A virtual power plant intelligent decision system and method based on hierarchical cooperative control

Abstract

This invention relates to the field of intelligent control technology for virtual power plants, and discloses an intelligent decision-making system and method for virtual power plants based on hierarchical collaborative control. The method includes: constructing a three-dimensional unified model; constructing an elastic virtual clock system to achieve collaborative operation of systems at different time scales; developing a time-aware topology reconstruction engine to optimize the control topology by analyzing the time characteristics of nodes; implementing a two-dimensional fault timing recovery mechanism to collaboratively handle node faults and time deviations; and constructing a spatiotemporal elasticity assessment system to achieve real-time monitoring of system status and comprehensive evaluation of spatiotemporal elasticity capabilities. This invention, by employing hierarchical collaborative control technology, unifies time and space dimensions within an elastic control framework, achieving dynamic adaptation of the virtual power plant's control systems at each level in terms of time characteristics and topology.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for virtual power plants, and more specifically, to an intelligent decision-making system and method for virtual power plants based on hierarchical collaborative control. Background Technology

[0002] Virtual power plants, as a comprehensive energy management system that integrates various heterogeneous energy resources such as distributed energy, controllable loads, and energy storage systems, have played an important role in improving the flexibility of power systems and the consumption of new energy sources in recent years. With the development of energy internet and smart grid technologies, the scale and complexity of virtual power plants are constantly increasing, and their efficient and stable operation places higher demands on the control system.

[0003] Existing virtual power plant control systems primarily employ a hierarchical collaborative control architecture, dividing the system into multiple layers such as equipment, control, and management. Each layer operates independently based on its own functions and time characteristics, achieving overall goals through a certain coordination mechanism. However, with the expansion of system scale and the increasing complexity of the operating environment, existing technologies exhibit significant shortcomings in the following aspects:

[0004] The control systems at different levels of a virtual power plant operate on different time scales. For example, the equipment level typically requires millisecond-level rapid response, while the market level has a decision-making cycle of hours. This significant difference in time scale leads to timing inconsistencies in cross-level collaborative control. Existing technologies often employ fixed time synchronization mechanisms or simple data interpolation methods to handle timing differences. However, in highly dynamic environments, these methods struggle to meet the system's requirements for real-time performance and flexibility, impacting the effectiveness of collaborative control. In actual operation, the control network structure of a virtual power plant may frequently change due to node failures, communication interruptions, or the dynamic access and exit of resources. Existing technologies typically employ static backup mechanisms or simple fault detection and recovery strategies, which are insufficient for flexible adaptation to different types and scales of faults. This is especially true in complex scenarios where multiple nodes fail simultaneously, compromising the system's topological stability and overall stability. Operational reliability is difficult to guarantee effectively. Existing technologies generally treat timing management and topology control as two independent problems, ignoring the inherent coupling relationship between them. In fact, changes in the topology directly affect the system's time synchronization performance, and deviations in time synchronization affect the effectiveness of topology reconstruction. This fragmented approach makes it difficult for the system to coordinate recovery strategies in both time and space dimensions under complex faults or extreme operating conditions, affecting overall recovery efficiency and system resilience. Existing hierarchical collaborative control methods often adopt a unified communication protocol and fixed interaction mode when dealing with heterogeneous systems, lacking the ability to dynamically adjust collaborative strategies according to the characteristics of different levels. This leads to low system resource utilization, especially in highly dynamic and uncertain operating scenarios, where collaborative control efficiency further declines, making it difficult to meet the flexibility and efficiency requirements of virtual power plants.

[0005] In summary, existing virtual power plant control systems have technical bottlenecks in terms of timing consistency, topology stability, spatiotemporal coordination, and collaborative control efficiency, and new technical solutions are urgently needed to improve the overall performance and operational reliability of the system. Summary of the Invention

[0006] This invention provides a virtual power plant intelligent decision-making system and method based on hierarchical collaborative control, which solves the technical problems of timing differences, topology stability, spatiotemporal coupling and low efficiency of collaborative control in related technologies.

[0007] This invention provides a smart decision-making method for virtual power plants based on hierarchical collaborative control, comprising:

[0008] Construct a unified three-dimensional model of time, topology, and control to integrate the time synchronization, topology, and control strategies of the hierarchical control system of the virtual power plant into a single framework;

[0009] Based on the output of a unified 3D model, a flexible virtual clock system is constructed to enable collaborative operation of systems with different time scales.

[0010] Based on the node time characteristics generated by the elastic virtual clock system, a time-aware topology reconstruction engine is developed to optimize the control topology by analyzing the time characteristics of nodes.

[0011] Based on the optimized control topology, a two-dimensional fault timing recovery mechanism is implemented to collaboratively handle node faults and time deviations.

[0012] By combining the operational results of a three-dimensional unified model, an elastic virtual clock system, a topology reconstruction engine, and a fault timing dual-dimensional recovery mechanism, a spatiotemporal elasticity assessment system is constructed to achieve real-time monitoring of system status and comprehensive assessment of spatiotemporal elasticity capabilities.

[0013] Furthermore, the steps for constructing the unified three-dimensional model of time, topology, and control include:

[0014] Establish a spatiotemporal elastic control mapping function to map system state, time mapping, and topology to control output;

[0015] A spatiotemporal elastic optimization objective function is constructed, comprising three parts: control performance loss, topology robustness assessment, and time synchronization loss.

[0016] Define mathematical representations of the contradictions between time consistency and system distribution, topological stability and dynamic adaptability, and local autonomy and global coordination;

[0017] Design a model training and update strategy to train the parameters of a three-dimensional unified model based on historical operation data of a virtual power plant.

[0018] Furthermore, the steps for constructing the flexible virtual clock system include:

[0019] A multi-scale time feature extraction module is constructed to extract the inherent time characteristic parameters of various control systems in the virtual power plant;

[0020] Design a flexible time mapping function to perform virtual time mapping on nodes in a virtual power plant based on their time characteristics;

[0021] Implement a virtual clock synchronization algorithm so that the virtual clocks of each node maintain the consistency of the system's logical time while preserving the time scale characteristics;

[0022] Develop a time scale conversion interface to achieve data consistency mapping between systems with different time scales.

[0023] Furthermore, the steps for developing the time-aware topology reconstruction engine include:

[0024] Construct a time-aware topology evaluation model, taking node temporal characteristics as an important factor in topology evaluation;

[0025] Implement an adaptive topology reconfiguration algorithm to dynamically adjust the control topology.

[0026] Establish a function migration and reorganization mechanism to migrate critical functions from faulty nodes to healthy nodes;

[0027] It enables predictive topology adjustments, predicting failure risks based on system operating status and historical data.

[0028] Furthermore, the steps for implementing the fault timing dual-dimensional recovery mechanism include:

[0029] Construct a fault timing joint diagnosis module to analyze system fault conditions and time synchronization status simultaneously;

[0030] Design a spatiotemporal collaborative optimization algorithm that takes time synchronization and topology reconstruction as unified optimization objectives;

[0031] Implement an adaptive recovery strategy selection mechanism to select the appropriate recovery strategy based on the fault type;

[0032] Construct a spatiotemporal self-healing capability assessment system to evaluate the spatiotemporal self-healing capability of a hierarchical collaborative control system for a virtual power plant.

[0033] Furthermore, the steps for constructing the spatiotemporal elasticity assessment system include:

[0034] Design a spatiotemporal elasticity measurement index system, including time elasticity index, spatial elasticity index, collaborative elasticity index, and recovery elasticity index;

[0035] Achieve multi-dimensional visualization of system health status and intuitively display the spatiotemporal elasticity of the system;

[0036] Construct a control strategy optimization module based on spatiotemporal elasticity to continuously optimize the control strategy based on the system's spatiotemporal elasticity assessment results;

[0037] Establish a spatiotemporal pattern library to collect and analyze the optimal spatiotemporal configuration of the system under different scenarios.

[0038] Furthermore, the elastic time mapping function is:

[0039] Node in physical time The corresponding virtual time is composed of the integral value of the node's time scaling function from the initial time to the current time plus the initial time mapping value;

[0040] The time scaling factor of a node is equal to its base time scaling factor multiplied by a dynamic adjustment function;

[0041] The dynamic adjustment function takes into account three factors: system state, topology, and node priority, and dynamically adjusts the base time scaling rate.

[0042] Furthermore, the time-aware topology evaluation model includes a time characteristic compatibility function. The time characteristic compatibility function evaluates the degree of time characteristic matching of nodes by calculating the negative exponential function value of the absolute difference between the basic time scaling rates of two nodes. This makes the compatibility function value of a node with smaller time characteristic differences closer to 1, and the compatibility function value of a node with larger time characteristic differences closer to 0.

[0043] Furthermore, the core optimization problem solved by the spatiotemporal collaborative optimization algorithm is:

[0044] Minimize the comprehensive loss function under the constraints of the topology set and the time synchronization state set;

[0045] A distributed solution strategy is adopted, combining the alternating direction multiplier method and the distributed gradient descent method to achieve efficient solution of the collaborative optimization problem.

[0046] This invention provides a virtual power plant intelligent decision-making system based on hierarchical collaborative control, used to execute the aforementioned virtual power plant intelligent decision-making method based on hierarchical collaborative control, comprising:

[0047] The three-dimensional unified model module for time, topology, and control is used to integrate the time synchronization, topology, and control strategies of the hierarchical control system of a virtual power plant into a single framework;

[0048] The flexible virtual clock system module is used to enable the coordinated operation of systems with different time scales;

[0049] The time-aware topology reconstruction engine module is used to optimize and control the topology structure based on the temporal characteristics of nodes;

[0050] A two-dimensional fault timing recovery mechanism module is used to collaboratively handle node failures and time deviations;

[0051] The spatiotemporal elasticity assessment system module is used to monitor the system status in real time and assess its spatiotemporal elasticity capability.

[0052] The beneficial effects of this invention are as follows: by adopting hierarchical collaborative control technology, the time and space dimensions are unified in the elastic control framework, realizing the dynamic adaptation of the control systems at each level of the virtual power plant in terms of time characteristics and topology;

[0053] Through spatiotemporal collaborative optimization technology and a two-dimensional recovery mechanism, the system achieves high reliability in complex environments;

[0054] By using a flexible virtual clock system and time-aware topology reconstruction technology, the efficiency of cross-level collaborative control has been effectively improved.

[0055] By unifying time synchronization and topology control strategies into a single control framework, the complexity of system design, debugging, and maintenance is reduced.

[0056] By introducing predictive maintenance technology, early warning and proactive prevention of system anomalies were achieved.

[0057] Through an adaptive evolution mechanism, the system achieves self-learning and continuous optimization capabilities during long-term operation. Attached Figure Description

[0058] Figure 1 This is a flowchart of a virtual power plant intelligent decision-making method based on hierarchical collaborative control in this invention;

[0059] Figure 2 This is a bar chart comparing the spatiotemporal elasticity method of this invention with traditional methods in terms of design complexity, number of debugging parameters, and maintenance cost.

[0060] Figure 3 It is a bar chart comparing the time scale differences of different levels of control systems in a simulated power plant;

[0061] Figure 4 This is a bar chart comparing the performance of spatiotemporal collaborative methods and traditional methods in terms of fault recovery. Detailed Implementation

[0062] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0063] At least one embodiment of the present invention discloses a smart decision-making method for virtual power plants based on hierarchical collaborative control, such as... Figure 1 As shown, it includes:

[0064] Step 1: Construct a unified three-dimensional model of time, topology, and control, integrating the time synchronization, topology, and control strategies of the virtual power plant hierarchical control system into a single framework;

[0065] This step utilizes spatiotemporal elastic control theory to construct a unified mathematical model, integrating the time synchronization, topology, and control strategies of the virtual power plant hierarchical control system into a single framework, laying a theoretical foundation for subsequent spatiotemporal collaborative optimization.

[0066] Specifically, it includes the following sub-steps:

[0067] Step 1.1: Establish the spatiotemporal elastic control mapping function;

[0068] The spatiotemporal elastic control mapping function is expressed as follows: At any given moment, the set of control commands generated by the system is determined by the system state, topology, and time mapping function. These three input variables are mapped to control outputs through spatiotemporal elastic control mapping.

[0069] Spatiotemporal elastic control mapping This represents the mapping relationship from system state, topology, and time to control output; system state It includes the operating parameters of each node in the virtual power plant; topology. ,express Node connections in a time-matter system; time mapping function It is used to coordinate the interaction between systems with different time scales.

[0070] function The specific implementation is as follows: First, the three input parameters of system state, topology, and time mapping are processed by feature extraction and encoding. Then, feature fusion is performed through a multi-layer neural network structure. The input layer receives three feature vectors. The encoder layer uses a multi-layer perceptron to extract features for system state, a graph convolutional network to extract features for topology, and a long short-term memory network to extract features for time mapping. The fusion layer uses an attention mechanism to weight and fuse the three features. Finally, the decoder layer converts the fused spatiotemporal feature vectors into a specific set of control commands. The entire mapping process can dynamically adjust the weights of the parameters in each layer according to the system operating state to achieve adaptive control strategy generation.

[0071] Spatiotemporal elastic control mapping It can be implemented as a deep learning model composed of multiple neural networks, specifically including:

[0072] Input layer: Receives system status Topology and time mapping eigenvectors;

[0073] Encoder layer: Extracts and encodes features from the three inputs respectively, generating corresponding feature representations;

[0074] Fusion layer: Fuses the three feature representations to form a unified spatiotemporal feature vector;

[0075] Decoder layer: Decodes the fused feature vectors into a set of control instructions. .

[0076] The feature extraction and fusion process is as follows:

[0077] System state encoding: Extracting system state features using a multi-layer perceptron (MLP):

[0078] ;

[0079] in, This represents the system state feature vector; This represents the multilayer perceptron encoding function for the system state; for System state parameters at time 1.

[0080] Topology encoding: Extracting topology features using a Graph Convolutional Network (GCN):

[0081] ;

[0082] in, It is a topological feature vector; For graph convolutional network encoding functions; for System topology information at any given time.

[0083] Temporal mapping encoding: Extracting temporal features using a Long Short-Term Memory (LSTM) network.

[0084] ;

[0085] in, This is a time-related feature vector; For Long Short-Term Memory (LSTM) network encoding functions; for The time mapping function at any given moment.

[0086] Feature fusion: Three features are fused through an attention mechanism.

[0087] ;

[0088] in, The fused spatiotemporal feature vector; This is the attention mechanism fusion function; This represents the system state feature vector; It is a topological feature vector; This is a time-related feature vector.

[0089] function The specific implementation is as follows: First, the similarity matrix between the three input feature vectors is calculated. Then, the similarity is converted into attention weights through the softmax function. Next, each feature vector is multiplied by its corresponding attention weight to obtain weighted features. Finally, all weighted features are linearly combined to obtain the fused spatiotemporal feature vector. This mechanism can automatically learn the importance of different features and dynamically adjust the contribution of each feature when the system state changes, thereby achieving a more accurate feature fusion effect.

[0090] Control output generation: Control commands are generated via the decoder.

[0091] ;

[0092] in, for A set of control commands generated at any given time; For decoder functions; This is the fused spatiotemporal feature vector.

[0093] The specific implementation is as follows: First, the fused spatiotemporal feature vector is transformed in dimension through a fully connected layer, then nonlinearly transformed through a multilayer perceptron structure, then the features are mapped to the control command space through the output layer, and finally the output is limited to a reasonable control range through an activation function. This decoder can generate corresponding control commands according to different combination modes of fused features, realizing the mapping and transformation from spatiotemporal features to specific control actions.

[0094] It should be noted that, in some embodiments, spatiotemporal elastic control mapping Other implementation methods are also possible, such as rule-based fuzzy control systems or hybrid control strategies. Especially in scenarios with limited computational resources, a less computationally complex implementation can be chosen.

[0095] Step 1.2: Construct the spatiotemporal elastic optimization objective function;

[0096] The system minimizes the comprehensive loss function by solving for the optimal topology and time mapping function, thereby achieving spatiotemporal co-optimization.

[0097] The comprehensive loss function consists of three parts: control performance loss, topology robustness assessment, and time synchronization loss. By introducing weighting coefficients, these three parts are weighted and summed to obtain the final optimization objective function. Control performance loss represents the deviation between the current control strategy and the ideal control effect; topology robustness assessment quantifies the ability of the current topology to resist failures; and time synchronization loss measures the degree of time consistency among nodes in the system.

[0098] Among them, control performance loss Indicates the deviation between the current control strategy and the ideal control effect; topological robustness assessment. Quantify the current topology's ability to withstand faults; time synchronization loss The weighting coefficients measure the time consistency of nodes within the system, including weights for control performance penalties. Weights for topological robustness evaluation Time synchronization loss weights It can be dynamically adjusted based on the system's operating status. For example, it can be increased when the system is running stably. Prioritize ensuring control performance; increase efficiency in high-failure-risk environments. To improve topological robustness; and to increase [the effectiveness of] cross-level collaborative control tasks. To ensure time synchronization performance.

[0099] It is important to note the weighting of control performance loss. Weights for topological robustness evaluation Time synchronization loss weights Normalization is required to ensure that their sum equals 1, avoiding numerical biases caused by loss functions with different dimensions during weighted calculation. Simultaneously, each loss function itself needs to be standardized to unify the indicators of different dimensions within the same numerical range, ensuring the effectiveness and comparability of weight adjustments.

[0100] Step 1.3: Define the mathematical representation of the three core contradictions;

[0101] The contradiction between time consistency and system distribution: using a flexible time mapping function Establish mapping relationships between systems with different time scales and introduce the concept of a virtual time domain to achieve logical time consistency while maintaining the distributed characteristics of the system;

[0102] The contradiction between topological stability and dynamic adaptability: Constructing a topology evaluation function Seeking a balance between stability and adaptability;

[0103] Conflict between local autonomy and global coordination: Design a coordination mechanism based on spatiotemporal constraints, and define the autonomous decision-making boundaries of nodes and global coordination rules.

[0104] Step 1.4, design model training and update strategies;

[0105] Based on historical operating data of virtual power plants, reinforcement learning methods are used to train the parameters of a three-dimensional unified model, and a dynamic update mechanism for the model based on changes in the operating environment is established to ensure that the model can adapt to long-term changes in the power system.

[0106] In some embodiments, model training can employ a distributed reinforcement learning architecture, where each level of control unit acts as an agent, and the overall system performance serves as a reward signal, continuously optimizing the control strategy through interactive learning. Furthermore, transfer learning techniques can be utilized to transfer model knowledge trained under one operating condition to a new operating condition, accelerating the model adaptation process.

[0107] like Figure 2 As shown, this invention improves system complexity by unifying time synchronization and topology control strategies into a single control framework. Compared to traditional methods (100%), the spatiotemporal elastic method reduces design complexity by 50%, the number of debugging parameters by 65%, and maintenance costs by 55%. These data verify that this invention, through a unified three-dimensional model of "time, topology, and control," reduces the complexity of system design, debugging, and maintenance, thereby improving the maintainability and practicality of the system.

[0108] Step 2: Based on the output of the three-dimensional unified model, construct a flexible virtual clock system to achieve coordinated operation of systems with different time scales;

[0109] According to one embodiment of this application, this step utilizes an elastic time mapping algorithm to construct a virtual clock system capable of enabling coordinated operation of systems with different time scales, thereby solving the timing difference problem of control systems at different levels in a virtual power plant and improving the time consistency of cross-level coordinated control.

[0110] Step 2.1: Construct a multi-scale temporal feature extraction module;

[0111] For various control systems within the virtual power plant, their inherent time characteristic parameters, including control cycle, response delay, and calculation time, are extracted to establish a time characteristic database, which serves as the basic data for the flexible virtual clock system.

[0112] like Figure 3 As shown, different levels of control systems in a virtual power plant exhibit differences in time scale. Equipment-level control systems typically require millisecond-level rapid responses, regional-level control systems have decision cycles in seconds, system-level control systems have control cycles in minutes, and market-level control systems have decision cycles in hours. This significant leap in time scale (from milliseconds to hours) validates the necessity of multi-time-scale collaborative control and provides a theoretical basis for the flexible virtual clock system of this invention.

[0113] Step 2.2, design the elastic time mapping function;

[0114] For any node in the virtual power plant Its elastic time mapping function is defined as follows: integrating the node's time scaling rate function from the initial time to the current time, and then adding the initial time mapping value. This definition allows virtual time to evolve continuously with physical time, while also dynamically adjusting the evolution rate according to the system state.

[0115] Among the nodes In physical time The corresponding virtual time is composed of the integral of the node's time scaling function from the initial time to the current time, plus the initial time mapping value. The time scaling function is calculated as follows:

[0116] A node's time scaling factor equals its base time scaling factor multiplied by a dynamic adjustment function. This adjustment function is based on three factors: the current system state, the topology, and the node priority, allowing the time scaling factor to adapt adaptively to the system's operating conditions.

[0117] Among them, the basic time scaling rate and the node The control cycle is related; the adjustment function dynamically adjusts the time scaling rate based on system state, topology and node priority.

[0118] The specific implementation of the elastic time mapping function includes:

[0119] Base time scaling Determination:

[0120] For device-level nodes, such as generator control systems, =0.001 (corresponding to a millisecond-level control cycle);

[0121] For regional level nodes, such as distribution network control systems, =0.1 (corresponding to a second-level control cycle);

[0122] For system-level nodes, such as scheduling systems, =1.0 (corresponding to minute-level control cycle);

[0123] For market-level nodes, such as trading systems, =10.0 (corresponding to hourly control cycle).

[0124] It is important to note that the base time scaling factor varies significantly across different node levels (from 0.001 to 10.0). This variation can lead to numerical instability when calculating the time compatibility function. Therefore, in practical applications, it is necessary to perform a logarithmic transformation or standardization on the base time scaling factor to unify time parameters of different dimensions into a similar numerical range, ensuring the stability and accuracy of the calculation.

[0125] Implementation of the adjustment function:

[0126] System state influence factor: First, calculate the difference between the current system state value and the normal state value. Then, divide this difference by the normal state value to obtain the system state deviation rate. Next, multiply this deviation rate by the system state weighting coefficient and add 1 to the result to obtain the system state influence factor. This factor reflects the degree of influence of the current system state on the time scaling factor. The greater the deviation of the system state from the normal value, the more significant the influence.

[0127] It is important to note that system state parameters typically include various types of indicators (such as voltage, power, and frequency), which have different dimensions and numerical ranges. Before calculating the state deviation rate, it is necessary to standardize each state parameter, unifying indicators with different dimensions to the same numerical range to ensure the accuracy and comparability of the state influence factor calculation.

[0128] Topology influence factor: First, calculate the difference between the current node's centrality index and the average centrality of all nodes in the system. Then, divide this difference by the average centrality to obtain the node centrality deviation rate. Next, multiply this deviation rate by the topology weight coefficient and add 1 to the result to obtain the topology influence factor. This factor reflects the degree to which the importance of a node in the topology affects the time scaling rate; nodes with higher centrality have a greater influence.

[0129] It is important to note that node centrality indices typically include various types such as degree centrality, betweenness centrality, and proximity centrality, each with different numerical ranges and distribution characteristics. Before calculating the centrality deviation rate, it is necessary to normalize each centrality index, unifying centrality indices with different dimensions into the range of 0 to 1, to ensure the stability and consistency of the calculation of topological influence factors.

[0130] Priority Influence Factor: First, calculate the difference between the current priority of a node and the average priority of all nodes in the system. Then, divide this difference by the average priority to obtain the priority deviation rate. Next, multiply this deviation rate by the priority weighting coefficient and add 1 to the result to obtain the priority influence factor. This factor reflects the degree to which node priority affects time scaling; nodes with higher priorities receive more time resources.

[0131] It is important to note that node priorities typically encompass multiple evaluation metrics (such as functional importance, resource requirements, and failure risk), which may have different data types and numerical ranges. Before calculating the priority deviation rate, each priority metric needs to be standardized to unify priorities with different dimensions into the same numerical range, ensuring the accuracy and fairness of the priority influence factor calculation.

[0132] The comprehensive adjustment function multiplies the system state influence factor, topology influence factor, and priority influence factor to obtain the comprehensive adjustment function value. This function dynamically adjusts the base time scaling rate by comprehensively considering system state, topology, and node priority, thereby achieving a reasonable allocation of time resources.

[0133] Discrete implementation of virtual time computation:

[0134] In practical systems, discrete-time approximation is used to calculate virtual time:

[0135] In actual calculations, the virtual time of the next moment is calculated as the virtual time of the current moment plus the product of the current time scaling factor and the physical time interval. That is, the virtual time of the current node is incremented in physical time according to its time scaling factor. This discrete calculation method facilitates the updating of virtual time in actual systems.

[0136] It should be understood that in different embodiments, different time mapping function implementations can be selected based on the specific application scenario and computing resources. For example, a simplified linear mapping function can be used for edge nodes with limited computing power, while a more complex nonlinear mapping function can be used for core control nodes to obtain more precise time control.

[0137] Step 2.3: Implement the virtual clock synchronization algorithm;

[0138] Design a distributed clock synchronization algorithm based on a consensus mechanism to ensure that the virtual clocks of each node maintain their own timescale characteristics while ensuring the consistency of the system's logical time. The algorithm mainly includes:

[0139] Local clock status broadcast: Each node periodically broadcasts its own virtual clock status;

[0140] Clock skew calculation: Calculate the local clock skew based on the received clock states of other nodes;

[0141] Time scaling adjustment: The time scaling is dynamically adjusted based on clock skew to reduce system time inconsistency.

[0142] Step 2.4, develop the time scale conversion interface;

[0143] Construct a cross-timescale data transformation interface to achieve data consistency mapping between systems with different timescales, including:

[0144] Downsampling transformation: an aggregation method for converting high-frequency data into low-frequency data;

[0145] Upward interpolation transformation: an interpolation method for converting low-frequency data into high-frequency data;

[0146] Time-series forecasting transformation: Predicting data values ​​at a specific point in time based on historical data.

[0147] Step 3: Based on the node time characteristics generated by the elastic virtual clock system, develop a time-aware topology reconstruction engine to optimize the control topology by analyzing the node time characteristics.

[0148] In one embodiment of this application, this step constructs a reconstructing engine that can optimize the control topology based on the node time characteristics, solves the topology stability problem of the virtual power plant control network under fault conditions, and improves the robustness of the system in complex environments.

[0149] Step 3.1: Construct a time-aware topology evaluation model;

[0150] By incorporating the temporal characteristics of nodes as an important factor in topology evaluation, a time-aware topology evaluation function is defined:

[0151] The topology evaluation function assesses the robustness of the overall topology by calculating the weighted connectivity values ​​between all pairs of nodes in the network. For each pair of nodes, the product of their connection weights and topological distance is first calculated. Then, this product is multiplied by the time compatibility function value between the nodes. Finally, the results for all pairs of nodes are summed to obtain the overall topology evaluation value.

[0152] The time-required compatibility function is defined as follows:

[0153] The specific implementation of the time characteristic compatibility function is as follows: First, calculate the absolute difference between the base time scaling rates of the two nodes, and then multiply this difference by the scaling factor. The compatibility index is obtained, and finally, the negative exponential function value of the index is taken. As a result of the compatibility function, this function is designed so that the compatibility value of nodes with smaller time characteristic differences is closer to 1, and the compatibility value of nodes with larger time characteristic differences is closer to 0. This allows for priority connection of nodes with similar time characteristics during topology reconstruction, thereby improving the timing consistency of the system.

[0154] in and They are nodes and The base time scaling factor This is the scaling factor; This represents an exponential function. Nodes with similar time characteristics have higher compatibility values ​​and are more suitable for maintaining connections in the topology.

[0155] In other possible embodiments, the time characteristic compatibility function can also take other forms, such as a function based on the node control cycle ratio. This function first calculates the ratio of the base time scaling rates of the two nodes, and then takes the smaller of this ratio and its reciprocal as the compatibility function value. This design ensures that the function result remains between 0 and 1 regardless of which node has a larger time scaling rate; the function value is 1 when the time characteristics of the two nodes are exactly the same, and gradually approaches 0 as the difference increases.

[0156] Step 3.2: Implement the adaptive topology reconstruction algorithm;

[0157] Based on a time-aware topology evaluation model, a reconstruction algorithm capable of dynamically adjusting and controlling the topology structure is designed. The algorithm mainly includes:

[0158] Fault detection and location: Quickly identify faulty nodes and connections in the network;

[0159] Candidate topology generation: Generates multiple feasible reconstruction schemes based on the current network state;

[0160] Topology evaluation and selection: The performance of each candidate topology is evaluated using a time-aware topology evaluation function, and the optimal solution is selected.

[0161] Smooth transition strategy: Design a smooth transition mechanism for the topology to avoid system instability caused by sudden changes.

[0162] The specific implementation of the adaptive topology reconstruction algorithm includes:

[0163] Fault detection and location:

[0164] Heartbeat detection mechanism: Nodes periodically send heartbeat messages; if no heartbeat is received within a preset time, the node is deemed to be malfunctioning.

[0165] Link quality monitoring: Real-time monitoring of link communication quality indicators, such as latency and packet loss rate; when an indicator exceeds a threshold, the link is deemed abnormal.

[0166] It's important to note that latency in link quality metrics is typically measured in milliseconds, while packet loss rate is expressed as a percentage. These two metrics, with their different units of measurement, require standardization when comprehensively evaluating link quality. Latency metrics need to be normalized, unifying latency values ​​across different links to the range of 0 to 1; packet loss rate metrics need to be converted, transforming percentage values ​​into decimals between 0 and 1, ensuring that both metrics have equal weight and comparability in link quality assessment.

[0167] Fault confirmation mechanism: Faults are confirmed through multi-path verification to avoid false positives;

[0168] Fault Scope Assessment: Determine the scope and severity of the fault's impact to provide a basis for subsequent refactoring.

[0169] Candidate topology generation:

[0170] A graph-based topology generation algorithm, considering the following constraints:

[0171] Connectivity constraints: ensure that network connectivity is not reduced;

[0172] Timing compatibility constraint: Prioritize connecting nodes with similar timing characteristics;

[0173] Load balancing constraints: to prevent some nodes from being overloaded with communication.

[0174] A genetic algorithm (GA) is used to generate multiple candidate topologies that satisfy the constraints.

[0175] For large networks, a partitioning strategy is adopted, and the affected areas are reconstructed locally first.

[0176] Topology evaluation and selection:

[0177] For each candidate topology Calculate the overall score:

[0178] ;

[0179] in, Representing candidate topology Overall score; , , These represent the weights for robustness, communication efficiency, and conversion cost, respectively. Representing candidate topology Robustness score; Indicates from the current topology Switch to Conversion cost score; Representing candidate topology The communication efficiency score is calculated by evaluating candidate topologies. The weighted average delay of all communication paths is obtained, and the specific implementation is as follows:

[0180] ;

[0181] in, Represents a set of nodes; Represents a node With nodes The weighting coefficients between them typically reflect the importance of communication or traffic demand; Indicating in candidate topology Next node and Communication delay between them; The summation symbol is used.

[0182] It is important to note that the comprehensive score calculation involves multiple scoring items with different dimensions (robustness score, communication efficiency score, and conversion cost score), which have different numerical ranges and distribution characteristics. Before performing weighted summation, each scoring item needs to be standardized to unify the scores of different dimensions to the same numerical range, ensuring that the weighting coefficients are within the same range. , , The validity and comparability of the data must be ensured. At the same time, the weighting coefficients themselves need to be normalized to ensure their sum equals 1, thus avoiding numerical bias.

[0183] function The specific implementation is as follows: First, the robustness score, communication efficiency score, and switching cost score of the candidate topology are calculated respectively. Then, the weight coefficient of each score item is determined according to the current system status and operation requirements. Next, each score is multiplied by its corresponding weight and summed to obtain the comprehensive score. This scoring function can comprehensively consider the stability of the topology, communication performance, and switching cost, providing a quantitative evaluation basis for topology reconstruction. The topology scheme with the highest score is selected as the final reconstruction scheme.

[0184] The conversion cost score represents the cost of switching from the current topology. Switch to new topology The required resource consumption is calculated as follows:

[0185] ;

[0186] in, Indicates the index of the edge; Representing topology and The symmetric difference set, i.e. the set of edges that need to be added or deleted; Indicates adding or deleting edges The cost; The summation symbol is used.

[0187] The topology scheme with the highest score is selected as the final reconstruction scheme.

[0188] Smooth transition strategy:

[0189] Phased topology switchover: Breaking down topology changes into multiple small steps and implementing them step by step;

[0190] Dual-path temporary transition: During the switchover process, the old path is maintained for a period of time to ensure uninterrupted data transmission;

[0191] Buffer mechanism: Set up a data buffer to ensure that data is not lost during the transition;

[0192] Rollback mechanism: Set checkpoints so that a quick rollback can be performed if a new topology anomaly is detected.

[0193] Step 3.3: Construct a functional migration and reorganization mechanism;

[0194] When a node fails in the network, a function migration and reconfiguration mechanism should be designed to migrate the critical functions of the failed node to healthy nodes, ensuring the overall functional integrity of the system. This mainly includes:

[0195] Functional importance assessment: Rank the functions carried by each node according to their importance;

[0196] Resource requirements analysis: Calculate the resource requirements of each function, including computing and communication.

[0197] Target node matching: Select the most suitable target node for the function to be migrated based on the importance of the function and resource requirements;

[0198] Functional code migration: Enables dynamic migration and deployment of functional modules.

[0199] It is important to note that feature importance assessment typically includes multiple dimensions of metrics (such as feature type, business impact, user dependence, etc.), which may contain both categorical and numerical data. For categorical data (such as feature type), one-hot encoding or label encoding is required to convert non-numerical data into numerical data. For numerical data, standardization is necessary to unify metrics with different dimensions to the same numerical range, ensuring the accuracy and consistency of feature importance assessment.

[0200] Step 3.4: Implement predictive topology adjustments;

[0201] Based on system operating status and historical data, potential fault risks are predicted, and topology adjustments are made in advance to avoid reactive responses after faults occur. This mainly includes:

[0202] Risk assessment model: assesses the failure risk of each node and connection based on monitoring data;

[0203] Preventive reconfiguration: Conducting local topology adjustments in advance for high-risk areas;

[0204] Emergency plan preparation: Develop emergency topology plans in advance for possible failure scenarios.

[0205] It is important to note that fault risk assessment typically involves various types of monitoring data (such as equipment status, environmental parameters, and historical fault records), which have different data types and numerical ranges. For numerical monitoring data, standardization is required to unify indicators with different dimensions into the same numerical range. For categorized data (such as equipment type, fault type, etc.), encoding is necessary to convert non-numerical data into numerical data, ensuring the consistency of input data format and the accuracy of calculations for the risk assessment model.

[0206] Step 4: Based on the optimized control topology, implement a two-dimensional fault timing recovery mechanism to collaboratively handle node faults and time deviations;

[0207] According to one embodiment of this application, this step develops a two-dimensional recovery mechanism capable of collaboratively handling node failures and time deviations, solving the system recovery problem under complex fault conditions, and improving the reliability and adaptability of the hierarchical collaborative control system of the virtual power plant.

[0208] Specifically, it includes the following sub-steps:

[0209] Step 4.1: Construct a fault timing joint diagnosis module;

[0210] Design a diagnostic module capable of simultaneously analyzing system fault conditions and time synchronization status to comprehensively assess the system's health. This mainly includes:

[0211] Fault timing correlation analysis: Identifying the causal relationship between topology faults and timing deviations;

[0212] Fault propagation model: predicting how faults affect time synchronization performance;

[0213] Impact assessment of timing deviations on topology: analyzing how time synchronization issues affect the stability of the topology.

[0214] like Figure 4 As shown, the proposed two-dimensional fault recovery mechanism improves fault recovery performance. Compared with traditional methods, the spatiotemporal collaborative method reduces fault recovery time from 15 minutes to 4 minutes and increases control recovery rate from 25% to 75%. These data verify that by using time synchronization and topology reconstruction as unified optimization objectives, the present invention can achieve rapid and efficient fault recovery, improving the reliability and adaptability of the hierarchical collaborative control system of the virtual power plant.

[0215] Step 4.2, design a spatiotemporal collaborative optimization algorithm;

[0216] Based on a unified three-dimensional model of "time, topology, and control," a collaborative optimization algorithm is constructed that takes time synchronization and topology reconstruction as unified optimization objectives. The core optimization problem is expressed as:

[0217] ;

[0218] ;

[0219] in For the joint loss function, Represents system state variables. This represents the topology to be optimized. This indicates the time synchronization parameters to be optimized. Indicates to and Find the minimum value by combining the results; For topological structure, The set of all feasible topologies. For time synchronization parameters, The set of all acceptable time synchronization states; This indicates a constraint.

[0220] The optimization algorithm employs a distributed solution strategy, combining the Alternating Direction Method of Multipliers (ADMM) and the Distributed Gradient Descent method to achieve efficient solutions to large-scale collaborative optimization problems.

[0221] It is important to note that for large-scale systems, a hierarchical solution strategy can be adopted. First, optimization is performed within local regions, and then the optimization results from each local region are coordinated to obtain the globally optimal solution. Furthermore, in time-sensitive scenarios, approximate solution methods can be used, sacrificing some optimization accuracy for faster solution speed.

[0222] Step 4.3: Implement the adaptive recovery strategy selection mechanism;

[0223] The system intelligently selects the most appropriate recovery strategy based on the type, scope, and severity of the fault. This includes:

[0224] Tiered recovery strategy library: Multiple preset recovery strategies are available for different types and severity of failures;

[0225] Preliminary assessment of recovery outcomes: Predicting the effectiveness of various recovery strategies based on the model;

[0226] Dynamic strategy selection: Select the optimal recovery strategy based on the current system status and pre-assessment results;

[0227] Adaptive strategy adjustment: The recovery strategy is dynamically adjusted based on real-time feedback during the recovery process.

[0228] It is important to note that fault type and severity typically include categorical data (such as fault type, scope of impact, etc.) and numerical data (such as fault duration, number of affected nodes, etc.). Categorical data requires encoding to convert non-numerical data into numerical data; numerical data requires standardization to unify indicators of different dimensions into the same numerical range, ensuring the accuracy and consistency of recovery strategy selection.

[0229] Step 4.4: Construct a spatiotemporal self-healing capability assessment system;

[0230] Develop an index system and evaluation method to assess the spatiotemporal self-healing capability of a hierarchical collaborative control system for virtual power plants, providing a basis for continuous system optimization. Key indicators include:

[0231] Time recovery metrics: speed, stability, and accuracy of time synchronization recovery;

[0232] Topology restoration metrics: integrity, stability, and efficiency of topology restoration;

[0233] Functional recovery indicators: the percentage and quality of system function recovery;

[0234] Collaborative recovery metrics: Time-based and topology-based collaborative recovery efficiency.

[0235] It is important to note that the assessment of spatiotemporal self-healing capabilities involves multiple indicators with different dimensions (such as recovery speed, recovery integrity, and recovery efficiency), which have different numerical ranges and distribution characteristics. Before conducting a comprehensive assessment, it is necessary to standardize each indicator, unifying indicators with different dimensions to the same numerical range to ensure the accuracy and comparability of the assessment results. Simultaneously, for indicators containing categorical data (such as recovery status and fault type), encoding is required to convert non-numerical data into numerical data.

[0236] Step 5: Combining the results of the operation of the three-dimensional unified model, the elastic virtual clock system, the topology reconstruction engine, and the fault timing dual-dimensional recovery mechanism, construct a spatiotemporal elasticity assessment system to achieve real-time monitoring of the system status and comprehensive assessment of spatiotemporal elasticity capabilities.

[0237] According to the embodiments of this application, this step establishes an evaluation system for real-time monitoring of system status and assessment of its spatiotemporal resilience, providing continuous monitoring and optimization support for the hierarchical collaborative control system of virtual power plants.

[0238] Specifically, it includes the following sub-steps:

[0239] Step 5.1: Design a spatiotemporal elasticity measurement index system;

[0240] Construct a comprehensive indicator system for evaluating the spatiotemporal resilience of the system, including:

[0241] Time resilience index: assesses a system's ability to adapt to changes at different time scales;

[0242] Spatial resilience index: assesses the system's ability to adapt to changes in topology;

[0243] Synergistic resilience index: assesses the ability to adapt synergistically across time and space dimensions;

[0244] Resilience metrics: assess a system’s ability to recover its original functionality after a failure.

[0245] It is important to note that the spatiotemporal resilience measurement index system includes multiple evaluation indicators with different dimensions (such as timescale adaptability, topological adaptability, and cooperative adaptability), which have different numerical ranges and distribution characteristics. Before conducting a comprehensive evaluation, it is necessary to standardize each resilience indicator, unifying indicators with different dimensions to the same numerical range to ensure the accuracy and comparability of resilience assessments. Simultaneously, for indicators containing categorical data (such as adaptation type and resilience level), encoding is required to convert non-numerical data into numerical data.

[0246] Step 5.2: Achieve multi-dimensional visualization of system health status;

[0247] Develop a visual interface that intuitively displays the spatiotemporal resilience of the system, enabling operations and maintenance personnel to quickly grasp the system's operational status. Key features include:

[0248] Spatiotemporal elasticity heatmap: Displays the spatiotemporal elasticity level of each region of the system in the form of a heatmap;

[0249] Timing performance trend chart: Displays the historical trend of system time synchronization performance;

[0250] Dynamic topology view: Displays real-time changes in the system's topology;

[0251] Fault Risk Warning Map: Identifies potentially high-risk areas in the system.

[0252] Step 5.3: Construct a control strategy optimization module based on spatiotemporal elasticity;

[0253] Based on the system's spatiotemporal elasticity assessment results, the control strategy is continuously optimized to improve the overall system performance. This mainly includes:

[0254] Elasticity-oriented parameter optimization: Automatically adjusting control parameters based on spatiotemporal elasticity assessment results;

[0255] Preventive reinforcement strategy: Implement preventive reinforcement measures for areas of the system with low resilience;

[0256] Long-term evolution optimization: Analyze long-term system operation data to discover opportunities for improving spatiotemporal elasticity.

[0257] Step 5.4: Establish a spatiotemporal pattern library;

[0258] The system's optimal spatiotemporal configurations under different scenarios are collected and analyzed to form an experience pattern library, providing a rapid response reference for similar scenarios in the future. The pattern library mainly includes:

[0259] Typical scenario mode: Record the optimal spatiotemporal configuration under different operating scenarios;

[0260] Fault response mode: Record effective response strategies for different fault types;

[0261] Load variation patterns: Record the system's optimal adaptation strategy under different load conditions;

[0262] Pattern similarity calculation: Evaluates the similarity between the current scene and historical patterns to assist in pattern selection.

[0263] A virtual power plant intelligent decision-making system based on hierarchical collaborative control, used to execute the aforementioned virtual power plant intelligent decision-making method based on hierarchical collaborative control, includes:

[0264] The three-dimensional unified model module for time, topology, and control is used to integrate the time synchronization, topology, and control strategies of the hierarchical control system of a virtual power plant into a single framework;

[0265] The flexible virtual clock system module is used to enable the coordinated operation of systems with different time scales;

[0266] The time-aware topology reconstruction engine module is used to optimize and control the topology structure based on the temporal characteristics of nodes;

[0267] A two-dimensional fault timing recovery mechanism module is used to collaboratively handle node failures and time deviations;

[0268] The spatiotemporal elasticity assessment system module is used to monitor the system status in real time and assess its spatiotemporal elasticity capability.

[0269] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A virtual power plant intelligent decision-making method based on hierarchical cooperative control, characterized in that, The application relates to a time, topology and control three-dimensional unified model, which integrates time synchronization, topology structure and control strategy of a virtual power plant hierarchical control system into a single framework. Based on the output of the three-dimensional unified model, an elastic virtual clock system is constructed to realize collaborative operation of different time scale systems. Relying on the node time characteristics generated by the elastic virtual clock system, a time sequence-aware topology reconstruction engine is developed to optimize the control topology structure by analyzing the time characteristics of the nodes. The step of constructing the elastic virtual clock system comprises the following steps. A multi-scale time characteristic extraction module is constructed to extract inherent time characteristic parameters of various control systems in the virtual power plant. An elastic time mapping function is designed to virtually map the nodes in the virtual power plant according to their time characteristics. A virtual clock synchronization algorithm is realized to make the virtual clocks of the nodes maintain the time scale characteristics while maintaining the system logical time consistency. A time scale conversion interface is developed to realize the data consistency mapping between different time scale systems. The step of developing the time sequence-aware topology reconstruction engine comprises the following steps. A time sequence-aware topology evaluation model is constructed to take the node time characteristics as an important factor for topology structure evaluation. An adaptive topology reconstruction algorithm is realized to dynamically adjust the control topology structure. A function migration and recombination mechanism is constructed to migrate the key functions of the fault nodes to the healthy nodes. Predictive topology adjustment is realized based on the system operation state and historical data to predict the fault risk. Based on the optimized control topology structure result, a fault time sequence two-dimensional recovery mechanism is realized to cooperatively process the node fault and time deviation. The running results of the three-dimensional unified model, the elastic virtual clock system, the topology reconstruction engine and the fault time sequence two-dimensional recovery mechanism are combined to construct a space-time elasticity evaluation system to realize real-time monitoring of the system state and comprehensive evaluation of the space-time elasticity capability. The step of constructing the time, topology and control three-dimensional unified model comprises the following steps. 2.The virtual power plant intelligent decision-making method based on hierarchical cooperative control according to claim 1, wherein, A space-time elasticity control mapping function is established to map the system state, time mapping and topology structure to the control output. A space-time elasticity optimization objective function is constructed, which comprises three parts of control performance loss, topology robustness evaluation and time synchronization loss. Mathematical expressions of contradictions between time consistency and system distribution, between topology stability and dynamic adaptability and between local autonomy and global coordination are defined. A model training and updating strategy is designed to train the parameters of the three-dimensional unified model based on the historical operation data of the virtual power plant. The step of realizing the fault time sequence two-dimensional recovery mechanism comprises the following steps. 3.The virtual power plant intelligent decision-making method based on hierarchical cooperative control according to claim 1, characterized in that, A fault time sequence joint diagnosis module is constructed to simultaneously analyze the system fault condition and the time synchronization state. A space-time collaborative optimization algorithm is designed to take the time synchronization and topology reconstruction as a unified optimization target. An adaptive recovery strategy selection mechanism is realized to select a suitable recovery strategy according to the fault type. A space-time self-healing capability evaluation system is constructed to evaluate the space-time self-healing capability of the virtual power plant hierarchical collaborative control system. The step of constructing the space-time elasticity evaluation system comprises the following steps.

4. The virtual power plant intelligent decision-making method based on hierarchical cooperative control according to claim 1, characterized in that, A space-time elasticity measurement index system is designed, which comprises time elasticity indexes, space elasticity indexes, collaborative elasticity indexes and recovery elasticity indexes. Multi-dimensional system health state visualization is realized to intuitively display the space-time elasticity state of the system. ​ A control strategy optimization module based on spatiotemporal elasticity is constructed to continuously optimize the control strategy according to the spatiotemporal elasticity evaluation result of the system; A spatiotemporal mode library is established to collect and analyze the optimal spatiotemporal configuration of the system under different scenarios.

5. The virtual power plant intelligent decision-making method based on hierarchical cooperative control according to claim 1, characterized in that, The elastic time mapping function is: The node is in physical time The corresponding virtual time is composed of the integral value of the time stretching function of the node from the initial time to the current time plus the initial time mapping value. The time stretching rate of a node is equal to its basic time stretching rate multiplied by a dynamic adjustment function; The dynamic adjustment function dynamically adjusts the basic time stretching rate by comprehensively considering the system state, topology structure and node priority.

6. The virtual power plant intelligent decision-making method based on hierarchical cooperative control according to claim 1, characterized in that, The time-aware topology evaluation model includes a time characteristic compatibility function, which evaluates the matching degree of the time characteristics of two nodes by calculating the negative exponential function value of the absolute difference of the basic time stretching rates of the two nodes, so that the smaller the time characteristic difference between the nodes, the closer the compatibility function value of the nodes to 1, and the greater the time characteristic difference between the nodes, the closer the compatibility function value of the nodes to 0.

7. The virtual power plant intelligent decision-making method based on hierarchical cooperative control according to claim 3, characterized in that, The core optimization problem solved by the spatiotemporal collaborative optimization algorithm is: Under the constraints of the topology structure set and the time synchronization state set, minimize the comprehensive loss function; A distributed solving strategy is adopted to realize efficient solving of the collaborative optimization problem by combining the alternating direction multiplier method and the distributed gradient descent method.

8. A virtual power plant intelligent decision system based on hierarchical cooperative control, characterized in that, A virtual power plant intelligent decision-making method based on hierarchical collaborative control for executing any one of claims 1-7, comprising: A time, topology and control three-dimensional unified model module for integrating the time synchronization, topology structure and control strategy of the hierarchical control system of the virtual power plant into a single framework; An elastic virtual clock system module for realizing collaborative operation of different time scale systems; A time-aware topology reconfiguration engine module for optimizing the topology structure according to the time characteristics of the nodes; A fault time sequence two-dimensional recovery mechanism module for cooperatively handling node faults and time deviations; A spatiotemporal elasticity evaluation system module for monitoring the system state in real time and evaluating its spatiotemporal elasticity.

Images