Virtual power plant cloud edge collaborative scheduling method and device based on multi-objective optimization and confidence coefficient screening

Abstract

The invention relates to a virtual power plant cloud edge collaborative scheduling method based on multi-objective optimization and confidence coefficient screening. The method comprises the following steps: obtaining related data of a virtual power plant transaction terminal; constructing a multi-objective function based on the related data, selecting a comprehensive optimal scheme under the condition that the multi-objective function is satisfied, and generating an execution strategy set covering global scheduling of the virtual power plant; and carrying out cloud edge cooperative scheduling on the virtual power plant based on the execution strategy set. According to the method, multi-stage time sequence multi-target scheduling optimization and a Pareto frontier-based strategy generation method are fused, a multi-VPP joint optimal scheduling model is constructed in a cloud set, a high-reliability solution set is screened through confidence and is distributed to each VPP unit, and an edge controller intelligently selects an adaptive solution according to a local state, so that the optimal scheduling efficiency is improved. And collaborative optimization of the response efficiency, the execution reliability and the overall benefit of the system is realized. The method has engineering realizability, and is suitable for multi-scene complex environments such as a region-level virtual power plant platform and a source-network-load-storage cooperative control system.

Description

Technical Field

[0001] This invention relates to the fields of automotive electronics and electronic measurement, and in particular to a virtual power plant cloud-edge collaborative scheduling method and device based on multi-objective optimization and confidence screening. Background Technology

[0002] With the diversification of energy structures and the increasing proportion of distributed energy integration, virtual power plants, as important carriers of load aggregation and flexible dispatch, play an increasingly crucial role in ensuring the safe and economical operation of the power system. Currently, virtual power plant dispatch systems generally face the following key challenges:

[0003] (1) Conflicts in scheduling objectives among multiple VPPs and lack of a unified coordination mechanism: There may be resource overlap and differences in regulation objectives among multiple virtual power plant units. Existing systems are mostly optimized independently, which leads to a decrease in overall scheduling efficiency.

[0004] (2) Single or static scheduling objectives: Most existing scheduling methods only target a single objective (such as economic efficiency), ignoring factors such as grid stability and user response differences, resulting in unstable scheduling effects.

[0005] (3) Scheduling strategies are difficult to implement: Although some optimized solutions are theoretically optimal, they are difficult to implement in real-world scenarios due to equipment constraints, user behavior, and other reasons, resulting in low execution confidence.

[0006] (4) Centralized optimization lacks local adaptability: The execution capabilities of edge terminals vary greatly, making it impossible to intelligently select the optimal strategy based on local conditions, which can easily lead to inefficient or delayed responses.

[0007] (5) Lack of multi-timescale linkage scheduling mechanism: The scheduling target weights and constraints should change dynamically under different time scales, but existing systems generally lack this adaptive capability. Summary of the Invention

[0008] The purpose of this invention is to at least address one of the shortcomings of the prior art by providing a virtual power plant cloud-edge collaborative scheduling method and apparatus based on multi-objective optimization and confidence screening.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] Specifically, a cloud-edge collaborative scheduling method for virtual power plants based on multi-objective optimization and confidence screening is proposed, including the following:

[0011] Obtain relevant data from the virtual power plant trading terminal;

[0012] Based on the relevant data, a multi-objective function is constructed. Under the condition of satisfying the multi-objective function, the comprehensive optimal solution is selected to generate an execution strategy set covering the global scheduling of the virtual power plant.

[0013] The virtual power plant is scheduled in a cloud-edge collaborative manner based on the aforementioned set of execution strategies.

[0014] Furthermore, specifically, the relevant data includes,

[0015] V1 Behavioral Habit Model, V2 Adjustable Capacity, V3 Response Time, V4 Response Duration, V5 Adjustment Accuracy, V6 System Ramp-up Capability, V7 Power Prediction, V8 Plan Execution Strategy;

[0016] Among them, the V1 behavior habit model is obtained based on data collection and feature extraction fitting, and is represented by a predictability score, which includes the standard deviation of the load curve and the scheduling response delay rate; the remaining V2-V6 content is obtained during system access testing and will be adjusted according to the actual situation during actual operation; the power prediction of V7 includes the load change of the region and the changes in energy storage power consumption and cycle; the V8 plan execution strategy is obtained by reading relevant data fed back to the cloud server.

[0017] Furthermore, specifically, the process of generating a set of execution policies for global scheduling covering the virtual power plant includes,

[0018] Key objective function fusion,

[0019] The multi-objective function is constructed by f1(x) representing the system's lumped revenue, f2(x) representing the impact of grid fluctuations, and f3(x) representing the response performance. The objective function is obtained by fusing these multi-objective functions. Where the weight w k (t) is dynamically set according to the time period, and the weight sum is 1;

[0020] Set the constraints as follows.

[0021] For each virtual power plant unit, i.e., VPP unit i, the following regulation capability constraints are satisfied:

[0022] Pimin≤P i (t)≤Pimax;

[0023] Where Pimax and Pimin represent the maximum and minimum adjustable power reported from VPP unit i, respectively, and Pi(t) represents the current power target allocated to VPP unit i by the scheduler.

[0024] At the same time, the overall scheduling scheme must meet power balance constraints:

[0025]

[0026] Where P target(t) represents the target total dispatch power value at time t, ΔP is the allowable net deviation threshold, i.e. the error range that should not lead to penalties from the power supply company; n represents the total number of VPP units i.

[0027] Set the iteration termination condition, which includes the convergence condition and the maximum number of iterations, and output the Pareto solution set.

[0028] Furthermore, the method also includes,

[0029] A correction factor ai(t) is added to the prediction result of the i-th VPP unit, and ai(t) is automatically updated and corrected based on the historical error of the model.

[0030] Furthermore, the method also includes,

[0031] A confidence model is introduced to filter and refine the set of execution strategies, specifically including...

[0032] Construct the confidence function:

[0033]

[0034] Where S i For VPP i The historical response success rate weight, ranging from [0,1], P match (x i VPP i ) is strategy x i With VPP i The degree of matching of regulatory capacity, P error For the historical execution error of this virtual power plant unit, SOH i For VPP i The health status of the equipment, u i For VPP i Operating environment influencing factors

[0035] The strategy of C(x) > b is written into the virtual power plant unit scheduling strategy set, where the threshold b ∈ [0,1].

[0036] Dynamic updates: Each time, the weights of each type are updated based on the execution results and historical data, such as:

[0037]

[0038] Where P success (x i ) is strategy x i The probability of success, where α is a preset weighting coefficient. This represents the response success rate weight before the update.

[0039] Furthermore, the method also includes,

[0040] When the virtual power plant unit at the edge receives the execution policy set, it autonomously adapts and executes the policy, then returns the result to the cloud server for further correction. Specifically,

[0041] The fitness function of the virtual power plant unit is:

[0042] K i =β*P match1 (x i VPP i )+γ*P match2 (x i VPP i )+δ*P score (VPP i )

[0043] Among them, P match1 (x i VPP i ) represents strategy x i With VPP i The degree of matching of adjustable resources; P match2 (x i VPP i ) indicates that at the current moment in policy x i Virtual Power Plant (VPP) under certain conditions i Resource availability; P score (VPP i (This refers to the current virtual power plant, VPP) i Adaptability score in edge scenarios;

[0044] Among them, β, γ, and δ are normalized weighting factors, which sum to 1 and can be adjusted based on historical execution feedback.

[0045] This invention also proposes a virtual power plant cloud-edge collaborative scheduling device based on multi-objective optimization and confidence screening, which applies the aforementioned virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening, including the following:

[0046] The data acquisition module is used to acquire relevant data from the virtual power plant trading terminal;

[0047] The strategy generation module is used to construct a multi-objective function based on the relevant data, select the comprehensive optimal solution under the multi-objective function, and generate an execution strategy set covering the global scheduling of the virtual power plant.

[0048] The operation scheduling module is used to perform cloud-edge collaborative scheduling of the virtual power plant based on the execution strategy set.

[0049] This invention also proposes a virtual power plant cloud-edge collaborative scheduling system based on multi-objective optimization and confidence screening, including:

[0050] The cloud server is equipped with a virtual power plant cloud-edge collaborative scheduling device based on multi-objective optimization and confidence screening as described in claim 7, which is used to construct a multi-objective optimization function and generate a Pareto scheduling strategy set based on the capability profiles of all VPP units.

[0051] The virtual power plant unit is used to collect terminal operation data, predict local terminal power change trends, model and classify terminal behavior characteristics, and finally perform adaptation analysis on the strategies sent by the cloud server to complete the final judgment and result feedback of the scheduling task.

[0052] Furthermore, the cloud server is also equipped with a confidence module, which is used to score the confidence of the generated scheduling strategy and filter the set of execution strategies.

[0053] The beneficial effects of this invention are as follows:

[0054] This invention provides a cloud-edge collaborative scheduling method for virtual power plants based on multi-objective optimization and confidence screening. It integrates multi-stage (day-ahead / hourly / real-time) time-series multi-objective scheduling optimization, a Pareto front-based strategy generation method, a confidence-driven strategy screening mechanism, a cloud-edge collaborative intelligent strategy distribution and execution structure, and a centralized scheduler coordination and control mechanism across multiple virtual power plant units.

[0055] This method centrally constructs a multi-VPP joint optimal scheduling model in the cloud, filters highly reliable solution sets through confidence and distributes them to each VPP unit, and the edge controller intelligently selects the appropriate solution based on the local state, thereby achieving synergistic optimization of response efficiency, execution reliability and overall system benefits.

[0056] This invention is engineering feasible and applicable to complex environments in various scenarios, such as regional virtual power plant platforms and source-grid-load-storage collaborative control systems. Attached Figure Description

[0057] The above and other features of this disclosure will become more apparent from the detailed description of the embodiments shown in conjunction with the accompanying drawings. In the accompanying drawings, the same reference numerals denote the same or similar output voltages. Obviously, the drawings described below are merely some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort. In the drawings:

[0058] Figure 1 The diagram shows a flowchart of a cloud-edge collaborative scheduling method for virtual power plants based on multi-objective optimization and confidence screening according to the present invention.

[0059] Figure 2 The diagram shown is a schematic diagram of the structure of a virtual power plant cloud-edge collaborative scheduling system based on multi-objective optimization and confidence screening according to the present invention. Detailed Implementation

[0060] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The same reference numerals used throughout the accompanying drawings indicate the same or similar parts.

[0061] Reference Figure 1 Example 1: This invention proposes a cloud-edge collaborative scheduling method for virtual power plants based on multi-objective optimization and confidence screening, comprising the following:

[0062] S1. Obtain relevant data from the virtual power plant trading terminal;

[0063] S2. Construct a multi-objective function based on the relevant data, select the comprehensive optimal solution under the multi-objective function, and generate an execution strategy set covering the global scheduling of the virtual power plant.

[0064] The virtual power plant is scheduled in a cloud-edge collaborative manner based on the aforementioned set of execution strategies.

[0065] In this embodiment 1, a multi-stage (day-ahead / hourly / real-time) time-series multi-objective scheduling optimization, a Pareto front-based strategy generation method, a confidence-driven strategy selection mechanism, a cloud-edge collaborative intelligent strategy distribution and execution structure, and a centralized scheduler coordination and control mechanism across multiple virtual power plant units are integrated.

[0066] This method centrally constructs a multi-VPP joint optimal scheduling model in the cloud, filters highly reliable solution sets through confidence and distributes them to each VPP unit, and the edge controller intelligently selects the appropriate solution based on the local state, thereby achieving synergistic optimization of response efficiency, execution reliability and overall system benefits.

[0067] This invention is engineering feasible and applicable to complex environments in various scenarios, such as regional virtual power plant platforms and source-grid-load-storage collaborative control systems.

[0068] This invention addresses how to achieve centralized and unified scheduling in multi-VPP scenarios, balancing global efficiency with unit autonomy.

[0069] Construct a multi-objective scheduling optimization model that can dynamically adapt to the differences in objectives at different scheduling stages;

[0070] Based on historical response and behavior data, a confidence evaluation function for the executability of scheduling strategies is established, and the model is dynamically optimized.

[0071] Improve system response speed and reliability through edge adaptation mechanisms.

[0072] In a preferred embodiment of the present invention, specifically, the relevant data includes...

[0073] V1 Behavioral habit model (such as historical load curve, usage cycle frequency), V2 Adjustable capacity, V3 Response time, V4 Response duration, V5 Adjustment accuracy, V6 System ramp-up capability, V7 Power prediction, V8 Plan execution strategy.

[0074] The V1 terminal behavior habit model is primarily based on data collection and feature extraction fitting, employing methods such as K-Means clustering, LSTM prediction, and Markov chains for modeling, and is represented by a predictability score (e.g., load curve standard deviation, scheduling response delay rate). This value measures the stability and predictability of terminal operation behavior. The remaining V2-V6 content is obtained during system access testing and will be adjusted according to actual conditions during operation. V7 power prediction includes load changes in the region and changes in energy storage power consumption and cycles. The V8 plan execution strategy is derived from steps S2, S3, and S4 and then fed back to the cloud server.

[0075] In a preferred embodiment of the present invention, specifically, the process of generating an execution strategy set covering the global scheduling of the virtual power plant involves constructing a multi-objective function based on the capability profile and power prediction results of the virtual power plant units collected in step S1, generating a Pareto front solution set using a multi-objective evolutionary algorithm (such as NSGA-II), and finally generating the strategy allocation logic for each virtual power plant unit. This includes...

[0076] Key objective function fusion,

[0077] The multi-objective function is constructed as follows: f1(x) represents the system's total revenue (including electricity purchase cost, dispatch revenue, etc.), f2(x) represents the impact of grid fluctuations (including grid voltage, frequency, etc.), and f3(x) represents the response performance (ramp-up capability, response time, duration, etc.). The objective function is obtained by fusing the multi-objective functions: Where the weight w k (t) is dynamically set according to the time period, and the weight sum is 1;

[0078] The constraints are set as follows.

[0079] For each virtual power plant unit, i.e., VPP unit i, the following regulation capability constraints are satisfied:

[0080] Pimin≤P i (t)≤Pimax;

[0081] Where Pimax and Pimin represent the maximum and minimum adjustable power reported from VPP unit i, respectively, and Pi(t) represents the current power target allocated to VPP unit i by the scheduler.

[0082] At the same time, the overall scheduling scheme must meet power balance constraints:

[0083]

[0084] Where P target (t) represents the target total dispatch power value at time t, ΔP is the allowable net deviation threshold, i.e. the error range that should not lead to penalties from the power supply company; n represents the total number of VPP units i.

[0085] Set the iteration termination condition, which includes the convergence condition and the maximum number of iterations, and output the Pareto solution set.

[0086] In practical application, it is assumed that there are 3 virtual power plant units: 3 industrial load aggregations (VPP1), 3 photovoltaic + energy storage systems (VPP2), and 3 residential flexible load groups (VPP3). After running NSGA-II for 100 generations, 50 Pareto solutions are obtained, resulting in strategy A = {VPP1: 100kW, VPP2: -50kW, VPP3: -30kW}, which possesses low cost and high reliability characteristics. Then, confidence evaluation and edge adaptation processing are performed.

[0087] In a preferred embodiment of the present invention, the method further includes,

[0088] A correction factor ai(t) is added to the prediction result of the i-th VPP unit, and ai(t) is automatically updated and corrected based on the historical error of the model.

[0089] In this preferred embodiment, if a correction factor ai(t) is introduced into the load forecast deviation, ai(t) is automatically updated and corrected based on the model's historical errors. Similarly, multiple correction factors can be added to capacity vectors such as capacity forecasting and electricity price assessment.

[0090] In a preferred embodiment of the present invention, the method further includes S3.

[0091] A confidence model is introduced to filter and refine the set of execution strategies, specifically including...

[0092] Construct the confidence function:

[0093]

[0094] Where S i For VPP i The historical response success rate weight, ranging from [0,1], P match(x i VPP i ) is strategy x i With VPP i The degree of matching of regulatory capacity, P error This is the historical virtual power plant unit, or VPP. i Execution error, SOH i For VPP i The health status of the equipment (this value can be set to select whether to put the equipment on or off), u i For VPP i Operating environment factors (such as power grid stability, communication stability, etc.)

[0095] The strategy of C(x) > b is written into the virtual power plant unit scheduling strategy set, where the threshold b ∈ [0,1].

[0096] Dynamic updates: Each time, the weights of each type are updated based on the execution results and historical data, such as:

[0097]

[0098] Where P success (x i ) is strategy x i The probability of success, where α is a preset weighting coefficient. This represents the response success rate weight before the update.

[0099] In a preferred embodiment of the present invention, the method further includes S4.

[0100] When the virtual power plant unit at the edge receives the execution policy set, it autonomously adapts and executes the policy, then returns the result to the cloud server for further correction. Specifically,

[0101] The fitness function of the virtual power plant unit is:

[0102] K i =β*P match1 (x i VPP i )+γ*P match2 (x i VPP i )+δ*P score (VPP i )

[0103] Among them, P match1 (x i VPP i ) represents strategy x i With VPP i The degree of matching of adjustable resources; P match2 (xi VPP i ) indicates that at the current moment in policy x i Virtual Power Plant (VPP) under certain conditions i Resource availability (e.g., SOC, capacity, status, etc.); P score (VPP i (This refers to the current virtual power plant, VPP) i Adaptability rating in edge scenarios (such as power grid reliability, efficiency, etc.);

[0104] Among them, β, γ, and δ are normalized weighting factors, which sum to 1 and can be adjusted based on historical execution feedback.

[0105] In practical applications, this preferred embodiment...

[0106] Assume the virtual power plant node (e.g., an energy storage system) currently has a battery SOC of 40% and a discharge capacity of 20kWh. Scheduling strategy x requires discharging 15kWh within 30 minutes. Based on the scoring, P can be obtained. match1 (x,VPP i ) = 1; P match2 (x,VPP i ) = 0.8 (considering the safety threshold, not completely venting); P score (x i The fitness score is 0.9 (local power grid is reliable and efficient), with weights set at β = 0.4, γ = 0.3, and δ = 0.3. The final fitness score is P. score (x i The score is calculated as follows: (0.4 * 1 + 0.3 * 0.8 + 0.3 * 0.9) = 0.91. A score ≥ 0.85 indicates a high strategy fit.

[0107] The system includes this strategy in the preferred execution list and has the ability to switch in case of strategy conflict or anomaly detection.

[0108] Example 2: This invention also proposes a virtual power plant cloud-edge collaborative scheduling device based on multi-objective optimization and confidence screening, which applies the above-mentioned virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening, including the following:

[0109] The data acquisition module is used to acquire relevant data from the virtual power plant trading terminal;

[0110] The strategy generation module is used to construct a multi-objective function based on the relevant data, select the comprehensive optimal solution under the multi-objective function, and generate an execution strategy set covering the global scheduling of the virtual power plant.

[0111] The operation scheduling module is used to perform cloud-edge collaborative scheduling of the virtual power plant based on the execution strategy set.

[0112] Reference Figure 2 Example 3: This invention also proposes a virtual power plant cloud-edge collaborative scheduling system based on multi-objective optimization and confidence screening, including:

[0113] The cloud server is equipped with a virtual power plant cloud-edge collaborative scheduling device based on multi-objective optimization and confidence screening as described in claim 7, which is used to construct a multi-objective optimization function and generate a Pareto scheduling strategy set based on the capability profiles of all VPP units.

[0114] The virtual power plant trading unit includes a data acquisition unit, a power prediction and status assessment module, and an edge decision-making unit. It is responsible for collecting terminal operation data (voltage, current, load power, status parameters, etc.), predicting local terminal power change trends (load and the virtual power plant unit itself), modeling and classifying terminal behavior characteristics, and finally performing adaptation analysis on the received cloud strategies to complete the final judgment and result feedback of the scheduling task.

[0115] The scheduling result feedback and correction module includes data reporting, model correction (updating predictor factors, confidence weights, and adaptation parameters), and cloud strategy distribution.

[0116] In a preferred embodiment of the present invention, the cloud server is further provided with a confidence module, which is used to score the confidence of the generated scheduling strategy and filter the execution strategy set.

[0117] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of the solution in this embodiment, depending on actual needs.

[0118] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0119] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0120] Although the description of the invention has been quite detailed and particularly of several described embodiments, it is not intended to limit it to any of these details or embodiments or any particular embodiment, but should be considered as providing a broad possible interpretation of the claims by referring to the appended claims and taking into account the prior art, thereby effectively covering the intended scope of the invention. Furthermore, the invention has been described above with respect to embodiments foreseeable by the inventors in order to provide a useful description, and non-substantial modifications to the invention that have not yet been foreseen may still represent equivalent modifications.

[0121] The above description is merely a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. Any embodiment that achieves the technical effects of the present invention using the same means should fall within the protection scope of the present invention. Within the protection scope of the present invention, various modifications and variations can be made to the technical solutions and / or implementation methods.

Claims

1. A virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening, characterized in that, Including the following: Obtain relevant data from the virtual power plant trading terminal; Based on the relevant data, a multi-objective function is constructed. Under the condition of satisfying the multi-objective function, the comprehensive optimal solution is selected to generate an execution strategy set covering the global scheduling of the virtual power plant. The virtual power plant is scheduled in a cloud-edge collaborative manner based on the aforementioned set of execution strategies.

2. The virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening according to claim 1, characterized in that, Specifically, the relevant data includes, V1 Behavioral Habit Model, V2 Adjustable Capacity, V3 Response Time, V4 Response Duration, V5 Adjustment Accuracy, V6 System Ramp-up Capability, V7 Power Prediction, V8 Plan Execution Strategy; Among them, the V1 behavior habit model is obtained based on data collection and feature extraction fitting, and is represented by a predictability score, which includes the standard deviation of the load curve and the scheduling response delay rate; the remaining V2-V6 content is obtained during system access testing and will be adjusted according to the actual situation during actual operation; the power prediction of V7 includes the load change of the region and the changes in energy storage power consumption and cycle; the V8 plan execution strategy is obtained by reading relevant data fed back to the cloud server.

3. The virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening according to claim 2, characterized in that, Specifically, the process of generating a set of execution policies for global scheduling covering the virtual power plant includes, Key objective function fusion, The multi-objective function is constructed by f1(x) representing the system's lumped revenue, f2(x) representing the impact of grid fluctuations, and f3(x) representing the response performance. The objective function is obtained by fusing these multi-objective functions. Where the weight w k (t) is dynamically set according to the time period, and the weight sum is 1; The constraints are set as follows. For each virtual power plant unit, i.e., VPP unit i, the following regulation capability constraints are satisfied: Pimin≤P i (t)≤Pimax; Where Pimax and Pimin represent the maximum and minimum adjustable power reported from VPP unit i, respectively, and Pi(t) represents the current power target allocated to VPP unit i by the scheduler. At the same time, the overall scheduling scheme must meet power balance constraints: Where P target (t) represents the target total dispatch power value at time t, ΔP is the allowable net deviation threshold, i.e., the error range that should not lead to penalties from the power supply company; n represents the total number of VPP units i. Set the iteration termination condition, which includes the convergence condition and the maximum number of iterations, and output the Pareto solution set.

4. The virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening according to claim 3, characterized in that, The method also includes, A correction factor ai(t) is added to the prediction result of the i-th VPP unit, and ai(t) is automatically updated and corrected based on the historical error of the model.

5. The virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening according to claim 1, characterized in that, The method also includes, A confidence model is introduced to filter and refine the set of execution strategies, specifically including... Construct the confidence function: Where S i For VPP i The historical response success rate weight, ranging from [0,1], P match (x i VPP i ) is strategy x i With VPP i The degree of matching of regulatory capacity, P error For the historical execution error of this virtual power plant unit, SOH i For VPP i The health status of the equipment, u i For VPP i Operating environment influencing factors The strategy of C(x) > b is written into the virtual power plant unit scheduling strategy set, where the threshold b ∈ [0,1]. Dynamic updates: Each time, the weights of each type are updated based on the execution results and historical data, such as: Where P success (x i ) is strategy x i The probability of success, where α is a preset weighting coefficient. This represents the response success rate weight before the update.

6. The virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening according to claim 5, characterized in that, The method also includes, When the virtual power plant unit at the edge receives the execution policy set, it autonomously adapts and executes the policy, then returns the result to the cloud server for further correction. Specifically, The fitness function of the virtual power plant unit is: K i =β*P match1 (x i ,VPP i )+γ*P match2 (x i ,VPP i )+δ*P score (VPP i ) Among them, P match1 (x i VPP i ) represents strategy x i With VPP i The degree of matching of adjustable resources; P match2 (x i VPP i ) indicates that at the current moment in policy x i Virtual Power Plant (VPP) under certain conditions i Resource availability; P score (VPP i (This refers to the current virtual power plant, VPP) i Adaptability score in edge scenarios; Among them, β, γ, and δ are normalized weighting factors, which sum to 1 and can be adjusted based on historical execution feedback.

7. A virtual power plant cloud-edge collaborative scheduling device based on multi-objective optimization and confidence screening, characterized in that, The virtual power plant cloud-edge collaborative scheduling method based on multi-objective optimization and confidence screening, which applies any one of claims 1-6, includes the following: The data acquisition module is used to acquire relevant data from the virtual power plant trading terminal; The strategy generation module is used to construct a multi-objective function based on the relevant data, select the comprehensive optimal solution under the multi-objective function, and generate an execution strategy set covering the global scheduling of the virtual power plant. The operation scheduling module is used to perform cloud-edge collaborative scheduling of the virtual power plant based on the execution strategy set.

8. A virtual power plant cloud-edge collaborative scheduling system based on multi-objective optimization and confidence screening, characterized in that, include, The cloud server is equipped with a virtual power plant cloud-edge collaborative scheduling device based on multi-objective optimization and confidence screening as described in claim 7, which is used to construct a multi-objective optimization function and generate a Pareto scheduling strategy set based on the capability profiles of all VPP units. The virtual power plant trading unit is used to collect terminal operation data, predict local terminal power change trends, model and classify terminal behavior characteristics, and finally perform adaptation analysis on the strategies sent by the cloud server to complete the final judgment and result feedback of the scheduling task.

9. A virtual power plant cloud-edge collaborative scheduling system based on multi-objective optimization and confidence screening as described in claim 8, characterized in that, The cloud server is also equipped with a confidence module, which is used to score the confidence of the generated scheduling strategy and filter the set of execution strategies.

Images