Dual-carbon power system optimization scheduling method

Abstract

The invention discloses a dual-carbon power system optimal scheduling method, which comprises the following steps of: acquiring carbon emission parameters and power flow data in a topological structure of a power system, so as to facilitate subsequent splitting and extraction. The method comprises the following steps: splitting all obtained carbon emission parameters and power flow data into multiple sections of data through a preset numerical range to avoid data disorder, and marking each section of data to form a first carbon value interval; next, abnormal values in each first carbon value interval are filtered, leaving values that can be analyzed, and the marking forms a second carbon value interval. Meanwhile, the carbon value of the second carbon value interval corresponds to a power generation side unit in the power system topological structure, the reward strength is evaluated according to the carbon value, and finally, different power generation side units are dispatched in the power system topological structure to distribute power of the whole network by adopting a preset dispatching model on the basis of the reward strength. Carbon emission flows of different power supplies and loads are analyzed, and low-carbon distribution of power generation resources is realized by taking reduction of carbon emission of the whole network as a target.

Description

Technical Field

[0001] This invention relates to the field of environmentally friendly power technology, specifically to an optimized scheduling method for a dual-carbon power system. Background Technology

[0002] With increasing global emphasis on environmental protection and sustainable development, intelligent control technology for power systems, as a key area of ​​energy consumption, is of great significance for promoting energy structure transformation, improving utilization efficiency, and reducing carbon emissions.

[0003] The fundamental path to achieving this goal lies in: significantly increasing the proportion of zero-carbon energy generation such as wind power and photovoltaic power, relying on ultra-high voltage power transmission technology to achieve cross-regional energy complementarity, and using advanced energy storage and intelligent dispatch to smooth out the volatility of new energy sources, thereby replacing fossil energy at the source and reducing or even eliminating carbon emissions.

[0004] However, although building a new power system based on new energy sources is an important measure to achieve dual carbon goals, the current power system still cannot monitor carbon emission data from the generation side to the user side in real time. The dispatch center has difficulty identifying the carbon emission sources of the load and cannot assess the marginal carbon emission increment generated when the generation unit is called. As a result, the power system cannot actively guide the power flow to the path with the lowest carbon emissions, which ultimately restricts the accurate low-carbon allocation of power generation resources. Summary of the Invention

[0005] The purpose of this invention is to provide an optimized scheduling method for a dual-carbon power system, which analyzes the carbon emissions of different power sources and loads, and aims to reduce the carbon emissions of the entire grid, thereby achieving a low-carbon allocation of power generation resources.

[0006] The technical solution of this invention is implemented as follows:

[0007] A method for optimizing the dispatch of a dual-carbon power system includes the following steps:

[0008] Step S1: Based on the power system topology, obtain carbon emission parameters and power flow data;

[0009] Step S2: Divide the carbon emission parameters and tidal data into multiple first carbon value intervals through a preset numerical range, and extract outliers in the first carbon value intervals and recombine them to generate multiple second carbon value intervals.

[0010] Step S3: Based on the generation-side unit of the power system topology corresponding to each carbon value in the second carbon value range, evaluate the reward intensity of the generation-side unit through an optimization mechanism;

[0011] Step S4: Based on the magnitude of the reward, a preset scheduling model is used to schedule different generation-side units in the power system topology to allocate the power of the entire network.

[0012] A further technical solution is that step S1 specifically includes:

[0013] Step S11: Based on the power system topology, determine the set of power generation nodes and the set of load nodes in the network, and establish the connection relationship between the set of power generation nodes and the set of load nodes;

[0014] Step S12: Obtain the carbon emission parameters of each generator unit in the set of power generation nodes. The carbon emission parameters include at least the unit type, fuel type, and carbon emission intensity per unit of power generation.

[0015] Step S13: Obtain power flow data based on the power grid during a specific operating period. The power flow data includes at least the injected power of each generator set and the power flow distribution of each branch.

[0016] Step S14: Associate the carbon emission parameters of the power generation node with the injected power of the corresponding generator set, and map the power flow data with the connection relationship to form a complete dataset for carbon flow analysis.

[0017] A further technical solution is that step S14 specifically includes:

[0018] Step S141: Correlate the carbon emission parameters of the power generation node with the injected power of the corresponding generator set, and calculate the carbon emission of each power generation node during the operating period.

[0019] Step S142: Map the branch power flow distribution in the power flow data to the connection relationship to determine the transmission path and distribution of power flow in the power system topology.

[0020] A further technical solution is that step S2 specifically includes:

[0021] Step S21: Determine the numerical range of the carbon emission parameters and tidal flow data, and set preset values ​​to define the boundaries of the carbon emission parameters and tidal flow data;

[0022] Step S22: Based on the preset values, the carbon emission parameters and tidal flow data are divided into multiple consecutive first carbon value intervals according to their numerical values;

[0023] Step S23: An anomaly propagation detection algorithm based on association consensus is used to identify and extract abnormal data points from the first carbon value interval;

[0024] Step S24: Collect the extracted abnormal data points and recombine them according to their numerical characteristics to form an abnormal data set;

[0025] Step S25: Define the abnormal data set together with the remaining first carbon value intervals as multiple second carbon value intervals.

[0026] A further technical solution is that step S22 specifically includes:

[0027] Step S231: Treat each data point in the first carbon value interval as a network node, and establish connection edges between the network nodes based on the temporal or numerical proximity of the data points.

[0028] Step S232: Traverse each of the connection edges, compare the values ​​of the two network nodes connected by the connection edge once, and record the logical relationship of the comparison level as the initial consensus label of the connection edge;

[0029] Step S233: Based on the initial consensus label, the consensus relationship is iteratively propagated along the connection edge, wherein, during the propagation process, it is detected whether the newly derived consensus relationship conflicts with the existing consensus relationship.

[0030] Step S234: During the propagation of the consensus relationship, network nodes that cannot maintain consistency with the mainstream consensus relationship, as well as network nodes located at the center of logical conflicts, are identified as abnormal data points and extracted.

[0031] A further technical solution is that step S233 specifically includes:

[0032] Step S2331: In the associated network, select a propagation path consisting of at least two connecting edges connected end to end, wherein the path contains at least three consecutive nodes;

[0033] Step S2332: Based on the known initial consensus labels on the propagation path, deduce the indirect consensus relationship between adjacent nodes through logical transitivity;

[0034] Step S2333: Compare the newly derived indirect consensus relationship with the existing direct consensus relationship in the network. If they are inconsistent, they are identified as logical conflicts.

[0035] A further technical solution is that step S3 specifically includes:

[0036] Step S31: Map each carbon value in the second carbon value range to the power generation unit that generates the carbon value in the power system topology;

[0037] Step S32: Configure a differentiated evaluation function based on the classification characteristics of the second carbon value range, wherein the evaluation function takes the carbon emission intensity and output level of the power generation unit as input and outputs its reward score.

[0038] Step S33: Based on the reward score calculated by the evaluation function, sort the power generation units belonging to different second carbon value intervals, and allocate differentiated reward levels according to the sorting results, with low-carbon and high-efficiency units receiving higher rewards.

[0039] A further technical solution is that step S32 specifically includes:

[0040] Step S321: Based on the classification characteristics of the second carbon value range, set a first evaluation strategy and a second evaluation strategy;

[0041] Step S322: Based on the first evaluation strategy, construct a first evaluation function whose output value monotonically increases as carbon emission intensity decreases and power output level increases; based on the second evaluation strategy, construct a second evaluation function to smooth out the impact of outliers or identify their potential value.

[0042] Step S323: Normalize the two input parameters, carbon emission intensity and power output level, of the power generation unit.

[0043] Step S324: Input the normalized carbon emission intensity and power output level parameters into the corresponding first evaluation function or second evaluation function to calculate the reward score for each power generation unit.

[0044] A further technical solution is that step S4 specifically includes:

[0045] Step S41: Quantify the reward intensity into scheduling priority weights;

[0046] Step S42: Combine the scheduling priority weights with the network security constraints and power balance constraints of the power system topology to calculate the optimal output plan for each of the generation-side units;

[0047] Step S43: According to the optimal output plan, issue a dispatching instruction to the corresponding power generation unit to implement power distribution throughout the entire network.

[0048] A further technical solution is that step S42 specifically includes:

[0049] Step S421: Integrate the scheduling priority weights into an optimization objective function, and mathematically model the network security constraints and power balance constraints to form an optimized scheduling model;

[0050] Step S422: Apply the mixed integer programming algorithm to numerically solve the optimization scheduling model, calculate and output the active power output setting values ​​of each generation-side unit that make the optimization objective function optimal under the network security constraints and power balance constraints.

[0051] The beneficial effects of this invention are as follows:

[0052] Within the overall power system topology, acquiring emission parameters and power flow data facilitates subsequent segmentation and extraction. All acquired emission parameters and power flow data are divided into multiple segments based on preset numerical ranges to avoid data clutter, and each segment is labeled to form a first carbon value interval. Next, outliers within each first carbon value interval are filtered, leaving only analyzable values, which are then labeled to form a second carbon value interval. Simultaneously, the carbon values ​​of the second carbon value intervals are mapped to generation-side units within the power system topology. The incentive level is assessed based on the carbon value, and finally, a preset scheduling model is used to allocate power across the entire network to different generation-side units within the power system topology based on the incentive level. This allows for the analysis of carbon emission flows from different power sources and loads, aiming to reduce overall network carbon emissions and achieve low-carbon allocation of power generation resources. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the overall method steps of the present invention;

[0054] Figure 2 This is a flowchart of the method steps in step S1 of the present invention;

[0055] Figure 3 This is a flowchart of the method steps for step S2 of the present invention;

[0056] Figure 4 This is a flowchart of the method steps in step S3 of the present invention;

[0057] Figure 5 This is a flowchart of the method steps for step S4 of the present invention. Detailed Implementation

[0058] To better understand the technical content of this invention, specific embodiments are provided below, and the invention will be further described in conjunction with the accompanying drawings.

[0059] See Figures 1 to 5 This invention provides an optimized scheduling method for a dual-carbon power system, comprising:

[0060] Step S1: Based on the power system topology, obtain carbon emission parameters and power flow data;

[0061] Step S2: Divide the carbon emission parameters and tidal data into multiple first carbon value intervals through a preset numerical range, and extract outliers in the carbon value intervals and recombine them to generate multiple second carbon value intervals.

[0062] Step S3: Based on the generation-side units of the power system topology corresponding to each carbon value in the second carbon value interval, evaluate the reward intensity of the generation-side units through an optimization mechanism;

[0063] Step S4: Based on the magnitude of the reward, a preset scheduling model is used to schedule different generation-side units in the power system topology to allocate the power of the entire network.

[0064] It should be noted that the power system topology can be a skeleton or layout in which power plants, substations, transmission lines, distribution lines, and users are connected together through nodes and lines. Carbon emission parameters can be how much carbon dioxide a power plant produces per kilowatt-hour of electricity generated, while power flow data can be the direction, path, and magnitude of electricity flow in the power grid at any given time.

[0065] The generating side unit can be a power plant, substation, transmission line, distribution line, or user.

[0066] Optimization mechanisms include adopting The weighting function formula is calculated for the power generation unit, with the primary goal of reducing the carbon emissions of the entire grid, thereby directly converting environmental performance into its priority level in dispatching decisions.

[0067] The pre-defined scheduling model refers to a mathematical optimization model that uses the scheduling weight transformed by the reward intensity as the objective function and is constrained by the safe operation of the power grid. It directly embeds the goal of minimizing carbon emissions into the model's decision. In the power grid, the carbon flow tracking results are transformed into executable scheduling weights, enabling power grid scheduling to transform from a purely economic model into a multi-objective decision-making model that considers safety, carbon emissions, and economy.

[0068] Specifically, the system collects data on the actual grid topology, carbon emissions per unit of power generation from each generator unit, and real-time power flow to provide a foundation for subsequent analysis. Next, the carbon emission data and real-time power flow data are divided into a first carbon value interval, and outliers are identified and reorganized to form a second carbon value interval, thus achieving a detailed understanding of the grid's carbon emissions. Then, the processed carbon data is mapped back to specific generation units, and based on their carbon emission performance, a built-in optimization mechanism assesses and rewards them, with low-carbon, high-efficiency units receiving higher scores. Finally, this score is used as a core weight and input into a scheduling model that considers the constraints of grid safety and stability, automatically generating generation commands and prioritizing the use of high-scoring green power sources, ultimately minimizing the overall carbon emissions of the entire grid's power allocation.

[0069] In this embodiment of the invention, emission parameters and power flow data are acquired within the overall power system topology to facilitate subsequent segmentation and extraction. All acquired emission parameters and power flow data are segmented into multiple data segments based on a preset numerical range to avoid data clutter, and each segment is labeled to form a first carbon value interval. Next, outliers in each first carbon value interval are filtered, leaving only analyzable values, which are then labeled to form a second carbon value interval. Simultaneously, the carbon values ​​of the second carbon value interval are mapped to generation-side units within the power system topology. The incentive level is assessed based on the carbon value, and finally, a preset scheduling model is used to allocate power across the entire network to different generation-side units within the power system topology based on the incentive level. This allows for the analysis of carbon emission flows from different power sources and loads, aiming to reduce overall network carbon emissions and achieve low-carbon allocation of power generation resources.

[0070] Preferably, step S1 specifically includes:

[0071] Step S11: Based on the power system topology, determine the set of generating nodes and the set of load nodes in the network, and establish the connection relationship between the set of generating nodes and the set of load nodes;

[0072] Step S12: Obtain the carbon emission parameters of each generator unit in the set of power generation nodes. The carbon emission parameters include at least the unit type, fuel type, and carbon emission intensity per unit of power generation.

[0073] Step S13: Obtain power flow data based on the power grid during a specific operating period. The power flow data shall include at least the injected power of each generator set and the power flow distribution of each branch.

[0074] Step S14: Correlate the carbon emission parameters of the power generation nodes with the injected power of the corresponding generator sets, and map the power flow data with the connection relationships to form a complete dataset for carbon flow analysis.

[0075] In this embodiment of the invention, the locations of all power generation units and electrical loads, along with their electrical connection paths, are clearly defined, constructing a complete network model. Subsequently, two key data points are collected: the inherent carbon emission attributes of each generator unit, and the actual output of each generator at a specific moment, as well as the distribution flow of electrical energy along each line. Finally, through data association technology, the carbon emission attributes of each generator unit are bound to its actual output data, and the line power flow data is accurately mapped to the corresponding connection paths based on the network model. This forms a dataset integrating the power grid structure, carbon emission sources, and real-time operating status, suitable for in-depth analysis. By combining previously isolated static carbon emission parameters with dynamic power grid power flow data under a unified network model, a data foundation is laid for subsequent calculations of carbon flow distribution along the entire transmission path from generation to consumption, enabling precise source tracing and spatiotemporal positioning of power grid carbon emissions.

[0076] Furthermore, step S14 specifically includes:

[0077] Step S141: Correlate the carbon emission parameters of the power generation nodes with the injected power of the corresponding generator sets to calculate the carbon emissions of each power generation node during its operating period.

[0078] Step S142: Map the branch power flow distribution and connection relationship in the power flow data to determine the transmission path and distribution of power flow in the power system topology.

[0079] In this invention, the inherent carbon emission intensity parameter of each power generation node—that is, the carbon emission per unit of power generation—is multiplied by its actual injected power (i.e., power generation) within a specific time period. This transforms the abstract emission intensity into a specific and quantifiable total carbon emission of the node within that time period. Next, the collected branch power flow data (i.e., current magnitude and direction) is compared and matched with preset connection relationships in the power grid topology diagram. This reveals the complete transmission trajectory and spatial distribution of electricity flowing from the power generation node through specific lines to the load node. Through association and mapping operations, static carbon emission attributes are combined with dynamic power grid operation data. This not only quantifies the total direct emissions of each power source but also reveals the path of carbon emission diffusion in the power grid along with power flow, providing a data foundation for subsequent carbon flow analysis that traces carbon sources from loads.

[0080] Preferably, step S2 specifically includes:

[0081] Step S21: Determine the numerical range of carbon emission parameters and tidal flow data, and set preset values ​​to define the boundaries of carbon emission parameters and tidal flow data;

[0082] Step S22: Based on preset values, the carbon emission parameters and tidal flow data are divided into multiple consecutive first carbon value intervals according to their numerical values;

[0083] Step S23: An anomaly propagation detection algorithm based on association consensus identifies and extracts abnormal data points from each first carbon value interval;

[0084] Step S24: Collect the extracted abnormal data points and recombine them according to their numerical characteristics to form an abnormal data set.

[0085] Step S25: Define the abnormal data set and the remaining first carbon value intervals together as multiple second carbon value intervals.

[0086] It should be noted that the anomaly propagation detection algorithm of the associated consensus refers to the method of logically inferring and marking anomalies that violate consistency by comparing the relative size relationship between data points (such as A>B, B>C). The difference between it and the threshold-based or complex statistical model-based methods is that it does not require numerical calculations and can achieve rapid identification by relying solely on logical relationship propagation. In the dual-carbon application of the power grid, it can detect abnormal carbon emissions or power flow data in real time with extremely low computational overhead, providing a speed advantage for adjusting scheduling strategies and plugging carbon emission loopholes in the first instance.

[0087] In this embodiment of the invention, reasonable upper and lower limits are set for the carbon emission parameters and power flow data of the power grid by analyzing historical data or industry standards, thus clarifying the boundaries of normal data. Next, within this numerical range, all data is evenly divided into several continuous intervals in ascending order of magnitude, completing the initial classification. Then, an anomaly propagation detection algorithm based on association consensus is used to identify anomalous data points that significantly deviate from the group pattern by comparing the relationship between each data point within an interval and its neighboring points. Subsequently, these identified anomalous points are extracted from the original intervals and re-aggregated into an independent anomaly set based on their similar numerical characteristics. Finally, the original normal data intervals and this newly formed anomaly set together constitute the interval system. This achieves automatic cleaning and intelligent grouping of massive amounts of power grid operation data, enabling the separation of obviously erroneous or highly specific anomalous data from mainstream normal data for targeted processing. This improves the data quality upon which subsequent carbon flow analysis depends, avoids interference from outliers on the overall analysis results, and ensures that operating condition data is not simply discarded but can be specifically analyzed.

[0088] Furthermore, step S22 specifically includes:

[0089] Step S231: Treat each data point in the first carbon value interval as a network node, and establish connection edges between network nodes based on the temporal sequence or numerical proximity of the data points.

[0090] Step S232: Traverse each connection edge, compare the values ​​of the two network nodes connected by the connection edge once, and record the logical relationship of the comparison level as the initial consensus label of the connection edge.

[0091] Step S233: Based on the initial consensus label, the consensus relationship is iteratively propagated along the connection edge. During the propagation process, it is detected whether the newly derived consensus relationship conflicts with the existing consensus relationship.

[0092] Step S234: Identify and extract network nodes that cannot maintain consistency with the mainstream consensus relationship during the consensus relationship propagation process, as well as network nodes located at the center of logical conflicts, as abnormal data points.

[0093] In an embodiment of the present invention, the carbon emissions or power flow values to be analyzed are virtualized into a network, where each data point is a node, and they are connected according to their proximity in time or value to form a relational network. Then, a one-time numerical comparison is performed on each edge connecting two nodes in the network, and basic logical relationships such as "greater than" and "less than" are recorded as the initial labels of the edge. Next, the labeled logical relationships are propagated in the network and inferred along the connection paths. For example, if A > B and B > C, then it can be inferred that A > C, and during the propagation process, it is checked whether the newly inferred relationship conflicts with the direct relationships already existing in the network. Finally, individual nodes that cannot reach logical consistency with most nodes during the propagation process or are at the center of contradictions are determined as outliers and screened out. By simulating the propagation of logical relationships, the rapid and accurate identification of outliers in massive power grid data is achieved. Without relying on complex mathematical models and a large amount of calculations, the goal can be achieved only through simple logical comparisons and consistency checks, reducing the consumption of computing resources.

[0094] Further, step S233 specifically includes:

[0095] Step S2331: In the association network, select a propagation path formed by at least two connecting edges connected end to end, and the path includes at least three consecutive nodes;

[0096] Step S2332: According to the known initial consensus labels on the propagation path, deduce the indirect consensus relationship between adjacent nodes through logical transitivity;

[0097] Step S2333: Compare the newly deduced indirect consensus relationship with the direct consensus relationship already existing in the network. If they are inconsistent, it is identified as a logical conflict.

[0098] In an embodiment of the present invention, in the established association network, first select an inference path formed by at least three nodes connected in sequence by connecting edges. Then, use the known direct magnitude relationships (i.e., initial consensus labels) between adjacent nodes on the path to deduce through logical transitivity. For example, if there are relationships A > B and B > C in the path A - B - C, then the indirect relationship A > C can be deduced. Finally, compare and verify this newly deduced indirect relationship with the direct relationship already recorded between nodes A and C in the network; if the direct relationship is A < C, it conflicts with the deduced A > C, and this conflict is identified as a logical conflict. Through the logical chain reasoning and verification mechanism, the local inconsistencies hidden in the complex data network can be detected efficiently and reliably. The logical conflict indicates the location of the abnormal data, providing a criterion for the subsequent positioning and screening of outliers, thus ensuring the accuracy and interpretability of the anomaly detection results.

[0099] Preferably, step S3 specifically includes:

[0100] Step S31: Map each carbon value in the second carbon value interval to the generation-side unit that generates the carbon value in the power system topology.

[0101] Step S32: Based on the classification characteristics of the second carbon value range, configure a differentiated evaluation function, wherein the evaluation function takes the carbon emission intensity and output level of the power generation unit as input and outputs its reward score.

[0102] Step S33: Based on the reward score calculated by the evaluation function, sort the power generation units belonging to different second carbon value intervals, and allocate differentiated reward levels according to the sorting results, with low-carbon and high-efficiency units receiving higher rewards.

[0103] In this embodiment of the invention, firstly, each carbon emission data point in the preprocessed second carbon value interval is mapped to a specific power plant in the power grid, clarifying the source of each emission data point. Next, different evaluation functions are activated based on the different characteristics of these data intervals, such as whether they are normal or abnormal. The evaluation functions use the power plant's carbon emission intensity and actual power generation output as core scoring indicators to calculate reward scores. All power plants are ranked according to their reward scores, and different priority dispatch rights are allocated based on the ranking results, ensuring that power plants that are both clean (low-carbon) and efficient (high-output) receive the greatest power generation incentives. By creating a market environment that can automatically identify and strongly incentivize low-carbon and efficient units, the dual-carbon goals are directly transformed into concrete and operable decision-making criteria in the daily dispatch of the power grid, guiding the optimal allocation of power generation resources towards low-carbon development, and ultimately promoting a reduction in the overall carbon emission level of the power grid from its root causes.

[0104] Furthermore, step S32 specifically includes:

[0105] Step S321: Based on the classification characteristics of the second carbon value range, set the first evaluation strategy and the second evaluation strategy;

[0106] Step S322: Based on the first evaluation strategy, construct a first evaluation function in which the output value monotonically increases as carbon emission intensity decreases and power output level increases; based on the second evaluation strategy, construct a second evaluation function to smooth out the impact of outliers or identify their potential value.

[0107] Step S323: Normalize the two input parameters, carbon emission intensity and power output level, of the power generation unit.

[0108] Step S324: Input the normalized carbon emission intensity and power output level parameters into the corresponding first evaluation function or second evaluation function to calculate the reward score for each power generation unit.

[0109] In this embodiment of the invention, different evaluation strategies are set according to the classification characteristics of whether the second carbon value interval is a normal or abnormal interval. For data within the normal interval, a first evaluation strategy that encourages low emissions and high output is adopted; for the abnormal interval, a second evaluation strategy aimed at correcting its impact or exploring its value is adopted. Next, specific mathematical scoring functions are constructed based on the strategies: the first evaluation function ensures that the score increases with decreasing carbon emissions and increasing output, while the second evaluation function specifically handles outliers. Then, to ensure that the two parameters of carbon emission intensity and output level with different dimensions can be calculated fairly in the function, they are first normalized and uniformly scaled to a comparable range. Finally, the processed parameters are input into the corresponding evaluation function, and the quantified reward score for each power generation unit is automatically calculated. This achieves fair and scientific performance evaluation of all types of generator units, whether in normal operation or abnormal conditions, avoiding homogeneous evaluation, and both strongly incentivizing the low-carbon and high-efficiency behavior of mainstream units and properly handling special data.

[0110] Preferably, step S4 specifically includes:

[0111] Step S41: Quantify the reward intensity into scheduling priority weights;

[0112] Step S42: Combine the scheduling priority weights with the network security constraints and power balance constraints of the power system topology to calculate the optimal output plan for each generation unit;

[0113] Step S43: Based on the optimal output plan, issue dispatch instructions to the corresponding power generation units to implement power distribution across the entire network.

[0114] It should be noted that network security constraints can be upper limits on line transmission security. Power balance constraints can be the total power generation of the entire power grid.

[0115] In this embodiment of the invention, the reward intensity reflecting the low-carbon performance of each power generation unit, calculated in the aforementioned steps, is transformed into a specific scheduling priority weight. A higher weight indicates that the unit is prioritized for deployment. Next, under the two rigid prerequisites of meeting the upper limit of line transmission safety determined by the power grid's physical structure and ensuring that the total power generation and consumption of the entire power grid are equal in real time, the aforementioned weights are used as the core optimization objective. A mathematical algorithm is then used to calculate the optimal power generation scheme that simultaneously achieves both low-carbon priority and power grid safety and stability—that is, the specific output value of each power generation unit. Finally, based on this scientifically calculated optimal scheme, the dispatch center automatically issues precise power generation instructions to each power plant, thereby completing the power distribution of the entire power grid.

[0116] Furthermore, step S42 specifically includes:

[0117] Step S421: Integrate the scheduling priority weights into an optimization objective function, and mathematically model the network security constraints and power balance constraints to form an optimized scheduling model;

[0118] Step S422: Apply the mixed integer programming algorithm to numerically solve the optimization scheduling model, calculate and output the active power output setting values ​​of each generator-side unit that make the optimization objective function optimal under the conditions of network security constraints and power balance constraints.

[0119] In this embodiment of the invention, the scheduling priority weight reflecting the low-carbon performance of each power generation unit is set as the objective function. Simultaneously, conditions such as the upper limit of line transmission power and real-time balance between power generation and consumption, which ensure the safe and stable operation of the power grid, are expressed using mathematical equations or inequalities, together forming a complete optimized scheduling model. Next, a mixed integer programming algorithm capable of handling unit start-up and shutdown states and output magnitude is used to automatically solve the optimized scheduling mode, calculating a set of specific power generation command values ​​for each power generation unit that achieve the optimal low-carbon scheduling objective while satisfying all conditions such as the upper limit of line transmission power and real-time balance between power generation and consumption. Through mathematical modeling and solving, the low-carbon priority scheduling strategy is transformed into an optimal power generation plan executable under hard safety rules. This achieves low-carbon allocation of power generation resources while ensuring power grid safety, ensuring the scientific and optimal nature of scheduling decisions.

[0120] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimal dispatching of a dual-carbon power system, characterized in that, Includes the following steps: Step S1: Based on the power system topology, obtain carbon emission parameters and power flow data; Step S2: Divide the carbon emission parameters and tidal data into multiple first carbon value intervals through a preset numerical range, and extract outliers in the carbon value intervals and recombine them to generate multiple second carbon value intervals. Step S3: Based on the generation-side unit of the power system topology corresponding to each carbon value in the second carbon value range, evaluate the reward intensity of the generation-side unit through an optimization mechanism; Step S4: Based on the magnitude of the reward, a preset scheduling model is used to schedule different generation-side units in the power system topology to allocate the power of the entire network.

2. The optimized scheduling method for a dual-carbon power system according to claim 1, characterized in that, Step S1 specifically includes: Step S11: Based on the power system topology, determine the set of power generation nodes and the set of load nodes in the network, and establish the connection relationship between the set of power generation nodes and the set of load nodes; Step S12: Obtain the carbon emission parameters of each generator unit in the set of power generation nodes. The carbon emission parameters include at least the unit type, fuel type, and carbon emission intensity per unit of power generation. Step S13: Obtain power flow data based on the power grid during a specific operating period. The power flow data includes at least the injected power of each generator set and the power flow distribution of each branch. Step S14: Associate the carbon emission parameters of the power generation node with the injected power of the corresponding generator set, and map the power flow data with the connection relationship to form a complete dataset for carbon flow analysis.

3. The method for optimal dispatching of a dual-carbon power system according to claim 2, characterized in that, Step S14 specifically includes: Step S141: Correlate the carbon emission parameters of the power generation node with the injected power of the corresponding generator set, and calculate the carbon emission of each power generation node during the operating period. Step S142: Map the branch power flow distribution in the power flow data to the connection relationship to determine the transmission path and distribution of power flow in the power system topology.

4. The optimized scheduling method for a dual-carbon power system according to claim 1, characterized in that, Step S2 specifically includes: Step S21: Determine the numerical range of the carbon emission parameters and tidal flow data, and set preset values ​​to define the boundaries of the carbon emission parameters and tidal flow data; Step S22: Based on the preset values, the carbon emission parameters and tidal flow data are divided into multiple consecutive first carbon value intervals according to their numerical values; Step S23: An anomaly propagation detection algorithm based on association consensus is used to identify and extract abnormal data points from the first carbon value interval; Step S24: Collect the extracted abnormal data points and recombine them according to their numerical characteristics to form an abnormal data set; Step S25: Define the abnormal data set together with the remaining first carbon value intervals as multiple second carbon value intervals.

5. The dual-carbon power system optimized dispatching method according to claim 4, characterized in that, Step S22 specifically includes: Step S231: Treat each data point in the first carbon value interval as a network node, and establish connection edges between the network nodes based on the temporal or numerical proximity of the data points. Step S232: Traverse each of the connection edges, compare the values ​​of the two network nodes connected by the connection edge once, and record the logical relationship of the comparison level as the initial consensus label of the connection edge; Step S233: Based on the initial consensus label, the consensus relationship is iteratively propagated along the connection edge, wherein, during the propagation process, it is detected whether the newly derived consensus relationship conflicts with the existing consensus relationship. Step S234: During the propagation of the consensus relationship, network nodes that cannot maintain consistency with the mainstream consensus relationship, as well as network nodes located at the center of logical conflicts, are identified as abnormal data points and extracted.

6. The optimized scheduling method for a dual-carbon power system according to claim 5, characterized in that, Step S233 specifically includes: Step S2331: In the associated network, select a propagation path consisting of at least two connecting edges connected end to end, wherein the path contains at least three consecutive nodes; Step S2332: Based on the known initial consensus labels on the propagation path, deduce the indirect consensus relationship between adjacent nodes through logical transitivity; Step S2333: Compare the newly derived indirect consensus relationship with the existing direct consensus relationship in the network. If they are inconsistent, they are identified as logical conflicts.

7. The optimized scheduling method for a dual-carbon power system according to claim 1, characterized in that, Step S3 specifically includes: Step S31: Map each carbon value in the second carbon value range to the power generation unit that generates the carbon value in the power system topology; Step S32: Configure a differentiated evaluation function based on the classification characteristics of the second carbon value range, wherein the evaluation function takes the carbon emission intensity and output level of the power generation unit as input and outputs its reward score. Step S33: Based on the reward score calculated by the evaluation function, sort the power generation units belonging to different second carbon value intervals, and allocate differentiated reward levels according to the sorting results, with low-carbon and high-efficiency units receiving higher rewards.

8. The optimized scheduling method for a dual-carbon power system according to claim 7, characterized in that, Step S32 specifically includes: Step S321: Based on the classification characteristics of the second carbon value range, set a first evaluation strategy and a second evaluation strategy; Step S322: Based on the first evaluation strategy, construct a first evaluation function whose output value monotonically increases as carbon emission intensity decreases and power output level increases; based on the second evaluation strategy, construct a second evaluation function to smooth out the impact of outliers or identify their potential value. Step S323: Normalize the two input parameters, carbon emission intensity and power output level, of the power generation unit. Step S324: Input the normalized carbon emission intensity and power output level parameters into the corresponding first evaluation function or second evaluation function to calculate the reward score for each power generation unit.

9. The optimized dispatching method for a dual-carbon power system according to claim 1, characterized in that, Step S4 specifically includes: Step S41: Quantify the reward intensity into scheduling priority weights; Step S42: Combine the scheduling priority weights with the network security constraints and power balance constraints of the power system topology to calculate the optimal output plan for each of the generation-side units; Step S43: According to the optimal output plan, issue a dispatching instruction to the corresponding power generation unit to implement power distribution throughout the entire network.

10. The optimized scheduling method for a dual-carbon power system according to claim 9, characterized in that, Step S42 specifically includes: Step S421: Integrate the scheduling priority weights into an optimization objective function, and mathematically model the network security constraints and power balance constraints to form an optimized scheduling model; Step S422: Apply the mixed integer programming algorithm to numerically solve the optimization scheduling model, calculate and output the active power output setting values ​​of each generation-side unit that make the optimization objective function optimal under the network security constraints and power balance constraints.

Images