment.

3. The hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources according to claim 1, characterized in that, The adaptive online clustering algorithm is an incremental DBSCAN or an improved spectral clustering algorithm, which can dynamically adjust the resource cluster to which the resource belongs based on the real-time behavior pattern drift of the resource.

4. The hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources according to claim 1, characterized in that, The physics engine layer embeds differential equations or constraints that describe the general physical laws of this type of resource; the data-driven correction layer uses a lightweight neural network, taking real-time state, environmental signals and historical behavior residuals as input, to predict the distribution of group behavior bias and uncertainty.

5. The hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources according to claim 1, characterized in that, In the master-slave-alliance hybrid game architecture, the virtual power plant central controller, as the master leader, issues total power regulation demand and price incentive signals to each resource cluster alliance; each resource cluster alliance, as the slave leader, allocates regulation tasks through cooperative game within the alliance and iteratively interacts with the central controller through the projection gradient method or consensus algorithm until a Stackelberg equilibrium is reached.

6. The hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources according to claim 1, characterized in that, The credible adjustment capability range includes: guaranteed output range, expected output baseline, and elastic potential range, along with corresponding confidence levels; the elastic potential range includes internal risk pricing information.

7. The hierarchical clustering-cooperative game aggregation and trusted interaction system for virtual power plants oriented towards heterogeneous resources according to claim 1, characterized in that, In the automatic performance verification and profit distribution module, the smart contract has pre-set assessment standards for the adjustment service. At the agreed verification time, it automatically compares the execution data stored on the chain with the standards and automatically triggers profit distribution according to the preset profit distribution rules.

8. The hierarchical clustering-cooperative game aggregation and trusted interaction method for virtual power plants oriented towards heterogeneous resources according to any one of claims 1-7, characterized in that, include: Multi-dimensional digital profiles are created for the heterogeneous resources connected to the virtual power plant, dynamic digital twin archives are constructed, and dynamic clustering is performed using an adaptive online clustering algorithm; For each resource cluster, a hybrid prediction model with enhanced physical information is constructed and updated to predict the deterministic trend and uncertainty distribution of its aggregation behavior. Based on a master-slave-alliance collaborative game architecture, distributed iterative optimization is performed between the virtual power plant central controller and each resource cluster alliance to generate collaborative optimal scheduling instructions. Online rolling execution of trusted regulation capability range assessment is performed, outputting a multi-dimensional trusted capability vector that includes the guaranteed output range, the expected output baseline, and the elastic potential range. Blockchain smart contracts are used to encrypt and store evidence, automatically verify, and distribute revenue throughout the entire regulation service process.

9. The method for hierarchical clustering-cooperative game aggregation and trusted interaction of virtual power plants for heterogeneous resources according to claim 8, characterized in that, In the dynamic clustering process, the clustering criteria include the physical attributes of resources, historical response delay, adjustment accuracy, reliability statistics, cost characteristics, and environmental correlation characteristics.

10. An electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the hierarchical clustering-cooperative game aggregation and trusted interaction method for virtual power plants oriented towards heterogeneous resources as described in claim 8.

Images