Method, device, equipment and storage medium for virtual power plant to regulate resources in seconds

Abstract

The present invention proposes a method, device, equipment and storage medium for regulating resources in a virtual power plant in seconds. By extracting key information features from the operating data of various adjustable resources of edge nodes and uploading them to the cloud. The cloud predicts the optimal regulation strategy and then feeds back the strategy to the edge node through low-latency communication technology. Before local execution, if there is no identical strategy in the cache, the received strategy will be decomposed into sub-strategies and stored. During execution, resource regulation is performed according to the sub-strategy queue in the cache. In addition, sub-strategies suitable for the current situation can be retrieved from local storage for application. The entire process optimizes the cloud-edge collaboration mechanism, reduces the time cost of data transmission and processing, and enables the powerful computing power of the cloud to be fully utilized for data analysis and intelligent decision-making, while the edge focuses on rapid response and instruction execution, jointly realizing efficient and accurate management and scheduling of virtual power plant resources.

Description

Technical Field

[0001] The present invention belongs to the technical field of power resource regulation and control, and specifically relates to a method, device, equipment and storage medium for regulating resources in a virtual power plant in seconds. Background Art

[0002] With the transformation of the global energy structure and the continuous advancement of intelligent technology, Virtual Power Plant (VPP), as a new type of power system operation and management method, is gradually becoming an important development direction of the power industry. Through advanced information and communication technology and software systems, virtual power plants integrate decentralized renewable energy, energy storage equipment, demand response and other resources to form a unified and coordinated power supply network. This integration can not only improve energy utilization efficiency, but also enhance the flexibility and reliability of the power grid, which is of great significance for addressing climate change and promoting sustainable energy development.

[0003] However, virtual power plants face a series of technical challenges in actual operation. First, there are many types of resources in virtual power plants, including distributed photovoltaics, wind power generation, energy storage systems, electric vehicle charging piles, and demand-side management. The operating characteristics and response speeds of these resources vary, which brings difficulties to real-time monitoring and regulation. Secondly, traditional centralized control methods have delays in data transmission and processing, which makes it difficult to meet the virtual power plant's demand for second-level response. In addition, the resource response of virtual power plants is uncertain, and traditional deterministic control methods are difficult to adapt to such changes. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method, device, equipment and storage medium for regulating resources in a virtual power plant in seconds, and to reduce data transmission delay and data calculation processing time by optimizing the cloud-edge collaboration mechanism. By utilizing the powerful computing power and data analysis technology of the cloud control center, the real-time data uploaded by the edge computing nodes is deeply mined and intelligently analyzed and processed to quickly generate the most optimized control strategy. At the same time, through the efficient cloud-edge collaboration mechanism, the cloud control center will quickly convey the generated control strategy to the edge computing node, and the edge computing node will be responsible for sending the control instructions directly to the terminal device for execution, thereby realizing rapid and accurate monitoring and control of virtual power plant resources.

[0005] In order to achieve the above object, the technical solution provided by the present invention is as follows:

[0006] In a first aspect, the present invention provides a method for regulating resources of a virtual power plant in seconds, comprising the following steps:

[0007] Monitor and collect the operating data of various adjustable resources of edge nodes, and extract key information features used to represent the operating status of various adjustable resources from the operating data;

[0008] Sending key information features to the cloud so that the cloud can predict the optimal control strategy based on the key information features using a pre-trained control strategy prediction model, and the optimal control strategy is used to control the adjustable resources;

[0009] Obtain the optimal control strategy from the cloud through low-latency communication technology;

[0010] The optimal control strategy is searched and executed according to the local cache and local storage. In the search stage, if the local cache does not have the same strategy as the optimal control strategy, the optimal control strategy is decomposed into multiple sub-strategies and cached and stored. In the execution stage, various adjustable resources are executed according to the sub-strategy queue in the local cache. The sub-strategies in the sub-strategy queue include sub-strategies obtained by decomposing the optimal control strategy according to the control instructions of the cloud, and sub-strategies searched from the local cache or storage according to local control needs.

[0011] Furthermore, each control strategy has a corresponding strategy tag, and the optimal control strategy is searched based on the local cache and local storage, including:

[0012] Calculate the hash value of the policy label of the optimal control policy, and use the hash search technology to search from the locally cached hash table;

[0013] If the corresponding hash value is found in the hash table, it is determined that there is a strategy in the hash table that is the same as the optimal control strategy and the sub-strategy of the same strategy is read from the cache;

[0014] If the corresponding hash value is not found in the hash table, it is determined that there is no strategy identical to the optimal control strategy, and the optimal control strategy is decomposed into multiple sub-strategies, and the decomposed sub-strategies are cached and stored; if the optimal control strategy cannot be decomposed, the optimal control strategy is directly cached and stored.

[0015] Furthermore, it also includes:

[0016] Monitor and analyze the control effect of the terminal equipment in real time, and upload feedback data reflecting the control effect to the cloud so that the cloud can optimize the control strategy prediction model; the cloud adds a strategy update mark to the optimal control strategy predicted by the optimized control strategy prediction model;

[0017] Receive the optimal control strategy from the cloud, and search and update the strategy based on the strategy label and strategy update identifier of the optimal control strategy.

[0018] Furthermore, the edge node records the data of the relevant end-side devices before each sub-strategy is executed and forms a policy execution data set;

[0019] When executing sub-strategies based on local regulatory needs, it includes:

[0020] Monitor and analyze the status data of end-side devices in real time;

[0021] If the current device data matches the data of the corresponding device in the recorded policy execution data set, a local control demand is generated and the policy to be executed for the current device is searched from the local cache and local storage in turn;

[0022] If the policy to be executed is found in the local cache, the sub-policy of the policy to be executed is directly executed from the sub-policy queue;

[0023] If the policy to be executed is found only in the local storage, it is executed immediately and added to the local cache from the local storage;

[0024] Mark the status of the strategy to be executed, and ignore the strategy that has been marked when receiving the control strategy sent by the cloud.

[0025] Furthermore, the optimal control strategy is obtained from the cloud through low-latency communication technology, including:

[0026] Use intelligent routing algorithms to select the optimal transmission path between edge nodes and the cloud;

[0027] Use time synchronization network technology to increase the processing priority of real-time sensitive data on the transmission path;

[0028] Software-defined network technology is used to optimize traffic scheduling and resource allocation for data on the transmission path.

[0029] Furthermore, a policy execution unit library is provided in the cloud, and the policy execution unit library includes a plurality of policy execution units, each of which has a policy decomposition identifier and a priority evaluation table, the policy decomposition identifier is used to perform policy decomposition on different policy execution units, and the priority evaluation table is used to evaluate the priority of the policy corresponding to the policy execution unit;

[0030] The optimal control strategy is predicted using the control strategy prediction model, including:

[0031] Using the output of the control strategy prediction model, searching for a strategy execution unit from a strategy execution unit library;

[0032] The found strategy execution units are combined to generate the optimal control strategy.

[0033] Furthermore, the optimal control strategy is decomposed into multiple sub-strategies and cached and stored, including:

[0034] Decomposing the optimal control strategy into several strategy execution units as sub-strategies according to the strategy decomposition marker;

[0035] Calculate the priority value of each sub-strategy according to the priority evaluation table of each strategy execution unit;

[0036] The sub-strategies are prioritized according to the priority values, and are cached and stored in the sub-strategy queue according to the priority numbers. Sub-strategies with the same priority have the same number and are executed in parallel using multi-threaded processing.

[0037] In a second aspect, the present invention provides a device for regulating resources of a virtual power plant in seconds, comprising:

[0038] The data collection and monitoring module is used to monitor and collect the operating data of various adjustable resources of the edge node, and extract key information features used to represent the operating status of various adjustable resources from the operating data;

[0039] A control strategy generation module is used to send key information features to the cloud so that the cloud can predict the optimal control strategy based on the key information features using a pre-trained control strategy prediction model. The optimal control strategy is used to control the adjustable resources.

[0040] Low-latency communication module, used to obtain the optimal control strategy from the cloud through low-latency communication technology;

[0041] The edge control policy execution module is used to search and execute the optimal control policy according to the local cache and local storage. In the search stage, if the local cache does not have the same policy as the optimal control policy, the optimal control policy is decomposed into multiple sub-policies and cached and stored. In the execution stage, the policy is executed on various adjustable resources according to the sub-policy queue in the local cache. The sub-policies in the sub-policy queue include the sub-policies obtained by decomposing the optimal control policy according to the control instructions of the cloud, and the sub-policies searched from the local cache or storage according to the local control needs.

[0042] In a third aspect, the present invention provides a computer device, the device comprising a processor and a memory:

[0043] The memory is used to store the computer program and send the instructions of the computer program to the processor;

[0044] The processor executes the method of regulating resources of a virtual power plant in seconds as described in the first aspect according to the instructions of the computer program.

[0045] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for regulating resources of a virtual power plant in seconds as in the first aspect.

[0046] In summary, the present invention proposes a method, device, equipment and storage medium for regulating resources in a virtual power plant in seconds. By collecting the operating data of various adjustable resources of the edge node, key information features are extracted and uploaded to the cloud. The cloud uses a pre-trained model to predict the optimal regulation strategy, and then feeds this strategy back to the edge node through low-latency communication technology. Before local execution, if there is no identical strategy in the cache, the received strategy will be decomposed into sub-strategies and stored. During execution, the relevant resources are controlled according to the sub-strategy queue in the cache. In addition, sub-strategies suitable for the current situation can be retrieved from local storage for application. The whole process optimizes the cloud-edge collaboration mechanism, reduces the time cost of data transmission and processing, and enables the powerful computing power of the cloud to be fully utilized for data analysis and intelligent decision-making, while the edge focuses on rapid response and instruction execution, jointly realizing efficient and accurate management and scheduling of virtual power plant resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 A flowchart of a method for regulating resources in a virtual power plant in seconds;

[0049] Figure 2 A block diagram of a virtual power plant device for regulating resources in seconds;

[0050] Figure 3 A block diagram of the composition of a computer device. DETAILED DESCRIPTION

[0051] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] See also Figure 1This embodiment provides a method for regulating resources of a virtual power plant in seconds, including the following steps:

[0053] Step 1: Monitor and collect the operating data of various adjustable resources of the edge node, and extract key information features used to represent the operating status of various adjustable resources from the operating data.

[0054] It should be noted that the operation data of various regulating resources such as distributed photovoltaics, energy storage, charging piles, and industrial and commercial adjustable loads are monitored and collected and analyzed, including electrical parameters, operating status parameters, environmental parameters, control parameters, etc. Key information features are extracted from the collected operating data, which can represent the operating status of various adjustable resources. This step uses data analysis techniques, such as machine learning or statistical analysis methods, to identify key parameters that have an important impact on regulation decisions from a large amount of data.

[0055] Step 2: Send the key information features to the cloud so that the cloud can predict the optimal control strategy based on the key information features using a pre-trained control strategy prediction model. The optimal control strategy is used to control the adjustable resources.

[0056] It should be noted that pre-trained machine learning models are deployed on cloud servers, which can predict the best scheduling plan based on the input data. In view of the adjustable potential and adjustable load characteristics of resources, including production load regulation (shutting down equipment, adjusting voltage and current), non-production load regulation (air conditioning load, lighting load), commercial building load regulation strategy (air conditioning load control strategy, lighting equipment control strategy, other equipment load control strategy), etc., the cloud generates the optimal control strategy through machine learning algorithms based on real-time data and historical data.

[0057] Step 3: Obtain the optimal control strategy from the cloud through low-latency communication technology.

[0058] It should be noted that in power systems, especially virtual power plants, rapid response is the key to ensuring stable operation of the system. Low-latency communication technology can ensure that the transmission of control strategies from the cloud to edge nodes is almost instantaneous, allowing the entire system to respond within seconds, which is very necessary for handling emergencies or making dynamic adjustments.

[0059] Step 4: Search and execute the optimal control strategy according to the local cache and local storage; wherein, in the search stage, if the local cache does not have the same strategy as the optimal control strategy, the optimal control strategy is decomposed into multiple sub-strategies and cached and stored; in the execution stage, the strategy is executed on various adjustable resources according to the sub-strategy queue in the local cache. The sub-strategies in the sub-strategy queue include the sub-strategies obtained by decomposing the optimal control strategy according to the control instructions of the cloud, and the sub-strategies searched from the local cache or storage according to the local control needs.

[0060] It should be noted that when searching and executing a strategy, if a similar situation has been encountered before and a solution has been found, the corresponding sub-strategy is directly called from the local cache to speed up the response time. For new situations, the overall strategy is split into several small tasks (sub-strategies) and then implemented one by one. Each sub-strategy is queued up for execution according to a certain priority; at the same time, the order is dynamically adjusted considering changes in actual conditions on site.

[0061] This embodiment provides a method for regulating resources in a virtual power plant in seconds, by collecting the operating data of various adjustable resources of the edge node, extracting key information features from them, and uploading them to the cloud. The cloud uses a pre-trained model to predict the optimal regulation strategy, and then feeds this strategy back to the edge node through low-latency communication technology. Before local execution, if there is no identical strategy in the cache, the received strategy will be decomposed into sub-strategies and stored. During execution, the relevant resources are controlled according to the sub-strategy queue in the cache. In addition, sub-strategies suitable for the current situation can be retrieved from local storage for application. The entire process optimizes the cloud-edge collaboration mechanism, reduces the time cost of data transmission and processing, and enables the powerful computing power of the cloud to be fully utilized for data analysis and intelligent decision-making, while the edge focuses on rapid response and command execution, jointly realizing efficient and accurate management and scheduling of virtual power plant resources.

[0062] In a possible implementation, step 1 of extracting key information features from the operating data includes the following:

[0063] After collecting the operating data, the powerful computing power of the edge computing nodes can be fully utilized to perform sophisticated and comprehensive pre-processing operations on the collected raw data. This key stage includes but is not limited to the following important steps:

[0064] (1) Data cleaning and noise elimination:

[0065] Denoising technology: Actively use sliding window methods or time series smoothing techniques to deal with noise in the data. For example, by using the moving average algorithm, the formula is:

[0066] ;

[0067] Significantly reduce the impact of short-term fluctuations, thereby successfully eliminating insignificant noise.

[0068] Filter application: Cleverly use the Kalman filter, the formula is: , deeply filter the data and effectively remove random noise and periodic interference.

[0069] (2) Outlier detection and removal:

[0070] Statistical methods: widely used statistical methods Law, if the data value X satisfies , it is considered an outlier, where is the mean, The standard deviation is used to accurately identify and decisively eliminate outliers that are beyond the normal data distribution range.

[0071] Machine learning method: Make full use of the machine learning algorithm Isolation Forest, and its anomaly score calculation formula is: , automatically and efficiently identify and exclude abnormal data points.

[0072] When extracting key information from operating data, it includes feature selection and time series analysis. Feature selection is to accurately extract key features that are decisive for model training from a large amount of preprocessed data, such as power generation, SOC (state of charge) of energy storage systems, grid load (current, voltage, active power, reactive power, power factor), etc. Time series analysis is to retain time series information that can reflect the dynamic changes of energy systems, such as the trend of power generation changes, the charge and discharge cycle of energy storage systems, etc. These preprocessing steps are designed to provide structured, high-quality data input for model training.

[0073] Specifically, key information extraction includes the following:

[0074] Data feature engineering: By applying advanced feature extraction algorithms, key features that are decisive for model training and strategy generation are extracted from pre-processed data. These features may include but are not limited to: trend analysis of power output, health of energy storage systems, and fluctuation patterns of grid load.

[0075] (1) Power output trend analysis:

[0076] Time series decomposition: Use time series decomposition techniques such as STL (Seasonally Decomposed Time Series) to decompose the power output data into trend, seasonal and residual components. The decomposition step includes determining the trend and seasonal components, and the formula is as follows:

[0077] ;

[0078] in is the trend component, It is a seasonal ingredient. is the residual component.

[0079] Autoregressive Moving Average Model (ARIMA): Model the time series data and analyze the long-term trend of power output through the ARIMA model. The model formula is as follows:

[0080] ;

[0081] in and are model parameters, is the white noise error.

[0082] (2) Health status of energy storage system:

[0083] Condition Monitoring and Diagnosis (SMD): SMD technology is used to evaluate the health of the energy storage system by monitoring key parameters (such as battery voltage, current, and temperature). Support vector machine (SVM) is used for classification, and the formula is as follows:

[0084] ;

[0085] Where x is the input feature, It's a label. is the kernel function, and are model parameters.

[0086] Degradation analysis: By analyzing the battery aging curve, the characteristics that characterize the degree of battery degradation are extracted. Nonlinear models, such as neural networks, are used to simulate the battery degradation process. The formula is as follows:

[0087] ;

[0088] Where W is the weight matrix, x is the input feature, b is the bias term, and f is the nonlinear activation function.

[0089] (3) Grid load fluctuation mode:

[0090] Wavelet analysis: Use wavelet transform to analyze the time-frequency characteristics of power grid load data, and extract features at different frequencies through multi-scale analysis. The basic formula of wavelet transform is as follows:

[0091] ;

[0092] in is the original signal, is the wavelet function, and are the wavelet parameters.

[0093] Hidden Markov Model (HMM): The fluctuation pattern of the power grid load is modeled by HMM, and the hidden states and their transition probabilities are identified. The formula is as follows:

[0094] , ;

[0095] in and It's the state. is the state transition probability, is the state emission probability.

[0096] Through these advanced feature extraction techniques, a rich set of features can be constructed from raw data, which are crucial for training efficient and accurate models and generating effective strategies.

[0097] This step ensures that the algorithm model obtains sufficient representative data during the training process, so that it can be more efficient and accurate when generating strategies.

[0098] In one embodiment, a policy execution unit library is provided in the cloud, and the policy execution unit library includes a plurality of policy execution units, each of which has a policy decomposition identifier and a priority evaluation table, the policy decomposition identifier is used to perform policy decomposition on different policy execution units, and the priority evaluation table is used to evaluate the priority of the policy corresponding to the policy execution unit;

[0099] The optimal control strategy is predicted using the control strategy prediction model, including:

[0100] Using the output of the control strategy prediction model, searching for a strategy execution unit from a strategy execution unit library;

[0101] The found strategy execution units are combined to generate the optimal control strategy.

[0102] Based on the above embodiment, in a possible implementation, the cloud-based strategy generation in step 2 includes two main steps: prediction model selection and strategy generation process. The two steps are introduced separately as follows:

[0103] (1) Prediction model selection:

[0104] Model selection: In order to capture the time series characteristics of power demand and grid status, the long short-term memory network (LSTM) model is adopted because it is good at processing sequence dependency and long-term memory information.

[0105] Model training: The LSTM model is trained using cleaned and preprocessed historical data to ensure that the model can learn seasonal patterns, trends, and cyclical fluctuations in electricity demand.

[0106] Real-time prediction: Deploy the trained LSTM model and use the latest real-time data to predict power demand and grid status in future periods, providing a data basis for strategy generation.

[0107] (2) Strategy generation process:

[0108] 1) Strategy initialization:

[0109] Forecast result integration: The forecast results of the LSTM model are integrated with historical data to form a comprehensive database, providing a comprehensive information perspective for strategy generation.

[0110] Strategy library construction: Based on the prediction results, a strategy execution unit library is constructed, which includes control measures such as peak load regulation and frequency regulation, and each execution strategy unit is evaluated for each priority factor value and weight to form a corresponding execution strategy unit priority evaluation comparison table. Then, the corresponding dependency table is generated based on the dependency relationship of the execution unit. At the same time, the dependency table and evaluation table are synchronously sent to the edge. The evaluation factors include the following:

[0111] ① Impact scope assessment value (IR) can be scored according to factors such as the number of system modules affected by the policy and the number of terminal devices involved, and the corresponding weight w1 can be set according to its relative importance.

[0112] ② The business objective impact assessment value (BOI) can be scored according to the degree of impact on key business indicators, and the corresponding weight w2 can be set.

[0113] ③ The time urgency assessment value (TU) is scored according to the degree of urgency and the corresponding weight w3 is set.

[0114] ④ The system stability impact assessment value (SSI) is scored according to the risk level that may cause system instability, and the corresponding weight w4 is set.

[0115] Determining the weight distribution of different evaluation values ​​usually requires comprehensive consideration of multiple factors. The relative importance of each factor can be determined by analyzing and summarizing past similar strategies. It is also possible to discuss and negotiate with experts in related fields and set weights based on their professional opinions. It is also possible to conduct some simulation experiments or data analysis to observe the impact of different weight distributions on the results, so as to gradually adjust and optimize the weight distribution to make it more in line with actual needs and expected results.

[0116] 2) Strategy optimization and adjustment:

[0117] The purpose of the virtual power plant control strategy is to optimize the allocation of power resources, improve the flexibility and stability of the power grid, ensure the balance of power supply and demand, while promoting the consumption of renewable energy, reducing carbon emissions, and achieving efficient use of energy and sustainable development.

[0118] When formulating a virtual power plant control strategy based on the above purpose, a comprehensive assessment can be conducted from the following aspects: resource characteristics and potential, technical feasibility and economics, grid demand and constraints, market environment and policy orientation, and social and environmental benefits;

[0119] Initial strategy optimization: Use the particle swarm optimization algorithm to adjust the initial strategy and find a solution that can achieve a balance in the multi-objective function.

[0120] Particle Swarm Optimization (PSO) is an optimization algorithm based on swarm intelligence. It simulates the foraging behavior of bird flocks to find the optimal solution. In multi-objective optimization problems, the PSO algorithm can find a solution that strikes a balance between multiple objective functions by adjusting the initial strategy.

[0121] The following are the general steps to use the PSO algorithm to adjust the initial strategy to find a multi-objective equilibrium solution:

[0122] Initialize the particle swarm: determine the initial parameters such as the number, position, and velocity of particles. Each particle represents a potential solution.

[0123] Define objective functions: Identify multiple objective functions to be optimized.

[0124] Calculate fitness: Evaluate the fitness of each particle according to the objective function.

[0125] Update the global optimal solution: Compare the fitness of each particle with the current global optimal solution. If a particle has a better fitness, update the global optimal solution.

[0126] Update particle position and velocity: Update the position and velocity of each particle according to the following formula:

[0127] Location Updates: ;

[0128] Speed ​​Update: ;

[0129] in, and They are The position and velocity of the particle at the iteration, and are the position and velocity of the current iteration, is the individual optimal position of the particle, is the global optimal position, is the inertia weight, and is the learning factor, and is a random number uniformly distributed between [0,1].

[0130] Repeat steps 3 to 5 until a stopping condition is reached: usually a maximum number of iterations is reached or a certain convergence criterion is met.

[0131] Output results: The final particle positions or other information representing the equilibrium solution can be used as output.

[0132] By adjusting the inertia weight, learning factor, and other parameters, the search behavior of the PSO algorithm and the equilibrium solution found can be influenced.

[0133] Policy decomposition identifier: In the process of generating policies, execution units are searched from the execution policy unit library, and the execution policies are generated by combining them. Each execution policy unit is separated by an identifier so that the edge node can decompose the execution policy.

[0134] Dynamic adjustment: During the optimization process, real-time data may change the state of the power grid, so the strategy needs to be adjusted dynamically to adapt to the new situation. Automatically adjust the strategy according to the changes in real-time data.

[0135] Iterative optimization: Based on real-time feedback and combined with historical data, iteratively optimize the strategy until the optimal or near-optimal solution is reached.

[0136] In addition, the cloud also includes strategy evaluation and feedback when generating strategies, including:

[0137] Performance evaluation: Evaluate the performance of the generated strategy in a simulated environment or on historical data.

[0138] Feedback loop: The actual operation results of the strategy are fed back into the system for further optimization and adjustment of the model.

[0139] In one embodiment, obtaining the optimal control strategy from the cloud through low-latency communication technology includes:

[0140] Use intelligent routing algorithms to select the optimal transmission path between edge nodes and the cloud;

[0141] Use time synchronization network technology to increase the processing priority of real-time sensitive data on the transmission path;

[0142] Software-defined network technology is used to optimize traffic scheduling and resource allocation for data on the transmission path.

[0143] Based on the above embodiment, in a possible implementation manner, the data interaction between the cloud and the edge node through the low-latency communication technology in step 3 includes:

[0144] Choose high-speed data transmission 5G network technology between the cloud and edge nodes to improve the speed and efficiency of data transmission. 5G network has the characteristics of low latency and high bandwidth, which can significantly reduce the latency of data transmission; at the same time, it adopts multiple input multiple output (MIMO) technology: MIMO technology uses multiple antennas at the sending and receiving ends to send and receive multiple data streams at the same time, thereby improving data transmission rate and reliability. In 5G communication, MIMO technology can further improve spectrum efficiency and reduce latency. For example, by using more antennas, the concurrent transmission of data can be increased and the waiting time of data packets can be reduced.

[0145] Select SRTP, a low-latency communication protocol specifically for cloud-edge collaboration. SRTP is an encryption extension for the Real-time Transport Protocol (RTP). It provides encryption and authentication of RTP packets to ensure the secure transmission of structured data. SRTP can be used in conjunction with other real-time communication applications to protect the security and low latency of communications.

[0146] With 5G slicing technology, network slicing can provide independent network resources for different applications to ensure that they get the required performance. By allocating dedicated network slices to low-latency applications, lower latency and higher reliability can be provided.

[0147] Utilize intelligent routing algorithms to select the optimal path for data transmission based on the real-time status of the network and data requirements, thereby reducing the path length and delay of data transmission.

[0148] The specific algorithm formula is as follows:

[0149] ;

[0150] in, , , , , , is the weight of each factor, , , , , is a function of various factors, , , , , are the values ​​of each factor, is a function of the degree of congestion, is the value of congestion level, is a function of other factors, are the values ​​of other factors.

[0151] Steps to select the best route:

[0152] 1. Collect the values ​​of various factors of each link in the network, including the number of hops, link quality, congestion level, traffic demand, node load, etc.

[0153] 2. For each destination node, calculate the routing scores to all other nodes.

[0154] 3. Comprehensively consider the influence of various factors, calculate the score of each factor through the function, multiply it by the weight and add it up to get the total routing score.

[0155] 4. Select the node with the smallest routing score as the best route.

[0156] The use of TSN and SDN technologies simultaneously can achieve millisecond-level latency in the data exchange rate between edge nodes and the central platform. TSN ensures the real-time and deterministic transmission of data, while SDN provides flexible control and management of the network, which helps optimize data transmission paths and improve network performance.

[0157] Combining intelligent routing algorithms with TSN and SDN technologies can achieve collaborative control. The specific steps are as follows:

[0158] 1. Intelligent routing algorithm selects the appropriate path based on network conditions and traffic demand.

[0159] 2.TSN technology ensures that real-time-sensitive traffic is given priority on the selected path, reducing latency.

[0160] 3. The SDN controller performs traffic scheduling and resource allocation optimization based on the decisions of the intelligent routing algorithm and the requirements of TSN.

[0161] 4. Through collaborative control, real-time sensitive traffic can be transmitted with lower latency, while non-real-time traffic can also be reasonably utilized to improve overall network performance.

[0162] In one embodiment, each control strategy has a corresponding strategy tag, and searching for the optimal control strategy based on the local cache and the local storage includes:

[0163] Calculate the hash value of the strategy label of the optimal control strategy, and use the hash search technology to search from the cached hash table;

[0164] If the corresponding hash value is found in the hash table, it is determined that there is a strategy in the hash table that is the same as the optimal control strategy and the sub-strategy of the same strategy is read from the cache;

[0165] If the corresponding hash value is not found in the hash table, it is determined that there is no strategy identical to the optimal control strategy, and the optimal control strategy is decomposed into multiple sub-strategies, and the decomposed sub-strategies are cached and stored; if the optimal control strategy cannot be decomposed, the optimal control strategy is directly cached and stored.

[0166] In one embodiment, it further includes:

[0167] Monitor and analyze the control effect of the terminal equipment in real time, and upload feedback data reflecting the control effect to the cloud so that the cloud can optimize the control strategy prediction model; the cloud adds a strategy update mark to the optimal control strategy predicted by the optimized control strategy prediction model;

[0168] Receive the optimal control strategy from the cloud, and search and update the strategy based on the strategy label and strategy update identifier of the optimal control strategy.

[0169] In one embodiment, the edge node records the data of the relevant end-side devices before each sub-policy is executed and forms a policy execution data set;

[0170] When executing sub-strategies based on local regulatory needs, it includes:

[0171] Monitor and analyze the status data of end-side devices in real time;

[0172] If the current device data matches the data of the corresponding device in the recorded policy execution data set, a local control demand is generated and the policy to be executed for the current device is searched from the local cache and local storage in turn;

[0173] If the policy to be executed is found in the local cache, the sub-policy of the policy to be executed is directly executed from the sub-policy queue;

[0174] If the policy to be executed is found only in the local storage, the sub-policy of the policy to be executed is added to the sub-policy queue from the local storage for execution;

[0175] Mark the status of the strategy to be executed, and ignore the strategy that has been marked when receiving the control strategy sent by the cloud.

[0176] In one embodiment, the optimal control strategy is decomposed to obtain multiple sub-strategies and cached and stored, including:

[0177] Decomposing the optimal control strategy into several strategy execution units as sub-strategies according to the strategy decomposition marker;

[0178] Calculate the priority value of each sub-strategy according to the priority evaluation table of each strategy execution unit;

[0179] The sub-strategies are prioritized according to the priority values, and are cached and stored in the sub-strategy queue according to the priority numbers. Sub-strategies with the same priority have the same number and are executed in parallel using multi-threaded processing.

[0180] Based on the above embodiment, in a possible implementation, for step 4, after the cloud generates the control strategy, it is quickly transmitted to the edge computing node through the cloud-edge low-latency communication module. After the edge computing node receives the control strategy, it will immediately process and execute it quickly. In order to achieve rapid execution of the strategy, the edge computing node adopts the following technical means:

[0181] (1) Strategy decomposition and caching:

[0182] Policy search and matching: When the cloud sends a control policy to the edge computing node, the edge computing node will first obtain the policy ID and perform a hash table search for the hash value. If it does not exist, it will use the preset hash function to calculate the hash value of the policy tag ID. The hash value is directly mapped to a specific location in the cache, and the edge node can quickly search for the policy at that location to determine whether the policy has been cached. If the policy exists in the cache, the node can directly read the corresponding sub-policy without reprocessing the entire policy, which will significantly reduce latency.

[0183] In order to achieve fast search, a hash table can be used to store the policy tag ID and its corresponding hash value. A hash table is a data structure based on a hash function that can quickly find and insert data. In the hash table, we use the policy tag ID as the key and its corresponding hash value as the value. When the edge computing node receives the control policy, it first looks up the hash value of the policy tag ID and uses this hash value as the key to look up in the hash table. If the corresponding key-value pair exists in the hash table, it means that the policy has been cached and the edge computing node can directly read the corresponding sub-policy. If the corresponding key-value pair does not exist in the hash table, it means that the policy has not been cached and the edge computing node needs to reprocess the entire policy.

[0184] Policy decomposition: If the control policy is not in the cache, the edge computing node will decompose the unmatched policy. Before decomposition, the characteristics and content of the policy are quickly identified to determine whether there are markers that can be used for policy decomposition. If decomposition is not possible, it will be directly executed and cached and stored on disk.

[0185] If the policy can be decomposed, the premise is that it can be decomposed into multiple executable units, then decompose it into the smallest executable unit. Then, according to the decomposed policy unit, search the local evaluation comparison table to obtain the corresponding evaluation factor value and corresponding weight, and then calculate the execution unit priority according to the evaluation factor value and weight value. The specific formula involved is as follows:

[0186] The impact range is estimated to be , whose weight is ; The business objective impact assessment value is , whose weight is ; The time urgency assessment value is , whose weight is ; The impact on system stability is evaluated as , whose weight is .

[0187] The priority value ;

[0188] The priority order is determined by the value of P. The smaller the value, the lower the priority; vice versa. The same priority can be processed and executed in parallel by multiple threads to improve efficiency. In this way, during the execution of the decomposed strategy, combined with the dependency relationship of the execution unit, the strategy with a high priority can be executed first while ensuring the dependency relationship, thereby reflecting the dependency order relationship and improving the overall execution efficiency.

[0189] Policy caching: After using the algorithm to quickly identify the policy, for policies that do not need to be decomposed, one-to-one caching and disk storage are directly performed. On the contrary, for the decomposed sub-policies, priority labels are assigned, and the starting value of the label is 0. The higher the priority, the smaller the label value. After the priority is calculated, a sub-policy execution queue is formed. The numerical labels of the same priority are consistent, and then the hash value is calculated using the policy tag ID and hash algorithm, and finally the one-to-many cache and disk storage of the decomposed policy is performed. At the same time, the default expiration time for cached data is set to 7 days, and dynamic setting of the cache time is supported. When the cached policy expires, the cache is automatically cleared.

[0190] When issuing policies for quick identification, if a policy update is found, first search for the policy in the cache. If it exists in the cache, delete it. Then search for the corresponding policy in the data disk and delete it. At the same time, the policy is decomposed as above before caching and disk storage operations are performed.

[0191] Policy execution: During the policy execution process, there are two situations: one is that the policy is issued and executed for the first time; the other is that the policy is automatically triggered and executed at the edge.

[0192] When a policy is issued for the first time, it will go through a series of processes such as policy identification and decomposition, caching, and disk storage. First, determine whether the policy meets the conditions for decomposition. If it cannot be decomposed, directly call the execution action to execute the edge policy and record the device, data, status and other information associated with the execution policy so that the edge node can determine whether to automatically trigger the policy execution when monitoring the end-side data, thereby improving the policy response rate.

[0193] When the edge automatically triggers execution, the edge monitors the status data of the end-side device in real time and analyzes the collected data in real time. During the analysis process, the policy execution record is searched. Once the current device data matches the device data in the policy execution record, for example, the device power consumption is greater than or equal to a certain value, the edge node automatically searches for the policy from the cache. If there is no policy in the cache, it searches from the disk storage. After the policy is found, the policy is executed. If the current policy does not exist in the cache, its policy is placed in the cache to increase the execution priority and facilitate subsequent rapid execution. At the same time, the current policy is marked in the cache and set to the executed state to avoid repeated execution of the policy when the cloud policy is issued. When this policy receives the execution instruction issued by the cloud, the corresponding cache search is performed. If the corresponding policy is found, check whether its status has been automatically executed by the edge node. If it has been executed, stop the subsequent execution operation and clear the executed state; otherwise, perform the edge execution operation. It is understandable that when all sub-policies of the policy to be executed are cached in the sub-policy execution queue of the local cache, the edge will form a real-time execution queue of the sub-policy based on the queue. When any sub-policy is executed, the real-time execution queue will be updated based on the sub-policy execution queue. Of course, any sub-policies of the control policy obtained according to the real-time control requirements of the cloud or local should be added to the real-time execution queue in time for immediate execution, and then the policy should be cached and stored accordingly to achieve a rapid response to the real-time control requirements.

[0194] (2) Real-time feedback optimization:

[0195] In the process of executing the control strategy, the edge computing node will monitor the control effect in real time and upload the feedback data to the cloud. The cloud will analyze and optimize based on the feedback data to adjust and improve the control strategy and improve the execution effect.

[0196] When the cloud policy is adjusted and improved, the policy ID will remain unchanged and the policy update identifier will be added. Once this policy is issued again, the edge will identify the policy accordingly. First, determine whether the policy has been updated. If there is an update, perform corresponding policy judgment, decomposition, caching, disk storage and other operations. Then, perform cache hash search by policy ID. If it exists, clean up the cache and delete the disk storage of this policy. After that, update the cache and disk storage while executing the policy. If there is no update, directly search the cache or disk. If there is a disk, directly read the policy for corresponding policy execution and update it to the cache. If it does not exist, perform corresponding policy judgment, decomposition, caching, storage and other operations.

[0197] The present invention proposes an innovative method for monitoring and controlling heterogeneous adjustable resources in virtual power plants in seconds. The method has made significant breakthroughs in the rapid generation and issuance of control strategies. By optimizing the cloud-edge collaboration mechanism, reducing data transmission delays, and reducing data calculation and processing time, the uncertainty characteristics of virtual power plant resource responses are targeted. The previous deterministic control method is changed, and a multi-level and multi-autonomous subject consistency and coordination control system is constructed by integrating cloud-edge collaboration and machine learning algorithms. The resource control capability of virtual power plants is improved, and low-cost, high-reliability, and secure access and interaction are achieved, meeting the second-level time control requirements of virtual power plants.

[0198] In terms of rapid generation of control strategies, the present invention uses the computing power and data analysis technology of the cloud control center to conduct in-depth mining and intelligent analysis and processing of real-time data uploaded by edge computing nodes. By integrating advanced time series prediction machine learning algorithms, the cloud can grasp the aggregate response characteristics of various resources and quickly generate the most optimized control strategy based on historical time series data and real-time status data. This process not only improves the accuracy of strategy generation, but also significantly shortens the time for strategy formulation, laying the foundation for subsequent rapid execution.

[0199] In terms of issuing and executing control strategies, the present invention adopts an efficient cloud-edge collaboration mechanism. The cloud control center quickly transmits the generated control strategy to the edge computing node through a high-speed data transmission network, and the edge computing node is responsible for issuing the control instructions directly to the terminal device for execution. In order to reduce data transmission delays, advanced network optimization technology and intelligent routing algorithms are used at the same time to ensure the efficiency and stability of data during transmission. In addition, by using strategy decomposition and caching sub-strategies at the edge computing node, relying on local computing processing capabilities, the processing time of data during transmission is greatly reduced, further improving the response speed of the system.

[0200] In addition, the present invention has also been deeply optimized in terms of data calculation and processing. As the first line of defense for data processing, edge computing nodes have powerful real-time data processing capabilities. They can quickly preprocess and preliminarily analyze the collected raw data and extract key information, thereby reducing the amount of data that needs to be transmitted to the cloud. This optimization not only reduces the load of data transmission, but also reduces the time for cloud data processing, allowing the entire system to respond more quickly to changes in the heterogeneous adjustable resources of the virtual power plant, and realize the collection and control of load-side resources such as distributed photovoltaics, energy storage, charging piles, and flexible loads.

[0201] In summary, the method for virtual power plant resource regulation in seconds proposed in the present invention has achieved remarkable results in the rapid generation and issuance of regulation strategies. By optimizing data transmission, reducing computing processing time and improving system response speed, this method provides strong technical support for real-time regulation and efficient management of virtual power plants.

[0202] Based on the same inventive concept, the embodiment of the present application also provides a device for virtual power plant second-level resource regulation for implementing the method for virtual power plant second-level resource regulation involved in the above-mentioned method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the embodiment of the device for virtual power plant second-level resource regulation provided below can be referred to the limitations of the method for virtual power plant second-level resource regulation above, and will not be repeated here.

[0203] See also Figure 2 This embodiment provides a device for regulating resources of a virtual power plant in seconds, including:

[0204] The data collection and monitoring module is used to monitor and collect the operating data of various adjustable resources of the edge node, and extract key information features used to represent the operating status of various adjustable resources from the operating data;

[0205] A control strategy generation module is used to send key information features to the cloud so that the cloud can predict the optimal control strategy based on the key information features using a pre-trained control strategy prediction model. The optimal control strategy is used to control the adjustable resources.

[0206] Low-latency communication module, used to obtain the optimal control strategy from the cloud through low-latency communication technology;

[0207] The edge control policy execution module is used to search and execute the optimal control policy according to the local cache and local storage. In the search stage, if the local cache does not have the same policy as the optimal control policy, the optimal control policy is decomposed into multiple sub-policies and cached and stored. In the execution stage, the policy is executed on various adjustable resources according to the sub-policy queue in the local cache. The sub-policies in the sub-policy queue include the sub-policies obtained by decomposing the optimal control policy according to the control instructions of the cloud, and the sub-policies searched from the local storage according to the local control needs.

[0208] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0209] Reference Figure 3 An embodiment of the present invention further provides a computer device, comprising: a memory and a processor and a computer program stored in the memory. When the computer program is executed on the processor, a method for regulating resources of a virtual power plant in seconds as described in any one of the above methods is implemented.

[0210] The computer device may be a desktop computer, a notebook, a PDA, a cloud server or other computing device. The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will understand that Figure 3 It is only an example of a computer device and does not constitute a limitation of the computer device. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0211] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0212] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device. Further, the memory may include both an internal storage unit and an external storage device of the computer device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory may also be used to temporarily store data that has been output or is to be output.

[0213] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for controlling resources of a virtual power plant in seconds as described in any one of the above methods is implemented.

[0214] In this embodiment, if the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, USB flash drive, mobile hard disk, disk or optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0215] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0216] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0217] In the embodiments disclosed in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0218] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for regulating resources in a virtual power plant in seconds, characterized in that: The steps include: Monitor and collect the operation data of various adjustable resources of the edge node, and extract key information features for indicating the operation status of various adjustable resources from the operation data; Sending the key information features to the cloud, so that the cloud can predict the optimal control strategy based on the key information features using a pre-trained control strategy prediction model, and the optimal control strategy is used to control the adjustable resources; Obtaining the optimal control strategy from the cloud through low-latency communication technology; The optimal control strategy is searched and executed according to the local cache and local storage; wherein, in the search stage, if the local cache has the same strategy as the optimal control strategy, the sub-strategy of the same strategy is directly called; if the local cache does not have the same strategy as the optimal control strategy, the optimal control strategy is decomposed to obtain multiple sub-strategies and cached and stored; During the execution phase, various adjustable resources are subject to policy execution according to the sub-policy queue in the local cache. The sub-policies in the sub-policy queue include sub-policies obtained by decomposing the optimal control strategy according to the control instructions of the cloud, and sub-policies obtained by searching from the local cache or storage according to local control needs.

2. The method for regulating resources of a virtual power plant in seconds according to claim 1, characterized in that: Each control strategy has a corresponding strategy tag, and searching for the optimal control strategy according to the local cache and local storage includes: Calculate the hash value of the strategy label of the optimal control strategy, and use hash search technology to search from the hash table of the local cache; If the corresponding hash value is found in the hash table, it is determined that there is a strategy identical to the optimal control strategy in the hash table and a sub-strategy of the same strategy is read from the cache; If the corresponding hash value is not found in the hash table, it is determined that there is no strategy that is the same as the optimal control strategy, and the optimal control strategy is decomposed into multiple sub-strategies, and the decomposed sub-strategies are cached and stored; if the optimal control strategy cannot be decomposed, the optimal control strategy is directly cached and stored.

3. The method for regulating resources of a virtual power plant in seconds according to claim 2, characterized in that: Also includes: Monitor and analyze the control effect of the terminal device in real time, and upload feedback data reflecting the control effect to the cloud so that the cloud can optimize the control strategy prediction model; The cloud adds a strategy update mark to the optimal control strategy predicted by the optimized control strategy prediction model; The optimal control strategy is received from the cloud, and the strategy is searched and updated according to the strategy tag of the optimal control strategy and the strategy update identifier.

4. The method for regulating resources of a virtual power plant in seconds according to claim 1, characterized in that: The edge node records the data of the relevant end-side devices before each of the sub-strategies is executed and forms a strategy execution data set; When executing sub-strategies based on local regulatory needs, it includes: Monitor and analyze the status data of end-side devices in real time; If the current device data is consistent with the data of the corresponding device in the recorded policy execution data set, a local control demand is generated and the policy to be executed for the current device is searched from the local cache and the local storage in turn; If the policy to be executed is found in the local cache, the sub-policy of the policy to be executed is directly executed from the sub-policy queue; If the policy to be executed is only found in the local storage, the policy to be executed is executed immediately and added to the local cache from the local storage; The strategy to be executed is marked with a status, and when the control strategy sent from the cloud is received, the strategy that has been marked with a status is ignored.

5. The method for regulating resources of a virtual power plant in seconds according to claim 1, characterized in that: Obtaining the optimal control strategy from the cloud through low-latency communication technology includes: Use intelligent routing algorithms to select the optimal transmission path between edge nodes and the cloud; Using time synchronization network technology to increase the processing priority of real-time sensitive data on the transmission path; Software-defined network technology is used to perform traffic scheduling and resource allocation optimization on the data on the transmission path.

6. The method for regulating resources of a virtual power plant in seconds according to claim 1, characterized in that: A policy execution unit library is provided in the cloud, and the policy execution unit library includes a plurality of policy execution units, each of which has a policy decomposition identifier and a priority evaluation table, the policy decomposition identifier is used to perform policy decomposition on different policy execution units, and the priority evaluation table is used to evaluate the priority of the policy corresponding to the policy execution unit; The optimal control strategy is predicted using the control strategy prediction model, including: Using the output of the control strategy prediction model to search for the strategy execution unit from the strategy execution unit library; The found strategy execution units are combined to generate the optimal control strategy.

7. The method for regulating resources of a virtual power plant in seconds according to claim 6, characterized in that: Decomposing the optimal control strategy to obtain multiple sub-strategies and caching and storing them includes: Decomposing the optimal control strategy into a plurality of strategy execution units according to the strategy decomposition identifier to serve as the sub-strategies; Calculating the priority value of each of the sub-strategies according to the priority evaluation table of each of the strategy execution units; The sub-strategies are prioritized according to the priority values, and are cached and stored in the sub-strategy queue according to the priority numbers. Sub-strategies of the same priority have the same number and are executed by multi-threaded parallel processing.

8. A device for regulating resources in a virtual power plant in seconds, characterized in that: include: A data collection and monitoring module is used to monitor and collect the operating data of various adjustable resources of the edge node, and extract key information features used to represent the operating status of various adjustable resources from the operating data; A control strategy generation module, used for sending the key information features to the cloud, so that the cloud can predict the optimal control strategy based on the key information features using a pre-trained control strategy prediction model, and the optimal control strategy is used to control the adjustable resources; A low-latency communication module, used for obtaining the optimal control strategy from the cloud through low-latency communication technology; An edge control strategy execution module is used to search and execute the optimal control strategy according to the local cache and local storage; wherein, in the search stage, if the local cache has the same strategy as the optimal control strategy, the sub-strategy of the same strategy is directly called; if the local cache does not have the same strategy as the optimal control strategy, the optimal control strategy is decomposed to obtain multiple sub-strategies and cached and stored; During the execution phase, various adjustable resources are subject to policy execution according to the sub-policy queue in the local cache. The sub-policies in the sub-policy queue include sub-policies obtained by decomposing the optimal control strategy according to the control instructions of the cloud, and sub-policies obtained by searching from the local cache or storage according to local control needs.

9. A computer device, characterized in that: The device comprises a processor and a memory: The memory is used to store a computer program and send instructions of the computer program to the processor; The processor executes the method for regulating resources of a virtual power plant in seconds as described in any one of claims 1 to 7 according to the instructions of the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for controlling resources of a virtual power plant in seconds as described in any one of claims 1 to 7 is implemented.

Images