Virtual power plant cooperative control method and device based on multi-level architecture, computer equipment, readable storage medium and program product

Abstract

The invention relates to a virtual power plant cooperative control method and device based on a multi-level architecture, computer equipment, a readable storage medium and a program product. The method comprises the following steps: acquiring power generation data of a renewable power supply in a virtual power plant, energy storage state data of energy storage equipment and power utilization data of a controllable load; performing data mining on the power generation data to obtain a first data mining result, performing data mining on the energy storage state data to obtain a second data mining result, and performing data mining on the power utilization data to obtain a third data mining result; inputting the first data mining result, the second data mining result and the third data mining result into a pre-constructed cooperative control strategy generation model, and outputting a target cooperative control strategy by adopting a particle swarm optimization algorithm; and sending a control instruction corresponding to the target cooperative control strategy to the corresponding renewable power supply, energy storage equipment or load terminal. By adopting the method, efficient collaborative management and accurate control of various energy resources can be realized.

Description

Technical Field

[0001] This application relates to the field of energy control technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for collaborative control of a virtual power plant based on a multi-level architecture. Background Technology

[0002] A Virtual Power Plant (VPP) is a new technology that aggregates distributed energy resources to participate in grid operation. Through advanced control architecture and communication technology, it integrates and coordinates widely distributed distributed power sources, energy storage systems, controllable loads and other resources, enabling them to participate in electricity market transactions and grid dispatch as a special power plant, thereby improving energy utilization efficiency and enhancing grid stability.

[0003] In traditional technologies, the control of virtual power plants often employs a relatively centralized, single architecture or focuses on local optimization for a specific type of resource. This approach typically relies on simple scheduling commands or preset rules to independently control various energy devices, lacking a system-level coordination mechanism.

[0004] However, traditional technologies have significant drawbacks: First, their simplistic architecture struggles to cope with the high uncertainty and volatility brought about by the integration of renewable energy sources, failing to achieve accurate perception and predictive coordination of intermittent power output such as wind and solar power. Second, due to the lack of multi-resource collaborative optimization mechanisms, different energy resources often operate independently or even conflict with each other, leading to a decline in the overall economic efficiency and limited regulation capacity of the virtual power plant. Therefore, traditional technologies cannot meet the demands of new power systems for efficient, reliable, and intelligent collaborative control of virtual power plants. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, device, computer equipment, computer-readable storage medium, and computer program product for collaborative control of virtual power plants based on a multi-level architecture, which can achieve efficient collaborative management and precise control of multiple energy resources within a virtual power plant, improve energy utilization efficiency, enhance power system stability, and promote the large-scale consumption of renewable energy, in response to the aforementioned technical problems.

[0006] Firstly, this application provides a collaborative control method for virtual power plants based on a multi-level architecture, including:

[0007] Based on various sensors deployed within the virtual power plant, the power generation data of renewable energy sources, the energy storage status data of energy storage devices, and the power consumption data of controllable loads within the virtual power plant are acquired.

[0008] Data mining is performed on the power generation data to obtain a first data mining result; data mining is performed on the energy storage status data to obtain a second data mining result; and data mining is performed on the electricity consumption data to obtain a third data mining result. The first data mining result includes the power generation prediction data of the renewable energy source; the second data mining result includes the charging and discharging characteristic data of the energy storage device; and the third data mining result includes the regulation characteristic data of the controllable load.

[0009] The first data mining result, the second data mining result, and the third data mining result are input into the pre-built collaborative control strategy generation model. The particle swarm optimization algorithm is used to output the target collaborative control strategy. The collaborative control strategy generation model takes at least one of the following as the optimization objective: the lowest total operating cost, the highest renewable energy absorption rate, and the lowest load fluctuation.

[0010] The control command corresponding to the target collaborative control strategy is sent to the corresponding renewable power source, energy storage device or load terminal, so that the renewable power source, the energy storage device or the load terminal adjusts its operating state based on the corresponding control command.

[0011] In one embodiment, the power generation data includes solar photovoltaic power generation data and wind power generation data, and the step of performing data mining on the power generation data to obtain a first data mining result includes:

[0012] Based on pre-built solar photovoltaic power generation prediction models and wind power generation models, data mining is performed on the solar photovoltaic power generation data and wind power generation data respectively to obtain a first data mining result. The first data mining result includes solar photovoltaic power generation prediction data and wind turbine power generation prediction data. The solar photovoltaic power generation prediction model is established using the support vector machine algorithm, and the wind power generation model is obtained by training a machine learning model based on the historical captured power of the wind turbine.

[0013] In one embodiment, the step of performing data mining on the energy storage state data to obtain a second data mining result includes:

[0014] Based on a pre-built energy storage device characteristic analysis model, data mining is performed on the energy storage state data to obtain a second data mining result. The second data mining result includes the efficiency characteristic data and state of charge change data of the energy storage device under different charge and discharge rates. The energy storage device characteristic analysis model is used to determine the efficiency characteristic data by fitting the relationship curve between charge and discharge efficiency and charge and discharge rate, and to determine the state of charge change data by using the ampere-hour integration method.

[0015] In one embodiment, the step of performing data mining on the electricity consumption data to obtain a third data mining result includes:

[0016] Based on a pre-built load characteristic database, data mining is performed on the electricity consumption data to obtain a third data mining result. The third data mining result includes the adjustable power range, load response speed, and load adjustment cost data of the controllable load. The load characteristic database is obtained by cluster analysis of the electricity consumption patterns of different types of controllable loads.

[0017] In one embodiment, the step of inputting the first data mining result, the second data mining result, and the third data mining result into a pre-built collaborative control strategy generation model, and using a particle swarm optimization algorithm to output a target collaborative control strategy includes:

[0018] The first data mining result and the second data mining result are input into the pre-built collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing power generation cost and balancing power supply and demand as optimization objectives, and is solved by particle swarm optimization algorithm to generate the first collaborative control strategy.

[0019] The first data mining result and the third data mining result are input into the collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing the total running cost as the optimization objective and is solved by the particle swarm optimization algorithm to generate a second collaborative control strategy.

[0020] The second data mining result and the third data mining result are input into the collaborative control strategy generation model so that the collaborative control strategy generation model takes the minimization of load fluctuation as the optimization objective and is solved by the particle swarm optimization algorithm to generate the third collaborative control strategy.

[0021] Based on the first collaborative control strategy, the second collaborative control strategy, and the third collaborative control strategy, a target collaborative control strategy is output.

[0022] In one embodiment, the solution obtained by particle swarm optimization includes:

[0023] Based on the decision variables and constraints in the cooperative control strategy model, an initial population containing multiple particles is randomly generated, where the position of each particle represents a candidate cooperative control strategy.

[0024] The fitness value of each particle is calculated based on the optimization objective function of the cooperative control strategy model.

[0025] Based on the fitness value of each particle, the target particle is selected from the initial population using the roulette wheel algorithm to obtain a new generation of population;

[0026] Based on the historical best position of the particles and the global best position of the population, the velocity and position of each particle in the new generation population are iteratively updated until the maximum number of iterations is reached or the change in the population's optimal fitness value is less than a set threshold. Then, the iteration stops and the current global best position is output as the cooperative control strategy.

[0027] Secondly, this application also provides a virtual power plant collaborative control device based on a multi-level architecture, comprising:

[0028] The data acquisition module is used to acquire power generation data of renewable energy sources, energy storage status data of energy storage devices, and power consumption data of controllable loads within the virtual power plant, based on various sensors deployed within the virtual power plant.

[0029] The data mining module is used to perform data mining on the power generation data to obtain a first data mining result, to perform data mining on the energy storage status data to obtain a second data mining result, and to perform data mining on the electricity consumption data to obtain a third data mining result. The first data mining result includes the power generation prediction data of the renewable energy source, the second data mining result includes the charging and discharging characteristic data of the energy storage device, and the third data mining result includes the regulation characteristic data of the controllable load.

[0030] The strategy generation module is used to input the first data mining result, the second data mining result and the third data mining result into the pre-built collaborative control strategy generation model, and use the particle swarm optimization algorithm to output the target collaborative control strategy. The collaborative control strategy generation model takes at least one of the following as the optimization objective: the lowest total operating cost, the highest renewable energy absorption rate and the lowest load fluctuation.

[0031] The collaborative control module is used to send the control command corresponding to the target collaborative control strategy to the corresponding renewable power source, energy storage device or load terminal, so that the renewable power source, the energy storage device or the load terminal adjusts its operating state based on the corresponding control command.

[0032] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described in any of the above-mentioned embodiments.

[0033] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0034] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any of the above claims.

[0035] The aforementioned multi-level architecture-based virtual power plant collaborative control method, device, computer equipment, computer-readable storage medium, and computer program product, by deploying various sensors to acquire power generation data from renewable energy sources, energy storage status data from energy storage devices, and electricity consumption data from controllable loads within the virtual power plant, can achieve comprehensive perception of the operational status of various energy resources within the virtual power plant, providing a complete and real-time data foundation for subsequent data analysis and collaborative control. By performing data mining on the power generation data, energy storage status data, and electricity consumption data respectively, the results obtained include renewable energy power generation prediction data, energy storage device charge and discharge characteristic data, and controllable load regulation characteristic data, enabling a deep understanding of the core characteristics of various energy resources and providing a basis for collaborative control strategies. This provides a precise basis for formulating strategies to effectively address the intermittency and volatility of renewable energy. By inputting data mining results into a pre-built collaborative control strategy generation model, the particle swarm optimization algorithm is used to output a target collaborative control strategy with at least one of the following optimization objectives: lowest total operating cost, highest renewable energy absorption rate, and minimum load fluctuation. This enables global optimization scheduling of energy resources within the virtual power plant, fully leveraging the synergistic advantages of various energy resources and improving energy utilization efficiency. The control commands corresponding to the target collaborative control strategy are sent to the relevant energy equipment or load terminals, enabling the equipment to adjust its operating status based on the commands. This achieves precise control and dynamic adjustment of energy resources, enhances the stability of the power system, and promotes the large-scale absorption of renewable energy. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating a multi-level architecture-based collaborative control method for virtual power plants in one embodiment.

[0038] Figure 2 This is a flowchart illustrating step S120 in one embodiment;

[0039] Figure 3 This is a flowchart illustrating step S130 in one embodiment;

[0040] Figure 4This is a flowchart illustrating the steps of solving the problem using the particle swarm optimization algorithm in one embodiment.

[0041] Figure 5 This is a structural block diagram of a virtual power plant collaborative control device based on a multi-level architecture in one embodiment;

[0042] Figure 6 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0044] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0045] The virtual power plant collaborative control method based on a multi-level architecture provided in this application can be applied to multi-level architecture models. The multi-level architecture model includes a perception layer, a data layer, a decision layer, and an execution layer. Wherein:

[0046] The perception layer is used to acquire power generation data of renewable energy sources, energy storage status data of energy storage devices, and power consumption data of controllable loads within the virtual power plant, based on various sensors deployed within the virtual power plant.

[0047] The data layer is used to perform data mining on power generation data to obtain the first data mining result, to perform data mining on energy storage status data to obtain the second data mining result, and to perform data mining on electricity consumption data to obtain the third data mining result. The first data mining result includes power generation prediction data of renewable energy sources, the second data mining result includes charging and discharging characteristic data of energy storage devices, and the third data mining result includes regulation characteristic data of controllable loads.

[0048] The decision layer is used to input the first, second, and third data mining results into the pre-built collaborative control strategy generation model. It adopts the particle swarm optimization algorithm to output the target collaborative control strategy. The collaborative control strategy generation model takes at least one of the following as its optimization objective: the lowest total operating cost, the highest renewable energy absorption rate, and the minimum load fluctuation.

[0049] The execution layer is used to send control commands corresponding to the target collaborative control strategy to the corresponding renewable power source, energy storage device or load terminal, so that the renewable power source, energy storage device or load terminal can adjust its operating status based on the corresponding control commands.

[0050] In one exemplary embodiment, such as Figure 1 As shown, a collaborative control method for a virtual power plant based on a multi-level architecture is provided, including the following steps S110 to S140. Wherein:

[0051] Step S110: Based on various sensors deployed in the virtual power plant, acquire power generation data of renewable energy sources, energy storage status data of energy storage devices, and power consumption data of controllable loads within the virtual power plant.

[0052] For example, the perception layer can deploy various sensors to collect real-time operational data of distributed renewable energy sources, energy storage devices, and controllable loads within the virtual power plant. Distributed renewable energy sources may include solar photovoltaic power plants, wind farms, etc. Energy storage devices may include battery energy storage systems, supercapacitors, etc. Controllable loads may include industrial loads, commercial loads, residential adjustable loads, etc. Real-time operational data may include power generation data from renewable energy sources, energy storage status data from energy storage devices, and electricity consumption data from controllable loads. Power generation data may include power generation capacity. Energy storage status data may include stored energy capacity. Electricity consumption data may include load power. Real-time operational data may also include device status.

[0053] Step S120: Perform data mining on the power generation data to obtain the first data mining result; perform data mining on the energy storage status data to obtain the second data mining result; perform data mining on the electricity consumption data to obtain the third data mining result. The first data mining result includes the power generation prediction data of renewable energy sources; the second data mining result includes the charging and discharging characteristic data of energy storage devices; and the third data mining result includes the regulation characteristic data of controllable loads.

[0054] For example, the raw data collected by the perception layer can be systematically stored, preprocessed, and deeply analyzed. A database system can be used to classify, store, and manage multi-source heterogeneous data, and data cleaning algorithms can be used to remove outliers and noisy data, improving data quality. Furthermore, data analysis algorithms (e.g., machine learning methods, time-series prediction models, cluster analysis, etc.) can be used to mine the patterns and characteristics hidden in the data. For instance, based on historical meteorological and power generation data, a renewable energy power generation prediction model can be constructed to predict its short-term and ultra-short-term power generation trends; cluster analysis of load power consumption data can be performed to identify the power consumption patterns and behavioral characteristics of different types of loads, providing data basis for controllable load regulation; and the operating status of energy storage devices can be modeled and analyzed to extract their charging and discharging efficiency, capacity decay characteristics, and SOC (State of Charge) variation patterns under different operating conditions, supporting the optimized control of the energy storage system.

[0055] Step S130: Input the first data mining result, the second data mining result, and the third data mining result into the pre-built collaborative control strategy generation model, use the particle swarm optimization algorithm to output the target collaborative control strategy, and take at least one of the following as the optimization objective: the lowest total operating cost, the highest renewable power absorption rate, and the minimum load fluctuation.

[0056] For example, the collaborative control strategy generation model can generate global or local collaborative control strategies based on the multi-dimensional analysis results provided by the data layer, and comprehensively consider factors such as real-time electricity prices in the electricity market, dispatch instructions, and system safety operation constraints (e.g., power balance, frequency stability, equipment operating limits, etc.). This model can be used to implement functions such as generation dispatch, energy storage management, and load regulation. Specifically, the model can optimize the output plans of various distributed power sources based on renewable energy power generation forecasts and load demand, thereby improving renewable energy absorption rates and reducing generation costs. It can also formulate reasonable charging and discharging strategies based on the charging and discharging characteristics and real-time status information of energy storage devices, smoothing renewable energy fluctuations and participating in system frequency regulation and peak shaving. Furthermore, based on load regulation characteristic data, the model can design demand response mechanisms to guide controllable loads to adjust their electricity consumption periods and power while ensuring user electricity comfort, promoting source-load interaction. Finally, particle swarm optimization algorithms can be used to solve the aforementioned multi-objective, multi-constraint optimization problem, efficiently searching for optimal or near-optimal collaborative control strategies in the complex solution space.

[0057] Step S140: Send the control command corresponding to the target collaborative control strategy to the corresponding renewable power source, energy storage device or load terminal, so that the renewable power source, energy storage device or load terminal can adjust its operating status based on the corresponding control command.

[0058] For example, after receiving the collaborative control strategy from the decision-making layer, the execution layer can parse the strategy into directly executable operation instructions through the power communication network or a dedicated control link. For instance, it can send power setpoints or start / stop commands to renewable energy generation units (e.g., photovoltaic inverters, wind turbine controllers, etc.) to adjust their output levels; send charging / discharging power commands or mode switching signals to energy storage systems to control their participation in system frequency regulation or energy time shifting; and send load adjustment commands to load control terminals (e.g., interruptible load controllers, smart power managers, etc.) to incentivize them to increase or decrease power consumption within a certain timeframe. Optionally, the execution layer can also collect equipment response results and actual operating parameters in real time and transmit this information back to the data layer to ensure the effective implementation of the control strategy and to dynamically adjust it according to actual operating conditions, thereby enhancing the system's reliability and adaptability.

[0059] For example, collaborative control decisions can be formulated based on the analysis results of the data layer, combined with electricity market demand and system operation constraints. The decision layer includes multiple sub-modules, such as a generation dispatch module, an energy storage management module, and a load regulation module. The generation dispatch module optimizes the generation plan of distributed power sources based on renewable energy generation forecasts and load demand; the energy storage management module determines the charging and discharging strategies of energy storage devices to balance electricity supply and demand and stabilize system frequency; and the load regulation module formulates load regulation schemes to guide controllable loads to participate in demand response.

[0060] The aforementioned multi-level architecture-based collaborative control method for virtual power plants utilizes various sensors to acquire power generation data from renewable energy sources, energy storage status data from energy storage devices, and electricity consumption data from controllable loads within the virtual power plant. This enables comprehensive perception of the operational status of various energy resources within the virtual power plant, providing a complete and real-time data foundation for subsequent data analysis and collaborative control. By performing data mining on the power generation data, energy storage status data, and electricity consumption data, results are obtained that include renewable energy power generation prediction data, energy storage device charge / discharge characteristic data, and controllable load regulation characteristic data. This allows for a deep understanding of the core characteristics of various energy resources, providing precise basis for formulating collaborative control strategies and effectively addressing the challenges. To address the intermittency and volatility of renewable energy, this approach inputs data mining results into a pre-built collaborative control strategy generation model. Using a particle swarm optimization algorithm, it outputs a target collaborative control strategy with at least one of the following optimization objectives: lowest total operating cost, highest renewable energy absorption rate, and minimum load fluctuation. This enables global optimization scheduling of energy resources within a virtual power plant, fully leveraging the synergistic advantages of various energy resources and improving energy utilization efficiency. Control commands corresponding to the target collaborative control strategy are sent to relevant energy equipment or load terminals, allowing the equipment to adjust its operating status based on the commands. This achieves precise control and dynamic adjustment of energy resources, enhances the stability of the power system, and promotes the large-scale absorption of renewable energy.

[0061] In one exemplary embodiment, the power generation data may include solar photovoltaic power generation data and wind power generation data, such as... Figure 2 As shown, step S120 above may include:

[0062] Step S201: Based on the pre-built solar photovoltaic power generation prediction model and wind power generation model, data mining is performed on the solar photovoltaic power generation data and wind power generation data respectively to obtain the first data mining result. The first data mining result includes the solar photovoltaic power generation prediction data and the wind power generation prediction data. The solar photovoltaic power generation prediction model is established using the support vector machine algorithm, and the wind power generation model is obtained by training the machine learning model based on the historical captured power of the wind turbine.

[0063] For example, a solar photovoltaic power generation prediction model can use historical sunlight intensity, ambient temperature, cloud cover, season, and weather type as input vectors, and historical photovoltaic power generation for the corresponding time period as output labels to construct a training sample set. Using the radial basis function (RBF) as the kernel function, an optimization algorithm is used to solve for the Lagrange multipliers and bias terms to establish a nonlinear regression prediction model for photovoltaic power generation, thereby achieving accurate prediction of power generation for future periods.

[0064] Specifically, a training sample set can be constructed first. , where the input vector It can include features related to photovoltaic power generation, such as historical light intensity, ambient temperature, cloud cover, season, and weather type, to fully consider the impact of cloud cover and seasonal changes on light intensity, and output labels. This can be the solar photovoltaic power generation capacity for the corresponding time period; then, the radial basis function (RBF) is selected as the kernel function, and its expression is:

[0065] (1)

[0066] in, The kernel function parameters are used; then, the optimal regression function is found through the support vector machine (SVM) algorithm, and the final photovoltaic power generation prediction model is established as follows:

[0067] (2)

[0068] in, Let b be a Lagrange multiplier and b be a bias term; finally, an optimization algorithm is used to solve the problem. The model is trained by setting b to minimize the prediction error on the training sample set. The model is then used to predict the solar photovoltaic power generation at different time scales, and the corresponding power generation prediction data is obtained.

[0069] For example, a wind power generation prediction model can be obtained by training a machine learning model based on historical captured power data of wind turbines, combined with parameters such as wind speed, wind direction, air density, rotor radius, and wind energy utilization coefficient. For instance, the theoretical captured power can be calculated first using Betz theory, and then the functional relationship between the actual wind energy utilization coefficient and wind speed can be fitted based on historical operating data to establish an accurate model of power-wind speed characteristics. This model can be trained and optimized using machine learning methods such as neural networks and random forests to capture the nonlinear and fluctuating characteristics of wind power generation and improve prediction accuracy.

[0070] Specifically, the theoretical capture power of the wind turbine can first be calculated based on the Betz theory, and the calculation formula is as follows:

[0071] (3)

[0072] in, Where R is the air density and R is the rotor radius. The wind energy utilization coefficient is defined as a function of wind speed v. Historical captured power data of wind turbines, corresponding wind speed, wind direction, wind shear data, and wind farm topography information are collected. Combined with the theoretical captured power, feature parameters related to wind power generation are extracted through numerical simulation and actual measurement data processing. These feature parameters are then input into a machine learning model (e.g., neural networks, random forests), using historical captured power as the output label to train the model. The focus is on optimizing the wind energy utilization coefficient through fitting a large amount of historical data. The model is designed to determine the functional relationship between wind speed v and wind power. During training, the model parameters are continuously adjusted to accurately capture the uncertainty and volatility of wind power generation. After training, a wind power generation prediction model is obtained, which can be used to output wind turbine power generation prediction data.

[0073] In this embodiment, a solar photovoltaic power generation prediction model is constructed by using a support vector machine algorithm. By combining factors such as sunlight and temperature with historical data, the complex relationship between power generation and influencing factors is accurately fitted, thereby improving the accuracy of photovoltaic power prediction. A wind power model is constructed by training a machine learning model based on the historical power captured by wind turbines. By combining Betz theory with historical data, the uncertainty characteristics of wind power are captured, thereby improving the accuracy of wind power prediction. Therefore, the renewable energy absorption capacity and power system stability are further improved.

[0074] In one exemplary embodiment, please continue to refer to Figure 2 The above step S120 may further include:

[0075] Step S202: Based on the pre-built energy storage device characteristic analysis model, data mining is performed on the energy storage state data to obtain the second data mining result. The second data mining result includes the efficiency characteristic data and state of charge change data of the energy storage device under different charge and discharge rates. The energy storage device characteristic analysis model is used to determine the efficiency characteristic data by fitting the relationship curve between charge and discharge efficiency and charge and discharge rate, and to determine the state of charge change data by using the ampere-hour integration method.

[0076] Specifically, the methods for determining efficiency characteristic data may include: first, obtaining the charging input power Q of the energy storage device at different charge / discharge rates C through experimental testing. in With discharge output power Q out Based on this, the charging efficiency η at the corresponding charge / discharge rate can be calculated. c Discharge efficiency η d (The calculation logic for discharge efficiency is consistent with that for charging efficiency); then, based on multiple sets of different charge / discharge rates C and their corresponding efficiencies (η)... c η d Based on the experimental data, the relationship curves between charging efficiency and charge / discharge rate were obtained by fitting the data.

[0077] (4)

[0078] And the relationship curve between discharge efficiency and charge / discharge rate:

[0079] (5)

[0080] Where a1, b1, and c1 are the charging efficiency fitting coefficients, and a2, b2, and c2 are the charging efficiency fitting coefficients and the discharging efficiency fitting coefficients, the efficiency characteristic data of the energy storage device under any charge and discharge rate are determined through the above relationship curves; at the same time, an equivalent circuit model of the energy storage device is established, and combined with the efficiency characteristic data, the voltage and current variation law under different charge and discharge rates is analyzed, thereby determining the optimal charge and discharge range of the energy storage device.

[0081] Methods for determining changes in state of charge can include: using the ampere-hour integration method, based on the formula:

[0082] (6)

[0083] Calculate the state of charge of the energy storage device, where, Let Q be the state of charge at time t, and Q be the rated capacity of the energy storage device. Let I be the charging and discharging current at time τ, where I is positive during charging and negative during discharging. Simultaneously, considering the self-discharge effect of the energy storage device, the above calculation results are corrected based on the self-discharge rate σ. The corrected formula is:

[0084] (7)

[0085] Obtain the final state of charge (SOC) change data; combine the optimal charge and discharge range to formulate corresponding SOC management strategies to extend the service life of energy storage devices.

[0086] In this embodiment, by fitting the relationship curve between charge / discharge efficiency and rate of the energy storage device characteristic analysis model, it is possible to accurately obtain efficiency characteristic data at different rates, clarify the optimal charge / discharge range, reduce energy loss, and improve energy storage utilization efficiency. By using the ampere-hour integral method to calculate the change in state of charge data and combining it with self-discharge correction, it is possible to accurately grasp the remaining power of the energy storage device in real time, providing a basis for the formulation of charge / discharge strategies and avoiding supply and demand imbalances caused by misjudgment of power capacity. Therefore, it is possible to extend the life of energy storage and further improve the stability of the power system.

[0087] In one exemplary embodiment, please continue to refer to Figure 2 The above step S120 may further include:

[0088] Step S203: Based on the pre-built load characteristic database, data mining is performed on the electricity consumption data to obtain the third data mining result. The third data mining result includes the adjustable power range of controllable loads, load response speed and load adjustment cost data. The load characteristic database is obtained by cluster analysis of the electricity consumption patterns of different types of controllable loads.

[0089] Specifically, controllable loads can first be classified into industrial loads, commercial loads, residential loads, etc., and historical electricity consumption data (including power consumption, electricity consumption time period, adjustment records, etc.) of each type of load can be collected. Then, clustering analysis algorithms (such as K-means algorithm) can be used to mine the electricity consumption patterns of different types of loads, identify typical electricity consumption characteristics of the same type of load (such as the continuous and stable electricity consumption pattern of industrial loads and the time-sharing fluctuation pattern of residential loads), form a load characteristic sub-library of the classification, and integrate them to build a complete load characteristic database.

[0090] In this embodiment, by constructing a load characteristic database based on cluster analysis, it is possible to accurately distinguish the electricity consumption patterns of different types of controllable loads such as industrial, commercial, and residential loads. By mining data such as adjustable power range, response speed, and adjustment cost, a basis is provided for formulating differentiated load control strategies. For example, less adjustment is needed for high-cost industrial loads, while precise scheduling is needed for high-potential residential loads. This can guide loads to use electricity during off-peak hours, balance power supply and demand, and thus further improve the renewable energy absorption capacity and power system stability.

[0091] In one exemplary embodiment, such as Figure 3 As shown, step S130 above may include:

[0092] Step S301: Input the first data mining result and the second data mining result into the pre-built collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing power generation cost and balancing power supply and demand as optimization objectives, and solves the problem through particle swarm optimization algorithm to generate the first collaborative control strategy.

[0093] Specifically, after the first data mining results (including renewable energy power generation prediction data) and the second data mining results (including energy storage device charging and discharging characteristics and SOC change data) are input into the collaborative control strategy to generate the model, the model takes minimizing system power generation cost and achieving power supply and demand balance as its dual optimization objectives. The objective function is shown in the following formula:

[0094] (8)

[0095] That is, the total cost is a function of the generation cost of the i-th power source in the distributed power source set N. ( The power generation capacity of the power source is a function of the charging and discharging cost of the j-th device in the set of energy storage devices M. ( The sum of the charging and discharging power of the device is included; simultaneously, the constraints are: the power generation of the distributed power source does not exceed its rated power upper limit and is not lower than its technical operating lower limit; the charging and discharging power of the energy storage device conforms to the power limit corresponding to the device's charging and discharging rate; and the SOC of the energy storage device is maintained within the safe operating range (e.g., 20%-80%). The model is solved using a particle swarm optimization algorithm, and the output first cooperative control strategy includes: when the predicted power generation of the renewable power source exceeds the load demand, the energy storage device charges at the optimal charging and discharging power; when the predicted power generation is insufficient or the load demand increases, the energy storage device discharges at the optimal charging and discharging power to supplement energy; and the corresponding distributed power source power generation scheduling plan.

[0096] Step S302: Input the first data mining result and the third data mining result into the collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing the total running cost as the optimization objective and solves it through the particle swarm optimization algorithm to generate the second collaborative control strategy.

[0097] Specifically, after the first data mining result (renewable power generation prediction data) and the third data mining result (including controllable load adjustable power range, response speed, and adjustment cost data) are input into the model, the model takes minimizing the total system operating cost as the optimization objective. The objective function is referenced in the following formula:

[0098] (9)

[0099] That is, the total operating cost is a function of the distributed generation cost and the adjustment cost of the k-th load in the controllable load set K. ( The sum of the load adjustment amounts is the sum of the controllable load adjustment amounts; constraints include: the controllable load adjustment amount does not exceed its adjustable power range, and the load adjustment response time meets the equipment technical requirements (e.g., the residential air conditioning adjustment response does not exceed 30 minutes). After being solved by the particle swarm optimization algorithm, the second cooperative control strategy includes: a distributed power generation plan based on power generation prediction data, and adjustment schemes for different types of controllable loads. For example, for residential loads with low adjustment costs, they are guided to increase electricity consumption during peak power generation periods and reduce electricity consumption during off-peak periods; for industrial loads with high adjustment costs, small-scale adjustments are only made when power generation is severely excessive or insufficient, and the adjustment method matches the adjustment rhythm of its production process parameters.

[0100] Step S303: Input the second data mining results and the third data mining results into the collaborative control strategy generation model so that the collaborative control strategy generation model takes the minimization of load fluctuation as the optimization objective and solves it through the particle swarm optimization algorithm to generate the third collaborative control strategy.

[0101] Specifically, after the second data mining results (energy storage device charge / discharge characteristics, SOC data) and the third data mining results (controllable load characteristic data) are input into the model, the model takes minimizing system load fluctuations as its optimization objective. The objective function is referenced in the following formula:

[0102] (10)

[0103] That is, the load fluctuation is the actual load power at each moment within the time period T. Average load power within the cycle The minimum sum of the absolute values ​​of the differences; constraints include the charging and discharging power of the energy storage device, SOC limits, and controllable load adjustment range and response speed limits. Solved by the particle swarm optimization algorithm, the third collaborative control strategy includes: when the load fluctuation exceeds a set threshold (e.g., 10%), the energy storage device charges and discharges at the optimal power according to the current SOC state to smooth the load fluctuation, while combining the controllable load adjustment potential—prioritizing rapid adjustment of commercial loads with faster response speeds (e.g., shopping mall lighting), and staggering peak scheduling for residential load clusters with high adjustment potential, forming a load curve smoothing scheme that coordinates energy storage and load.

[0104] Step S304: Based on the first collaborative control strategy, the second collaborative control strategy, and the third collaborative control strategy, output the target collaborative control strategy.

[0105] The coordinated control strategy includes at least one of the following: a scheduling plan for renewable power sources, a charging and discharging strategy for energy storage devices, and a regulation scheme for controllable loads.

[0106] Specifically, when the power system is in peak electricity demand and renewable energy generation is insufficient, the energy storage discharge scheme in the first coordinated control strategy and the load reduction regulation scheme in the second coordinated control strategy can be prioritized, supplemented by the load fluctuation smoothing measures in the third coordinated control strategy. When the system is in low electricity demand and renewable energy generation is excessive, the energy storage charging scheme in the first coordinated control strategy and the load increment regulation scheme in the second coordinated control strategy should be prioritized, combined with the third coordinated control strategy to optimize the load curve. During normal operation, the optimal parameters of generation dispatch, energy storage charging and discharging, and load regulation in the three strategies are combined to form a comprehensive control strategy that takes into account cost, supply and demand, and fluctuations. The final output target coordinated control strategy specifies the specific power generation of renewable energy sources, the charging and discharging time and power of energy storage devices, and the regulation objects and regulation amounts of controllable loads in each time period.

[0107] In this embodiment, three types of collaborative control strategies are generated by inputting different data combinations for different scenarios. These strategies accurately match the collaborative needs of power generation and energy storage, load and power generation, and energy storage and load, avoiding the problem of poor adaptability of a single strategy and improving the targeting of control. By using particle swarm optimization algorithm to efficiently solve problems with objectives such as minimizing power generation costs and achieving supply-demand balance, the optimal solution is quickly found, reducing system operating costs. By integrating the three types of strategies to output the target strategy, multi-objective optimization is achieved, giving full play to the synergistic advantages of various energy resources. Therefore, the renewable energy absorption capacity and power system stability are further improved.

[0108] In one exemplary embodiment, such as Figure 4 As shown, the steps for solving the problem using the particle swarm optimization algorithm may include:

[0109] Step A1: Based on the decision variables and constraints of the decision variables in the cooperative control strategy model, an initial population containing multiple particles is randomly generated. The position of each particle in the multiple particles represents a candidate cooperative control strategy.

[0110] Specifically, the decision variables in the collaborative control strategy model include: the power generation allocation of distributed power sources (such as photovoltaic power plants and wind farms). The charging and discharging power of energy storage devices (such as battery energy storage systems and supercapacitors) Adjustment amount of controllable loads (such as industrial loads and residential adjustable loads) The constraints on the decision variables include: That is, the power generation capacity of distributed power sources shall not exceed their rated power limit and shall not be lower than their minimum technical operating power. This means that the charging and discharging power of the energy storage device meets the power limit corresponding to its charging and discharging rate (such as the maximum charging and discharging power determined based on the charging and discharging efficiency curve), and the state of charge (SOC) of the energy storage device is maintained within a safe range. ; This means that the controllable load adjustment amount does not exceed its adjustable power range. Within the above constraints, N particles are randomly generated to form an initial population, and the position vector of each particle i is:

[0111] (11)

[0112] This position vector represents a set of candidate cooperative control strategies; at the same time, each particle is assigned an initial velocity vector:

[0113] (12)

[0114] Furthermore, the speed parameters are randomly set within a reasonable range.

[0115] Step A2: Calculate the fitness value of each particle based on the optimization objective function of the cooperative control strategy model.

[0116] Specifically, if the optimization objective is to minimize the system's power generation cost and achieve a balance between power supply and demand (corresponding to the generation of the first collaborative control strategy), the objective function is based on the above formula (8), i.e., the total cost is the power generation cost function of all distributed power sources. Cost function of charging and discharging all energy storage devices The sum of these values, substituted into the objective function, yields the fitness value of particle i.

[0117] (13)

[0118] If the optimization objective is to minimize the total operating cost of the system (corresponding to the generation of the second collaborative control strategy), the objective function is referenced by formula (9). The total cost is the sum of the distributed power generation cost and the controllable load adjustment cost function. Substitute these values ​​into the formula to calculate the fitness value. If the optimization objective is to minimize the system load fluctuation (corresponding to the generation of the third collaborative control strategy), the objective function is referenced by formula (10). That is, the load fluctuation is the sum of the absolute values ​​of the difference between the actual load power and the average load power at each moment within the time period. Substitute these values ​​into the formula to calculate the load fluctuation, which is the fitness value (the smaller the value, the better the fitness).

[0119] Step A3: Based on the fitness value of each particle, the target particle is selected from the initial population using the roulette wheel algorithm to obtain a new generation of population.

[0120] Specifically, first, the sum of the fitness values ​​of all particles is calculated, and then the selection probability of each particle is calculated:

[0121] (14)

[0122] If the optimization objective is to minimize load fluctuation, the fitness value needs to be normalized first, converting small fluctuation into high probability; then, a random number r between [0,1] is generated. If the j-th particle is selected to enter the new generation population, the random selection process is repeated until the size of the new generation population reaches the initial population size N.

[0123] Step A4: Based on the historical best position of the particles and the global best position of the population, iteratively update the velocity and position of each particle in the new generation of the population until the maximum number of iterations is reached or the change in the population's optimal fitness value is less than a set threshold, then stop iterating and output the current global best position as the cooperative control strategy.

[0124] Specifically, first record the historical best position of each particle (i.e., the position with the best fitness value during the iteration process of that particle) and the global best position of the entire population (i.e., the position with the best fitness value during the iteration process of all particles); then update according to the velocity formula:

[0125] (15)

[0126] Update the particle velocity, where ω is the inertia weight (used to balance global and local search capabilities). and As a learning factor, and A random number between [0,1] This represents the historical best position of particle i. The optimal position for the population is determined; then the position is updated using the following formula:

[0127] (16)

[0128] Update the particle position. After updating, check if the particle position meets the constraints of the decision variables. If it exceeds the constraints, truncate and correct it. Repeat the process of calculating the fitness value, updating the historical optimum and global optimum, and updating the velocity and position until the termination condition is met (e.g., reaching the maximum number of iterations). If the optimal fitness value of the population changes less than a set threshold ϵ after n consecutive iterations, the global optimal position of the current population after stopping iteration is the corresponding optimal cooperative control strategy.

[0129] In this embodiment, an initial population is generated based on decision variables and constraints, with each particle corresponding to a candidate strategy. This ensures the diversity of solutions, avoids getting trapped in local optima, and provides sufficient candidate schemes for subsequent optimization. By combining the objective function to calculate the fitness value and using the roulette wheel algorithm to screen particles, high-quality candidate strategies are accurately retained, improving the quality of the population and laying the foundation for efficient solution. By iteratively updating the speed and position based on the particle's historical best and the population's global best, and stopping the iteration based on the termination condition, the optimal solution can be found quickly while ensuring the optimality of the collaborative control strategy. Therefore, the operating efficiency and stability of the virtual power plant are further improved.

[0130] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0131] Based on the same inventive concept, this application also provides a multi-level architecture-based virtual power plant collaborative control device for implementing the multi-level architecture-based virtual power plant collaborative control method described above. The solution provided by this device is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more multi-level architecture-based virtual power plant collaborative control device embodiments provided below can be found in the limitations of the multi-level architecture-based virtual power plant collaborative control method described above, and will not be repeated here.

[0132] In one exemplary embodiment, such as Figure 5 As shown, a virtual power plant collaborative control device 400 based on a multi-level architecture is provided, including: a data acquisition module 401, a data mining module 402, and a strategy generation module 403, wherein:

[0133] The data acquisition module is used to acquire power generation data of renewable energy sources, energy storage status data of energy storage devices, and power consumption data of controllable loads within the virtual power plant, based on various sensors deployed within the virtual power plant.

[0134] The data mining module is used to perform data mining on power generation data to obtain the first data mining result, to perform data mining on energy storage status data to obtain the second data mining result, and to perform data mining on electricity consumption data to obtain the third data mining result. The first data mining result includes power generation prediction data of renewable energy sources, the second data mining result includes charging and discharging characteristic data of energy storage devices, and the third data mining result includes regulation characteristic data of controllable loads.

[0135] The strategy generation module is used to input the first data mining result, the second data mining result, and the third data mining result into the pre-built collaborative control strategy generation model. It adopts the particle swarm optimization algorithm to output the target collaborative control strategy. The collaborative control strategy generation model takes at least one of the following as the optimization objective: the lowest total operating cost, the highest renewable energy absorption rate, and the minimum load fluctuation.

[0136] The collaborative control module is used to send control commands corresponding to the target collaborative control strategy to the corresponding renewable power source, energy storage device or load terminal, so that the renewable power source, energy storage device or load terminal can adjust its operating status based on the corresponding control commands.

[0137] In one embodiment, the power generation data includes solar photovoltaic power generation data and wind power generation data. Data mining is performed on the power generation data, and the aforementioned data mining module is further used for:

[0138] Based on pre-built solar photovoltaic power generation prediction models and wind power generation models, data mining is performed on solar photovoltaic power generation data and wind power generation data respectively to obtain the first data mining results. The first data mining results include solar photovoltaic power generation prediction data and wind turbine power generation prediction data. The solar photovoltaic power generation prediction model is established using the support vector machine algorithm, and the wind power generation model is obtained by training a machine learning model based on the historical captured power of the wind turbine.

[0139] In one embodiment, the data mining module is further configured to:

[0140] Based on the pre-built energy storage device characteristic analysis model, data mining is performed on the energy storage state data to obtain the second data mining result. The second data mining result includes the efficiency characteristic data and state of charge change data of the energy storage device under different charge and discharge rates. The energy storage device characteristic analysis model is used to determine the efficiency characteristic data by fitting the relationship curve between charge and discharge efficiency and charge and discharge rate, and to determine the state of charge change data by using the ampere-hour integration method.

[0141] In one embodiment, the data mining module is further configured to:

[0142] Based on a pre-built load characteristic database, data mining is performed on electricity consumption data to obtain a third data mining result. The third data mining result includes the adjustable power range of controllable loads, load response speed, and load adjustment cost data. The load characteristic database is obtained by cluster analysis of the electricity consumption patterns of different types of controllable loads.

[0143] In one embodiment, the strategy generation module is further configured to:

[0144] The first and second data mining results are input into the pre-built collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing power generation cost and balancing power supply and demand as optimization objectives, and is solved by particle swarm optimization algorithm to generate the first collaborative control strategy.

[0145] The first and third data mining results are input into the collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing the total running cost as the optimization objective and is solved by the particle swarm optimization algorithm to generate the second collaborative control strategy.

[0146] The second and third data mining results are input into the collaborative control strategy generation model so that the collaborative control strategy generation model takes the minimization of load fluctuation as the optimization objective and is solved by the particle swarm optimization algorithm to generate the third collaborative control strategy.

[0147] Based on the first, second, and third collaborative control strategies, the target collaborative control strategy is output.

[0148] In one embodiment, the strategy generation module is further configured to:

[0149] Based on the decision variables and constraints of the decision variables in the cooperative control strategy model, an initial population containing multiple particles is randomly generated, and the position of each particle in the multiple particles represents a candidate cooperative control strategy.

[0150] Calculate the fitness value of each particle based on the optimization objective function of the cooperative control strategy model;

[0151] Based on the fitness value of each particle, the roulette wheel algorithm is used to select target particles from the initial population to obtain a new generation of population;

[0152] Based on the historical best position of the particles and the global best position of the population, the velocity and position of each particle in the new generation of the population are iteratively updated until the maximum number of iterations is reached or the change in the population's optimal fitness value is less than a set threshold. Then, the iteration stops and the current global best position is output as the cooperative control strategy.

[0153] The modules in the aforementioned multi-level architecture-based virtual power plant collaborative control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0154] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a multi-level architecture-based virtual power plant collaborative control method.

[0155] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0156] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0157] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0158] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0159] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0160] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0161] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A collaborative control method for a virtual power plant based on a multi-level architecture, characterized in that, The method includes: Based on various sensors deployed within the virtual power plant, the power generation data of renewable energy sources, the energy storage status data of energy storage devices, and the power consumption data of controllable loads within the virtual power plant are acquired. Data mining is performed on the power generation data to obtain a first data mining result; data mining is performed on the energy storage status data to obtain a second data mining result; and data mining is performed on the electricity consumption data to obtain a third data mining result. The first data mining result includes the power generation prediction data of the renewable energy source; the second data mining result includes the charging and discharging characteristic data of the energy storage device; and the third data mining result includes the regulation characteristic data of the controllable load. The first data mining result, the second data mining result, and the third data mining result are input into the pre-built collaborative control strategy generation model. The particle swarm optimization algorithm is used to output the target collaborative control strategy. The collaborative control strategy generation model takes at least one of the following as the optimization objective: the lowest total operating cost, the highest renewable energy absorption rate, and the lowest load fluctuation. The control command corresponding to the target collaborative control strategy is sent to the corresponding renewable power source, energy storage device or load terminal, so that the renewable power source, the energy storage device or the load terminal adjusts its operating state based on the corresponding control command.

2. The method according to claim 1, characterized in that, The power generation data includes solar photovoltaic power generation data and wind power generation data. The data mining of the power generation data to obtain a first data mining result includes: Based on pre-built solar photovoltaic power generation prediction models and wind power generation models, data mining is performed on the solar photovoltaic power generation data and wind power generation data respectively to obtain a first data mining result. The first data mining result includes solar photovoltaic power generation prediction data and wind turbine power generation prediction data. The solar photovoltaic power generation prediction model is established using the support vector machine algorithm, and the wind power generation model is obtained by training a machine learning model based on the historical captured power of the wind turbine.

3. The method according to claim 1, characterized in that, The step of performing data mining on the energy storage status data to obtain a second data mining result includes: Based on a pre-built energy storage device characteristic analysis model, data mining is performed on the energy storage state data to obtain a second data mining result. The second data mining result includes the efficiency characteristic data and state of charge change data of the energy storage device under different charge and discharge rates. The energy storage device characteristic analysis model is used to determine the efficiency characteristic data by fitting the relationship curve between charge and discharge efficiency and charge and discharge rate, and to determine the state of charge change data by using the ampere-hour integration method.

4. The method according to claim 1, characterized in that, The data mining of the electricity consumption data to obtain a third data mining result includes: Based on a pre-built load characteristic database, data mining is performed on the electricity consumption data to obtain a third data mining result. The third data mining result includes the adjustable power range, load response speed, and load adjustment cost data of the controllable load. The load characteristic database is obtained by cluster analysis of the electricity consumption patterns of different types of controllable loads.

5. The method according to any one of claims 1 to 4, characterized in that, The step of inputting the first data mining result, the second data mining result, and the third data mining result into a pre-constructed collaborative control strategy generation model, and using a particle swarm optimization algorithm to output a target collaborative control strategy includes: The first data mining result and the second data mining result are input into the pre-built collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing power generation cost and balancing power supply and demand as optimization objectives, and is solved by particle swarm optimization algorithm to generate the first collaborative control strategy. The first data mining result and the third data mining result are input into the collaborative control strategy generation model so that the collaborative control strategy generation model takes minimizing the total running cost as the optimization objective and is solved by the particle swarm optimization algorithm to generate a second collaborative control strategy. The second data mining result and the third data mining result are input into the collaborative control strategy generation model so that the collaborative control strategy generation model takes the minimization of load fluctuation as the optimization objective and is solved by the particle swarm optimization algorithm to generate the third collaborative control strategy. Based on the first collaborative control strategy, the second collaborative control strategy, and the third collaborative control strategy, a target collaborative control strategy is output.

6. The method according to claim 5, characterized in that, The solution obtained through particle swarm optimization includes: Based on the decision variables and constraints in the cooperative control strategy model, an initial population containing multiple particles is randomly generated, where the position of each particle represents a candidate cooperative control strategy. The fitness value of each particle is calculated based on the optimization objective function of the cooperative control strategy model. Based on the fitness value of each particle, the target particle is selected from the initial population using the roulette wheel algorithm to obtain a new generation of population; Based on the historical best position of the particles and the global best position of the population, the velocity and position of each particle in the new generation population are iteratively updated until the maximum number of iterations is reached or the change in the population's optimal fitness value is less than a set threshold. Then, the iteration stops and the current global best position is output as the cooperative control strategy.

7. A virtual power plant collaborative control device based on a multi-level architecture, characterized in that, The device includes: The data acquisition module is used to acquire power generation data of renewable energy sources, energy storage status data of energy storage devices, and power consumption data of controllable loads within the virtual power plant, based on various sensors deployed within the virtual power plant. The data mining module is used to perform data mining on the power generation data to obtain a first data mining result, to perform data mining on the energy storage status data to obtain a second data mining result, and to perform data mining on the electricity consumption data to obtain a third data mining result. The first data mining result includes the power generation prediction data of the renewable energy source, the second data mining result includes the charging and discharging characteristic data of the energy storage device, and the third data mining result includes the regulation characteristic data of the controllable load. The strategy generation module is used to input the first data mining result, the second data mining result and the third data mining result into the pre-built collaborative control strategy generation model, and use the particle swarm optimization algorithm to output the target collaborative control strategy. The collaborative control strategy generation model takes at least one of the following as the optimization objective: the lowest total operating cost, the highest renewable energy absorption rate and the lowest load fluctuation. The collaborative control module is used to send the control command corresponding to the target collaborative control strategy to the corresponding renewable power source, energy storage device or load terminal, so that the renewable power source, the energy storage device or the load terminal adjusts its operating state based on the corresponding control command.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images