Novel flexible carbon reduction resource allocation optimization method for power distribution and utilization system

Abstract

The invention discloses a novel flexible carbon reduction resource allocation optimization method for a power distribution and utilization system, and relates to the technical field of power distribution and utilization, and the method comprises the steps: collecting the multi-source data of the power distribution and utilization system in real time; wherein the multi-source data comprises energy attribute data, load dynamic data and power grid interaction data; deep analysis is carried out on the multi-source data, a load dynamic influence coefficient and a power grid interaction coordination coefficient are obtained, operation characteristics of the power distribution and utilization system to be analyzed are extracted, and the operation characteristics comprise energy output, a load fluctuation rate and a resource dynamic response speed; presetting a carbon reduction optimization judgment condition, comparing the operation characteristics with the judgment condition to generate an optimization scheme, and outputting a resource configuration result; wherein the judgment conditions comprise a first condition for identifying a high carbon emission risk and a second condition for activating collaborative carbon reduction; according to the invention, the efficiency of carbon reduction resource allocation is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution, in particular to a novel flexible carbon reduction resource allocation optimization method for power distribution systems. BACKGROUND

[0002] The power distribution system is a key link in the power system responsible for power distribution and terminal user power supply, covering all chain facilities from high-voltage distribution networks to low-voltage user sides; its core function is to convert high-voltage power transmitted by the power transmission network into voltage levels suitable for user use, and ensure that power is safely, reliably and economically distributed to various loads; with the development of technology, the power distribution system urgently needs to achieve carbon reduction through flexible resource optimization allocation;

[0003] However, the traditional carbon reduction resource allocation flexible resource is scattered and independently operated, lacking a cross-level coordination mechanism, resulting in low renewable energy consumption and large fluctuations in carbon emissions; at the same time, the traditional method relies on fixed thresholds (such as carbon emission upper limit), which cannot adapt to the minute-level fluctuations of new energy; if the carbon flow of any grid node exceeds its carrying threshold, flexibility is reduced, and resource dynamic response speed is greatly reduced, resulting in that carbon reduction measures cannot be executed in time. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a novel flexible carbon reduction resource allocation optimization method for power distribution systems, which collects real-time energy attributes, load dynamics and grid interaction data, constructs multi-dimensional operation characteristics, combines load dynamic influence coefficients and grid interaction coordination coefficients generated by deep analysis, covers source-network-load-storage whole link data, and avoids the limitations of traditional single data sources; by dynamically quantifying the interaction between load and grid, the driving type is analyzed, and double judgment conditions are further preset, solving the problems raised in the background art.

[0005] To achieve the above purpose, the present application realizes the following technical solutions:

[0006] The present application provides a novel flexible carbon reduction resource allocation optimization method for power distribution systems, which comprises:

[0007] Real-time collection of multi-source data of the power distribution system; wherein the multi-source data includes energy attribute data, load dynamic data and grid interaction data;

[0008] Deep analysis of the multi-source data to obtain load dynamic influence coefficients and grid interaction coordination coefficients, extract the operation characteristics of the power distribution system to be analyzed, and the operation characteristics include energy output, load fluctuation rate and resource dynamic response speed;

[0009] The system presets carbon reduction optimization judgment conditions, compares the operating characteristics with the judgment conditions to generate optimization schemes, and outputs resource allocation results. The judgment conditions include a first condition for identifying high carbon emission risks and a second condition for activating synergistic carbon reduction.

[0010] Furthermore, energy attribute data includes, but is not limited to, energy type, energy location, and energy capacity; load dynamic data includes, but is not limited to, CPU utilization, bandwidth, and response latency; and grid interaction data includes, but is not limited to, node voltage deviation, power generation, and node carbon flow path.

[0011] Furthermore, in-depth analysis of multi-source data includes:

[0012] Load dynamic data and power grid interaction data are imported into a preset state analysis model, which outputs load dynamic influence coefficient and power grid interaction coordination coefficient. Based on the ratio of load dynamic influence coefficient and power grid interaction coordination coefficient, the driving type is analyzed.

[0013] Based on the mapping of driving type to monitoring nodes, the node pair relationship of power distribution and consumption is constructed, and the energy attribute data and node carbon flow path are matched and associated. An operation spatiotemporal coordinate system is established, and multidimensional spatiotemporal encoding is performed on energy location and node carbon flow path. Graph convolutional network is used to extract the topological relationship features between energy types. Energy attribute data is fused through attention mechanism to generate spatiotemporal constraint matrix and extract operation features. Among them, the monitoring nodes include power source side point, grid side point, load side point, and storage side point.

[0014] Furthermore, the design method of the preset state analysis model is as follows:

[0015] The input information is labeled as a parameter set, which contains several element values. Each element value is assigned a corresponding weight coefficient. The influence coefficient of the parameter set is output by multiplying the element value by the corresponding weight coefficient and summing the results.

[0016] Load dynamic data and power grid interaction data are imported into the state analysis model respectively, and load dynamic influence coefficient and power grid interaction coordination coefficient are generated accordingly.

[0017] Furthermore, preset carbon reduction optimization judgment conditions are established, and the operating characteristics are compared with the judgment conditions, including:

[0018] If the operating characteristics meet the first condition, then it is further determined whether the second condition is met. If the second condition is met, the first resource allocation result is output; if the second condition is not met, the emergency resource allocation result is output.

[0019] If the operating characteristics do not meet the first condition, the second resource allocation result is output; if the second resource allocation result is output, the carbon emission constraint mechanism is triggered and the resource scheduling strategy is updated.

[0020] Furthermore, the execution steps of the first condition include:

[0021] Within a preset time period, the operating characteristics under several time series are obtained. For any time series, a feature space is established with energy output as the X-axis, load fluctuation rate as the Y-axis, and resource dynamic response speed as the Z-axis. A standard feature space is then drawn on the feature space.

[0022] The current feature space is superimposed with the standard feature space to obtain the volume offset. The gradient range of the volume offset is set. If the volume offset is within the gradient range, it is determined that the first condition is met, and it is determined whether the second condition is met. If the volume offset exceeds the gradient range, it is determined that the second condition is not met, and the second resource allocation result is output. The second resource allocation result indicates that there is a high carbon emission risk.

[0023] Furthermore, the correlation between energy types and carbon flow paths is retrieved to construct the collaborative directions of different monitoring nodes, including the source-grid direction, the grid-load direction, and the load-storage direction.

[0024] Extract the average and fluctuation values ​​of the corresponding operating characteristics in all collaborative directions, construct several judgment vectors based on the average and fluctuation values, input any judgment vector into the preset judgment model, and determine whether it meets the third condition. If it does, mark it as a risk.

[0025] Collect the number of times risks occur under all collaborative directions, determine carbon reduction priorities based on the number of occurrences, with higher occurrences indicating higher priorities; if no risk is found, continue to analyze and monitor the nodes.

[0026] This invention provides a novel method for optimizing resource allocation in power distribution systems to reduce carbon emissions and improve flexibility, which has the following advantages:

[0027] This invention constructs multi-dimensional operational characteristics by collecting real-time energy attributes, load dynamics, and grid interaction data. Combined with load dynamic impact coefficients and grid interaction coordination coefficients generated through in-depth analysis, it covers data from the entire process of source-grid-load-storage, avoiding the limitations of traditional single data sources. By dynamically quantifying the interaction between load and grid, it analyzes the dominant driving type and further presets dual judgment conditions, including the first condition: high carbon emission risk identification; and the second condition: collaborative carbon reduction activation. Based on feature comparison, it generates optimized solutions, adapts to complex scenarios, improves resource utilization efficiency, accelerates carbon reduction configuration, and enhances flexibility. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0030] Example:

[0031] This invention provides a novel method for optimizing resource allocation to reduce carbon emissions in power distribution systems. Figure 1 This is a schematic diagram of the steps of the present invention; please refer to it. Figure 1 The method includes the following steps:

[0032] Deploy IoT monitoring terminals to collect multi-source data from the power distribution system in real time; the multi-source data includes energy attribute data, load dynamic data, and power grid interaction data.

[0033] Energy attribute data includes, but is not limited to, energy type, energy location, and energy capacity;

[0034] Energy type: refers to the form of energy accessed in the power distribution system, including distributed new energy sources (such as photovoltaic and wind power), energy storage devices (such as lithium batteries and lead-acid batteries), traditional power sources (such as thermal power and hydropower) and other auxiliary energy sources (such as biomass energy);

[0035] Data collection method: For new energy equipment, the equipment ledger information (e.g., photovoltaic inverter model GW1500, energy storage battery pack type LFP) is entered through the IoT terminal during the connection process, and the type is automatically identified by associating a unique device ID; for existing equipment, the type identifier in the device communication protocol (e.g., the device type register in the Modbus protocol) is read through the edge computing node, and automatically matched with the preset energy type coding library (e.g., 1-photovoltaic, 2-wind power, 3-energy storage).

[0036] Energy location: The physical installation location of energy equipment, used to locate the energy access point and its association with the power grid topology;

[0037] Data collection method: GPS or Beidou positioning module is integrated in the equipment control box to upload latitude and longitude coordinates in real time. Edge nodes map the coordinates to the corresponding transformer area or line on the power grid GIS map. Location information (such as 220kV XX substation, 10kV XX line photovoltaic grid connection point) is associated through the preset power grid topology ledger, and accurate positioning is achieved by combining the tower number and cabinet number.

[0038] Energy capacity: The rated output capacity and real-time available capacity of energy equipment, including rated power, current actual output, and remaining energy storage capacity;

[0039] Data acquisition method: Data is acquired via a power sensor;

[0040] Load dynamic data includes, but is not limited to, CPU utilization, bandwidth, and response latency;

[0041] CPU utilization: refers to the real-time occupancy rate of the central processing unit (CPU) of industrial control equipment and intelligent terminals (such as power distribution automation terminals), reflecting the computing load pressure of the equipment;

[0042] Data collection method: Deploy edge computing gateways in industrial user power distribution rooms and power distribution network automation cabinets, and connect them to the operating systems of industrial control computers and smart terminals via Ethernet interfaces; the gateway periodically sends instructions to read the CPU utilization register of the devices, filters invalid values, and then uploads them;

[0043] Bandwidth: refers to the real-time data transmission rate of the load-side communication network, reflecting the communication load between the load and the power grid dispatching system and cloud platform;

[0044] Data collection method: Deploy network traffic probes on the mirror port of the load-side communication switch to capture bidirectional data packets in real time; the probes use deep packet inspection (DPI) technology to parse packet length and timestamps and calculate bandwidth utilization per unit time (bandwidth = total packet size / collection period, unit Mbps);

[0045] Response delay: refers to the delay in the execution of power grid dispatch instructions by load equipment, that is, the time difference from the issuance of the instruction to the change of equipment status, reflecting the flexibility of load regulation;

[0046] Data collection method: The distribution network dispatching platform periodically sends test commands to the load equipment, and the commands carry precise timestamps; after receiving the commands, the load equipment records the reception time through its built-in IoT terminal, and feeds back the status change time after execution; response delay = status change time - command issuance time, which is automatically calculated and uploaded by the terminal;

[0047] Grid interaction data includes, but is not limited to, node voltage deviation, power generation, and carbon flow paths;

[0048] Node voltage deviation: refers to the percentage deviation between the actual voltage of a grid node and its rated voltage, used to assess the stability of grid operation;

[0049] Data acquisition method: Real-time voltage is acquired through a voltage transformer. The calculation method is: Voltage deviation = |Actual voltage - Rated voltage| / Rated voltage × 100%;

[0050] Power generation: refers to the electrical energy output of each power node (such as photovoltaic power station, thermal power plant, distributed generator) in the power distribution system per unit time, including instantaneous power generation (kW) and cumulative power generation (kWh), which is measured by power sensors;

[0051] Nodal carbon flow path: refers to the flow trajectory of carbon emissions during the transmission of electrical energy from power generation nodes to load nodes, used to quantify the carbon emission contribution of each node;

[0052] Deep analysis of multi-source data yields load dynamic impact coefficients and power grid interaction coordination coefficients, and extracts the operating characteristics of the power distribution system to be analyzed.

[0053] In-depth analysis of multi-source data, including:

[0054] Load dynamic data and grid interaction data are imported into a preset state analysis model, which outputs load dynamic influence coefficient and grid interaction coordination coefficient. Based on the ratio of load dynamic influence coefficient and grid interaction coordination coefficient, the driving type is analyzed. Among them, load dynamic influence coefficient is used to evaluate the comprehensive impact of load on system stability and response efficiency, and grid interaction coordination coefficient is used to evaluate the electrical coordination and low-carbon performance of grid nodes.

[0055] The design method of the preset state analysis model is as follows:

[0056] The input information is labeled as parameter set U, which contains Q element values. Any element value is labeled as Xs. Then, the Q element values ​​are assigned corresponding weight coefficients φe. By multiplying the element value by the corresponding weight coefficient and summing the results, the influence coefficient Ue of parameter set U is output: Ue=∑(Xs*Φe).

[0057] It should be noted that the assigned weight coefficients were obtained during training based on the dominant driver type.

[0058] Load dynamic data and grid interaction data are imported into the state analysis model to generate load dynamic influence coefficient U1 and grid interaction coordination coefficient U2.

[0059] The load dynamic impact coefficient U1 = CPU utilization * first weight + bandwidth * second weight + response delay * third weight; it quantifies the dynamic response capability of the load side (such as adjustable load, energy storage control system) and is used to determine whether load resources have the potential to quickly participate in carbon reduction regulation.

[0060] The grid interaction coordination coefficient U2 = voltage deviation * fourth weight + power generation * fifth weight; the matching degree between real-time power generation and demand affects carbon intensity (e.g., excessive generation of gas turbines leads to a surge in carbon emissions);

[0061] The load dynamic impact coefficient and the grid interaction coordination coefficient are used to assess the grid operation status and energy supply and demand balance level, and to make decisions on whether cross-regional coordinated carbon reduction is needed.

[0062] Based on the ratio of the load dynamic influence coefficient and the power grid interaction coordination coefficient, the ratio and the ratio interval [bz] are used to... min bz max Comparison to determine the dominant driver type, including:

[0063] If the ratio is greater than or equal to bz max Marked as load-driven, this indicates that the load dynamic influence coefficient is much greater than the grid interaction coordination coefficient, suggesting that the system risk is mainly driven by excessive load demand, and the risk needs to be controlled by reducing the load requirements on the system.

[0064] If the ratio is less than or equal to bz max And the ratio is greater than bz min This is marked as grid-limited drive, indicating that the load demand is close to the grid's carrying capacity but not completely unbalanced. The risk is mainly caused by insufficient grid carrying capacity (e.g., low grid voltage stability, excessive carbon flow), which needs to be mitigated by improving the grid interaction coordination coefficient.

[0065] If the ratio is less than or equal to bz min The term "cooperative adaptation type" indicates that the load demand and the grid carrying capacity are well matched, the risk is low, and the dominant type of drive is the cooperative adaptation between the load and the grid, which can maintain normal operation.

[0066] Before extracting runtime features, multi-source data preprocessing is also performed, including:

[0067] Outlier handling: The isolated forest algorithm is used to identify outlier data. For continuous data, cubic spline interpolation is used for repair, and for discrete data, a state transition probability model based on Markov chains is used for filling in the gaps.

[0068] Spatiotemporal alignment: Based on timestamp synchronization technology, data with different sampling frequencies are unified to the same time scale, and multi-source data are associated with monitoring nodes through spatial coordinate mapping;

[0069] Based on the mapping of driving type to monitoring nodes, a node-to-node relationship for power distribution is constructed, and energy attribute data and node carbon flow paths are matched and associated. An operational spatiotemporal coordinate system is established, and multidimensional spatiotemporal encoding is performed on energy locations and node carbon flow paths. A graph convolutional network is used to extract topological relationship features between energy types, and energy attribute data is fused through an attention mechanism to generate a spatiotemporal constraint matrix and extract operational features. Among them, monitoring nodes include power source side points, grid side points, load side points, and storage side points.

[0070] Establish the spatiotemporal coordinate system for the operation, including:

[0071] The X-axis is the direction along the main transmission line (such as the east-west direction of the UHVDC line), the Y-axis is the direction perpendicular to the main line (north-south direction), the Z-axis is the normalized voltage, and the time dimension T is introduced to form a four-dimensional operation time-space coordinate system.

[0072] Energy Location Code: C node =Hash(E type ||X bus Y bus Z volt ||T sync In the formula, C node Indicates energy location code, E type Indicates the type of goods, (X) bus Y bus () indicates the location of the goods, which is the center coordinate of the bottom surface of the goods, Z. volt T represents voltage. sync Represents a timestamp;

[0073] Node carbon flow path coding: Discretize and sample the operational trajectories of equipment at power source, grid, load, and storage points to generate a spatiotemporal sequence, represented by: S traj ={(S type ||X t Y t Z t T t )}, and t=1、2……N; where S traj S represents the node carbon flow path encoding. type Indicates the monitoring node type, (X) t ,Y t Z represents the latitude and longitude coordinates at a certain sampling time. t T represents elevation. t This indicates a specific sampling time, and the sampling interval is dynamically adjusted according to the device type. It can be adjusted based on the PID control algorithm. The specific implementation method will not be described in detail.

[0074] Based on the driving force type, a regression model is built within each driving force type interval to determine the corresponding training set. An appropriate training algorithm is then used to train the weight coefficients, including:

[0075] If the load-driven model is used, R1 (load dynamic data, ratio) is used for weight training, and the gradient descent algorithm is used for weight training.

[0076] Specifically, the input load dynamic data includes CPU utilization (normalized to 0-1), bandwidth utilization (normalized to 0-1), and response latency (normalized to 0-1); the target variable is the ratio, which is standardized (corresponding to the high interval); by learning the correlation between load dynamic data and high ratios R, the weight coefficients of load features on risk are trained, quantifying the contribution of each load feature to high risk; gradient descent has a strong fitting ability for high-dimensional features and can quickly converge to the optimal weights, making it suitable for scenarios with a strong correlation between load dynamic data and high risk; the larger the weight value, the more significant the impact of the load feature on the high ratio R;

[0077] If it is a grid-constrained drive, use R2 (grid interaction data, ratio) for weight training, and use the Lasso regression algorithm for weight training;

[0078] Specifically, the input grid interaction data includes node voltage deviation (standardized to 0-0.05), power generation (standardized to 0-1), and node carbon flow path (standardized to 0-1); the target variable is the ratio, which is standardized and corresponds to the median interval. By learning the correlation between grid interaction data and the median ratio R, the weight coefficients of grid characteristics on risk are trained to identify key factors restricting grid carrying capacity. Lasso regression, through L1 regularization, makes some weight coefficients approach 0 (sparseness), which can automatically filter out grid characteristics (such as voltage deviation) that have a significant impact on risk and remove redundant features.

[0079] If it is a collaborative adaptation type of drive, R(3 load dynamic data, grid interaction data, ratio) is used for weight training, and the weight training adopts the ridge regression algorithm.

[0080] The training process incorporates fused load dynamics data and grid interaction data, including both original features (CPU, voltage deviation, etc.) and cross-features (e.g., CPU utilization × voltage deviation, bandwidth × carbon flow), totaling 6-8 dimensions. The target variable is the ratio, which is standardized and corresponds to the low-range. By learning the impact of the synergistic relationship between load and grid on the low ratio R, the weight coefficients of the fused features are trained, quantifying the supporting role of the compatibility between the two in low-risk scenarios. Ridge regression alleviates multicollinearity of the fused features (e.g., the correlation between load and grid features) through L2 regularization, ensuring stable weight coefficients and balancing the synergistic effects of the two.

[0081] It should be noted that R(·) is a simplified expression of a scenario-based regression model. By clarifying the input features under different scenarios, the training of weight coefficients can be more focused on the core influencing factors. The mathematical logic of the regression model itself (solving the weights by fitting the correlation between features and the target) determines that it can be effectively used for training weight coefficients.

[0082] The system presets carbon reduction optimization judgment conditions, compares the operating characteristics with the judgment conditions to generate optimization schemes, and outputs resource allocation results; among them, the judgment conditions include a first condition for identifying high carbon emission risks and a second condition for activating synergistic carbon reduction mechanisms.

[0083] Preset carbon reduction optimization judgment conditions, and compare the operating characteristics with the judgment conditions, including:

[0084] If the operating characteristics meet the first condition, then it is further determined whether the second condition is met. If the second condition is met, the first resource allocation result is output; if the second condition is not met, the emergency resource allocation result is output.

[0085] If the operating characteristics do not meet the first condition, the second resource allocation result is output; if the second resource allocation result is output, the carbon emission constraint mechanism is triggered and the resource scheduling strategy is updated.

[0086] The steps to fulfill the first condition include:

[0087] Within a preset time period, operational characteristics under several time series are acquired. For any given time series, a feature space is established with energy output as the X-axis, load fluctuation rate as the Y-axis, and resource dynamic response speed as the Z-axis. A standard feature space is drawn on the feature space. The current feature space and the standard feature space are superimposed to obtain overlapping and non-overlapping volumes. The volume offset is obtained by adding the overlapping and non-overlapping volumes and dividing the overlapping volume by the sum. A gradient range for the volume offset is set. If the volume offset is within the gradient range, the first condition is satisfied, and the second condition is also determined. If the volume offset exceeds the gradient range, the second condition is not satisfied, and a second resource allocation result is output. The second resource allocation result indicates a high carbon emission risk.

[0088] Trigger the carbon emission constraint mechanism and update the resource scheduling strategy: record the proportion of the second resource allocation result and the number of resource scheduling within a preset time period, and use the Pareto algorithm to solve the power supply by minimizing the proportion and minimizing the number of resource scheduling as constraints, combined with the driving type.

[0089] The steps to execute the second condition include:

[0090] The correlation between energy types and carbon flow paths is retrieved to construct the collaborative direction of different monitoring nodes, including the source-grid direction, the grid-load direction, and the load-storage direction.

[0091] Extract the average value and fluctuation value of the corresponding operating characteristics in all collaborative directions, construct several judgment vectors based on the average value and fluctuation value, input any judgment vector into the preset judgment model, and judge whether it meets the third condition. If it does, it is marked as a risk. There are 4 types of judgment vectors constructed under each collaborative direction, including: average value + fluctuation value, average value - fluctuation value, average value × fluctuation value, and average value ÷ fluctuation value.

[0092] Collect the number of times risks occur under all collaborative directions, determine carbon reduction priorities based on the number of occurrences, with higher occurrences indicating higher priorities; if the risk does not meet the criteria, continuously analyze and monitor the nodes.

[0093] The third condition is a preset judgment threshold. The output result is compared with the judgment threshold. If the output result is less than the judgment threshold, it is marked as meeting the third condition and marked as a risk.

[0094] This invention constructs multi-dimensional operational characteristics by collecting real-time energy attributes, load dynamics, and grid interaction data. Combined with load dynamic impact coefficients and grid interaction coordination coefficients generated through in-depth analysis, it covers data from the entire process of source-grid-load-storage, avoiding the limitations of traditional single data sources. By dynamically quantifying the interaction between load and grid, it analyzes the dominant driving type and further presets dual judgment conditions, including the first condition: high carbon emission risk identification; and the second condition: collaborative carbon reduction activation. Based on feature comparison, it generates optimization schemes to adapt to complex scenarios and improve resource utilization efficiency.

[0095] In the application, the various formulas mentioned are all calculated by removing dimensions and taking their numerical values. The formulas are derived from the most recent real-world situation by collecting a large amount of data and simulating it with software. The formulas are set by those skilled in the art according to the actual situation.

[0096] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0097] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A novel method for optimizing resource allocation in power distribution systems to reduce carbon emissions, characterized in that: The method includes: Real-time acquisition of multi-source data from the power distribution system; including energy attribute data, load dynamic data, and grid interaction data; Deep analysis of multi-source data is conducted to obtain the load dynamic impact coefficient and the power grid interaction coordination coefficient, and to extract the operating characteristics of the power distribution system to be analyzed, including energy output, load fluctuation rate and resource dynamic response speed. The system presets carbon reduction optimization judgment conditions, compares the operating characteristics with the judgment conditions to generate optimization schemes, and outputs resource allocation results. The judgment conditions include a first condition for identifying high carbon emission risks and a second condition for activating synergistic carbon reduction.

2. The novel method for optimizing resource allocation in a power distribution system to reduce carbon emissions and improve flexibility, as described in claim 1, is characterized in that: The energy attribute data includes, but is not limited to, energy type, energy location, and energy capacity; The load dynamic data includes, but is not limited to, CPU utilization, bandwidth, and response latency. The grid interaction data includes, but is not limited to, node voltage deviation, power generation, and node carbon flow paths.

3. The novel method for optimizing resource allocation in a power distribution system to reduce carbon emissions and improve flexibility, as described in claim 1, is characterized in that... The in-depth analysis of multi-source data includes: Load dynamic data and power grid interaction data are imported into a preset state analysis model, which outputs load dynamic influence coefficient and power grid interaction coordination coefficient. Based on the ratio of load dynamic influence coefficient and power grid interaction coordination coefficient, the driving type is analyzed. Based on the mapping of driving type to monitoring nodes, the node pair relationship of power distribution and consumption is constructed, and the energy attribute data and node carbon flow path are matched and associated. An operation spatiotemporal coordinate system is established, and multidimensional spatiotemporal encoding is performed on energy location and node carbon flow path. Graph convolutional network is used to extract the topological relationship features between energy types. Energy attribute data is fused through attention mechanism to generate spatiotemporal constraint matrix and extract operation features. Among them, the monitoring nodes include power source side point, grid side point, load side point, and storage side point.

4. A novel method for optimizing resource allocation in a power distribution system to reduce carbon emissions and improve flexibility, as described in claim 3, is characterized in that... The design method of the preset state analysis model is as follows: The input information is labeled as a parameter set, which contains several element values. Each element value is assigned a corresponding weight coefficient. The influence coefficient of the parameter set is output by multiplying the element value by the corresponding weight coefficient and summing the results. Load dynamic data and power grid interaction data are imported into the state analysis model respectively, and load dynamic influence coefficient and power grid interaction coordination coefficient are generated accordingly.

5. A novel method for optimizing resource allocation in a power distribution system to reduce carbon emissions and improve flexibility, as described in claim 1, is characterized in that... Based on the driving type, a regression model is established within each driving type interval to determine the corresponding training set, and an appropriate training algorithm is called to train the weight coefficients.

6. The novel method for optimizing resource allocation in a power distribution system to reduce carbon emissions and improve flexibility, as described in claim 1, is characterized in that... The preset carbon reduction optimization judgment conditions compare the operating characteristics with the judgment conditions, including: If the operating characteristics meet the first condition, then it is further determined whether the second condition is met. If the second condition is met, the first resource allocation result is output; if the second condition is not met, the emergency resource allocation result is output. If the operating characteristics do not meet the first condition, the second resource allocation result is output; if the second resource allocation result is output, the carbon emission constraint mechanism is triggered and the resource scheduling strategy is updated.

7. A novel method for optimizing resource allocation in a power distribution system to reduce carbon emissions and improve flexibility, as described in claim 1, is characterized in that... The execution steps of the first condition include: Within a preset time period, the operating characteristics under several time series are obtained. For any time series, a feature space is established with energy output as the X-axis, load fluctuation rate as the Y-axis, and resource dynamic response speed as the Z-axis. A standard feature space is then drawn on the feature space. The current feature space is superimposed with the standard feature space to obtain the volume offset. The gradient range of the volume offset is set. If the volume offset is within the gradient range, it is determined that the first condition is met, and it is determined whether the second condition is met. If the volume offset exceeds the gradient range, it is determined that the second condition is not met, and the second resource allocation result is output. The second resource allocation result indicates that there is a high carbon emission risk.

8. A novel method for optimizing resource allocation in a power distribution system to reduce carbon emissions and improve flexibility, as described in claim 1, is characterized in that... The execution steps of the second condition include: The correlation between energy types and carbon flow paths is retrieved to construct the collaborative direction of different monitoring nodes, including the source-grid direction, the grid-load direction, and the load-storage direction. Extract the average and fluctuation values ​​of the corresponding operating characteristics in all collaborative directions, construct several judgment vectors based on the average and fluctuation values, input any judgment vector into the preset judgment model, and determine whether it meets the third condition. If it does, mark it as a risk. Collect the number of times risks occur under all collaborative directions, determine carbon reduction priorities based on the number of occurrences, with higher occurrences indicating higher priorities; if no risk is found, continue to analyze and monitor the nodes.

Images