Energy supply side optimization management system and implementation method

Abstract

The application discloses an energy supply side optimization management system and an implementation method, and relates to the technical field of energy management. The method comprises the following steps: collecting source end data of an energy supply side, analyzing the source end data through a multi-level protocol adaptation engine and a dynamic rule matching engine, and extracting various energy characteristics and GIS information of the energy supply side; constructing a multi-energy flow topology constraint matrix based on the energy characteristics, performing fine-grained geographical grid division based on the GIS information of the energy supply side, and determining grid-level energy dynamic parameters; combining the multi-energy flow topology constraint matrix to perform dynamic trend adjustment on the grid-level energy dynamic parameters, and determining a grid-level energy dynamic prediction atlas; finding a best trade-off solution set through a population aggregation and dispersion search algorithm; combining a safe operation boundary to perform constraint solving on the best trade-off solution set, and determining an optimal solution of energy dispatching. The application improves the accuracy and foresight of energy management, accurately traces loss links, and drives the system to continuously evolve in the direction of low carbon and high efficiency.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to an energy supply-side optimization management system and its implementation method. Background Technology

[0002] As the proportion of intermittent and highly volatile renewable energy sources such as wind and solar power in the energy mix continues to rise, unprecedented challenges have been brought to the safe, stable, and efficient operation of traditional energy systems. Traditional energy supply-side management systems typically employ a centralized dispatching model based on "source follows load," which faces numerous bottlenecks: First, the accuracy of renewable energy output prediction is insufficient, leading to significant deviations between dispatch plans and actual conditions; second, optimized dispatching and safety verification are often separated, resulting in low computational efficiency and difficulty in adapting to real-time changes; third, the system lacks proactive defense and autonomous recovery capabilities, relying on manual intervention in the face of sudden failures or extreme disturbances, resulting in slow response times and a high risk of escalating accidents. Existing systems also suffer from extensive energy efficiency management, making it difficult to achieve accurate source tracing and closed-loop optimization of carbon flow and energy loss. This invention aims to overcome the problems of insufficient renewable energy absorption capacity, high reliance on manual intervention, slow response times, and extensive energy efficiency management in existing energy management systems. Summary of the Invention

[0003] This invention provides an energy supply-side optimization management method, comprising:

[0004] Collect energy supply-side source data, and parse the source data through a multi-level protocol adaptation engine and a dynamic rule matching engine to extract various energy characteristics and energy supply-side GIS information;

[0005] A multi-energy flow topology constraint matrix is ​​constructed based on energy characteristics, and fine-grained geographic grids are divided based on energy supply-side GIS information to determine grid-level energy dynamic parameters.

[0006] By combining the multi-energy flow topological constraint matrix, the grid-level energy dynamic parameters are dynamically adjusted to determine the grid-level energy dynamic prediction spectrum;

[0007] Based on the grid-level energy dynamic prediction map, the optimal trade-off solution set is found through a population aggregation and dispersion search algorithm;

[0008] By combining the safety operation boundary with the constraint solution set of the best trade-off, the optimal solution for energy scheduling is determined.

[0009] The aforementioned energy supply-side optimization management method collects energy supply-side source data, parses the source data through a multi-level protocol adaptation engine and a dynamic rule matching engine, and extracts various energy characteristics and energy supply-side GIS information, including:

[0010] By analyzing the characteristics of source data, an updatable quality governance rule base is built using a dynamic rule engine driven by machine learning and knowledge graphs.

[0011] The system collects source data in real time through a multi-level protocol adaptation engine, calls the quality governance rule base to extract various energy characteristics and energy supply-side GIS information, and constructs a standard source database with a hierarchical storage strategy.

[0012] The aforementioned energy supply-side optimization management method constructs a multi-energy flow topology constraint matrix based on energy characteristics and performs fine-grained geographic grid division based on energy supply-side GIS information to determine grid-level energy dynamic parameters, including:

[0013] Energy data points are extracted from the source database, and fine-grained geographic grids are divided based on energy supply-side GIS information to generate a spatial distribution field. Based on the preliminary gridded features, the topological structure data of the energy network is extracted to construct a multi-energy flow topological constraint matrix.

[0014] Based on the multi-energy flow topological constraint matrix as the hidden layer regularization term, and using the extracted spatial features as input, a grid-level energy dynamic prediction model is constructed, and grid-level energy dynamic parameters are output.

[0015] The aforementioned energy supply-side optimization management method, which combines a safe operating boundary with constraint solving of the optimal trade-off solution set to determine the optimal energy dispatch solution, includes:

[0016] Based on the optimal trade-off solution set, security bottlenecks are identified, and the safe operating boundary is obtained.

[0017] By combining the safe operation boundary, and adding safety constraints to the feasible domain based on the best trade-off solution set, the final scheduling scheme is obtained by re-optimizing the calculation.

[0018] The aforementioned energy supply-side optimization management method identifies safety bottlenecks based on the optimal trade-off solution set and obtains a safe operating boundary, including:

[0019] Extract a three-dimensional multi-dimensional feature set from the source data, mine implicit associations, and construct a graph neural network to build a constraint system;

[0020] By simulating the system state under all potential failure scenarios in parallel, the limiting conditions for maintaining safe operation are solved in reverse.

[0021] An energy supply-side optimization management system, comprising:

[0022] The wide-area data acquisition module is used to collect energy supply-side source data. It parses the source data through a multi-level protocol adaptation engine and a dynamic rule matching engine to extract various energy characteristics and energy supply-side GIS information.

[0023] The fine-grained energy parameter partitioning module is used to construct a multi-energy flow topology constraint matrix based on energy characteristics and to perform fine-grained geographic grid partitioning based on energy supply-side GIS information to determine grid-level energy dynamic parameters.

[0024] The energy prediction map construction module is used to dynamically adjust the grid-level energy dynamic parameters by combining the multi-energy flow topological constraint matrix, and determine the grid-level energy dynamic prediction map.

[0025] The optimal trade-off set generation module is used to find the optimal trade-off solution set based on the grid-level energy dynamic prediction map using a population aggregation and dispersion search algorithm.

[0026] The energy scheduling optimal solution generation module is used to combine the safe operation boundary to perform constraint solving on the best trade-off solution set and determine the optimal energy scheduling solution.

[0027] The beneficial effects achieved by this invention are as follows:

[0028] It enhances the precision and foresight of energy management, providing far more accurate data input for optimized dispatching than traditional methods, thus reducing uncertainty risks at the source. By internalizing grid security constraints as an inherent part of the optimization process, it achieves secure, embedded dispatch plan generation, significantly improving computational efficiency and economic benefits while ensuring the physical security of the grid. The construction of a layered defense system encompassing "pre-warning, in-process correction, and post-event protection" greatly reduces the risk of large-scale power outages and significantly improves power supply reliability. Precisely tracing the source of losses drives the system's continuous evolution towards low-carbon and high-efficiency directions. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0030] Figure 1 This is a flowchart of an energy supply-side optimization management implementation method provided in Embodiment 1 of this application.

[0031] Figure 2 This is a schematic diagram of an energy supply-side optimization management system provided in Embodiment 2 of this application. Detailed Implementation

[0032] 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, not all, of the embodiments of the present invention. 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.

[0033] Example 1

[0034] like Figure 1 As shown, Embodiment 1 of this application provides an energy supply-side optimization management method, including:

[0035] S1: Collect energy supply-side source data, parse the source data through a multi-level protocol adaptation engine and a dynamic rule matching engine, and extract various energy characteristics and energy supply-side GIS information;

[0036] The process involves collecting energy supply-side source data, parsing the source data using a multi-level protocol adaptation engine and a dynamic rule matching engine, and extracting various energy characteristics and energy supply-side GIS information. This includes the following sub-steps:

[0037] S11: By analyzing the characteristics of source data, an updatable quality governance rule base is built using a dynamic rule engine driven by machine learning and knowledge graphs.

[0038] Real-time acquisition of source data, including operational data of energy units, renewable energy power plants, wide-area measurement data, and meteorological and environmental data collected from multiple sources such as sensors, satellite radar, etc. Multi-dimensional features of the source data are extracted, including data range, fluctuation characteristics, temporal correlation, and missing data patterns. A knowledge graph is used to construct a network of relationships between data entities to characterize the complex constraints on data quality. Abnormal data patterns are automatically identified, and the root causes of anomalies are traced using a knowledge graph inference engine. Based on the learned patterns and root causes, data quality assessment rules (such as dynamic threshold rules and context consistency rules) and repair rules are generated. Rule parameters and confidence levels are continuously optimized based on feedback from repair effects, ultimately outputting a real-time updatable and interpretable quality governance rule base.

[0039] S12: Real-time data collection from the source end through a multi-level protocol adaptation engine, extraction of various energy characteristics and energy supply-side GIS information from the quality governance rule base, and construction of a standard source end database using a hierarchical storage strategy.

[0040] A multi-level protocol adaptation engine is activated. The bottom layer loads decoders adapted to common protocols and connects to source devices using standard protocols to acquire data. The middle layer uses a proprietary protocol parsing tool to process non-standard data from older devices and complete protocol conversion. The top layer activates the preprocessing function of edge nodes to compress and resume high-frequency data. Then, a quality governance rule base is invoked to perform standardization processing on the adapted source data, including format unification, outlier filtering, and missing value completion. Based on a hierarchical storage strategy, the processed data is categorized by access frequency and timeliness, allocated to corresponding storage media, and data indexes and access interfaces are established, ultimately constructing a standard source database.

[0041] S2: Construct a multi-energy flow topology constraint matrix based on energy characteristics, and perform fine-grained geographic grid division based on energy supply-side GIS information to determine grid-level energy dynamic parameters;

[0042] S21: Extract energy data points from the source database, and perform fine-grained geographic grid division based on energy supply-side GIS information to generate a spatial distribution field. Based on the preliminary gridded features, extract the topological structure data of the energy network to construct a multi-energy flow topological constraint matrix.

[0043] From the standard source database, data such as source energy output and multi-energy flow (electricity-gas-heat) operating parameters are extracted. Energy data points with geographic coordinates are selected. For data-sparse areas, energy parameter values ​​of unknown grid points are estimated based on the spatial correlation of known points to generate a continuous spatial distribution field, transforming discrete data into preliminary gridded features covering the entire region. Based on the preliminary gridded features, topological data of the energy network, such as line connection relationships, are extracted. A multi-energy flow topological constraint matrix is ​​constructed using graph theory methods to quantify the adjacency relationships between nodes and energy transmission constraints. Its expression is:

[0044] Based on the preliminary features of the grid, the topological structure data of the energy network is extracted. A multi-energy flow topological constraint matrix is ​​constructed using graph theory methods to quantify the adjacency relationships between nodes and energy transmission constraints. Its expression is as follows: Represents grid nodes With nodes Between Quantization values ​​of topological constraints for energy flows such as electricity, gas, and heat; Represents the adjacency discrimination coefficient, when node Between Quantization values ​​of topological constraints for energy flows such as electricity, gas, and heat; Represents the adjacency discrimination coefficient, when node and There exists a first When physical connections are made for energy flow, When there is no connection, ; Indicates the first Energy flow in nodes and The transmission capacity limit between; Represents a node and Between The number of segments in a power flow transmission path; Indicates the first Transmission efficiency of segment paths; It is the first Energy flow from nodes arrive The unit vector of the transmission direction; It is a node arrive Between The potential gradient vector of energy flow; It is a node and The three-dimensional Euclidean distance; It is the first The distance attenuation factor of energy flux varies depending on the type of energy flux, such as thermal energy. Typically greater than electrical energy; Represents partial derivative terms; It is the first Energy flow in nodes and Transmission loss function between; It is the first The transmission angle parameter of energy flow; Represents a node remove Other than The set of adjacent nodes of the energy flow class; It is a node Indirect coupling coefficient between them; express Inter-transmission capacity; express Inter-transmission capacity, the ratio of the two reflects Other connections to it Connection capacity preemption effect.

[0045] S22: Based on the multi-energy flow topological constraint matrix as the hidden layer regularization term, and using the extracted spatial features as input, a grid-level energy dynamic prediction model is constructed, and grid-level energy dynamic parameters are output.

[0046] The constraint matrix is ​​integrated into the model structure design, using extracted spatial features as the input layer, multi-energy flow topological constraints as hidden layer regularization terms, and grid-level energy dynamic parameters as the output layer. The model is trained by dividing the dataset. First, the spatial features of the input layer are expanded along the time-series dimension, incorporating historical energy dynamic data. Short-term fluctuation features are extracted through a temporal convolutional layer, and then graph convolutional layers utilize the multi-energy flow topological constraint matrix to define the adjacency relationships of graph nodes. This ensures that the activation values ​​of hidden layer neurons are preferentially transmitted to topologically connected grid nodes, strengthening the learning of energy coupling patterns between physically connected nodes. The output layer assigns dynamic weights to features of different grids and at different times, such as assigning higher weights to recent features of load center grids, ultimately outputting grid-level energy parameter predictions for future periods. During training, the temporal error between the predicted and actual values ​​is used as the loss function. During gradient descent, hidden layer regularization terms penalize weight updates that violate topological constraints, such as excessive weights between non-connected nodes, forcing the model's predictions to conform to the physical transmission patterns of the energy network. After multiple iterations, a high-precision grid-level energy dynamic prediction is constructed.

[0047] S3: Combine the multi-energy flow topological constraint matrix to dynamically adjust the grid-level energy dynamic parameters and determine the grid-level energy dynamic prediction map;

[0048] A spatial granularity prediction model is invoked, dividing the entire region into fine-grained geographic grids based on energy node distribution and load density using a geographic information system. The latest real-time source data, historical data from similar time periods, and environmental correlation data are extracted from a standard source database and input into the model. The model captures short-term dynamic trends through temporal convolutional layers, enhances the collaborative characteristics of topologically related grids through graph convolutional layers, and focuses the influence weights of key grids and time periods through a spatiotemporal attention mechanism. The physical rationality of energy parameters for each grid is verified by combining a multi-energy flow topological constraint matrix, generating grid-level energy dynamic parameters for future multi-time periods. Finally, the prediction results for each time period and each grid are integrated into a visualized energy dynamic prediction map, which includes continuous spatiotemporal trends and extreme value warnings for key nodes.

[0049] S4: Based on the grid-level energy dynamic prediction map, the optimal trade-off solution set is found through the population aggregation and dispersion search algorithm;

[0050] Based on the grid-level energy dynamic prediction map, the energy supply and demand forecasts, transmission path capacity, and key equipment operating parameters of each grid are extracted as inputs. A multi-objective optimization system is determined, covering economy, low carbon emissions, and reliability, and priority weights are set based on the conflict relationships between objectives. First, a population of decision variables including power output adjustments and load allocation schemes is initialized, and the fitness value of each individual is calculated through the objective function. The search strategy is dynamically adjusted according to the population distribution: when the solution space is dispersed, the search range is expanded to explore potential optimal regions; when solutions are clustered, the step size is reduced for fine optimization. Individuals that violate basic operating constraints are filtered out, and after multiple iterations, a comprehensive optimal solution is selected from the solution set based on the real-time operating scenario, generating the best trade-off solution set.

[0051] S5: Combine the safety operation boundary to solve the optimal trade-off solution set under constraints, and determine the optimal solution for energy scheduling.

[0052] The process of constraining the optimal trade-off solution set by combining the safe operation boundary to determine the optimal energy dispatch solution includes the following sub-steps:

[0053] S51: Identify security bottlenecks based on the optimal trade-off solution set to obtain the safe operating boundary;

[0054] Among these, the safety bottleneck is identified based on the optimal trade-off solution set, and the safe operating boundary is obtained, including:

[0055] S511: Extract a three-dimensional multi-dimensional feature set from the source data, mine implicit associations, and construct a graph neural network to build a constraint system;

[0056] Explicit safety and economic constraints, such as power balance formulas and equipment rated limits, are extracted from standard source data. Then, based on a geographic information system, each parameter is anchored to its corresponding grid topology node, marking the physical connections between nodes. Parameters are segmented into time series within fixed time windows, and statistical features such as mean, volatility, and extreme values ​​are extracted for each time period. By associating features from adjacent time periods, a three-dimensional dataset—spatial coordinates, topological relationships, and temporal features—is generated, i.e., a multi-dimensional feature set. A time decay factor is introduced on top of the traditional frequent itemset tree, assigning higher weight to recent data to enhance timeliness. Safety event triggers (such as line overload or voltage exceeding limits) are set as the target itemset, with the multi-dimensional feature set prior to the event as the antecedent. By dynamically adjusting the support (adaptively changing with event sparsity) and confidence threshold, implicit associations are uncovered—for example, when the photovoltaic output volatility in a certain region exceeds a threshold and the load in adjacent regions suddenly increases, the confidence of transformer overload in subsequent time periods must meet preset conditions. Subsequently, a graph neural network is constructed: a three-level node system is used, consisting of a system layer (power balancing nodes across the entire network), a region layer (regional control nodes), and a device layer (specific component nodes). Explicit constraints are used as node attributes, and implicit association patterns (such as discovered temporal coupling relationships) are used as directed edges between nodes. Graph convolutional layers learn the interactions between nodes within each layer, and graph attention layers strengthen the transmission of constraints between layers. Finally, the constraint parameters corresponding to each layer are output, forming a structured constraint system.

[0057] S512: By simulating the system state under all potential failure scenarios in parallel, the limit conditions for maintaining safe operation are solved in reverse.

[0058] Based on a structured constraint system, the decision variables of the optimal trade-off solution set are substituted into the constraint network. The current satisfaction state of each level of constraint (e.g., whether it is close to the power imbalance threshold) is calculated using a constraint satisfaction algorithm. Based on constraint nodes that are not satisfied or are critically satisfied, targeted fault scenarios are automatically generated. The scenario evolution process is then simulated in a digital twin environment. Inverse limit optimization is used to solve for the constraint critical values, as shown in the following formula:

[0059]

[0060] It is a reverse optimization operator, whose function is to find the critical value C from the set of positive real numbers that minimizes the safety deviation of the subsequent fraction under all fault scenarios; This represents the set of potential failure scenarios in parallel simulation; It is a scene The greater the impact weight, the more severe the consequences of the failure. The larger the value; This is the time decay coefficient; It is a scene The time interval from the current moment; For the scene The number of time series deduction steps; It is the first The attenuation coefficient of the step; yes Norm operations This indicates that when the constraint parameter is At that time, in the scene In the Step-by-step system security indicators; This is the baseline value of the safety indicator corresponding to this scenario; the norm result quantifies the degree of deviation between the safety indicator and the baseline value. It is a smoothing coefficient. The larger its value, the stronger the penalty for the deviation between the critical value and the initial plan, thus avoiding the optimization result from deviating from the best trade-off solution set. These are the initial constraint values ​​corresponding to the optimal trade-off solution set; The physical upper limit of the constraint parameters; where To constrain the number of dimensions, such as device, region, energy flow type, etc.; It is a dimension The weighting of core equipment or key areas is higher. It is a dimension The ideal critical value.

[0061] Simultaneously, a feedback channel is established between real-time measurement data and the constraint network: when actual operation data shows a deviation of a certain implicit association rule, the weight of the rule in the constraint network is dynamically adjusted, the boundary threshold of the corresponding scenario is corrected synchronously, and finally a safe operation boundary that dynamically matches the constraint system is generated.

[0062] S52: Combining the safe operation boundary, the feasible domain with added safety constraints based on the best trade-off solution set is re-optimized to obtain the final scheduling scheme.

[0063] The constraints of the safe operating boundary are transformed into inequality constraints in the optimization model, establishing a mapping relationship with the decision variables in the optimal trade-off solution set. With economy, low carbon emissions, and reliability as optimization objectives, the reliability objective weight is automatically increased when the safety boundary tightens (e.g., the transmission limit of a certain line decreases). During the search for feasible solutions in the solution space, a constraint verification module compares the fit between the decision variables and the safety boundary in real time, eliminating solutions that exceed the limits and imposing penalty factors on critical solutions, guiding the search towards the safe and feasible region. Simultaneously, a real-time operational data feedback channel is introduced, inputting real-time parameters such as current equipment status and source-load fluctuations into the optimization model. A rolling optimization mechanism updates the decision variables at fixed intervals, ensuring dynamic matching between the scheduling scheme and the actual system operating state. Finally, the optimal comprehensive solution is selected from the optimized solution set and transformed into executable scheduling instructions for each distributed entity, forming the final scheduling scheme.

[0064] The real-time execution data of the final scheduling scheme is synchronized with the wide-area measurement data in the source data to construct a real-time operation matrix containing execution deviations. An improved dynamic stability assessment algorithm is then invoked: first, using a deviation quantification model, the difference between the actual parameters and the expected scheduling parameters in the operation matrix (such as output lag and load adjustment deviation rate) is used as a correction term, incorporating indicators such as power angle stability margin and voltage support capability; then, a dynamic response characteristic library of equipment is embedded, combining the response delay and ramp rate of new energy units and energy storage equipment to dynamically adjust the stability threshold (e.g., relaxing short-term frequency deviation tolerance for slow-response units); simultaneously, subsequent scheduling execution plans are rolled over in a fixed window, incorporating future short-term instruction changes into the evaluation, and predicting the cumulative impact of execution rhythm through time-series correlation coefficients. Finally, the corrected stability indicators are output, combined with ultra-short-term trend extrapolation to predict frequency fluctuations and voltage evolution paths, identifying the type and level of instability risk.

[0065] Based on the type and level of instability risk, a dynamic disturbance map is constructed for low-to-medium power imbalance risk signals and disturbance trend data. The predicted frequency / voltage exceedance trend is decomposed into three-dimensional features of "time-space-amplitude". Topologically sensitive nodes within the risk-affected area are simultaneously marked. Agile resources (energy storage, adjustable loads, etc.) within the area are treated as distributed nodes. A compensation model is collaboratively trained through a federated learning mechanism without sharing the original data. Each node generates an initial compensation scheme based on its local resource characteristics, which is then aggregated into a global optimization strategy after encrypted parameter interaction, solving the latency problem of centralized decision-making. A two-stage "pre-response-actual response" mechanism is introduced: In the first stage, a digital twin is invoked to simulate the disturbance evolution, and pre-trigger commands are issued to resource nodes based on the simulation results; in the second stage, when the disturbance actually occurs, the compensation amount is dynamically adjusted according to the real-time measurement deviation, avoiding the fixed deviation of traditional look-ahead compensation. At the same time, a resource efficiency self-learning library is established to record the response latency and accuracy error of each compensation, and subsequent resource matching weights are optimized through gradient descent.

[0066] Upon receiving a high instability risk signal, the dynamic role switching mechanism of distributed nodes is immediately activated, enabling each device to autonomously claim the role of "coordinating node" or "executing node" based on real-time over-limit parameters. The device with the lowest over-limit severity automatically becomes the coordinating node, while the rest become executing nodes; role allocation is completed without central instructions. The coordinating node aggregates the local operating data of each executing node through an improved consensus algorithm (integrating power grid topology correlation weights), generating a preliminary repair direction without relying on a global model, and broadcasting this direction as encrypted constraint parameters to the executing nodes. Based on local resource characteristics, the executing nodes generate specific repair plans within the constraints using a self-organizing game algorithm. After the coordinating node verifies any conflicts, each node executes the adjustment actions in parallel. Simultaneously, holographic monitoring of the repair effect is initiated, constructing a repair trajectory map through real-time status interaction between nodes. If the repair in a certain area does not meet expectations, the coordinating node automatically triggers a secondary repair chain (activating backup resources) until the global parameters return to a safe range.

[0067] Upon detecting an extreme collapse risk signal, the core load dynamic locking mechanism is immediately activated. This mechanism real-time filters core load clusters requiring preservation and marks the source network equipment upon which their power supply depends, generating a "core load-supporting equipment" mapping map to avoid missing dynamically critical loads using traditional fixed lists. Combining the current frequency / voltage collapse rate with the grid topology, the propagation path and impact range of the collapse wave are calculated in real-time, dynamically defining the islanding separation boundary—not relying on preset separation points, but rather using "core load cluster power supply integrity" as the objective to determine the non-core areas to be separated and the scope of the islanded subnets to be preserved. Then, millisecond-level separation signals are sent to the boundary circuit breakers to ensure consistent action timing at each breakpoint, avoiding inrush currents during separation. Simultaneously, the control mode of the islanded subnet is automatically switched from global scheduling to island autonomy, activating the source-load-storage collaborative balancing algorithm within the subnet. Through parallel actions such as fine-tuning of source output, rapid charging and discharging of energy storage, and temporary reduction of non-core loads, power balance within the island is maintained. The system monitors the island's operating parameters in real time. If a new stable fluctuation occurs, a secondary survival strategy is immediately triggered, and the data of this disengagement and switchover is recorded, and the island solution optimization library is updated.

[0068] The periodic execution data of the synchronized energy dispatch scheme and the real-time measurement data of each node are linked to various indicators and dispatch scheme decisions (such as the impact of the renewable energy consumption ratio on collaborative efficiency) through a dynamic weight allocation mechanism, generating periodic performance evaluation results. Simultaneously, combining the energy input type (such as renewable energy, thermal power) and transmission losses of each node, the node carbon potential intensity is calculated. Based on the correlation between load electricity consumption and node carbon potential, the carbon consumption of each load is derived, forming a node carbon distribution map. Through a loss chain tracing algorithm, key energy loss links are located from the carbon distribution map and energy efficiency data, such as abnormal transmission loss on a certain line segment or excessive operating loss of a certain equipment. The correlation logic between loss and dispatch scheme decisions is analyzed, a source tracing report is generated, and synchronously fed back to the self-optimizing dispatch module.

[0069] Example 2

[0070] like Figure 2 As shown, Embodiment 2 of this application provides an energy supply-side optimization management system, including:

[0071] Wide-area data acquisition module 21: Collects energy supply-side source data, parses the source data through a multi-level protocol adaptation engine and a dynamic rule matching engine, and extracts various energy characteristics and energy supply-side GIS information;

[0072] Fine-grained energy parameter partitioning module 22: Construct a multi-energy flow topology constraint matrix based on energy characteristics, and perform fine-grained geographic grid partitioning based on energy supply-side GIS information to determine grid-level energy dynamic parameters;

[0073] Energy Prediction Map Construction Module 23: Combines the multi-energy flow topology constraint matrix to dynamically adjust the grid-level energy dynamic parameters and determine the grid-level energy dynamic prediction map;

[0074] Optimal trade-off set generation module 24: Based on the grid-level energy dynamic prediction map, it uses a population aggregation and dispersion search algorithm to find the optimal trade-off solution set;

[0075] Energy scheduling optimal solution generation module 25: Combines the safe operation boundary to perform constraint solving on the best trade-off solution set to determine the optimal energy scheduling solution.

[0076] Corresponding to the above embodiments, the present invention provides a computer storage medium, including: at least one memory and at least one processor;

[0077] The memory is used to store one or more program instructions;

[0078] A processor is used to run one or more program instructions to implement an energy supply-side optimization management method.

[0079] Corresponding to the above embodiments, this embodiment of the invention provides a computer-readable storage medium containing one or more program instructions, which are executed by a processor to implement an energy supply-side optimization management method.

[0080] The embodiments disclosed in this invention provide a computer-readable storage medium storing computer program instructions that, when executed on a computer, cause the computer to perform the aforementioned energy supply-side optimization management method.

[0081] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0082] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0083] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0084] Among them, non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0085] Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus RAM (DRRAM).

[0086] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0087] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0088] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing energy supply-side management, characterized in that, include: Collect energy supply-side source data, and parse the source data through a multi-level protocol adaptation engine and a dynamic rule matching engine to extract various energy characteristics and energy supply-side GIS information; A multi-energy flow topology constraint matrix is ​​constructed based on energy characteristics, and fine-grained geographic grids are divided based on energy supply-side GIS information to determine grid-level energy dynamic parameters. The expression for the multi-energy flow topology constraint matrix is: , Represents grid nodes With nodes Between Quantization values ​​of topological constraints for energy flow-like structures; Represents the adjacency discrimination coefficient, when node and There exists a first When physical connections are made for energy flow, When there is no connection, Indicates the first Energy flow in nodes and The transmission capacity limit between; Represents a node and Between The number of segments in a power flow transmission path; Indicates the first Transmission efficiency of segment paths; It is the first Energy flow from nodes arrive The unit vector of the transmission direction; It is a node and Between Potential gradient vector similar to energy flow; It is a node and The three-dimensional Euclidean distance; It is the first The distance attenuation factor of energy flow varies depending on the type of energy flow. Represents partial derivative terms; It is the first Energy flow in nodes and Transmission loss function between; It is the first The transmission angle parameter of energy flow; Represents a node remove Other than The set of adjacent nodes of the energy flow class; It is a node Indirect coupling coefficient between them; express Inter-transmission capacity; express Inter-transmission capacity, the ratio of the two reflects Other connections to it Connection capacity preemption effect; This study uses a multi-energy flow topology constraint matrix to dynamically adjust grid-level energy dynamic parameters and determine a grid-level energy dynamic prediction map. Specifically, it involves: calling a spatial granularity prediction model; dividing the entire region into fine-grained geographic grids based on energy node distribution and load density using a geographic information system; extracting the latest real-time source data, historical similar time period data, and environmental correlation data from a standard source database and inputting them into the model; the model captures short-term dynamic trends through temporal convolutional layers; graph convolutional layers enhance the collaborative characteristics of topologically related grids; and a spatiotemporal attention mechanism focuses on the influence weights of key grids and time periods. The study then verifies the physical rationality of energy parameters for each grid using the multi-energy flow topology constraint matrix, generating grid-level energy dynamic parameters for future multiple time periods. Finally, the prediction results for each time period and each grid are integrated into a visualized energy dynamic prediction map, which includes continuous spatiotemporal trends and extreme value warnings for key nodes. Based on the grid-level energy dynamic prediction map, the optimal trade-off solution set is found through a population aggregation and dispersion search algorithm; By combining the safety operation boundary with the constraint solution set of the best trade-off, the optimal solution for energy scheduling is determined.

2. The energy supply-side optimization management method according to claim 1, characterized in that, Data from the energy supply side's source end is collected, and the source end data is parsed through a multi-level protocol adaptation engine and a dynamic rule matching engine to extract various energy characteristics and energy supply side GIS information, including: By analyzing the characteristics of source data, an updatable quality governance rule base is built using a dynamic rule engine driven by machine learning and knowledge graphs. The system collects source data in real time through a multi-level protocol adaptation engine, calls the quality governance rule base to extract various energy characteristics and energy supply-side GIS information, and constructs a standard source database with a hierarchical storage strategy.

3. The energy supply-side optimization management method according to claim 1, characterized in that, A multi-energy flow topological constraint matrix is ​​constructed based on energy characteristics, and fine-grained geographic grids are generated based on energy supply-side GIS information to determine grid-level energy dynamic parameters, including: Energy data points are extracted from the source database, and fine-grained geographic grids are divided based on energy supply-side GIS information to generate a spatial distribution field. Based on the preliminary gridded features, the topological structure data of the energy network is extracted to construct a multi-energy flow topological constraint matrix. Based on the multi-energy flow topological constraint matrix as the hidden layer regularization term, and using the extracted spatial features as input, a grid-level energy dynamic prediction model is constructed, and grid-level energy dynamic parameters are output.

4. The energy supply-side optimization management method according to claim 1, characterized in that, By combining the safety operation boundary with the constraint solution set of the optimal trade-offs, the optimal solution for energy dispatch is determined, including: Based on the optimal trade-off solution set, security bottlenecks are identified, and the safe operating boundary is obtained. By combining the safe operation boundary, and adding safety constraints to the feasible domain based on the best trade-off solution set, the final scheduling scheme is obtained by re-optimizing the calculation.

5. The energy supply-side optimization management method according to claim 4, characterized in that, Based on the optimal trade-off solution set, security bottlenecks are identified, and the safe operating boundary is obtained, including: Extract a three-dimensional multi-dimensional feature set from the source data, mine implicit associations, and construct a graph neural network to build a constraint system; By simulating the system state under all potential failure scenarios in parallel, the limiting conditions for maintaining safe operation are solved in reverse.

6. An energy supply-side optimization management system, characterized in that, include: The wide-area data acquisition module is used to collect energy supply-side source data. It parses the source data through a multi-level protocol adaptation engine and a dynamic rule matching engine to extract various energy characteristics and energy supply-side GIS information. The fine-grained energy parameter partitioning module is used to construct a multi-energy flow topology constraint matrix based on energy characteristics and to perform fine-grained geographic grid partitioning based on energy supply-side GIS information to determine grid-level energy dynamic parameters. The expression for the multi-energy flow topology constraint matrix is: , Represents grid nodes With nodes Between Quantization values ​​of topological constraints for energy flow-like structures; Represents the adjacency discrimination coefficient, when node and There exists a first When physical connections are made for energy flow, When there is no connection, Indicates the first Energy flow in nodes and The transmission capacity limit between; Represents a node and Between The number of segments in a power flow transmission path; Indicates the first Transmission efficiency of segment paths; It is the first Energy flow from nodes arrive The unit vector of the transmission direction; It is a node and Between Potential gradient vector similar to energy flow; It is a node and The three-dimensional Euclidean distance; It is the first The distance attenuation factor of energy flow varies depending on the type of energy flow. Represents partial derivative terms; It is the first Energy flow in nodes and Transmission loss function between; It is the first The transmission angle parameter of energy flow; Represents a node remove Other than The set of adjacent nodes of the energy flow class; It is a node Indirect coupling coefficient between them; express Inter-transmission capacity; express Inter-transmission capacity, the ratio of the two reflects Other connections to it Connection capacity preemption effect; The energy prediction map construction module is used to dynamically adjust grid-level energy dynamic parameters by combining a multi-energy flow topological constraint matrix, and determine the grid-level energy dynamic prediction map. Specifically, it includes: calling a spatial granularity prediction model; dividing the entire region into fine-grained geographic grids based on energy node distribution and load density using a geographic information system; extracting the latest real-time source data, historical similar time period data, and environmental correlation data from a standard source database and inputting them into the model; the model captures short-term dynamic trends through temporal convolutional layers, strengthens the collaborative characteristics of topologically related grids through graph convolutional layers, and focuses the influence weights of key grids and time periods through a spatiotemporal attention mechanism; verifying the physical rationality of energy parameters in each grid by combining the multi-energy flow topological constraint matrix, and generating grid-level energy dynamic parameters for future multiple time periods; finally, integrating the prediction results of each time period and each grid into a visualized energy dynamic prediction map, which includes continuous spatiotemporal change trends and extreme value warnings for key nodes. The optimal trade-off set generation module is used to find the optimal trade-off solution set based on the grid-level energy dynamic prediction map using a population aggregation and dispersion search algorithm. The energy scheduling optimal solution generation module is used to combine the safe operation boundary to perform constraint solving on the best trade-off solution set and determine the optimal energy scheduling solution.

7. A computer-readable storage medium, characterized in that, It includes one or more program instructions, which are executed by a processor as described in any one of claims 1-5, to provide an energy supply-side optimization management method.

Images