Distributed collaborative optimization and dispatching system based on energy demand side management

Abstract

This invention belongs to the field of power system dispatching technology, and particularly to a distributed collaborative optimization dispatching system based on energy demand-side management, comprising an edge sensing module, a distributed collaboration module, and a global dispatching module. The edge sensing module acquires demand-side resource data and generates a global database of adjustable demand-side resource potential; the distributed collaboration module generates local preliminary dispatching strategies and cross-regional resource complementary allocation strategies based on prediction results, and performs differential privacy-enhancing processing on regional operation data; the global dispatching module generates a unified global optimization dispatching strategy.

Description

Technical Field

[0001] This invention relates to the field of power system dispatching technology, specifically to a distributed collaborative optimization dispatching system based on energy demand-side management. Background Technology

[0002] The new power system integrates demand-side resources, primarily distributed photovoltaic and wind power, on a large scale. These demand-side resources are characterized by decentralization and fragmentation, making unified management and control difficult.

[0003] Existing technologies generally adopt a centralized energy dispatch model to uniformly manage demand-side resources.

[0004] However, centralized scheduling has drawbacks such as high computing power pressure, high response latency, and poor fault tolerance. In addition, it lacks data privacy protection capabilities during data sharing, has high barriers to collaborative scheduling, and lacks the adaptive capability of scheduling strategies. It is difficult to adapt to complex dynamic working conditions, resulting in technical problems such as difficulty in unified management and control of massive fragmented demand-side resources, difficulty in data sharing, and low scheduling accuracy under complex dynamic working conditions.

[0005] It should be noted that the above content falls within the inventor's technical knowledge and does not necessarily constitute prior art. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a distributed collaborative optimization scheduling system based on energy demand-side management, which solves the current problems.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a distributed collaborative optimization scheduling system based on energy demand-side management, comprising steps A, B, C and D, wherein step A includes: acquiring demand-side resource data, classifying and clustering the demand-side resource data, and generating a global demand-side resource adjustable potential database; The step B includes: inputting the global demand-side resource adjustable potential database into the long short-term memory network model and outputting load and new energy output prediction results; generating a local preliminary scheduling strategy based on the load and new energy output prediction results, and generating a cross-regional resource complementary allocation strategy through horizontal data interaction between sub-nodes. The step C includes: inputting the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy into the improved particle swarm optimization model, and outputting a unified global optimization scheduling strategy with the goal of grid peak-valley balance and maximum absorption of new energy. Step D includes: distributing the global unified optimization scheduling strategy to the edge terminal for execution and obtaining execution status feedback data.

[0008] In some embodiments, the demand-side resource data includes rigid load data, flexible adjustable load data, energy storage charging and discharging power data, distributed new energy output data, equipment operation constraint data, and electricity price signal data; the classification and clustering are divided according to equipment type and adjustment characteristics, wherein when a record corresponds to a rigid load, the record is identified as a rigid load category; when a record corresponds to a flexible adjustable load, the record is identified as a flexible adjustable load category.

[0009] In some embodiments, the global demand-side resource adjustability potential database includes the adjustable capacity, adjustment response speed, and operating constraints of each device.

[0010] In some embodiments, the input layer of the Long Short-Term Memory (LSTM) network model receives historical electricity consumption data and meteorological data from the global demand-side resource adjustable potential database, and the output layer is a fully connected layer that outputs short-term and medium-to-long-term load forecasts, photovoltaic power output forecasts, and wind power output forecasts. During training, the mean squared error is used as the loss function, and the Adam optimization algorithm is used to update the parameters of the LTM network model. Before inputting the global demand-side resource adjustable potential database into the LTM network model, the historical electricity consumption data and the meteorological data are respectively subjected to Z-score standardization.

[0011] In some embodiments, when obtaining the load and new energy output prediction results, user electricity consumption data is also used as an additional input to dynamically update the hidden layer node weights of the long short-term memory network model.

[0012] In some embodiments, the local preliminary scheduling strategy includes flexible load peak shaving and valley filling instructions, energy storage charging and discharging optimization instructions, and new energy local consumption instructions; when generating the cross-regional resource complementary allocation strategy, the generation task of the cross-regional resource complementary allocation strategy is decomposed into multiple regional sub-tasks, which are independently and in parallel by each sub-node.

[0013] In some embodiments, the improved particle swarm optimization model uses the power regulation of each device as the decision variable. Under the premise of satisfying grid voltage constraints, frequency constraints, maximum device regulation power constraints, and user electricity consumption constraints, the optimization objectives are to minimize the grid peak-valley difference and maximize the absorption of new energy. After iterative solution, the power regulation command value corresponding to each device is output. A distributed iterative algorithm is used to update the parameters of the improved particle swarm optimization model in real time, and a minute-level dynamic update strategy is generated based on load fluctuations and changes in new energy output.

[0014] In some embodiments, step D further includes: transmitting the execution status feedback data to the distributed coordination layer and the global scheduling layer to trigger the next round of perception, prediction, decision execution, and feedback process; the execution status feedback data includes the current operating power of the device and the device on / off status data.

[0015] In some embodiments, step E is further included: acquiring regional operation data of each sub-node, processing the regional operation data using a differential privacy desensitization algorithm to obtain desensitized aggregated data, and uploading the desensitized aggregated data to the global scheduling layer; wherein, the differential privacy desensitization algorithm superimposes noise following a Laplace distribution onto the regional operation data, and the scale parameter of the Laplace distribution is the ratio of the global sensitivity of the regional operation data to the privacy budget; and the variance parameter of the Laplace distribution is dynamically adjusted according to the data sensitivity level of each sub-node.

[0016] This invention discloses a distributed collaborative optimization scheduling system for executing the aforementioned distributed collaborative optimization scheduling method.

[0017] The edge awareness module is used to acquire demand-side resource data, classify and cluster the demand-side resource data, and generate a global demand-side resource adjustable potential database; it is also used to parse the global unified optimization scheduling strategy, generate equipment control commands and execute them, and acquire execution status feedback data. The distributed coordination module is used to obtain load and renewable energy output prediction results through a long short-term memory network model based on the global demand-side resource adjustable potential database; it is used to generate local preliminary scheduling strategies and cross-regional resource complementary allocation strategies based on the load and renewable energy output prediction results; and it is used to perform differential privacy desensitization processing on regional operation data, obtain desensitized aggregated data, and upload it to the global scheduling layer. The global scheduling module is used to input the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy into the improved particle swarm optimization model, output a global unified optimization scheduling strategy, and distribute the global unified optimization scheduling strategy to the edge perception module. Compared with existing technologies, this invention provides a distributed collaborative optimization scheduling system based on energy demand-side management, which has the following advantages: This distributed collaborative optimization scheduling system based on energy demand-side management adopts a distributed parallel computing architecture. It decomposes the task of generating cross-regional resource complementary allocation strategies into multiple regional sub-tasks, which are independently and concurrently computed by each sub-node. This solves the technical problems of the difficulty in unified management of massive fragmented demand-side resources and poor system fault tolerance, achieving the technical effects of reducing the computing power pressure on single nodes and improving scheduling response speed. The invention uses a differential privacy desensitization algorithm to process regional operational data, uploading only the desensitized aggregated data, solving the technical problem of difficult data sharing and achieving the technical effect of protecting user data privacy while ensuring multi-entity collaborative scheduling. The invention uses a distributed iterative algorithm to update the parameters of the improved particle swarm optimization model in real time, solving the technical problem of low scheduling accuracy under complex dynamic conditions, and achieving the technical effect of generating minute-level dynamically updated strategies to adapt to load fluctuations and changes in new energy output. Attached Figure Description

[0018] Figure 1 A flowchart of the distributed cooperative optimization scheduling method provided in the embodiments of the present invention; Figure 2 This is a sub-flowchart for obtaining demand-side resource data and generating a global demand-side resource adjustability potential database in the distributed collaborative optimization scheduling method provided in the embodiments of the present invention. Figure 3 This is a sub-flowchart of the distributed collaborative optimization scheduling method provided in this embodiment of the invention, which obtains load and new energy output prediction results based on a global demand-side resource adjustable potential database. Figure 4 The following is a sub-flowchart for generating a local preliminary scheduling strategy and a cross-regional resource complementary allocation strategy based on load and new energy output prediction results in the distributed collaborative optimization scheduling method provided in this embodiment of the invention. Figure 5 This is a sub-flowchart of the distributed collaborative optimization scheduling method provided in this embodiment of the invention, which generates a unified optimization scheduling strategy for the entire domain based on a local preliminary scheduling strategy and a cross-regional resource complementary allocation strategy. Figure 6 This is a sub-flowchart of the distributed collaborative optimization scheduling method provided in this embodiment of the invention, which distributes the unified optimization scheduling strategy across the entire domain to the edge terminal and obtains execution status feedback data. Figure 7 This is a sub-flowchart of the distributed collaborative optimization scheduling method provided in this embodiment of the invention, which performs differential privacy desensitization processing on regional operation data to obtain desensitized aggregated data. Detailed Implementation

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

[0020] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0021] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0023] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0024] Please see Figure 1-7 Example 1: A new power system integrates large-scale demand-side resources, primarily distributed photovoltaic and wind power. Traditional centralized energy dispatching models are ill-suited to the operational characteristics of this new energy network. Demand-side resources are dispersed and fragmented, making unified management difficult. Centralized dispatching suffers from high computational pressure, long response times, and poor fault tolerance. Source-load mismatch and large peak-valley differences lead to high grid operating costs. Data privacy is poor, and collaborative dispatching faces high barriers. Dispatching strategies are simplistic and lack adaptability.

[0025] The hardware environment of this implementation comprises a three-layer structure: an edge awareness layer, a distributed collaboration layer, and a global scheduling layer. The edge awareness layer is deployed on user terminals and other devices. The distributed collaboration layer includes multiple sub-nodes. The global scheduling layer includes a top-level server.

[0026] A distributed cooperative optimization scheduling method according to an embodiment of the present invention includes the following steps: Step 1: Obtain demand-side resource data and generate a global database of adjustable demand-side resource potential.

[0027] Step 101: Obtain demand-side resource data.

[0028] Step 102: Classify and cluster the demand-side resource data to generate a global database of adjustable demand-side resource potential.

[0029] It should be noted that the aforementioned demand-side resource data includes rigid load data, flexible adjustable load data, energy storage charging and discharging power data, distributed renewable energy output data, equipment operation constraint data, and electricity price signal data. The above classification and clustering are based on equipment type and regulation characteristics. When a record corresponds to a rigid load, it is identified as a rigid load; when a record corresponds to a flexible adjustable load, it is identified as a flexible adjustable load.

[0030] In this embodiment of the application, in order to accurately assess the adjustment capability of terminal devices, when generating the global demand-side resource adjustment potential database, the adjustable capacity, adjustment response speed and operating constraints of each device are also calculated, thereby forming a global demand-side resource adjustment potential database containing the above parameters.

[0031] Step 2: Based on the global demand-side resource adjustability potential database, obtain the load and new energy output forecast results.

[0032] Step 201: Obtain the database of adjustable potential resources on the demand side across the entire domain.

[0033] Step 202: Input the database of adjustable potential resources on the demand side of the entire domain into the long short-term memory network model, and output the load and new energy output prediction results.

[0034] It should be noted that the input layer of the aforementioned Long Short-Term Memory (LSTM) network model receives historical electricity consumption data and meteorological data from the global demand-side resource adjustability potential database. The output layer is a fully connected layer, outputting short-term and medium-to-long-term load forecasts, photovoltaic power output forecasts, and wind power output forecasts. During training, the mean squared error is used as the loss function, and the Adam optimization algorithm is employed to update the parameters of the LTM network model.

[0035] Furthermore, since the dimensions of historical electricity consumption data and meteorological data are inconsistent, Z-score standardization is performed on historical electricity consumption data and meteorological data respectively before inputting the database of adjustable potential of demand-side resources across the entire domain into the long short-term memory network model to eliminate the impact of the difference in dimensions on the training of the long short-term memory network model.

[0036] In this embodiment of the application, in order to improve the accuracy of prediction, when obtaining the load and renewable energy output prediction results, user electricity consumption data is also used as an additional input to dynamically update the hidden layer node weights of the long short-term memory network model, thereby obtaining load and renewable energy output prediction results that are adapted to meteorological changes and sudden load changes.

[0037] Step 3: Based on the load and new energy output forecast results, generate a local preliminary scheduling strategy and a cross-regional resource complementary allocation strategy.

[0038] Step 301: Based on the load and new energy output forecast results, and with the goal of regional load balance, generate a preliminary local scheduling strategy.

[0039] Step 302: Each child node interacts horizontally with other nodes through a distributed message queue to obtain resource data from neighboring child nodes and generate a cross-regional resource complementarity allocation strategy.

[0040] It should be noted that the aforementioned preliminary local dispatch strategy includes instructions for flexible load peak shaving and valley filling, instructions for optimizing energy storage charging and discharging, and instructions for local consumption of new energy.

[0041] In this embodiment of the application, in order to reduce the computing power pressure of a single node and improve the generation speed, a distributed parallel computing architecture is introduced when generating cross-regional resource complementary allocation strategies. The generation task of cross-regional resource complementary allocation strategies is decomposed into multiple regional sub-tasks, which are independently and parallelly computed by each sub-node, thereby breaking the regional resource scheduling barrier.

[0042] Step 4: Based on the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy, generate a unified optimized scheduling strategy for the entire domain.

[0043] Step 401: Obtain the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy.

[0044] Step 402: Input the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy into the improved particle swarm optimization model, and output a unified optimization scheduling strategy for the whole domain with the goal of grid peak-valley balance and maximum absorption of new energy.

[0045] It should be noted that the inputs to the improved particle swarm optimization model are the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy, and the output is a unified global optimization scheduling strategy that satisfies multiple constraints. These multiple constraints include grid voltage constraints, frequency constraints, equipment maximum regulating power constraints, and user electricity consumption constraints.

[0046] Furthermore, the unified optimization scheduling strategy across the entire domain takes the power adjustment amount of each device as the decision variable. Under the premise of satisfying the above multiple constraints, the optimization objectives are to minimize the peak-valley difference of the power grid and maximize the absorption of new energy. The improved particle swarm optimization model iteratively solves the problem and outputs the corresponding power adjustment command value of each device, which serves as the basis for the execution of edge terminals in subsequent steps.

[0047] In this embodiment of the application, in order to adapt to complex dynamic working conditions, when generating a unified optimization scheduling strategy across the entire domain, a distributed iterative algorithm is used to update the parameters of the improved particle swarm optimization model in real time, and a minute-level dynamic update strategy is generated based on load fluctuations and changes in new energy output, thereby solving the problem of poor adaptability of the fixed scheduling model.

[0048] For data that has passed step 4, step 5 can be performed. Step 5 may specifically include: Step 5: Distribute the unified optimization scheduling strategy across the entire domain to the edge terminals and obtain execution status feedback data.

[0049] Step 501: Distribute the unified optimization scheduling strategy across the entire domain to the edge perception layer in layers.

[0050] Step 502: The edge perception layer parses the global unified optimization scheduling strategy, generates device control commands and executes them, and obtains execution status feedback data.

[0051] It should be noted that the aforementioned equipment control commands include flexible load peak-shaving start / stop commands, power adjustment commands, and energy storage charging / discharging period optimization commands. The aforementioned execution status feedback data includes the current operating power of the equipment and the equipment switching status data.

[0052] In this embodiment of the application, in order to form a closed-loop scheduling, after obtaining the execution status feedback data, the execution status feedback data is also transmitted to the distributed coordination layer and the global scheduling layer, thereby triggering the next round of perception, prediction, decision-making, and execution feedback process, and realizing the dynamic adaptive correction of the scheduling strategy.

[0053] Step 6: Perform differential privacy desensitization on the regional operation data to obtain the desensitized aggregated data.

[0054] Step 601: Obtain the regional operation data of each child node.

[0055] Step 602: The differential privacy de-identification algorithm is used to process the regional operation data to obtain the de-identified aggregated data, and the de-identified aggregated data is uploaded to the global scheduling layer.

[0056] It should be noted that the aforementioned regional operational data includes total load data and total renewable energy output data within the region. The processing formula for the aforementioned differential privacy desensitization algorithm is as follows: ,in For regional operational data, For data aggregation after anonymization, To ensure global sensitivity of regional operational data, For privacy budget, This represents the Laplace distribution.

[0057] In this embodiment of the application, in order to balance data availability and privacy security, when acquiring the de-identified aggregated data, the variance parameter of the Laplace distribution is dynamically adjusted according to the data sensitivity level of each sub-node, thereby ensuring user data privacy protection while guaranteeing multi-subject collaborative scheduling.

[0058] On the other hand, the present invention also proposes a distributed cooperative optimization scheduling system, comprising: The edge perception module is used to acquire demand-side resource data and generate a global database of adjustable potential demand-side resources. It also distributes the global unified optimization scheduling strategy to the edge perception layer, parses the global unified optimization scheduling strategy, generates and executes equipment control commands, and obtains execution status feedback data.

[0059] The distributed coordination module is used to obtain load and renewable energy output forecast results based on the global demand-side resource adjustability potential database; generate local preliminary scheduling strategies and cross-regional resource complementary allocation strategies based on the load and renewable energy output forecast results; perform differential privacy desensitization processing on regional operation data, obtain desensitized aggregated data, and upload the desensitized aggregated data to the global scheduling layer.

[0060] The global scheduling module is used to generate a unified optimized scheduling strategy for the entire domain based on the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy.

[0061] By way of example, the system proposed in this invention may also include various features and combinations thereof in the method embodiments, which will not be elaborated here.

[0062] In the application, several formulas are calculated by removing dimensions and taking their numerical values. The formulas are established by collecting a large amount of data and simulating the most recent real situation. Some coefficients or weights in the formulas are set by those skilled in the art according to the actual situation, so they will not be elaborated here.

[0063] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0064] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A distributed collaborative optimization scheduling system based on energy demand-side management, characterized in that: It includes steps A, B, C and D, wherein step A includes: acquiring demand-side resource data, classifying and clustering the demand-side resource data, and generating a global demand-side resource adjustable potential database; The step B includes: inputting the global demand-side resource adjustable potential database into the long short-term memory network model and outputting load and new energy output prediction results; generating a local preliminary scheduling strategy based on the load and new energy output prediction results, and generating a cross-regional resource complementary allocation strategy through horizontal data interaction between sub-nodes. The step C includes: inputting the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy into the improved particle swarm optimization model, and outputting a unified global optimization scheduling strategy with the goal of grid peak-valley balance and maximum absorption of new energy. Step D includes: distributing the global unified optimization scheduling strategy to the edge terminal for execution and obtaining execution status feedback data.

2. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: The demand-side resource data includes rigid load data, flexible adjustable load data, energy storage charging and discharging power data, distributed new energy output data, equipment operation constraint data, and electricity price signal data. The classification and clustering are divided according to equipment type and adjustment characteristics. When a record corresponds to a rigid load, the record is identified as a rigid load category; when a record corresponds to a flexible adjustable load, the record is identified as a flexible adjustable load category.

3. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: The global demand-side resource adjustability potential database includes the adjustable capacity, adjustment response speed, and operating constraints of each device.

4. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: The input layer of the Long Short-Term Memory (LSTM) network model receives historical electricity consumption data and meteorological data from the global demand-side resource adjustable potential database. The output layer is a fully connected layer that outputs short-term and medium-to-long-term load forecasts, photovoltaic power output forecasts, and wind power output forecasts. During training, the mean squared error is used as the loss function, and the Adam optimization algorithm is used to update the parameters of the LTM network model. Before inputting the global demand-side resource adjustable potential database into the LTM network model, the historical electricity consumption data and the meteorological data are respectively subjected to Z-score standardization.

5. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: When obtaining the load and new energy output prediction results, user electricity consumption data is also used as an additional input to dynamically update the hidden layer node weights of the long short-term memory network model.

6. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: The local preliminary scheduling strategy includes flexible load peak shaving and valley filling instructions, energy storage charging and discharging optimization instructions, and new energy local consumption instructions; when generating the cross-regional resource complementary allocation strategy, the generation task of the cross-regional resource complementary allocation strategy is decomposed into multiple regional sub-tasks, which are independently and in parallel by each sub-node.

7. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: The improved particle swarm optimization model uses the power regulation of each device as the decision variable. Under the premise of satisfying the grid voltage constraint, frequency constraint, maximum regulation power constraint of the device, and user electricity consumption constraint, the optimization objectives are to minimize the peak-valley difference of the grid and maximize the absorption of new energy. After iterative solution, the power regulation command value corresponding to each device is output. A distributed iterative algorithm is used to update the parameters of the improved particle swarm optimization model in real time, and a minute-level dynamic update strategy is generated according to load fluctuations and changes in new energy output.

8. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: Step D further includes: transmitting the execution status feedback data to the distributed collaboration layer and the global scheduling layer to trigger the next round of perception, prediction, decision-making, and execution feedback process; the execution status feedback data includes the current operating power of the device and the device's on / off status data.

9. The distributed collaborative optimization scheduling system based on energy demand-side management according to claim 1, characterized in that: The method also includes step E: acquiring regional operation data of each sub-node, processing the regional operation data using a differential privacy desensitization algorithm to obtain desensitized aggregated data, and uploading the desensitized aggregated data to the global scheduling layer; wherein, the differential privacy desensitization algorithm superimposes noise following a Laplace distribution onto the regional operation data, and the scale parameter of the Laplace distribution is the ratio of the global sensitivity of the regional operation data to the privacy budget; and the variance parameter of the Laplace distribution is dynamically adjusted according to the data sensitivity level of each sub-node.

10. A distributed cooperative optimization scheduling system, used to execute the distributed cooperative optimization scheduling method according to any one of claims 1 to 9, characterized in that, include: The edge awareness module is used to acquire demand-side resource data, classify and cluster the demand-side resource data, and generate a global demand-side resource adjustable potential database. It is used to parse the global unified optimization scheduling strategy, generate and execute device control commands, and obtain execution status feedback data; The distributed coordination module is used to obtain load and renewable energy output prediction results through a long short-term memory network model based on the global demand-side resource adjustable potential database; it is used to generate local preliminary scheduling strategies and cross-regional resource complementary allocation strategies based on the load and renewable energy output prediction results; and it is used to perform differential privacy desensitization processing on regional operation data, obtain desensitized aggregated data, and upload it to the global scheduling layer. The global scheduling module is used to input the local preliminary scheduling strategy and the cross-regional resource complementary allocation strategy into the improved particle swarm optimization model, output a global unified optimization scheduling strategy, and distribute the global unified optimization scheduling strategy to the edge perception module.

Images