Multi-source heterogeneous distributed new energy aggregation regulation and coordination method, device, equipment, medium and product

Abstract

The application discloses a multi-source heterogeneous distributed new energy aggregation regulation and coordination method and device, equipment, medium and product, and relates to the field of new energy aggregation regulation and coordination. The method comprises the following steps: acquiring multi-source data full-dimensional feature information and performing preset standardization preprocessing; performing dynamic aggregation modeling based on time attention and improved DBN of Gaussian regularization; solving a double-objective optimization function based on an improved MOPSO algorithm, constraint conditions and a Lagrange penalty term processing constraint, and screening an optimal solution considering power grid stability and income; determining a millimeter-level control instruction corresponding to an edge gateway based on an output strategy, performing Kalman filtering complex data synchronization, docking a power system to perform bid synchronization and execution decomposition, performing effect evaluation and feedback processing, and iteratively optimizing a decision model. The application can realize efficient aggregation and accurate regulation of distributed new energy, and realize aggregation regulation and marketization coordination of distributed new energy.

Description

Technical Field

[0001] This application relates to the field of new energy aggregation, regulation and coordination, and in particular to a method, device, equipment, medium and product for multi-source heterogeneous distributed new energy aggregation, regulation and coordination. Background Technology

[0002] Traditional aggregation models use static equivalent circuit modeling, which only considers the output of a single resource and does not integrate time series, environmental coupling, and grid state characteristics. The modeling accuracy is less than 85%, and it cannot adapt to the dynamic changes of multiple types of heterogeneous resources. The aggregation capacity assessment error is greater than 10%.

[0003] The current system only supports single grid control or single power trading, and has not established a synergistic optimization mechanism between grid security constraints and market-based benefits. It either sacrifices benefits to ensure grid stability or sacrifices stability to pursue benefits, making it impossible to achieve a win-win situation.

[0004] The real-time performance and reliability of the control do not meet the requirements of the power grid. The transmission delay of traditional cloud-based centralized control is >200ms, which cannot meet the millisecond-level response requirements such as primary frequency regulation and rapid backup. After the network is disconnected, the equipment loses control and the data is lost. There is no synchronization mechanism after reconnection, and the system availability is <99%.

[0005] Market-based operations lack a complete closed-loop process. The reporting of quantities and prices, bidding and execution, effect evaluation and strategy iteration rely on manual operation, resulting in low transaction efficiency, large revenue loss and inability to adapt to the real-time bidding and rapid clearing rules of the power market. Summary of the Invention

[0006] The purpose of this application is to provide a method, device, equipment, medium, and product for the coordinated regulation and control of multi-source heterogeneous distributed new energy, which can realize efficient aggregation and precise regulation and control of distributed new energy, and achieve coordinated regulation and control of distributed new energy in the market.

[0007] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a method for the coordinated regulation and control of multi-source heterogeneous distributed new energy sources, including: Obtain multi-dimensional feature information from multi-source data and perform pre-processing with preset standardization to obtain processed data; Based on the processed data, dynamic aggregation modeling is performed using time attention and Gaussian regularization-improved DBN to construct a decision model; the decision model includes a bi-objective optimization function, constraints, and Lagrange penalty term processing constraints. Based on the improved MOPSO algorithm, the bi-objective optimization function is solved according to the constraints and the Lagrange penalty term. The Pareto optimal solution screening method is used to screen the optimal solution that takes into account both grid stability and revenue, and the output strategy is determined. Based on the device control strategy included in the output strategy, the millimeter-level control command corresponding to the edge gateway is determined and Kalman filtering is performed to reconnect and synchronize the data. Also, based on the application strategy included in the output strategy, the power system is connected and the corresponding application data is automatically submitted to perform bid synchronization and execution decomposition, and the automated instruction execution data is determined. The automated instruction execution data includes: device-level execution plan, device execution status, and real-time operation data. Based on the automated instruction execution data, effect evaluation and feedback processing are performed to iteratively optimize the decision model, so that the execution deviation corresponding to the effect evaluation is within a preset deviation range.

[0008] Secondly, this application provides a multi-source heterogeneous distributed new energy aggregation, regulation, and coordination device, comprising: The data acquisition and processing module is used to acquire multi-dimensional feature information of multi-source data and perform preset standardization preprocessing to obtain processed data; The model building module is used to construct a decision model based on the processed data, dynamic aggregation modeling based on time attention and Gaussian regularized improved DBN; the decision model includes a bi-objective optimization function, constraints, and Lagrange penalty term processing constraints. The output strategy determination module is used to solve the bi-objective optimization function based on the improved MOPSO algorithm, according to the constraints and the Lagrange penalty term, and to select the optimal solution that balances grid stability and revenue by using the Pareto optimal solution screening method to determine the output strategy. An automated instruction execution module is used to determine millimeter-level control instructions corresponding to the edge gateway based on the device control strategy included in the output strategy and to perform Kalman filtering and multiplexing data synchronization. It is also used to connect to the power system based on the application strategy included in the output strategy and automatically submit the corresponding application data for bid synchronization and execution decomposition, thereby determining automated instruction execution data. The automated instruction execution data includes: device-level execution plan, device execution status, and real-time operating data. The closed-loop feedback module is used to evaluate the effect and process the feedback based on the automated instruction execution data, so as to iteratively optimize the decision model and keep the execution deviation corresponding to the effect evaluation within a preset deviation range.

[0009] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described multi-source heterogeneous distributed new energy aggregation, regulation, and coordination method.

[0010] Fourthly, this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the aforementioned multi-source heterogeneous distributed new energy aggregation, regulation, and coordination method.

[0011] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned multi-source heterogeneous distributed new energy aggregation, regulation, and coordination method.

[0012] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method, device, equipment, medium, and product for the coordinated regulation and control of multi-source heterogeneous distributed renewable energy. It acquires multi-dimensional feature information from multi-source data and performs pre-processing with pre-defined standardization. Dynamic aggregation modeling is performed based on time attention and an improved Gaussian regularized DBN to enhance the accuracy and stability of multi-source heterogeneous resource aggregation. An improved MOPSO algorithm is used to solve the dual-objective optimization function based on constraints and Lagrange penalty terms, and the optimal solution that balances grid stability and profitability is selected, achieving optimal synergy between grid security and market profitability. Millimeter-level control commands corresponding to edge gateways are determined based on output strategies, and Kalman filtering and data synchronization are performed. This data is then connected to the power system for bid synchronization and execution decomposition, achieving millisecond-level regulation response speeds to meet the primary frequency regulation requirements of the power grid. Furthermore, reliability is improved, enabling self-regulation during grid outages and error-free reconnection. Effect evaluation and feedback processing are performed based on automated command execution data to iteratively optimize the decision model, ensuring that the execution deviation corresponding to the effect evaluation is within a preset deviation range. Through participation in the power market, a fully automated closed-loop process is achieved, improving efficiency and compliance. Therefore, this application can realize the efficient aggregation and precise control of distributed new energy sources, and achieve the aggregation and control of distributed new energy sources and market-based collaboration. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart of a multi-source heterogeneous distributed renewable energy aggregation, regulation, and synergistic method; Figure 2 This is a structural diagram of a multi-source heterogeneous distributed renewable energy aggregation, regulation, and coordination device; Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0016] This application proposes a complete technical solution for multi-source heterogeneous aggregation, coordinated regulation and trading, cloud-edge collaborative execution, and market-oriented closed-loop operation, which realizes efficient aggregation, precise regulation, market-oriented profitability, and grid-friendly operation of distributed new energy, filling the technical gap in the coordination of distributed new energy aggregation regulation and market-oriented trading.

[0017] This application addresses the problems of the inability to accurately and dynamically aggregate multi-source heterogeneous distributed resources, the conflict between the dual objectives of regulation security and market benefits, insufficient real-time regulation, unreliability issues due to network outages, complex power market participation processes, and the inability to automate closed-loop operations. It also solves the engineering implementation problems of infeasible and non-convergent optimization strategies under multiple constraints.

[0018] The technical solution of this application can achieve the following: aggregate dispersed resources to form a large-scale regulation capability that is dispatchable, tradable, and regulated; simultaneously meet the requirements of power grid safety and stability and maximize user benefits; enable autonomous control during grid outages and synchronous operation during grid connections; achieve fully automated and closed-loop operation, reducing manual intervention by more than 90%; support the compliant participation of distributed renewable energy in ancillary services and spot trading, increasing asset returns by 10%–15%; and improve the renewable energy consumption rate and reduce the pressure on power grid regulation.

[0019] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] This application adopts a cloud-edge collaborative microservice architecture, which is divided into a perception and acquisition layer, an edge computing layer, a cloud decision-making layer, a transaction docking layer, and a visualization display layer. It supports multi-protocol access such as Modbus / OPCUA / IEC104 and Docker containerized deployment.

[0021] The proposed solution comprises a cloud server, edge gateway, and terminal devices working collaboratively within a new energy collaborative control and aggregation regulation system. The overall technical approach follows a complete roadmap of "multi-source data perception - feature preprocessing - dynamic aggregation modeling - dual-objective optimization decision-making - cloud-edge collaborative execution - market-based transaction closed loop - model iterative optimization." By constructing a high-precision AI model and optimization algorithms in the cloud and achieving millisecond-level real-time control at the edge, a collaborative operation mode of "globally optimal + locally reliable" is formed. All steps revolve around the aggregation capabilities, real-time regulation, transaction profitability, and system reliability of distributed new energy. Through standardized data processing, model-based intelligent decision-making, automated command execution, and closed-loop effect feedback, unified access, dynamic aggregation, precise regulation, and efficient market-based operation of multi-source heterogeneous distributed resources are achieved, with stable operation throughout the entire process requiring no manual intervention.

[0022] Perception layer (perception and acquisition layer): Components: Distributed photovoltaic inverter, energy storage PCS, wind power converter, controllable load controller, multi-protocol adapter module.

[0023] Functions: Data acquisition, command execution, Modbus / OPCUA / IEC104 protocol integration.

[0024] Deployment: Distributed deployment at the customer's site.

[0025] Edge layer (edge ​​computing layer): Components: Custom edge gateway, TensorFlowLite lightweight AI model, local controller.

[0026] Functions: Real-time command generation, network outage autonomy, data preprocessing, and Kalman synchronization.

[0027] Deployment: Deployed near the customer's site.

[0028] Cloud layer (cloud decision-making layer): Components: Spring Cloud microservice cluster, AI algorithm engine, InfluxDB time-series database, data platform.

[0029] Functions: Aggregate modeling, bi-objective optimization, market decision-making, and model iteration.

[0030] Deployment: Containerized deployment on the company's private cloud / public cloud.

[0031] Interface Layer (Transaction Interface Layer): Components: Power trading system API, power grid dispatching system IEC104 interface, message middleware RabbitMQ.

[0032] Functions: Data interoperability, business collaboration, and transaction reporting.

[0033] Presentation layer (visual presentation layer): Components: Vue3 + ElementPlus + ECharts Web management platform.

[0034] Features: Status monitoring, strategy management, revenue visualization, and report generation.

[0035] In one exemplary embodiment, such as Figure 1 As shown, a method for coordinated regulation and control of multi-source heterogeneous distributed new energy sources is provided, including: Step 100: Obtain full-dimensional feature information from multi-source data and perform preset standardization preprocessing to obtain processed data.

[0036] Specifically, this step corresponds to multi-dimensional feature acquisition and standardized preprocessing, which is executed by the perception acquisition layer and the preprocessing module in the edge gateway.

[0037] Multi-source data, full-dimensional feature information collection: Resource characteristics collected: Distributed photovoltaic power output Wind power output Energy storage capacity , ,Voltage Current Adjustment rate Response time .

[0038] Collect power grid characteristics: power grid frequency Bus voltage Regional load Dispatch instructions .

[0039] Market characteristics collected: spot electricity price Ancillary service prices Transaction rule parameters.

[0040] Environmental characteristics of the data collection: Irradiance Wind speed Ambient temperature .

[0041] Data collection frequency: resource / environment data once per second, power grid data once per millisecond, and market data once per minute.

[0042] Abnormal data cleaning: Outliers are removed using the 3σ criterion: |x-μ|>3σ is considered an anomaly and is filled in using linear interpolation; where x is the data variable and μ is the expected value.

[0043] Remove duplicate data and invalid data segments with more than 10% missing values.

[0044] Z-Score standardization is used to obtain standardized feature data. : .

[0045] These are the original eigenvalues; For the first Mean of dimensional features; For the first Dimensional characteristic standard deviation.

[0046] Function: Eliminate dimensional differences and map features to the interval [-1,1].

[0047] Dimensionality reduction using Kernel Principal Component Analysis (KPCA): Selecting the radial basis kernel function: , .

[0048] in, This represents the radial basis function (RBF) value. For the first One original input sample; For the first One original input sample; For kernel parameters (bandwidth); It is the square of the Euclidean distance between the two sample vectors.

[0049] Extract core features with a cumulative contribution rate ≥95%, reduce the 18-dimensional features to 8-dimensional features, and output a high-dimensional aggregated feature matrix. : .in, This is a resource feature submatrix; For the characteristic submatrix of the power grid; This is a submatrix of environmental features.

[0050] Output: Preprocessed standardized core feature set .

[0051] Technical function: Eliminates data noise, dimensional differences, and feature redundancy, providing high-quality input for aggregation modeling.

[0052] Step 200: Based on the processed data, perform dynamic aggregation modeling using time attention and Gaussian regularization-improved DBN to construct a decision model. The decision model includes a bi-objective optimization function, constraints, and Lagrange penalty term handling constraints.

[0053] This step is a dynamic aggregation modeling based on temporal attention and Gaussian regularization to improve DBN, and its execution is carried out by a cloud AI algorithm engine in a cloud server.

[0054] The expression corresponding to the weights of temporal attention is: .

[0055] in, Weights for time-based attention; This is the attention decay coefficient; This represents the total duration of the time window. This is the current time step; This refers to a time window time step. In one embodiment, ; .

[0056] Function: Enhances the impact of data within the last 10 minutes on the model, improving the accuracy of dynamic modeling.

[0057] The network structure of the Gaussian-regularized DBN includes an input layer, three hidden layers, and an output layer; the activation function for the hidden layers is ReLU with Gaussian noise regularization, as shown in the formula: .

[0058] Output layer: Sigmoid activation, output aggregate equivalent parameters [ , , ].

[0059] Model training (weighted MSE + L2 regularization), the network loss function corresponding to Gaussian regularization improved DBN is: .

[0060] in, The activation function corresponding to the hidden layer; For activation functions; For the first The weight matrix of the hidden layer; To process the standardized core feature set in the data; The Hadamard product, i.e., element-wise multiplication, represents the weighting of temporal attention. With feature vectors, which are the standardized core feature sets in the data being processed. Perform element-wise weighted summation; For the first The bias vector of the hidden layer; This is a Gaussian noise regularization term; The network loss function; This represents the number of training samples in a batch. For sample index; For the first The aggregated true value of active power for each sample; For the first The aggregated active power model prediction for each sample; For the first Aggregate reserve capacity for each real value and each sample; For the first The aggregated reserve capacity model prediction for each sample; For the first The true value of the aggregation response time or adjustment rate for each sample; For the first The aggregate response time model prediction for each sample; It is the set of all trainable weight parameters of the network; The square of the Frobenius norm of the weight matrix, i.e., the sum of squares of all weights, is used as the L2 regularization term, with a coefficient of 0.001 to control the regularization strength.

[0061] The values ​​0.5, 0.3, and 0.2 represent the output / capacity / response time weights; the value 0.001 represents the L2 regularization coefficient.

[0062] Optimizer: Adam, learning rate η=0.001, batch size 32, 100 iterations.

[0063] Dynamic model update: Data is re-collected every 5 minutes, and model fine-tuning is performed once to adapt to real-time changes in resource status.

[0064] Aggregate parameter output: Output: Aggregated equivalent output Aggregate regulation capacity Aggregate response time .

[0065] Output: Accurate aggregation of equivalent parameter sets.

[0066] Technical function: To achieve unified modeling of heterogeneous resources, with an aggregation accuracy of ≥96% and a prediction bias of ≤3%.

[0067] The bi-objective optimization solution based on improved MOPSO + Lagrange penalty is implemented by the cloud decision engine in the cloud decision layer. The expression for the bi-objective optimization function is: Objective 1: Minimize power grid regulation deviation: .

[0068] Objective 2: Maximize market-based benefits: .

[0069] in, The target optimization function for power grid regulation deviation; This represents the total duration of the time window. This is the current time step; To contribute to the equivalent of aggregation; Power for scheduling commands; This is a vector representing the change in power regulation. The square of the L2 norm; Optimize the function to achieve market-based profit targets; The spot market electricity price; The actual amount of electricity traded online (power); To compensate for prices in the ancillary services market; The actual capacity of ancillary services provided; To adjust the deviation cost coefficient; Based on the planned output value; This is a vector representing the change in ancillary service capacity.

[0070] The constraints include: resource constraints, energy storage constraints, transaction constraints, and response constraints.

[0071] The expression corresponding to the resource constraint is: .

[0072] .

[0073] The expression corresponding to the energy storage constraint is: .

[0074] .

[0075] The expression corresponding to the transaction constraint is: .

[0076] The expression corresponding to the response constraint is: .

[0077] in, To contribute to the equivalent of aggregation; This represents the upper limit of the aggregate equivalent output. The actual capacity of ancillary services provided; To regulate the polymerization capacity; For energy storage capacity; Energy storage capacity; The maximum charge and discharge power limit for energy storage systems; The actual amount of electricity traded online (power); Reserved power for future adjustments (standby capacity); It is a unit of time, namely "second"; It contributes to the aggregation equivalent of the previous time step.

[0078] The expression corresponding to the Lagrange penalty term handling constraint is: ; in, The total penalty value (scalar); Index for constraints (c=1, 2, 3, 4, 5); This is the penalty coefficient; For the first A function for the amount of constraint violation; A vector of decision variables; This is an indicator function.

[0079] In one embodiment, The indicator function is 1 if the constraint is violated, and 0 otherwise.

[0080] Step 300: Based on the improved MOPSO algorithm, solve the bi-objective optimization function according to the constraints and the Lagrange penalty term, and use the Pareto optimal solution screening method to screen the optimal solution that takes into account both grid stability and revenue, and determine the output strategy.

[0081] In one embodiment, based on the improved MOPSO algorithm, the bi-objective optimization function is solved according to the constraints and the Lagrange penalty term. The Pareto optimal solution screening method is then used to select the optimal solution that balances grid stability and profitability, determining the output strategy, including: Based on the inertia weights corresponding to the improved MOPSO algorithm, as well as the preset number of iterations and learning factor, the biobjective optimization function is solved according to the constraints and the Lagrange penalty term, yielding the solution result; where the expression corresponding to the inertia weights is: .

[0082] =0.9, =0.4.

[0083] For the first The inertial weight of the particle in the next iteration; The maximum value of the inertial weight is 0.9. The minimum inertial weight is 0.4. This represents the maximum total number of iterations for the algorithm. This represents the current iteration step.

[0084] Improved MOPSO algorithm parameter settings: number of particles: 100; number of iterations: 200; learning factors: c1=2.0, c2=2.0.

[0085] The Pareto optimal solution screening method is used to perform non-dominated sorting and congestion calculation to screen out the optimal solution that balances grid stability and revenue, so as to obtain the output strategy.

[0086] Output strategies: The optimal strategy for regulation and trading coordination includes: equipment regulation instructions, spot trading reporting strategies, and auxiliary service reporting strategies.

[0087] Technical benefits: Achieve grid regulation deviation ≤2%, and increase market revenue by 10%-15%.

[0088] Step 400: Based on the device control strategy included in the output strategy, determine the millimeter-level control commands corresponding to the edge gateway and perform Kalman filtering and data synchronization. Also, based on the application strategy included in the output strategy, connect to the power system and automatically submit the corresponding application data for bid synchronization and execution decomposition, determining the automation command execution data. The automation command execution data includes: device-level execution plan, device execution status, and real-time operating data.

[0089] This step involves cloud-edge collaborative instruction execution and Kalman data synchronization, with the execution entity being the device controller in the edge gateway and terminal device.

[0090] Cloud-based policy delivery: Policies are delivered to the edge gateway via the MQTT protocol, with a transmission latency of <10ms.

[0091] Edge-local real-time command generation: The edge gateway runs a TensorFlowLite lightweight model to generate millisecond-level control commands; the command response time is ≤50ms, which meets the requirements for one-time frequency adjustment.

[0092] Autonomous operation and data caching after network outage: The edge gateway operates independently locally after network outage, caching running data; caching capacity: supports 7×24h data storage with no data loss.

[0093] Kalman filter multiplexing data synchronization: .

[0094] in, For the first The state estimate of the step; This is the state transition matrix; For the previous step The state estimate of the step; To control the input matrix; For control vectors; Kalman gain; These are actual measured data; This is the observation matrix.

[0095] Synchronization error ≤ 0.1%, data is error-free.

[0096] Output results: Device execution status, real-time operating data Technical function: Ensures real-time control and regulation, and system availability ≥99.9%.

[0097] Step 500: Based on the data from the execution of automated instructions, conduct effect evaluation and feedback processing to iteratively optimize the decision model so that the execution deviation corresponding to the effect evaluation is within the preset deviation range.

[0098] This step involves market-based transaction execution and closed-loop feedback iteration, with the transaction service module in the transaction interface layer serving as the execution entity.

[0099] Automated execution of quantity and price quotation: It connects to the power trading system API and automatically submits application data without manual operation.

[0100] Synchronization and execution decomposition of bidding results: Real-time acquisition of bidding results; decomposition of the bid-winning capacity into equipment-level execution plans.

[0101] Performance evaluation and iteration: Calculate the execution deviation: |Actual execution - Planned execution| / Planned execution; if the deviation is greater than 5%, trigger the model to iterate and optimize again; generate a revenue accounting report and an execution evaluation report.

[0102] Outputs: Transaction settlement data, evaluation report, and iterated model; Technical benefits: Achieves full-process automation, reducing manual intervention by 90%.

[0103] This application is based on a dynamic aggregation model of multi-source heterogeneous new energy sources using time attention and Gaussian regularization-improved DBN: By integrating 18 high-dimensional features of distributed resources, power grid, and environment, and using a deep belief network with time attention mechanism and Gaussian noise regularization, dynamic equivalent modeling, real-time capacity assessment, and accurate response time calculation of heterogeneous resources such as photovoltaic, wind power, energy storage, and controllable loads are achieved, solving the problems of poor static modeling adaptation and inaccurate aggregation of heterogeneous resources.

[0104] This application is based on dynamic optimization of the regulation-trading dual objectives using an improved MOPSO+Lagrange penalty function: A dual-objective optimization model is constructed to minimize grid regulation deviation and maximize market-based revenue. A nonlinear constraint Lagrange penalty term and adaptive inertia weight are introduced to solve for the Pareto solution set that satisfies the optimal balance between grid security and revenue, thereby generating regulation instructions and trading application strategies.

[0105] This application is based on a cloud-edge collaborative regulation and autonomous execution mechanism for network outages using Kalman filtering. The edge gateway performs real-time command generation, state estimation, and network outage caching locally, while the cloud performs global optimization, market decision-making, and model iteration. Kalman filtering is used to achieve accurate data synchronization, no loss, and no deviation after network outage and reconnection.

[0106] This application utilizes a closed-loop operation across the entire process of distributed new energy marketization: By connecting to the power trading / dispatch system, it can automate the reporting and bidding process, synchronize the bidding results, decompose the execution plan, accurately calculate the revenue, and iterate and optimize the strategy, thus realizing a complete closed loop of "aggregation-control-trading-feedback".

[0107] The benefits of this application are: The accuracy and stability of multi-source heterogeneous resource aggregation have been significantly improved: Performance: Multi-type distributed resource aggregation modeling accuracy ≥96%, equivalent output prediction deviation ≤3%, supports stable access for 2000+ terminals, and aggregation capacity assessment error <3%.

[0108] Technical sources: Improved DBN dynamic aggregation modeling method based on temporal attention, 18-dimensional high-dimensional feature fusion, KPCA dimensionality reduction and Gaussian regularization training.

[0109] Cause analysis: Traditional models only use static equivalent circuits and do not incorporate temporal and environmental features; this application strengthens the weight of recent data through a time attention mechanism and uses a deep belief network to capture nonlinear coupling relationships, while preventing overfitting through Gaussian noise regularization. Therefore, it still maintains extremely high accuracy in heterogeneous, fluctuating, and time-varying scenarios.

[0110] The control response speed reaches the millisecond level, meeting the primary frequency regulation requirements of the power grid. Performance: End-to-end response to control commands ≤50ms, power grid dispatch command tracking error ≤2%, and can stably participate in primary frequency regulation and rapid standby.

[0111] Technology source: Cloud-edge collaborative architecture + edge local real-time control + Kalman filter state estimation.

[0112] Cause analysis: Traditional centralized architecture requires uploading to the cloud and then distributing, resulting in long links and high latency; this application pushes real-time control down to the edge gateway, with the cloud only performing global optimization and transaction decisions, and the edge directly generating execution instructions, thus reducing latency to the millisecond level.

[0113] Achieving optimal synergy between the dual objectives of power grid security and market-based benefits: Performance: Under the premise of meeting grid constraints, market-based returns are 10% to 15% higher than traditional single strategies, and regulation deviation and economic benefits are simultaneously optimized.

[0114] Technical source: Improved MOPSO algorithm + Lagrange penalty function + dynamic weight allocation mechanism.

[0115] Analysis of the reasons: Traditional methods are either single-objective optimizations or weighted fixed methods that cannot adapt. This application constructs a dual-objective optimization model and searches for the Pareto front through non-dominated sorting. At the same time, it strictly satisfies the constraints with Lagrange penalty, which can automatically switch preferences in different scenarios such as peak power grid / high market price to achieve a win-win situation.

[0116] System reliability has been greatly improved; it can self-regulate after a network outage and reconnect without errors. Performance: The system can operate autonomously 24 / 7 even when the network is disconnected at the edge, and the data synchronization error after reconnection is ≤0.1%, with an overall system availability of ≥99.9%.

[0117] Technology source: edge local caching + Kalman filter synchronization + key module cluster deployment.

[0118] Cause analysis: Traditional systems lose control and data when the network is disconnected; this application has independent control, local storage, and state estimation capabilities at the edge. When reconnected, the timing data is smoothed and calibrated through Kalman filtering, so there is no data loss, no data corruption, and no impact.

[0119] The electricity market participation process is fully automated and closed-loop, significantly improving efficiency and compliance: Performance: The entire process from aggregation, prediction, volume reporting, bidding, execution, and accounting is fully automated, reducing human intervention by 90% and improving transaction execution efficiency by 80%, fully complying with transaction and scheduling interface specifications.

[0120] Technology source: Market-oriented operation closed-loop method + standard interface integration + automatic strategy iteration mechanism.

[0121] Analysis of causes: Traditional processes rely on manual form filling, manual declaration, and manual verification; this application models all transaction rules, scheduling instructions, execution plans, and effect evaluations, and the system operates automatically in a closed loop, thus being faster, more accurate, and more compliant.

[0122] In complex scenarios with multiple constraints, the optimization strategy is feasible, implementable, and does not exceed the limits: Performance: Under multiple constraints such as SOC, power, response speed, and trading boundaries, the strategy is 100% executable, with no overstepping of limits, no violations, and no equipment damage.

[0123] Technical sources: Multi-constraint modeling + Lagrange penalty + improved MOPSO convergence enhancement.

[0124] Cause analysis: Traditional algorithms often fail to implement strategies due to constraint overruns; this application transforms all constraints into penalty terms and embeds them into the optimization objective, allowing particles to search within the feasible region, thus enabling the output strategy to be directly deployed and executed.

[0125] The system boasts strong scalability and compatibility, enabling rapid integration with various devices and systems. Features: Supports multiple protocols such as Modbus / OPCUA / IEC104, and can be connected to mainstream trading systems, scheduling systems, inverters / PCS / load controllers without the need for hardware modification.

[0126] Technology source: layered microservice architecture + protocol adaptation layer + standardized interface design.

[0127] Cause analysis: Traditional systems have high coupling and closed protocols; this application uses a unified adapter layer to shield the differences in equipment and modularizes and plugs into services, thus having extremely strong compatibility and scalability.

[0128] In one exemplary embodiment, such as Figure 2 As shown, a multi-source heterogeneous distributed renewable energy aggregation, regulation, and coordination device is provided, comprising: The data acquisition and processing module is used to acquire multi-dimensional feature information from multiple sources and perform preset standardized preprocessing to obtain processed data.

[0129] The model building module is used to construct a decision model based on the processed data, using dynamic aggregation modeling based on time attention and Gaussian regularized improved DBN; the decision model includes a bi-objective optimization function, constraints, and Lagrange penalty term processing constraints.

[0130] The output strategy determination module is used to solve the bi-objective optimization function based on the improved MOPSO algorithm, according to the constraints and the Lagrange penalty term, and to select the optimal solution that balances grid stability and revenue by using the Pareto optimal solution screening method, and to determine the output strategy.

[0131] The automated instruction execution module is used to determine the millimeter-level control instructions corresponding to the edge gateway based on the device control strategy included in the output strategy and to perform Kalman filtering and multiplexing data synchronization. It is also used to connect to the power system based on the application strategy included in the output strategy and automatically submit the corresponding application data for bid synchronization and execution decomposition to determine the automated instruction execution data. The automated instruction execution data includes: device-level execution plan, device execution status, and real-time operation data.

[0132] The closed-loop feedback module is used to evaluate the effect and process the feedback based on the automated instruction execution data, so as to iteratively optimize the decision model and keep the execution deviation corresponding to the effect evaluation within a preset deviation range.

[0133] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, 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 operating system and computer programs in the non-volatile storage media to run. The database stores multi-source heterogeneous distributed renewable energy aggregation, regulation, and coordination data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a multi-source heterogeneous distributed renewable energy aggregation, regulation, and coordination method.

[0134] Those skilled in the art will understand that Figure 3 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.

[0135] In one exemplary embodiment, a computer device is also 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.

[0136] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0137] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0138] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0139] 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 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).

[0140] 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, data processing logic devices, etc., and are not limited to these.

[0141] 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 specification.

[0142] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for the coordinated regulation and control of multi-source heterogeneous distributed new energy sources, characterized in that, include: Obtain multi-dimensional feature information from multi-source data and perform pre-processing with preset standardization to obtain processed data; Based on the processed data, dynamic aggregation modeling is performed using time attention and Gaussian regularization-improved DBN to construct a decision model; the decision model includes a bi-objective optimization function, constraints, and Lagrange penalty term processing constraints. Based on the improved MOPSO algorithm, the bi-objective optimization function is solved according to the constraints and the Lagrange penalty term. The Pareto optimal solution screening method is used to screen the optimal solution that takes into account both grid stability and revenue, and the output strategy is determined. Based on the device control strategy included in the output strategy, the millimeter-level control command corresponding to the edge gateway is determined and Kalman filtering is performed to reconnect the data. Also, based on the application strategy included in the output strategy, the power system is connected and the corresponding application data is automatically submitted to perform bid synchronization and execution decomposition, and the automated command execution data is determined. The automated instruction execution data includes: device-level execution plan, device execution status, and real-time operation data; Based on the automated instruction execution data, effect evaluation and feedback processing are performed to iteratively optimize the decision model, so that the execution deviation corresponding to the effect evaluation is within a preset deviation range.

2. The multi-source heterogeneous distributed new energy aggregation, regulation, and coordination method according to claim 1, characterized in that, The expression corresponding to the weights of the time attention is: ； in, Weights for time-based attention; This is the attention decay coefficient; This represents the total duration of the time window. This is the current time step; For time windows, time steps; The network structure corresponding to the Gaussian regularized improved DBN includes: an input layer, three hidden layers, and an output layer; The activation function corresponding to the hidden layer is: ； The network loss function corresponding to the Gaussian regularized improved DBN is: ； in, The activation function corresponding to the hidden layer; For activation functions; For the first The weight matrix of the hidden layer; To process the standardized core feature set in the data; For Hadamard product; For the first The bias vector of the hidden layer; This is a Gaussian noise regularization term; The network loss function; This represents the number of training samples in a batch. For sample index; For the first The aggregated true value of active power for each sample; For the first The aggregated active power model prediction for each sample; For the first Aggregate reserve capacity for each real value and each sample; For the first The aggregated reserve capacity model prediction for each sample; For the first The true value of the aggregation response time or adjustment rate for each sample; For the first The aggregate response time model prediction for each sample; It is the set of all trainable weight parameters of the network; The square of the Frobenius norm of the weight matrix is ​​used as the L2 regularization term, with a coefficient of 0.001 to control the regularization strength.

3. The multi-source heterogeneous distributed new energy aggregation, regulation, and synergistic method according to claim 1, characterized in that, The expression corresponding to the bi-objective optimization function is: ； ； in, The target optimization function for power grid regulation deviation; This represents the total duration of the time window. This is the current time step; To contribute to the equivalent of aggregation; Power for scheduling commands; This is a vector representing the change in power regulation. The square of the L2 norm; Optimize the function to achieve market-based profit targets; The spot market electricity price; This refers to the actual amount of electricity traded online. To compensate for prices in the ancillary services market; The actual capacity of ancillary services provided; To adjust the deviation cost coefficient; Based on the planned output value; This is a vector representing the change in ancillary service capacity.

4. The multi-source heterogeneous distributed new energy aggregation, regulation, and coordination method according to claim 1, characterized in that, The constraints include: resource constraints, energy storage constraints, transaction constraints, and response constraints; The expression corresponding to the resource constraint is: ； ； The expression corresponding to the energy storage constraint is: ； ； The expression corresponding to the transaction constraint is: ； The expression corresponding to the response constraint is: ； in, To contribute to the equivalent of aggregation; This represents the upper limit of the aggregate equivalent output. The actual capacity of ancillary services provided; To regulate the polymerization capacity; For energy storage capacity; Energy storage capacity; The maximum charge and discharge power limit for energy storage systems; This refers to the actual amount of electricity traded online. Reserved power for future adjustments (standby capacity); A unit of time; It contributes to the aggregation equivalent of the previous time step.

5. The method for coordinated regulation and control of multi-source heterogeneous distributed new energy sources according to claim 1, characterized in that, The expression corresponding to the constraint handled by the Lagrange penalty term is: ； in, This is the total penalty value; Index for constraints; This is the penalty coefficient; For the first A function for the amount of constraint violation; A vector of decision variables; This is an indicator function.

6. The multi-source heterogeneous distributed new energy aggregation, regulation, and coordination method according to claim 1, characterized in that, Based on the improved MOPSO algorithm, the bi-objective optimization function is solved according to the constraints and the Lagrange penalty term. The Pareto optimal solution selection method is then used to select the optimal solution that balances grid stability and profitability, determining the output strategy, including: Based on the inertia weights corresponding to the improved MOPSO algorithm, as well as the preset number of iterations and learning factor, the biobjective optimization function is solved according to the constraints and the Lagrange penalty term, yielding the solution result; where the expression corresponding to the inertia weights is: ； =0.9， =0.4； For the first The inertial weight of the particle in the next iteration; This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. This represents the maximum total number of iterations for the algorithm. This represents the current iteration step. The Pareto optimal solution screening method is used to perform non-dominated sorting and congestion calculation to screen out the optimal solution that balances grid stability and revenue, so as to obtain the output strategy.

7. A multi-source heterogeneous distributed new energy aggregation, regulation, and coordination device, characterized in that, include: The data acquisition and processing module is used to acquire multi-dimensional feature information of multi-source data and perform preset standardization preprocessing to obtain processed data; The model building module is used to construct a decision model based on the processed data, dynamic aggregation modeling based on time attention and Gaussian regularized improved DBN; the decision model includes a bi-objective optimization function, constraints, and Lagrange penalty term processing constraints. The output strategy determination module is used to solve the bi-objective optimization function based on the improved MOPSO algorithm, according to the constraints and the Lagrange penalty term, and to select the optimal solution that balances grid stability and revenue by using the Pareto optimal solution screening method to determine the output strategy. The automated instruction execution module is used to determine the millimeter-level control instructions corresponding to the edge gateway based on the device control strategy included in the output strategy and to perform Kalman filtering and multiplexing data synchronization; and to connect to the power system based on the application strategy included in the output strategy and automatically submit the corresponding application data to perform bid synchronization and execution decomposition, and determine the automated instruction execution data. The automated instruction execution data includes: device-level execution plan, device execution status, and real-time operation data; The closed-loop feedback module is used to evaluate the effect and process the feedback based on the automated instruction execution data, so as to iteratively optimize the decision model and keep the execution deviation corresponding to the effect evaluation within a preset deviation range.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the multi-source heterogeneous distributed new energy aggregation, regulation, and coordination method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the multi-source heterogeneous distributed new energy aggregation, regulation and coordination method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the multi-source heterogeneous distributed new energy aggregation, regulation and coordination method as described in any one of claims 1-6.

Images