A distribution network demand response method and system based on estimation of the marginal contribution of carbon emission reduction at closed nodes

Abstract

This invention belongs to the field of power system low-carbon operation and active distribution network demand response technology. It discloses a distribution network demand response method and system based on the estimation of the marginal contribution of carbon emission reduction at closed nodes. Utilizing the topological characteristics of radial distribution networks and the optimality conditions of economic dispatch KKT, a closed-form analytical formula for NMCAC is derived. The marginal response of nodes is calculated by weighting the quadratic coefficient of the unit cost, reducing the complexity to O(N·depth). Based on this, a two-way tiered pricing mechanism with four segments—high-order incentives, low-order incentives, zero-incentive dead zone, and penalty zone—is proposed. Nodes with negative NMCAC are subject to a blocking response or reverse pricing, fundamentally preventing carbon-increasing responses.

Description

Technical Field

[0001] This invention belongs to the field of low-carbon operation of power systems and active distribution network demand response technology, specifically involving a distribution network demand response method and system based on the estimation of the marginal contribution of carbon emission reduction at closed nodes. Background Technology

[0002] Against the backdrop of the power system's transition to low-carbon development, carbon-sensing demand response has become one of the key technologies for achieving low-carbon operation in active distribution networks. Existing methods for carbon-sensing demand response are mainly divided into three categories: The first is the demand response incentive method based on nodal average carbon intensity (CI), which obtains the nodal average carbon intensity CI by allocating generation-side carbon emissions to each load node according to the power flow ratio. i Then based on CI i The methods employed include: 1) providing incentive subsidies to users based on deviations from the system's average carbon intensity; 2) a precise calculation method based on nodal marginal carbon emissions (LMCE), using the first-order necessary condition for optimality of the optimization problem (Karush-Kuhn-Tucker, KKT) to solve for nodal-level sensitivity and achieve accurate LMCE calculation; and 3) a carbon-sensing demand response pricing method based on linear incentives, using LMCE as the core incentive signal and mapping it to a subsidy price in a linear proportion, represented by the following formula: For negative LMCE nodes, existing solutions typically use a method of directly setting them to zero or ignoring them.

[0003] While the aforementioned technologies offer some insights into carbon sensing demand response, they still suffer from three key drawbacks when adapting to the actual operational needs of distribution networks. First, centralized LMCE (Limited Power Response) has high computational complexity, making it difficult to meet the demands of real-time distribution network dispatch. Existing LMCE / implicit differential methods all rely on the network-wide power flow equations and KKT matrix inversion operations at the dispatch center, resulting in a computational complexity of O(N). 3 The distribution network typically comprises hundreds to thousands of nodes, and in actual operation, it requires rolling dispatching at the 5-minute level. Centralized computing models suffer from significant timeliness bottlenecks. Secondly, existing technologies do not fully utilize the physical characteristics of the radial topology of the distribution network. Existing LMCE / LMCI methods are mostly geared towards general power grid designs with transmission network mesh structures, relying on the construction and inversion of PTDF and KKT matrices, and the computation is still essentially constrained by the matrix size. However, radial distribution networks possess the core characteristics of a unique path from load nodes to the root node and a fixed power flow direction. This characteristic has not been explicitly utilized by existing methods to further simplify the computational process, resulting in a lack of targeted optimization of computational resources. Finally, existing linear incentive mechanisms cannot effectively address the negative NMCAC node problem. In actual operating scenarios where network congestion and unit operation constraints coexist, some nodes exhibit the abnormal phenomenon of "reducing load but increasing total system carbon emissions" (i.e.,...). For such nodes, existing technologies either directly set them to zero or still provide micro-subsidies according to linear rules, neither of which can fundamentally prevent carbon-increasing response behavior. In fact, they may even amplify the risk of carbon emission reduction, which is not conducive to achieving the goal of low-carbon operation of the distribution network. Summary of the Invention

[0004] To address the aforementioned technical challenges, this invention provides a distribution network demand response method and system based on DistFlow closed-loop node carbon emission reduction marginal contribution estimation and bidirectional tiered incentive-penalty pricing. This method effectively reduces computational complexity and adapts to the real-time scheduling requirements of the distribution network. A closed-loop formula for the node carbon emission reduction marginal contribution is provided, improving computational accuracy and reliability. A four-stage bidirectional tiered incentive-penalty integrated pricing mechanism is designed to fundamentally prevent carbon-increasing response behavior and ensure the smooth achievement of the distribution network's low-carbon operation goals.

[0005] The present invention discloses a distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes. This method is applicable to distribution networks with a radial topology (i.e., each load node in the network has a unique path to the root node of the substation), and includes the following steps: Step 1, NMCAC Closed-Form Estimation: Perform closed-form estimation of the marginal contribution of carbon emission reduction (NMCAC) of each load node in the radial distribution network, and output a vector containing the NMCAC values ​​of all load nodes in the entire network. Step 2, Two-way tiered incentive-penalty pricing: Based on the NMCAC estimation results obtained in Step 1, implement two-way tiered incentive-penalty integrated demand response pricing for each load node, and output differentiated incentive price signals for each load node; Step 3: Based on the incentive price signal obtained in Step 2, perform differentiated execution control on the demand response behavior of each load node, wherein the demand response requests of the penalty zone nodes are blocked, and the actual load adjustment results after system scheduling are output.

[0006] Furthermore, in step 1, a closed-form estimation of the marginal contribution of the node to carbon emission reduction is performed, including: Step 1-1: Constructing the radial tree: Using the root node of the distribution network substation as the root, a parent-child node mapping parent[i] and a subtree set subtree[i] are established through breadth-first search to form a radial topology; the specific implementation steps are as follows: Step i: Adjacency list construction; Distribution network node set Node 0 is the root node of the substation; the distribution network branch set E is derived from the network topology database of the SCADA / EMS system, and each branch records the node numbers at both ends. Construct an adjacency list adj[v] based on the branch set E; for each branch... Add v to adj[u] and u to adj[v], so that the adjacency list reflects the bidirectional connection relationship between nodes; Step ii: Breadth-first search (BFS) determines the parent node mapping; ① Initialize the first-in-first-out queue Q, and enqueue the root node 0; initialize the access flag set. Set the parent[i] of all non-root nodes to null for now;

[0007] ② When Q is not empty, execute the following loop: remove node v from the head of Q; traverse each neighbor node u in adj[v], and if u is not in visited, then let Add u to the visited queue and add u to the tail of the Q queue; ③ After the loop ends, the parent[i] of each non-root node i is uniquely determined, and the parent[0] of the root node is defined as empty (NULL); Step iii: Recursively construct the subtree set; ① Initialize the list of child nodes for each node, children[v], to be an empty set; ② Iterate through all non-root nodes i, add i to children[parent[i]], and obtain the list of child nodes for each node; ③ Starting from the root node, recursively calculate the subtree set of each node b using a depth-first approach: ; Output: The parent node mapping parent[i] records the unique upstream node of each non-root node; the subtree set subtree[b] records the set of all downstream nodes rooted at b; Step 1-2, Solving the baseline optimal power flow: With the goal of minimizing the total generation cost of the system, solve for the DC optimal power flow that satisfies power balance, upper and lower limits of unit output, and line capacity constraints, to obtain the output P of each unit. G,g Branch Road Current P ij and line load factor ρ ij ; Steps 1-3: Identification of freely adjustable units: Based on the unit output from Step 1-2, calculate the upward adjustment of standby power for each unit (g). and lowering the reserve ; will simultaneously satisfy and The units are labeled as freely adjustable units, forming a set of freely adjustable units G. free ; Steps 1-4: Line Congestion Identification: Based on the line load rate output in Step 1-2, lines that meet the threshold conditions will be identified. A line with a blocking threshold (typically 0.95) is marked as blocked. Steps 1-5: Closed-loop NMCAC calculation: For each load node i, based on the radial tree structure obtained in Step 1-1 and the set of freely adjustable units G obtained in Step 1-3... free And the line blockage markers obtained in steps 1-4, execute in sequence: (1) Following the parent[i] relationship obtained in step 1-1, trace back from node i to the root node to obtain the transmission path Path of node i. i ; (2) Starting from node i, find the nearest blocked line in the path that is closest to i, and denote the downstream end node of the blocked line as the cut-off point b. i If there are no blocked lines on the path, take b. i It is the root node; (3) Determine the set of available units, i.e., the freely adjustable units within the cutoff point subtree, i.e. ; (4) Calculate the NMCAC value of node i using the following closed formula: , Where Ui is the set of available units corresponding to node i, and c 2,g Let ef be the coefficient of the quadratic term in the power generation cost function of unit g. g The carbon emission factor of the unit is g; Steps 1-6, Negative NMCAC Correction: For each NMCAC output from Steps 1-5 i Check if a zero-carbon unit g* exists in the downstream subtree of load node i, i.e. The generator sets, including photovoltaic, wind power, energy storage discharge, and other power sources that do not produce CO2 emissions, and g* simultaneously satisfy: (a) the upward reserve of unit g* is less than the threshold; (b) there is a blocked line marked in steps 1-4 on the path from unit g* to node i; (c) node i is on the path between the zero-carbon unit and the root node; if all three conditions are satisfied, then NMCAC will be set to NMCAC. i Revised to ,in It is a negative correction factor.

[0008] Furthermore, in steps 1-3, a numerical stability threshold δ is introduced to determine the free-regulating unit: only when... and (Typical value 0.001) is when unit g is included in G. free This avoids misjudgment of unit status due to numerical fluctuations.

[0009] Furthermore, in steps 1-4, the line congestion identification adopts an adaptive threshold strategy: if the 95th quantile of the overall network load distribution... Then take Otherwise take .

[0010] Furthermore, in steps 1-6, negative NMCAC detection introduces a temporal continuity criterion: only when K consecutive... neg Time period (typical value) Only when all three conditions for negative NMCAC are met will the corresponding node be marked as a negative NMCAC node, thus avoiding misjudgment caused by numerical fluctuations in a single time period.

[0011] Furthermore, in step 2, the system average carbon emission factor It is calculated using the following formula: , E total It is the total carbon emissions, P G,g It is the unit output obtained from steps 1-2; For two-way tiered incentive-penalty pricing, the following steps are included: Step 2-1: Adaptive threshold calculation: using the system average carbon emission factor Based on this, calculate the four-level segmentation threshold: , , , Where, α high This represents the higher-order excitation threshold, used to delineate the boundary between higher-order and lower-order excitation regions; when the NMCAC value of a node is greater than α... high At this point, the node enters the higher-order incentive zone and is priced according to a higher incentive coefficient; α low This represents the low-order excitation threshold, used to delineate the boundary between the low-order excitation region and the zero-excitation dead zone. When the NMCAC value of a node is between α... high With α low When the node's NMCAC value is less than -ε, it enters the low-order excitation region and is priced according to a low excitation factor; ε is the zero excitation threshold, used to delineate the boundary between the zero excitation dead zone and the penalty zone. When the node's NMCAC value is less than -ε, the node enters the penalty zone; all three conditions must be met. Together, they divided the NMCAC number line into four segments: high-order excitation region, low-order excitation region, zero-excitation dead zone, and penalty region. These are preset threshold scaling factors, all of which are dimensionless positive numbers, where β high This represents the set higher-order threshold scaling factor, β. low β is the low-order threshold scaling factor set. ε The zero-excitation threshold scaling factor is used; the threshold dynamically scales with the real-time carbon emission level of the power grid to avoid the mismatch problem of fixed threshold operating conditions. Step 2-2, Four-segment pricing: Based on the NMCAC obtained in Step 1 i π is determined by the following nonlinear piecewise function. i : (a) When At that time, higher-order incentive pricing is adopted. ; (b) when At that time, low-order incentive pricing is adopted. ; (c) when When set to zero-excitation dead zone ; (d) when At that time, enter the penalty zone and set a negative penalty price: ; in k high k is the higher-order excitation ratio factor. low This is a low-order excitation ratio factor. This is the penalty multiplier. Based on the carbon price.

[0012] Furthermore, step 3 involves differentiated execution control of the demand response behavior of each load node, including: (a) Regarding incentive prices The node triggers user load reduction based on the price elasticity model; (b) Regarding incentive prices The nodes do not send excitation signals and operate at the baseline load; (c) Regarding incentive prices The node blocks its demand response requests.

[0013] Furthermore, in step 3, the incentive price... The node blocking of its demand response request is executed in one of the following three modes, depending on the distribution network operation scenario and user contract terms: (a) Blocking mode: Directly rejects the request to respond to the demand, Applicable scenarios: Regular scheduling under standard DR contracts; (b) Reverse electricity pricing model: Incentivize users to increase their electricity consumption by absorbing local zero-carbon power sources and their output; Applicable scenarios: Users have signed a two-way incentive contract with the power distribution network and have the physical ability to increase their electricity consumption. (c) Penalty pricing model: charging a penalty for spontaneous load shedding. Penalty for electricity charges; Applicable scenario: When a user violates the DR contract and voluntarily reduces their load.

[0014] Furthermore, in step 2-2, the segmented pricing function employs linear transition smoothing at the boundaries of each segment to avoid threshold jumps causing oscillations in user response.

[0015] Furthermore, an adaptive adjustment step for the incentive parameters can be set between step 2 and step 3: based on the incentive price signal obtained in step 2, the total carbon emission reduction rate η of the system in the current period is predicted. pred If η pred Below the preset minimum emission reduction target η min Then the base carbon price C is adjusted according to the penalty function method. price With higher-order excitation threshold α high And return to step 2 to re-execute pricing until η pred Reaching η min Or it may reach the maximum number of iterations.

[0016] This invention also provides a distribution network demand response system based on the estimation of the marginal contribution of carbon emission reduction at closed nodes. The system is used in the above method and includes: NMCAC estimation module: used to perform closed-form estimation of the marginal contribution of carbon emission reduction of each load node in the radial distribution network, and output the NMCAC estimation vector of the load nodes of the entire network; Tiered pricing module: Used to receive NMCAC estimation vector, perform bidirectional tiered incentive-penalty integrated demand response pricing for each load node, and output incentive price signals for each node; Response Execution Module: Used to receive incentive price signals and send incentive signals to the demand response aggregator platform.

[0017] Furthermore, the distribution network demand response system can be deployed at the active distribution network dispatching master station and interact with distribution automation terminals and demand response aggregator platforms via the IEC 61850 standard protocol, thereby enabling the online distribution of NMCAC signals within 5 minutes. The IEC 61850 protocol is a standard communication protocol in the field of power system automation, and those skilled in the art can implement data interaction between this system and external terminals based on this standard protocol.

[0018] The beneficial effects of this invention are as follows: 1) This invention utilizes the topological properties of a radial distribution network from the root node to the subtree to transform the estimation of the marginal contribution of node carbon emission reduction into a word traversal along the radial tree. For each load node, a transmission path is formed by traversing the parent node relationship to the root node. After determining the set of available generating units, the NMCAC value is calculated directly by weighted summation using a closed-form formula, without requiring matrix inversion operations throughout the process. Simultaneously, the physical criterion of "freely adjustable generating units" replaces the mathematical conditions based on the optimizer's dual variables, avoiding repeated calls to the OPF solver to output dual variables. This method reduces the computational complexity from O(N) to O(N) 3) down to In a 24-hour simulation of a radial distribution network based on the IEEE 69-bus standard, only 48 calls to the OPF solver are required, compared to 1224 calls for the existing perturbation method, resulting in a 96.1% reduction in computation time. The computation time of NMCAC is reduced to the second level, thus supporting 5-minute online rolling decisions at the distribution network dispatching master station, filling the gap in existing technology capabilities for real-time carbon sensing dispatching in distribution networks. 2) This invention models the distribution network as a rooted tree structure with the substation root node as the root. It establishes a parent-child node mapping `parent[i]` and a subtree set `subtree[i]` through breadth-first search, explicitly encoding the physical topology characteristics of the distribution network into a traversable data structure. Based on the fact that "any path from any node to the root node is unique" in the radial tree, the transmission path is obtained directly by traversing the parent node relationship, eliminating the need to solve linear equations. Furthermore, based on the physical fact that the power flow direction is unique, blocking cut-off points on the path are defined, and the set of units that can respond to node disturbances is directly determined by the intersection operation of the cut-off point subtree and the set of freely adjustable units, transforming the process that originally required sensitivity analysis into a simple set operation. The core calculation process of this invention does not rely on PTDF matrices, KKT matrices, or any sensitivity matrix; it only requires tree structure traversal and set operations. The closed-loop NMCAC estimation method of this invention has a high statistical correlation with the accurate LMCE calculation results based on the perturbation method, with small estimation errors. While significantly reducing computational complexity, it can ensure engineering-grade estimation accuracy. 3) This invention designs a complete handling system for negative NMCAC nodes, consisting of three layers: detection, decision-making, and execution. The detection layer proposes an analytical criterion for negative NMCAC based on three independent physical conditions: the existence of a zero-carbon unit in the downstream subtree of the node, the zero-carbon unit having reached its output limit, and a blocked line on the path from the unit to the node. When all three conditions are met simultaneously, the negative NMCAC node is identified during the calculation phase, eliminating the need for post-simulation observation. The decision-making layer designs a four-segmented pricing function, dividing the node into a high-order excitation region, a low-order excitation region, a zero-excitation dead zone, and a penalty region based on the NMCAC value, replacing the existing single-segment mapping of linear excitation plus a negative value to zero. The execution layer employs one of the following modes for the penalty region: a blocking rejection mode, a reverse electricity price mode, or a penalty electricity price mode, fundamentally preventing carbon-increasing response behavior. Simultaneously, a time continuity criterion is introduced, requiring that the three negative NMCAC conditions be met for multiple consecutive time periods before a node is marked, avoiding misjudgments caused by single-value fluctuations. By directly prohibiting demand response requests from nodes in the penalty zone using a blocking rejection mode, this invention fundamentally eliminates the conditions for carbon-increasing responses. The carbon emission reduction effect is significantly better than existing linear incentive methods. Simultaneously, the zero-incentive dead-zone mechanism does not issue incentive signals to nodes with near-zero marginal contribution to carbon emission reduction, avoiding ineffective incentive expenditures and improving the efficiency of incentive fund utilization. The execution of the blocking rejection mode does not require users to instantly increase electricity consumption, add new bidirectional incentive contracts, or modify user-side equipment; it can be directly executed by the distribution network dispatching master station within the existing contract framework, resulting in low deployment costs and strong engineering feasibility. Attached Figure Description

[0019] Figure 1 This is a network structure diagram of the IEEE 69-bus simulation system; Figure 2 This is a flowchart of the closed-loop NMCAC calculation process; Figure 3 This is a scatter plot comparing closed-loop NMCAC versus precise LMCE. Figure 4 It is a comparison curve of carbon emissions over 24 hours. Detailed Implementation

[0020] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0021] like Figure 2 As shown, the closed-form estimation method for the marginal contribution of node carbon emission reduction in a radial distribution network according to the present invention includes the following steps: Step 1, NMCAC Closed-Form Estimation: Perform closed-form estimation of the marginal contribution of carbon emission reduction (NMCAC) of each load node in the radial distribution network, and output a vector containing the NMCAC values ​​of all load nodes in the entire network. Step 2, Two-way tiered incentive-penalty pricing: Based on the NMCAC estimation results obtained in Step 1, implement two-way tiered incentive-penalty integrated demand response pricing for each load node, and output differentiated incentive price signals for each load node; Step 3: Based on the incentive price signal, perform differentiated execution control on the demand response behavior of each load node, block the demand response requests of nodes in the penalty zone, and output the actual load adjustment results after system scheduling.

[0022] Step 1 is as follows: S1-1, Radial Tree Construction: Using the root node of the substation in the distribution network as the tree root, breadth-first search is used to establish the parent-child node mapping relationship of the entire network and the set of subtrees of each node, thus constructing the radial topology of the distribution network. For a radial distribution network, the root node (substation) is denoted as node 0; by power conservation, the active power flow of any branch (i,j) is equal to the total net load within the subtree of node j: (1) That is, the flow rate of branch (i,j) is equal to the net active power consumption in the subtree of node j. Equation (1) shows that the power flow direction is uniquely determined under the radial structure, and the path from each node to the root node is unique.

[0023] S1-2, Solving the baseline DC optimal power flow: With the goal of minimizing the total system generation cost, a cost function containing quadratic terms is used to construct and solve the DC optimal power flow model that considers power balance constraints, upper and lower limits of unit output constraints, and line capacity constraints. The model outputs the output of each unit, the equivalent output of the main grid, branch power flow, and line load factor. Specifically: The system scheduling aims to minimize the total power generation cost. (2a) Constraints: (2b) (2c) (2d) Lagrange function with respect to P G,g The KKT stability condition is: (3)

[0024] S1-3, Unit Status Classification: Calculate the upward and downward reserve for each generator unit, and set a numerical stability threshold δ; units with both upward and downward reserve greater than δ are classified as freely adjustable units, and these are aggregated to form a set of freely adjustable units; specifically: Define the upward and downward reserve of unit g: (4a) (4b) when and This means that the unit has not triggered the upper or lower limit constraints (i.e., ( ), are called freely adjustable units. The set of all freely adjustable units is denoted as G. free .

[0025] Furthermore, if there are no blocked paths on the path from node i to the root node, then the set of responsive units for node i is... (Full-network freely adjustable unit); If there is a blocked line on the path, then the downstream end of the blocked line closest to i is defined as the cut-off point b. i And order: (5) That is, only the freely adjustable units within the point tree can respond to load disturbances at node i.

[0026] Consider the load disturbance at node i Regarding both sides of equation (3) with respect to P L,i By taking the partial derivative and combining it with the power balance constraint and the second-order KKT condition, the output response sensitivity of unit g can be obtained: (6) Where the weighting coefficient α i,g The closed-form solution is given by the second-order KKT conditions: (7) The physical meaning of equation (7): load reduction disturbance In the free-regulating unit U i Internal sharing, the proportion borne by each unit is inversely proportional to the coefficient c of its quadratic term of cost. 2,g This is a direct result of the principle of equal marginal costs in economic regulation.

[0027] The weights satisfy the normalization property: (8)

[0028] S1-4, Line Blockage Adaptive Identification: Mark lines with a load rate greater than or equal to the current blocking threshold as blocked lines, and the corresponding line dual variable is non-zero;

[0029] S1-5, Node-by-node closed-loop NMCAC estimation; For each load node i, perform the following steps sequentially: S1-5-1. Traverse the radial topology upwards to obtain the unique transmission path from load node i to the root node; S1-5-2. Search upwards in the transmission path for the nearest blocked line to node i, and set the downstream node of the blocked line as the cut-off point; if there is no blocked line in the path, the root node is directly used as the cut-off point. S1-5-3. Define the available unit set as all freely adjustable units within the cutoff point subtree range; if the available unit set is empty, then activate the global rollback strategy: ,in For global Weighted normalization coefficients are used to avoid loss of excitation signals; S1-5-4. Based on the DistFlow power flow model and the second-order condition of optimality of the KKT economic scheduling, the marginal contribution of node i to carbon emission reduction, NMCACᵢ, is calculated using a closed analytical formula: The total carbon emissions of the system are: (9) Taking the partial derivative with respect to load node i, we obtain LMCE: (10) Since load reduction and load increase are symmetrical on LMCE, NMCAC (semantics for load reduction scenarios) is defined: (11) Final core formula: Substituting equation (7) into equation (10), we obtain the core closed-form expression of this invention: , In the formula: U i The set of responsive units determined by equation (5); c 2,g It is the coefficient of the quadratic term of the cost of unit g; ef g It is the carbon emission factor of unit g.

[0030] S1-6, Negative NMCAC Detection: When a zero-carbon unit g* exists downstream of node i (i.e., NMCAC will be valid if the following three conditions are met. i Negative: (a) Zero-carbon units have reached their upper limit: ; (b) There is a blocked path on the path from g* to node i. ); (c) Node i is on the path between g* and the root node.

[0031] At this time, it is ordered that: (12) in It is a negative correction factor, typically ranging from 0.6 to 0.8.

[0032] Step 2 is as follows: 1) Adaptive segmented threshold calculation: Define the system average carbon emission factor: (13) Based on this, the four-segment stratification thresholds are adaptively determined: (14a) (14b) (14c) Typical parameters , , .

[0033] 2) Four-stage tiered pricing and boundary smoothing: NMCAC for node i i The incentive price π is determined using the following piecewise function. i As shown in Table 1:

[0034] Table 1. Four-Tier Pricing

[0035] Typical parameters , , .

[0036] 3) Differentiated Requirement Response Execution Strategy by Region: User response behavior is handled differently based on their hierarchical level: (a) Excitation region (π) i >0): Load shedding triggered by the price elasticity model, adjustment amount satisfy: (15a) Constraints: , of which DR max This represents the maximum reduction ratio (typical value 0.20). (b) Punishment area ( ): Block the node's demand response requests, that is, make (15b) (c) Zero-excitation zone ( ): No excitation signal is issued, and the node operates at the baseline load.

[0037] 4) Minimum carbon emission reduction constraint guarantee: If the total carbon emission reduction rate predicted based on current parameters does not reach the minimum target η min Then, adaptive adjustment is initiated: optimization is performed using a weighted objective function. (16) That is, to seek a balance between total subsidy expenditure and default on minimum emission reduction constraints, where γ is the default penalty coefficient.

[0038] (a) Subsidy expenditure: This invention applies subsidies to all incentive zone nodes during a certain period. The total amount of incentives issued shall be calculated according to the following formula:

[0039] in Let be the load reduction amount of node i under the excitation.

[0040] (b) Minimum emission reduction constraint: Suppose that the expected emission reduction rate of the system during this period is η. pred The preset minimum emission reduction target is η min Then the amount of the default constraint is: .

[0041] (c) Objective function of the penalty function method: Where γ is the penalty coefficient for breach of contract.

[0042] By adjusting C price With α high The goal is to minimize this objective function so that the cost of subsidies is balanced with the emission reduction target.

[0043] The present invention also provides a carbon sensing demand response incentive pricing system for distribution networks, including an upstream NMCAC closed-loop estimation module, a downstream two-way tiered pricing module, a data interaction interface, and a field communication interaction unit;

[0044] The upstream NMCAC closed estimation module is used to perform a closed estimation method for the marginal contribution of node carbon emission reduction in a radial distribution network, and outputs the NMCAC vector of all nodes in the network at a fixed time interval of 5 minutes.

[0045] The downstream bidirectional tiered pricing module receives the network-wide NMCAC vector output by the upstream module, executes the bidirectional tiered incentive-penalty integrated demand response pricing method, and generates differentiated incentive / penalty price signals for each node in real time.

[0046] The system performs parameter calibration every K time intervals, where K≥6. The perturbation method is used to solve for the accurate LMCE values ​​of key nodes, which are then used to correct the parameters of the closed-loop estimation module.

[0047] Furthermore, when the correlation coefficient between the calibrated closed-loop NMCAC and the accurate LMCE is lower than a preset threshold ρ, the system automatically shortens the calibration interval K and enables the linear regression correction model. The coefficients a and b are obtained by fitting calibration samples from the recent M time periods, thus achieving dynamic correction of the closed-form estimation results.

[0048] To verify the effectiveness of the method of the present invention, such as Figure 1 As shown, a simulation was performed on the IEEE 69-node standard radial distribution network example. This example includes 69 nodes, 68 branches, and 49 load nodes, with a simulation period of 24 time periods. This example represents a typical radial distribution network, exhibiting a tree-like structure and loop-free topology, which matches the characteristics of a radial distribution network as defined in the claims of this invention.

[0049] The distributed power supply configurations are shown in Table 2 below:

[0050] Table 2 Distributed Power Configuration

[0051] Number of time periods Hours; Price elasticity Maximum reduction ratio Basic carbon price Blocking threshold Incentive multiplier , , .

[0052] The comparison results are shown in Table 3:

[0053] Table 3 Comparison of Emission Reduction and Economic Indicators of Different Methods

[0054] like Figure 3 As shown, the NMCAC estimated by the closed-form formula of this invention and the LMCE precisely solved by the perturbation method are compared in... Comparison on individual samples, data points along The baseline distribution visually demonstrates a high degree of agreement between the closed-form NMCAC estimation results and the exact LMCE solution results: Pearson correlation coefficient ; Mean Absolute Error ; Root mean square error ; Computational overhead is reduced by 96.1% (48 OPF vs 1224 OPF).

[0055] As shown in Table 3: (1) The closed-form formula of this invention has a strong correlation with the exact LMCE. The accuracy meets engineering requirements; (2) The computational complexity is greatly reduced, making it suitable for real-time rolling scheduling of power distribution networks at the 5-minute level; (3) The carbon emission reduction effect (−26.06%) brought about by the two-way step incentive-penalty mechanism is significantly better than the traditional CI method (−3.46%) and the precise LMCE+ linear incentive method (−21.31%). Even the method of the present invention using the estimated value is still better than the existing method using the precise value, which proves the key value of the penalty zone mechanism in preventing carbon increase response.

[0056] like Figure 4 As shown in the figure, this figure is a comparison curve of carbon emissions of the power distribution network over 24 hours under four scenarios. The horizontal axis represents the number of operating hours and the vertical axis represents the CO2 emissions. It presents the carbon emission variation over time for the baseline scenario, the traditional CI method, the precise LMCE method, and the method of this invention. Carbon emissions in all scenarios show a bimodal characteristic of morning and evening. Moreover, the carbon emission level of the method of this invention is always the lowest throughout the entire period. The 24-hour carbon emission curve of the scheme of this invention is lower than that of the other three schemes, and the total daily carbon emissions are reduced to 47.87 tons of carbon dioxide, which shows the best carbon emission reduction effect compared with the other three methods.

[0057] Table 4 shows a comparison between the method of this invention and the prior art:

[0058] Table 4. Multidimensional Comparison of Different Carbon Emission Reduction Methods

[0059] As shown in Table 4, the multi-dimensional comparison results between this invention and traditional CI methods, exact LMCE methods, and linear excitation methods demonstrate that this invention fully utilizes the topological characteristics of radial distribution networks, reducing the computational complexity of node marginal carbon emissions to O(N·depth), exhibiting excellent real-time scheduling adaptability. Furthermore, its unique negative marginal carbon emission blocking / penalty mechanism fills a gap in existing technologies. Simulation results from the IEEE 69-bus standard show that this invention achieves a daily carbon emission reduction rate of 26.06%, significantly outperforming various traditional methods. It achieves a synergistic improvement in computational efficiency, operational safety, and carbon emission reduction effectiveness, providing effective technical support for real-time low-carbon scheduling of distribution networks.

[0060] The above description is merely a preferred embodiment of the present invention and is not intended to further limit the present invention. All equivalent changes made based on the description and drawings of the present invention are within the protection scope of the present invention.

Claims

1. A distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes, characterized in that, Includes the following steps: Step 1: Perform closed-form estimation of the marginal contribution of carbon emission reduction (NMCAC) of each load node in the radial distribution network, and output a vector containing the NMCAC values ​​of all load nodes in the entire network. Step 2: Based on the NMCAC estimation results, implement bidirectional tiered incentive-penalty integrated demand response pricing for each load node, and output differentiated incentive price signals for each load node; Step 3: Based on the incentive price signal obtained in Step 2, perform differentiated execution control on the demand response behavior of each load node, block the demand response requests of nodes in the penalty zone, and output the actual load adjustment results after system scheduling.

2. The distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes as described in claim 1, characterized in that, In step 1, a closed-form estimation of the marginal contribution of carbon emission reduction at each load node is performed, including: Step 1-1: Constructing a radial tree: Using the root node of the distribution network substation as the root, establish the parent-child node mapping parent[i] and the subtree set subtree[i] through breadth-first search to form a radial topology. Step 1-2, Solving the baseline optimal power flow: With the goal of minimizing the total generation cost of the system, solve for the DC optimal power flow that satisfies power balance, upper and lower limits of unit output, and line capacity constraints, to obtain the output P of each unit. G,g Branch Flow P ij and line load factor ρ ij ; Steps 1-3: Identification of freely adjustable units: Based on the unit output from Step 1-2, calculate the upward adjustment of standby power for each unit (g). and lowering the reserve Among them, P G,g,max and P G,g,min This represents the upper limit of the maximum active power output and the lower limit of the minimum active power output of unit g; Will simultaneously satisfy and The units are labeled as freely adjustable units, forming a set of freely adjustable units G. free ; Steps 1-4: Line Congestion Identification: Based on the line load rate output in Step 1-2, lines that meet the threshold conditions will be identified. The line is marked as blocked; where λ bind This is a preset line congestion threshold, indicating that the line is considered to be in a congested state when the line load rate reaches this proportion. Steps 1-5, Closed-loop NMCAC calculation: For each load node i, execute the following sequentially: (1) Following the parent[i] relationship obtained in step 1-1, trace back from node i to the root node to obtain the transmission path Path of node i. i ; (2) Starting from node i, find the nearest blocked line in the path and denote the downstream end node of the blocked line as the cut-off point b. i If there are no blocked lines on the path, take b. i It is the root node; (3) Determine the set of available units, i.e., the freely adjustable units within the cutoff point subtree, i.e. ; (4) Calculate the NMCAC value of node i using the following closed formula: ， Where Ui is the set of available units corresponding to node i, and c 2,g Let ef be the coefficient of the quadratic term in the power generation cost function of unit g. g The carbon emission factor of the unit is g; Steps 1-6, Negative NMCAC Correction: For each NMCAC output from Steps 1-5 i Check if a zero-carbon unit g* exists in the downstream subtree of load node i, i.e. The generator set, and g* simultaneously satisfies: (a) the upward reserve of unit g* is less than the threshold; (b) there is a blocked line marked in steps 1-4 on the path from unit g* to node i; (c) node i is on the path between the zero-carbon unit and the root node; if all three conditions are met, then NMCAC will be set to NMCAC. i Revised to ,in It is a negative correction factor.

3. The distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes as described in claim 2, characterized in that, In steps 1-3, a numerical stability threshold δ is further introduced to determine the free-regulating unit: only when and The time will include unit g in G free .

4. The distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes as described in claim 2, characterized in that, In steps 1-4, line congestion identification employs an adaptive threshold strategy: if the 95th quantile of the overall network load distribution... Then take Otherwise take .

5. The distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes as described in claim 2, characterized in that, In steps 1-6, negative NMCAC detection further introduces a temporal continuity criterion: only when K consecutive... neg Only when all three conditions for negative NMCAC are met in each time period will the corresponding node be marked as a negative NMCAC node, thus avoiding misjudgment caused by numerical fluctuations in a single time period.

6. The distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes as described in claim 1, characterized in that, In step 2, the system average carbon emission factor It is calculated using the following formula: ， E total It is the total carbon emissions, P G,g It is the unit output obtained from steps 1-2; For two-way tiered incentive-penalty pricing, the following steps are included: Step 2-1: Adaptive threshold calculation: using the system average carbon emission factor Based on this, calculate the four-level segmentation threshold: ， ， ， Where, α high This represents the higher-order excitation threshold, when the NMCAC value of the load node is greater than α. high At this time, the load node enters the high-order incentive zone and is priced according to the high-multiplier incentive coefficient; α low This represents the low-order excitation threshold, when the NMCAC value of a node is between α and α. high With α low When the load node is in the middle, it enters the low-order excitation region and is priced according to the low excitation factor; ε is the zero excitation threshold. When the NMCAC value of the node is less than -ε, the load node enters the penalty region; all three conditions must be met. Together, they divided the NMCAC number line into four segments: high-order excitation region, low-order excitation region, zero-excitation dead zone, and penalty region. ,β high This represents the set higher-order threshold scaling factor, β. low β is the low-order threshold scaling factor set. ε The scaling factor is zero for the excitation threshold. Step 2-2, Four-segment pricing: Based on the NMCAC obtained in Step 1 i π is determined by the following nonlinear piecewise function. i : (a) When At that time, higher-order incentive pricing is adopted. ; (b) when At that time, low-order incentive pricing is adopted. ; (c) when When set to zero-excitation dead zone ; (d) when At that time, enter the penalty zone and set a negative penalty price: ； in k high k is the higher-order excitation ratio factor. low This is a low-order excitation ratio factor. This is the penalty multiplier. Based on the carbon price.

7. The distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes as described in claim 1, characterized in that, Step 3 involves differentiated execution and control of the demand response behavior of each load node, including: (a) Regarding incentive prices The node triggers user load reduction based on the price elasticity model; (b) Regarding incentive prices The nodes do not send excitation signals and operate at the baseline load; (c) Regarding incentive prices The node blocks its demand response requests.

8. The distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes as described in claim 7, characterized in that, In step 3, the incentive price is... The node blocks its demand response request, and the execution adopts one of the following three modes according to the distribution network operation scenario and user contract terms: (a) Blocking mode: Directly rejects the request to respond to the demand, ; Applicable scenarios: Regular scheduling under standard DR contracts; (b) Reverse electricity pricing model: Inducing users to increase their electricity consumption and absorb local zero-carbon power supply output. Applicable scenarios: Users have signed a two-way incentive contract with the power distribution network and have the physical ability to increase electricity consumption; (c) Penalty pricing model: charging a penalty for spontaneous load shedding. Penalty for electricity bill; Applicable scenario: When a user violates the DR contract and voluntarily reduces their load.

9. A distribution network demand response method based on the estimation of the marginal contribution of carbon emission reduction at closed nodes, as described in claim 6, is characterized in that... The system further includes an adaptive adjustment step for incentive parameters: based on the incentive price signal obtained in step 2, the total carbon emission reduction rate η of the system in the current period is predicted. pred If η pred Below the preset minimum emission reduction target η min Then the base carbon price C is adjusted according to the penalty function method. price With higher-order excitation threshold α high And return to step 2 to re-execute pricing until η pred Reaching η min Or it may reach the maximum number of iterations.

10. A distribution network demand response system based on the estimation of the marginal contribution of carbon emission reduction at closed nodes, characterized in that, The system is used to perform the method according to any one of claims 1-9, comprising: NMCAC estimation module: used to perform closed-form estimation of the marginal contribution of carbon emission reduction of each load node in the radial distribution network, and output the NMCAC estimation vector of the load nodes of the entire network; Tiered pricing module: Used to receive NMCAC estimation vector, perform bidirectional tiered incentive-penalty integrated demand response pricing for each load node, and output incentive price signals for each node; Response Execution Module: Used to receive incentive price signals and send incentive signals to the demand response aggregator platform.

Images