A zero-carbon park-oriented energy exchange and fusion charging load and industrial load collaborative optimization method

Abstract

The application discloses a zero-carbon park inter-energy fusion-oriented charging load and industrial load collaborative optimization method, relates to the technical field of energy internet and power system optimization control, acquires park multi-source data, constructs a double-network model of a traffic network and a power distribution network, and takes a charging station as a shared node; a site side and a power grid side risk field is inferred based on a space-time graph neural network; an inter-energy consistency checker IEC is constructed to generate a zero setting gate matrix to form a feasible candidate set; a multi-objective mixed integer programming collaborative scheduling model is solved under the constraint of the feasible candidate set, a bilateral trigger type safety valve slack variable is introduced to generate a bottleneck evidence bag; a key bottleneck object set is determined based on a constraint shadow price distribution and a zero setting ratio distribution, an expansion candidate set is generated, an evidence chain expansion planning model is constructed to obtain an expansion scheme, a feasible region improvement constraint judgment is executed, and rewriting is performed into rolling optimization.

Description

Technical Field

[0001] This invention relates to the field of energy internet and power system optimization control technology, specifically a method for the coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration. Background Technology

[0002] Zero-carbon industrial parks typically simultaneously handle the concentrated charging demands of electric vehicles, the electricity needs of industrial production, and the fluctuating supply of renewable energy within the park. Charging stations within the park serve as both important service nodes in the transportation network and load access points in the distribution network. Charging loads exhibit significant spatiotemporal aggregation and uncertainty, while industrial loads are characterized by significant plateau periods, strong process constraints, and difficulty in arbitrarily shifting peak loads. The combined effect of these two factors can easily trigger capacity constraints at distribution nodes and current-carrying limits on power lines during localized periods, leading to a decrease in the park's energy supply security margin. Simultaneously, renewable energy output is prone to insufficient absorption and low green electricity utilization efficiency during high-output periods. Therefore, the park's operators not only need to achieve coordinated scheduling of charging and industrial loads but also need to establish an executable and unified constraint system among transportation accessibility, grid capacity, critical line margins, and green electricity matching requirements, and form an auditable expansion and optimization closed loop after bottlenecks are identified.

[0003] Existing technologies typically achieve local optimization through time-of-use pricing, charging reservation queuing strategies, single-dimensional charging station queuing prediction, load shaving optimization based on distribution network power flow constraints, or peak shifting control based on the transferability of industrial loads. Among these, some schemes use traffic-side prediction results to guide charging diversion, some schemes use distribution network boundary constraints to limit charging power, and some schemes use planning models for charging facility expansion site selection and capacity configuration. However, most of the above-mentioned solutions adopt separate modeling and phased solution methods for the transportation side and the power grid side. They often only handle infeasible combinations in the optimization stage with penalty terms or posterior verification methods, lacking a pre-verification mechanism for energy consistency. This leads to candidate solutions being unfeasible or requiring a large amount of manual parameter adjustment when there is a conflict between transportation accessibility and power grid carrying capacity. At the same time, common collaborative scheduling only focuses on peak shaving or cost targets, making it difficult to ensure green electricity matching threshold constraints while meeting the time-period conflict threshold constraints of industrial load platforms. Furthermore, it is difficult to interpret the line-sensitive constraints caused by the set of critical lines and line margins. In addition, existing expansion plans are mostly based on static load forecasting and lack an evidence chain planning mechanism supported by the constraint shadow price distribution and zero-ratio distribution output by the scheduling solution. This makes the determination of the expansion candidate set easy to rely on experience, and after the expansion, there is a lack of feasible domain improvement constraint judgment to verify whether the feasible candidate set has been truly expanded and the key bottleneck objects have been eliminated, making it difficult to form a closed loop of continuous optimization.

[0004] Existing technologies achieve park load management through time-sharing guidance, predictive scheduling, and planned site selection, but they still have certain limitations. For example, they cannot incorporate the site-side risk field and the grid-side risk field into a unified energy consistency checker (IEC) and form a consistency check result matrix to generate a zero-gating matrix; they are difficult to remove candidate time periods and candidate charging station combinations that do not meet the consistency threshold from the variable domain of the multi-objective mixed integer programming collaborative scheduling model to form a feasible candidate set; they are difficult to simultaneously satisfy the industrial load platform time period conflict threshold constraint and green electricity matching threshold constraint within the same closed loop, while taking into account the set of critical lines and line margin; they are difficult to use the site-side triggered safety valve relaxation variables and the grid-side triggered safety valve relaxation variables to form an auditable bottleneck evidence package and determine the set of critical bottleneck objects accordingly; and they are also difficult to output expansion schemes based on the expansion candidate set and evidence chain expansion planning model and achieve closed-loop verification of the expansion effect through feasible domain improvement constraint judgment.

[0005] Therefore, there is an urgent need for a collaborative optimization method for charging load and industrial load in zero-carbon industrial parks that integrates charging and energy. This method can achieve executable scheduling by constructing a dual-network model, a spatiotemporal synchronous graph neural network, an energy consistency checker (IEC), a zero-gating matrix, and a feasible candidate set. Furthermore, it can form a closed-loop mechanism for continuous optimization by combining a bottleneck evidence package, a set of key bottleneck objects, a set of expansion candidates, an evidence chain expansion planning model, and feasible domain improvement constraints. This will improve the safety of park operation, the capacity for green electricity utilization, and the relevance of expansion decisions. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and propose a method for the coordinated optimization of charging load and industrial load in zero-carbon industrial parks oriented towards energy integration, so as to solve the above-mentioned problems.

[0007] The objective of this invention is achieved through the following technical solution: a method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, comprising the following steps: S1, acquiring park traffic data, charging station operation data, industrial load data, renewable energy output data, and distribution network operation boundary data, aligning them with a unified time granularity, constructing a dual-network model including the traffic network and the distribution network, and setting charging stations as shared nodes, while determining the set of critical lines and line margins based on the distribution network operation boundary data; S2, constructing a spatiotemporal graph based on charging stations and inferring the site-side risk field and the grid-side risk field using a spatiotemporal synchronization graph neural network; S3, constructing an energy integration consistency checker (IEC) and performing a check on each... First, a set of candidate time periods and a set of candidate charging stations are generated based on charging demand. Consistency check values ​​are calculated based on the site-side risk field, grid-side risk field, vehicle arrival time window, charging station service capacity, distribution node capacity redundancy, and line margin. These values ​​are then compared with a consistency threshold to obtain a consistency check result matrix. The consistency check values ​​are jointly determined by traffic accessibility consistency, distribution capacity consistency, and line sensitivity consistency. Second, a zeroing gating matrix is ​​generated from the consistency check result matrix. Assignment variables corresponding to candidate time periods and candidate charging station combinations that do not meet the consistency threshold are zeroed and removed from the variable domain of the multi-objective mixed-integer programming collaborative scheduling model to form a feasible candidate set. Third, a feasible candidate set is established under the constraints of the feasible candidate set. The multi-objective mixed-integer programming collaborative scheduling model is solved. This model includes constraints on unique assignment of charging demand, charging station capacity, distribution node capacity, line current carrying boundary, industrial load platform time-period conflict threshold, and green electricity matching threshold. Site-side triggered safety valve relaxation variables and grid-side triggered safety valve relaxation variables are introduced into the charging station capacity constraint and distribution node capacity constraint, respectively. These relaxation variables are only allowed to be positive during periods when renewable energy output exceeds the first output threshold and the industrial load platform time-period conflict threshold constraint is met, generating a bottleneck evidence package. S6, Output Bottleneck. The evidence package includes the trigger trajectory, constraint shadow price distribution, and zero-ratio distribution. Based on the constraint shadow price distribution and zero-ratio distribution, the set of key bottleneck objects is determined. S7: Based on the bottleneck evidence package, an expansion candidate set is generated and an evidence chain expansion planning model is constructed. The expansion candidate set is jointly determined by the trigger number threshold, tightness threshold, and green electricity matching gap threshold. The green electricity matching gap threshold is determined based on the proportion of unassigned charging demand energy to the total charging demand energy during the high renewable energy output period. The expansion site selection and capacity determination of charging facilities are carried out only within the expansion candidate set, and the expansion scheme is obtained by embedding the distribution network voltage boundary constraints, line current boundary constraints, and distribution node capacity boundary constraints.S8, execute the feasible region improvement constraint determination, including updating the charging station service capacity, distribution node capacity margin, and line margin based on the expansion plan, and re-executing steps S3 and S4 to obtain the updated zeroing gate matrix and the updated feasible candidate set. When the size of the updated feasible candidate set is larger than the size of the original feasible candidate set and the zeroing percentage of the key bottleneck object set in the updated zeroing gate matrix is ​​less than the zeroing percentage threshold, the expansion plan is determined to have passed the feasible region improvement constraint determination, and the expansion plan is written into the spatiotemporal graph and the dual-network model to enter the next round of rolling collaborative optimization.

[0008] Traffic accessibility consistency is determined by judging the arrival time window satisfaction of vehicles corresponding to charging demand at candidate charging stations during the candidate time period. Power distribution capacity consistency is determined by judging that the capacity margin of the power distribution nodes connected to the candidate charging station is not less than the capacity margin threshold during the candidate time period. Line sensitivity consistency is determined by judging that the line margin of the critical lines in the critical line set is not less than the line margin threshold during the candidate time period.

[0009] The zeroing gating matrix is ​​obtained by binarizing the consistency check result matrix. The binarization includes marking the candidate time period and candidate charging station combination with the consistency check value not less than the consistency threshold as feasible and marking the candidate time period and candidate charging station combination with the consistency check value less than the consistency threshold as zero.

[0010] The site-side risk field is generated by traffic flow density, betweenness centrality of key traffic nodes, and historical queue length sequence of charging stations, while the grid-side risk field is generated by industrial load platform intensity sequence, distribution node load centrality, and line power flow sensitivity sequence.

[0011] The conflict threshold constraint for industrial load platform periods is achieved by calculating the proportion of energy that the charging load falls into during the industrial load platform period and limiting the proportion of energy to not exceed the conflict threshold.

[0012] The green electricity matching threshold constraint is achieved by calculating the proportion of charging load matching during the period when the renewable energy output is greater than the first output threshold and limiting the proportion of charging load matching to not less than the matching threshold. Furthermore, the satisfaction of the charging load matching proportion is based on the premise of simultaneously satisfying the distribution node capacity constraint and the line current carrying boundary constraint.

[0013] The relaxation variables of the site-side triggered safety valve and the grid-side triggered safety valve are respectively set with relaxation upper limit thresholds, and relaxation penalty coefficients are set in the objective function of the multi-objective mixed integer programming collaborative scheduling model, and the relaxation penalty coefficients are not less than the penalty coefficient thresholds.

[0014] The constrained shadow price distribution is composed of the constrained shadow prices output during the solution process of the multi-objective mixed integer programming collaborative scheduling model. The zero-set ratio distribution is obtained by statistically analyzing the proportion of candidate time periods and candidate charging station combinations that are set to zero in the zero-set gating matrix to the total number of candidate time periods and candidate charging station combinations, and summarizing them by time period and spatial location.

[0015] The set of key bottleneck objects includes charging stations with a trigger count greater than the trigger count threshold and distribution nodes with a constraint shadow price greater than the tightness threshold.

[0016] The expansion candidate set includes expansion candidate charging stations, expansion candidate distribution nodes, and expansion candidate regions. Expansion candidate charging stations are determined by charging stations whose trigger count is greater than the trigger count threshold. Expansion candidate distribution nodes are determined by distribution nodes whose constraint shadow price is greater than the tightness threshold. Expansion candidate regions are determined by the green electricity matching gap threshold.

[0017] The beneficial effects of this invention are: This invention establishes a dual-network model encompassing both transportation and power distribution networks in a zero-carbon industrial park scenario, designating charging stations as shared nodes. This enables the allocation of charging load time periods and power configurations to move beyond local decisions based solely on available resources at the charging station. Instead, it achieves executable global collaborative optimization under the constraints of the power distribution network's operational boundaries, critical line sets, and line margins. This avoids the overreach of power distribution nodes and the triggering of line current-carrying boundaries caused by the concentrated superposition of charging loads during industrial load periods, thereby enhancing the safety margin and operational stability of the park's energy supply side.

[0018] This invention introduces a site-side risk field and a grid-side risk field, and uses traffic reachability consistency, distribution capacity consistency, and line sensitivity consistency in the energy consistency checker (IEC) to calculate the consistency check value and compare it with the consistency threshold to obtain a consistency check result matrix. This allows the feasibility determination of the candidate time period set and the candidate charging station set to be constrained by both traffic congestion risk and grid over-limit risk, eliminating high-risk combinations from the source and avoiding repeated searches in infeasible areas by the subsequent multi-objective mixed integer programming collaborative scheduling model, thereby improving the feasibility of the solution and the engineering feasibility of the scheduling results.

[0019] This invention establishes and solves a multi-objective mixed integer programming collaborative scheduling model under feasible candidate set constraints. The model simultaneously includes unique assignment constraints for charging demand, charging station capacity constraints, distribution node capacity constraints, line current carrying boundary constraints, industrial load platform time-period conflict threshold constraints, and green electricity matching threshold constraints. This clearly quantifies the conflict between charging load and industrial load in the spatiotemporal dimensions and controls it under threshold constraints. At the same time, it matches charging behavior with renewable energy output constraints, so that charging demand during high renewable energy output periods can be prioritized for consumption while meeting distribution node capacity constraints and line current carrying boundary constraints. This improves the green electricity utilization efficiency within the park and reduces the peak risk of external power purchases.

[0020] This invention introduces site-side triggered safety valve relaxation variables and grid-side triggered safety valve relaxation variables into the charging station capacity constraints and distribution node capacity constraints, respectively. Controlled triggering is achieved through relaxation upper limit threshold, relaxation penalty coefficient and penalty coefficient threshold. This enables the model to have controllable buffering capacity when facing sudden arrival deviations, short-term fluctuations in industrial load or prediction errors. At the same time, it avoids the abuse of relaxation variables to mask the real bottlenecks and ensures that the scheduling scheme has higher robustness and continuous operation capability within the safety boundary.

[0021] This invention generates a bottleneck evidence package from the output of collaborative scheduling solutions. The bottleneck evidence package includes trigger trajectories, constraint shadow price distributions, and zero-ratio distributions. Based on this, a set of key bottleneck objects is determined, transforming bottleneck identification from subjective judgment into a traceable chain of evidence. This clearly distinguishes whether the bottleneck originates from insufficient charging station service capacity, insufficient power distribution node capacity, or insufficient margin of critical lines, providing an auditable basis for subsequent expansion decisions and preventing expansion planning from falling into an uncontrollable path based solely on experience-based site selection and capacity configuration.

[0022] This invention generates an expansion candidate set based on bottleneck evidence packages and constructs an evidence chain expansion planning model. The expansion candidate set is jointly determined by trigger count threshold, tightness threshold, and green electricity matching gap threshold. The site selection and capacity determination of charging facility expansion are carried out only within the expansion candidate set, and the distribution network voltage boundary constraints, line current carrying boundary constraints, and distribution node capacity boundary constraints are embedded. This strictly limits the expansion scope to an implementable area where the bottleneck evidence can be explained and the distribution boundary can be satisfied, reducing the risk of ineffective expansion and over-construction, and improving the targeting and feasibility of expansion investment.

[0023] This invention sets up a feasible domain to improve constraint judgment and re-executes the power consistency checker IEC and zero-gating matrix generation after updating the charging station service capacity, distribution node capacity margin and line margin in the expansion plan. By comparing the updated feasible candidate set size with the zero-ratio of the key bottleneck object set, the expansion plan is verified in a closed loop. This ensures that the expansion plan must simultaneously expand the feasible candidate set and reduce the zero-ratio of key bottlenecks before it can be written back to enter the next round of rolling collaborative optimization, thus forming a closed-loop mechanism for continuous optimization. This avoids the problem of expansion planning and scheduling optimization being separated, resulting in effective planning but infeasible scheduling or feasible scheduling but unresolved bottlenecks.

[0024] This invention generates a zero-gating matrix from the consistency check result matrix and sets the assignment variables corresponding to candidate time periods and candidate charging station combinations that do not meet the consistency threshold to zero and removes them from the variable domain of the multi-objective mixed integer programming collaborative scheduling model. It directly shrinks the feasible candidate set through gating pruning, making the energy exchange consistency checker (IEC) a pre-generator of the feasible domain of the scheduling model. This forms a closed constraint chain of risk field - consistency check - zero-gating - variable domain removal. Mechanistically, it avoids the out-of-bounds solutions and unauditable empirical parameter tuning caused by soft constraints implemented only by the objective function penalty term, and reduces the risk of being interpreted as conventional constraint preprocessing.

[0025] This invention forms a closed link between prediction, scheduling, and planning by employing a dual-network model, risk field inference, energy consistency checker (IEC), zero-gating matrix, feasible candidate set, multi-objective mixed integer programming collaborative scheduling model, site-side triggered safety valve relaxation variables, grid-side triggered safety valve relaxation variables, bottleneck evidence package, key bottleneck object set, expansion candidate set, evidence chain expansion planning model, and the causal relationship and closed loop between feasible domain improvement constraint determination. This enables the distribution network to maintain controllable operation boundaries and achieve priority consumption of renewable energy while meeting the industrial load platform time-period conflict threshold constraints and green electricity matching threshold constraints. It has strong engineering applicability and auditability. Attached Figure Description

[0026] Figure 1 The process of this invention Figure 1 ; Figure 2 The process of this invention Figure 2 ; Figure 3 The process of this invention Figure 3 . Detailed Implementation

[0027] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Example 1

[0029] like Figure 1 As shown in Example 1, a closed-loop collaborative optimization process for a zero-carbon integrated energy and transportation park is presented. Park traffic data, charging station operation data, industrial load data, renewable energy output data, and distribution network operation boundary data are aligned with a unified time granularity (15 minutes in this example) to form a discrete time period set. It repeatedly executes the risk field-IEC-zero gating-coordinated scheduling-evidence package-expansion-writeback in a rolling window manner.

[0030] Building a transportation network With power distribution network A dual-network model is used, with charging stations as shared nodes. A set of critical routes is selected based on route margin. , ,in, For the line During the period Margin (A); For the line During the period The upper limit of the current (A); For the line During the period The historical current (A).

[0031] when When the line is less than the margin threshold, Included .

[0032] The identification of critical nodes in a dual-network system uses betweenness centrality and load centrality indicators: Betweenness centrality This is used to identify key traffic nodes. For nodes Betweenness centrality (dimensionless). For transportation network The starting and ending points (and time period indicators) in distinguish); From arrive The number of shortest paths; From arrive And passing through nodes The number of shortest paths.

[0033] Load Centrality This is used to identify critical nodes in power distribution. For power distribution nodes Load centrality (dimensionless). For nodes Industrial load power (kW, or power index taken as average / peak value within the window) within a rolling window. For nodes Charging load power (kW, or power index based on average / peak value within the window) within the rolling window. For power distribution network The node index in the data.

[0034] The Spatiotemporal Synchronization Graph Neural Network (STSGCN) constructs a spatiotemporal graph using charging stations as nodes. Output site side risk field Risk field on the power grid side This will be considered a subsequent IEC penalty.

[0035] In this embodiment, the risk field calculation is generated as follows: for each charging station With time period Traffic flow density index Betweenness centrality summary index Normalized index of station queue length Construct a site-side risk field; for each distribution node With time period The strength of the industrial load platform Load centrality of power distribution nodes Combined sensitivity with line power flow Construct the risk field on the power grid side. The risk field is constructed by normalized weighted summation and truncated to [value missing]. : in, For charging stations During the period Site-side risk field (dimensionless, value range is) ); For truncation functions, the value is restricted to a certain range. ; The weighted coefficients are normalizable and satisfy the following conditions: (This example uses...) ); For the site During the period Traffic flow density index For the site The corresponding betweenness centrality summary index, For the site During the period The normalized index of queue length, among which Historical / predicted queue length (vehicles). This is a time-limited dummy variable; The calculation is shown in Example 2.

[0036] in, For power distribution nodes During the period The grid-side risk field (dimensionless, with a value range of) ); This is a truncation function; The weighted coefficients are normalizable and satisfy the following conditions: (This example uses...) ); The strength index of the industrial load platform (dimensionless, see Example 2); For power distribution nodes Load centrality (dimensionless). For nodes During the period The overall sensitivity index of power flow in the line.

[0037] This embodiment is for each charging demand. Generate a set of candidate sites With candidate time set : in, For charging needs A set of candidate charging stations; For charging stations; For demand From the vehicle's starting point to the charging station The road network distance (km); The service radius threshold (km, in this embodiment, is taken as...) km).

[0038] in, For charging needs The set of candidate time periods; A set of discrete time periods; For demand Acceptable lower and upper bounds of arrival time windows (time slot number or minutes, must be consistent with...) (The time granularity is consistent). Therefore, the candidate combination set is: Subsequently, the IEC calculates the consistency check value on this candidate combination set.

[0039] The power consistency checker (IEC) checks each charging demand. Candidate combinations Calculate the consistency check value: in, Index for charging demand (belonging to the charging demand set) ), For charging station index (belonging to the charging station collection) ), Index for discrete time periods (belonging to a set of time periods) ); , which is the weighting coefficient (dimensionless); For traffic accessibility consistency indication or rating (dimensionless). For distribution load consistency indication or scoring (dimensionless). For line sensitivity consistency indication or scoring (dimensionless). The risk field on the site side (dimensionless). For the grid-side risk field (dimensionless); For charging stations Mapped to the distribution node it connects to The mapping function.

[0040] This embodiment uses a binary determination rule for definition. , and : Traffic Accessibility Consistency in, For traffic accessibility consistency indication (0 or 1); For demand The lower and upper bounds of the acceptable arrival time window; For demand During the period Arrival Station at Departure / Decision The predicted arrival time (can be obtained from the shortest travel time of the road network and congestion correction).

[0041] Power distribution load consistency in, For power distribution load consistency indication (0 or 1); For charging stations Mapped to its access distribution node The mapping function; For power distribution nodes During the period Capacity margin (kW); For demand If on the site Time period Charging power requirement during service (kW); The load capacity threshold (kW, in this embodiment, is taken as) kW).

[0042] Line sensitivity consistency in, For line-sensitive consistency indication (0 or 1); For the set of critical paths; For this direction of the line The resulting current increment (A, which can be estimated by the sensitivity coefficient or linearized power flow); For the line During the period Margin (A); The line margin threshold (A, in this embodiment, is taken as) A).

[0043] In generating the zero-gated matrix, the consistency check result matrix is ​​binarized to generate the zero-gated matrix. The assignment variables corresponding to candidate combinations that fail the verification are removed from the domain of subsequent model variables, and the zero-gating matrix is ​​defined as follows: in, The zeroing gate matrix element (0 or 1) indicates that the candidate combination passed the check, and 0 indicates that it failed. The consistency threshold (dimensionless) has a default value of 0.5. This is the consistency check value.

[0044] And define the feasible candidate set as This represents the set of all candidate time-site-demand combinations that have passed the consistency check. At this point, the assignment decision in the scheduling model only... Combinations that do not pass the IEC standard are not included in the solution, thus ensuring that the subsequent output has a consistent and executable basis from a structural perspective.

[0045] In constructing a multi-objective mixed-integer programming collaborative scheduling model, the assignment variable is defined. And limited to only Modeling; defining the relaxation variables of the site-side triggered safety valve Relaxation variable of grid-side triggered safety valve And set a relaxation upper limit. Set a first output threshold. And define the set of high-output periods. The industrial platform time period set is as follows: .

[0046] The constraint expression of the cooperative scheduling model in this embodiment. Let... Then the unique assignment constraint is: in, For demand The set of feasible candidate combinations; Assign variables; A collection of charging needs.

[0047] set up For charging stations During the period Given the maximum available service capacity (kW), the charging station capacity constraint (including site-side relaxation) is: in, For charging stations During the period The maximum available service capacity (kW); For the site-side triggered safety valve relaxation variable (kW), and only aggregated during the permitted relaxation period. Positive values ​​are allowed within the range.

[0048] set up For power distribution nodes During the period Given the industrial load power (kW), the distribution node capacity constraint (including grid-side relaxation) is: in, For power distribution nodes During the period Industrial load power (kW); For grid-side triggered safety valve relaxation variable (kW), and only aggregated during permitted relaxation periods. Positive values ​​are allowed within the range.

[0049] The critical path current-carrying boundary constraint is expressed in the form of path margin as follows: in, For assignment For the line The resulting current increment (A); For the line During the period Available margin (A); This is the set of critical paths.

[0050] Slack variables are allowed to be positive only during periods when renewable energy output exceeds the first output threshold and the industrial load platform time-period conflict threshold constraint is met. This embodiment defines the set of allowed slack periods as follows: in, For high renewable energy output periods; For time period The renewable energy output power (kW); The first output threshold (kW); For industrial load platform time periods; For a set of time periods that allow for relaxation.

[0051] And on Apply ,right Positive values ​​are allowed within the upper limit. The model includes: unique assignment constraints for charging demand, charging station capacity constraints, distribution node capacity constraints, critical line current carrying boundary constraints, industrial load platform time-period conflict threshold constraints, and green electricity matching threshold constraints. The conflict ratio is defined as: in, The percentage of charging energy during the industrial platform period (dimensionless). For charging needs Total charging energy demand (kWh) within the rolling window; Assigning variables (only for) (Definition); The numerator represents the total charging energy assigned during the platform period, and the denominator represents the total charging energy assigned throughout the entire period.

[0052] Requirements and constraints (This embodiment takes) The high-output periods are set as follows: The green electricity matching ratio is defined as: in, The proportion of charging energy during periods when high renewable energy output is provided (green electricity matching proportion, dimensionless). This represents the set of high-output periods; the remaining symbols are... Consistent with the above.

[0053] Requirements and constraints (This embodiment takes) Furthermore, the green electricity matching requirement is based on the premise of simultaneously satisfying the capacity constraints of distribution nodes and the current-carrying boundary constraints of lines.

[0054] The objective function adopts a multi-objective weighted summation form, simultaneously considering distribution network security (peak shaving and load shedding at critical nodes), renewable energy utilization (green electricity matching), industrial load conflict suppression, and constraint relaxation penalties, to ensure that an executable scheduling scheme is output while satisfying all constraints. The complete objective function is defined as follows: in, The combined objective value after weighting multiple objectives; The target weight coefficient (dimensionless); Total charging power for the entire park within the scrolling window Peak value (kW); For time period Normalized intensity of industrial load (dimensionless); For time period Total charging power of the entire park (kW); For time period charging station The charging power (kW); For charging stations Whether it belongs to a critical node; For time period Normalized output of renewable energy (dimensionless, note the risk field) distinguish); The relaxation variable (kW) of the site-side triggered safety valve. The relaxation variable (kW) of the grid-side triggered safety valve is collected during the allowable relaxation period. Internally constrained by an upper limit (in this embodiment, we take...) Example values ​​for the weighting coefficient: .

[0055] in, and The corresponding relaxation penalty coefficient satisfies and (This embodiment takes) ), used to suppress long-term dependent relaxation.

[0056] Regarding the bottleneck evidence package output, the trigger trajectory for the slack variable to become positive is recorded. and Simultaneously construct the linear slack problem of the collaborative scheduling model within the same rolling window, and read the dual information corresponding to the distribution node capacity constraint and the line current carrying boundary constraint in the linear slack problem as the constraint shadow price distribution; and set the zero-gating matrix. The distribution of the percentage of zero-set points is statistically analyzed. The trigger trajectory, shadow price distribution, and zero-set point percentage distribution are encapsulated into a bottleneck evidence package, and the key bottleneck object set (charging stations with trigger counts exceeding the threshold and distribution nodes with shadow prices exceeding the tightness threshold) are selected based on this.

[0057] To facilitate threshold determination, this embodiment provides statistical formulas for the distribution of the proportion of zero-set nodes and the comprehensive shadow price of nodes: in, For the site Time period The proportion of zero (dimensionless). For zeroing gated matrix elements; For the indicator function, the denominator is used to count the number of stations. With time period Actual candidate combinations.

[0058] in, For power distribution nodes The comprehensive shadow price index (yuan / kW); and These are the dual variables (yuan / kW) of the nodal capacity constraints and the line current-carrying boundary constraints in the linear slack subproblem, respectively. For nodes Related line sets; For time period index, This is a route index.

[0059] Regarding the evidence chain expansion planning model, an expansion candidate set is generated based on the bottleneck evidence package. The candidate set is jointly determined by the trigger number threshold, the tightness threshold, and the green electricity matching gap threshold. Expansion site selection and capacity determination are carried out only within the candidate set, and the expansion scheme is output by embedding the distribution network voltage boundary constraints, line current carrying boundary constraints, and distribution node capacity boundary constraints.

[0060] Among them, the proportion of green electricity matching gap Used to identify candidate areas for expansion: in, ;when (This embodiment takes) When the voltage boundary constraint is applied, the corresponding region will be included in the expansion candidate region set. The voltage boundary constraint can be expressed as... (This embodiment takes) pu、 pu), of which Calculated from power distribution flow or its linearized model.

[0061] Regarding the determination of constraints for improving the feasible region, the service capacity of charging stations is updated based on the expansion plan. Distribution node capacity redundancy With line margin And within the same scrolling window, re-execute IEC and zero-gating to obtain the updated... The updated feasible candidate set is used; when the size of the updated feasible candidate set increases and the proportion of key bottleneck objects set to zero is lower than the threshold, the expansion plan is determined to be approved and written back to the spatiotemporal graph and the dual network model to enter the next round of rolling.

[0062] The determination of an increase in the size of the feasible candidate set can be made by using... (This embodiment takes) The threshold for setting the percentage of critical bottleneck objects to zero can be taken as follows: .

[0063] Through the above embodiments, closed-loop optimization can be achieved in the park scenario, including energy consistency verification and pruning, collaborative scheduling solution, bottleneck evidence output, and evidence chain expansion and rewriting.

[0064] Example 2

[0065] like Figure 1 and Figure 2 As shown, Example 2, based on Example 1, supplements the disclosure with the generation details of the site-side risk field and the power grid-side risk field, as well as the IEC's three-consistency determination and zero-gating matrix binarization rules.

[0066] Risk field generation at charging stations: This is achieved using traffic flow density, betweenness centrality of key traffic nodes, and historical queue length sequences at charging stations. Traffic flow density indices: in, For charging stations During the period Traffic flow density risk indicators (dimensionless or normalized according to a unified dimension). For charging station index (belonging to the charging station collection) ), Index for discrete time periods (belonging to a set of time periods) ); Index for roads / road segments; In order to connect with charging stations The set of roads associated with the service area; For roads For the site The influence weight (dimensionless, can be set and normalized according to distance or OD share rate). For time period the way Traffic flow (vehicles / hour or vehicles / 15 minutes). For time period the way Average vehicle speed (km / h); For roads Length (km).

[0067] Betweenness centrality summary index: in, For charging stations The corresponding summary index of betweenness centrality of key traffic nodes (dimensionless). For indexing traffic network nodes; In order to connect with charging stations The set of transportation network nodes associated with the service area; For nodes For the site Contribution weights (dimensionless, normalized to satisfy) ); For traffic nodes Betweenness centrality (dimensionless).

[0068] Normalized metrics for station queue length: in, For time period charging station Historical or predicted queue length (vehicles); This is a time-limited dummy variable; .

[0069] Accordingly, the site-side risk field is calculated using a normalized weighted sum and truncated to the nearest whole number. : in, This is a truncation function; And it can be normalized to satisfy (This example uses...) ).

[0070] Grid-side risk field generation: Employing industrial load plateau strength sequences, distribution node load centrality, and line power flow sensitivity sequences. Industrial load plateau strength: in, For time period Industrial load platform strength (dimensionless, value) ); For time period Total industrial load power of the park (kW); This is a time-limited dummy variable; This represents the maximum industrial load power (kW) within the rolling window.

[0071] Comprehensive sensitivity index (to avoid slack variables) Confusion, denoted as ): in, For power distribution nodes During the period The line power flow comprehensive sensitivity index (dimensionless or normalized according to a unified dimension). For route indexing; For the set of critical paths; For time period Next node Injected power to the line Sensitivity coefficient of current / power flow (e.g.) It can be obtained from power distribution power flow linearization or sensitivity analysis, and the unit can be A / kW).

[0072] Accordingly, the risk field on the power grid side is calculated using a normalized weighted sum and truncated to [value missing]. : in, And it can be normalized to satisfy (This example uses...) ); For power distribution nodes The load centrality (the calculation method is shown in Example 1).

[0073] Three-consistency determination rule: Traffic accessibility consistency: in, For charging needs Assigned to station And during the period Traffic accessibility consistency indicator for the service (0 or 1); For demand Acceptable lower and upper bounds of arrival time windows (minutes or time slot numbers, must be consistent with...) unified); For the time period Departure requirements Arrival Station The predicted arrival time (obtained from the shortest travel time of the road network or dynamic traffic conditions).

[0074] Consistency of power distribution capacity: in, For power distribution load consistency indication (0 or 1); For charging stations Mapping of accessed power distribution nodes; For power distribution nodes During the period Available capacity margin (kW); For charging needs If on the site Time period The corresponding charging power requirement (kW) during service can be determined by energy demand. (calculated based on service duration) This represents the node capacity redundancy threshold (kW).

[0075] Line sensitivity consistency: in, For line-sensitive consistency indication (0 or 1); For the set of critical paths; For the line During the period Available margin (A); When assigning the charging demand to this station, the line The resulting current increment (A); The line margin threshold (A) is used.

[0076] Consistency check value synthesis: The risk field is written as a penalty term into the check value, which adopts a directly calculable linear form. in, For charging needs In candidate combinations The cross-energy consistency check value (dimensionless). , which is the weighting coefficient (dimensionless); These are respectively indicators for traffic accessibility consistency, power distribution capacity consistency, and line sensitivity consistency (0 or 1). The risk field on the site side (dimensionless). For the grid-side risk field (dimensionless); For charging stations Mapped to its access distribution node The mapping function.

[0077] Set consistency threshold (In this embodiment, the value is 0.5), when The candidate combination is then determined to pass the consistency check.

[0078] In terms of generating the zero-gated matrix and eliminating variables from their domains, the consistency check results are binarized to obtain the zero-gated matrix, and a feasible candidate set is formed based on this matrix: in, The zeroing gate matrix element (0 or 1) indicates that the candidate combination has passed the check, and 0 indicates that it has been zeroed.

[0079] And based on this, a feasible candidate set is formed. The subsequent collaborative scheduling model only applies to Define assignment variables .

[0080] Industrial load platform time period set is The percentage of charging energy during a given time period is used to characterize the intensity of the conflict, and is defined as follows: in, The percentage of charging energy during the industrial platform period (dimensionless). For industrial load platform time periods; For charging needs Total charging energy demand (kWh) within the rolling window; Assigning variables (only for) definition).

[0081] Requirements and constraints (In this embodiment, the value is 0.2).

[0082] The set of high renewable energy output periods is The percentage of charging energy during high-output periods is defined as the green energy matching percentage: in, The proportion of charging energy during periods when high renewable energy output is provided (green electricity matching proportion, dimensionless). This refers to the collection of high-output periods; For time period Renewable energy output power (kW); The first output threshold (kW); the remaining symbols are... Consistent with the above.

[0083] Requirements and constraints (The value in this embodiment is 0.8), and it is based on the premise of simultaneously satisfying the distribution node capacity constraint and the line current carrying boundary constraint. Through the supplementary disclosure of this embodiment, the continuous output of each rolling window includes at least: the site-side risk field. Risk field on the power grid side Consistency check value Zero-gated matrix Feasible candidate set Conflict percentage The proportion of green electricity This is used for the collaborative scheduling solution and subsequent evidence packet generation in Example 1.

[0084] Example 3

[0085] like Figures 1 to 3 As shown, Example 3, based on Examples 1 and 2, further discloses the controlled triggering mechanism of the trigger-based slack variable, the bottleneck evidence package generation, and the rules for determining the expanded candidate set.

[0086] Triggered slack variable control: Define a set of high-output periods. The industrial platform time period set is And define the set of allowed relaxation periods. .right Apply and Constraints; Apply a relaxation cap: in, For charging stations During the period The station-side triggered safety valve relaxation variable (kW); For power distribution nodes During the period The relaxation variable (kW) of the grid-side triggered safety valve. The upper limit of relaxation on the site side (kW). The upper limit of grid-side relaxation (kW); Index of charging stations (belonging to) ), For power distribution node index, For time period index (belonging to) ).

[0087] Bottleneck Evidence Package Generation: The constraint shadow price distribution can be constructed from the dual information of the output of the linear programming subproblem obtained by linearly relaxing the cooperative scheduling model within a given rolling window. Let the dual variables of the distribution node capacity constraint and the line current-carrying boundary constraint in the linear relaxation subproblem be respectively... and Then the node's overall shadow price: in, For power distribution nodes The comprehensive shadow price index (yuan / kW); For nodes in the linear slack subproblem During the period The capacity constraint dual variable (yuan / kW); For the linear relaxation subproblem, the circuit During the period The dual variable of the current-carrying constraint (yuan / kW); For nodes A set of associated routes; For route indexing; This is a time-period index.

[0088] Zero-load percentage statistics by charging station and time period: in, For charging stations During the period The proportion of zero (dimensionless). For charging demand set; The elements of the zeroing gate matrix are (1 indicates that the candidate combination passes the consistency check, and 0 indicates that it is set to zero). This is an indicator function; the denominator is used to count the number of points for that station. With time period The actual number of candidate combinations corresponds to the required quantity, avoiding inconsistencies in the denominator due to missing candidate sets.

[0089] Regarding the controlled triggering mechanism of the trigger-type safety valve, this embodiment limits the positive conditions for the site-side trigger-type safety valve relaxation variable ({s,t}) and the grid-side trigger-type safety valve relaxation variable ({g,t}) to be high renewable energy output and within the allowable relaxation period set. Furthermore, the relaxation upper limit threshold and relaxation penalty coefficient are further solidified.

[0090] in, For high renewable energy output periods; For time period Renewable energy output power (kW); The first output threshold (kW); For industrial load platform time periods; To allow for the set of relaxation periods.

[0091] And for all Apply and Hard constraints; and simultaneously for all Apply a relaxed upper limit threshold constraint: in, The upper limit of relaxation on the site side (kW). Let be the upper limit of grid-side relaxation (kW). To suppress the long-term positive slack, a relaxation penalty term is introduced into the objective function, and the penalty coefficient is set to be no less than the threshold. The relaxation-related part of the objective function can be written as: in, These are the relaxation penalty coefficients for the site side and the grid side, respectively (objective function weights, the dimension of which can be regarded as yuan / kW or a dimensionless weighted coefficient, and the scale must be consistent with the scale of other objective items). This represents the set of nodes in the power distribution network; the other symbols are the same as above.

[0092] Furthermore, to meet the requirement that the penalty coefficient is not less than the penalty coefficient threshold, this embodiment sets... and (Example) ).

[0093] in, For charging station collection, This represents the set of power distribution nodes. Therefore, the solution in the output... and The positive trajectory constitutes evidence of a bottleneck trigger.

[0094] Regarding the generation method of the constraint shadow price distribution, the linear slack problem of the multi-objective mixed integer programming collaborative scheduling model is constructed synchronously within the same rolling window, and the dual variables of the distribution node capacity constraint in the linear slack problem are read. Dual variables of line current-carrying boundary constraints And summarized as , as a constraint on the shadow price distribution.

[0095] This embodiment further consolidates the shadow prices on the line side by node. Let nodes... The associated line set is The comprehensive shadow price index of a node is then defined as: in, For power distribution nodes The comprehensive shadow price index (yuan / kW). For nodes in the linear slack subproblem During the period The capacity constraint dual variable (yuan / kW). For the linear relaxation subproblem, the circuit During the period The dual variable of the current-carrying constraint (yuan / kW). For nodes A set of associated routes, This is a time-period indicator.

[0096] Regarding the generation method of the zero-set percentage distribution, this embodiment directly uses the zero-set gating matrix output in Embodiment 2. Statistical analysis was conducted, and the distribution of the percentage of charging stations with zero charges was aggregated at two levels: charging station-time period and charging station-entire lifecycle. Let the charging demand set be... For any charging station With time period The percentage of zeros during this period is defined as follows: in, For charging stations During the period The proportion of zero (dimensionless). For charging needs At the charging station Time period The zeroing gate value (0 or 1). This is an indicator function used to count the total number of candidate combinations for the charging station during the current time period, thus avoiding inconsistencies in the denominator due to missing candidate sets.

[0097] For any charging station The overall percentage of zero-valued elements within the scrolling window is defined as follows: in, For charging stations The maximum percentage of zero-point periods (dimensionless) is used to identify the long-term bottleneck level of the charging station. Through the above statistics, the distribution of zero-point percentages can not only be used to identify long-term infeasible areas, but also, together with shadow prices, to confirm that the bottleneck truly stems from grid capacity or transportation accessibility constraints rather than accidental data gaps.

[0098] This embodiment limits the set of critical bottleneck objects to a combination of critical bottleneck charging stations and critical bottleneck power distribution nodes. The critical bottleneck charging stations are filtered by a trigger count threshold, and the critical bottleneck power distribution nodes are filtered by a tightness threshold. The trigger count is derived from the slack variable trigger trajectory. Let the charging station... The set of time periods during which the relaxation variable of the site-side triggered safety valve is positive within the scrolling window is as follows: in, For charging stations The set of time periods in which the site-side slack variables are positive within the scrolling window; For site-side slack variables (kW).

[0099] The number of triggers is defined as follows: in, For charging stations The number of relaxation triggers (dimensionless, count). This is the set of triggering time periods for the charging station. This is a site-side relaxation variable. A threshold for the number of triggers is set. ,when Time to charge station It has been identified as a key bottleneck charging station.

[0100] Tightness threshold based on node-based comprehensive shadow price index Determination. Set tightness threshold. ,when Distribution nodes It has been identified as a critical bottleneck power distribution node.

[0101] This yields the set of key bottleneck objects: in, A collection of key bottleneck objects; This is a collection of charging stations for key bottlenecks; This refers to a set of key bottleneck power distribution nodes. This represents the union of sets.

[0102] in, For key bottleneck charging station clusters (meeting) A collection of charging stations. For the set of key bottleneck distribution nodes (satisfying) (A collection of power distribution nodes). This embodiment will... , The bottleneck evidence package is encapsulated along with the set of objects, providing a traceable data foundation for the formation of the key bottleneck object set.

[0103] Regarding the rules for determining the ternary structure of the expanded candidate set, the expanded candidate set... From expanding candidate charging stations Expanding candidate power distribution nodes With expansion candidate areas Composition, in which The candidate areas for expansion are determined by the green energy matching gap threshold, which is defined as follows: in, The proportion of green electricity matching gap (dimensionless) ), For charging demand aggregation, This is a collection of periods of high output from renewable energy sources. For charging station collection, For charging needs Total energy demand (kWh) For demand On the site Time period The assignment variable (0 or 1). As an indicator function, when the demand The value is 1 if no high-output period is assigned, otherwise it is 0.

[0104] And set a green electricity matching gap threshold. ,when Determine the set of candidate areas for expansion at that time The final output of the expanded candidate set is: in, To expand the candidate charging station set (equal to) ), To expand the candidate distribution node set (equal to ), This provides a set of candidate expansion areas. In this embodiment, the output is a structured list. The list fields include at least the object type, object identifier, trigger count or tightness indicator or gap indicator, corresponding threshold, and the mapping relationship between associated charging stations and associated power distribution nodes, ensuring the traceability of expansion decisions. Facility planning objective function and decision variables: only applicable to the expansion candidate set. Expansion site selection and capacity determination are conducted within the specified scope. Decision variables include expansion site selection variables. Expanding capacity variables Access variables The objective function can be: in, The overall target value for the expansion planning phase; For each objective item, the weight coefficients are (dimensionless or dimensionless after being standardized). To expand the pool of candidate charging stations, To expand the set of candidate distribution nodes; For the site Fixed construction costs (RMB); For the site Site selection variables for whether to construct / expand; For the site Unit capacity cost (RMB / kW); For the site Expansion capacity (kW); For the site after adopting the expansion plan The zero-ratio assessment quantity (dimensionless). For the distribution nodes after adopting the expansion plan The shadow price assessment quantity.

[0105] in, The comprehensive shadow price index can be obtained by constructing a linear slack subproblem on the cooperative scheduling model within the same rolling window after updating the boundary of the expansion scheme. .

[0106] Key constraints include at least expansion capacity constraints, access constraints, and node capacity boundary constraints, and embed line current-carrying boundary constraints and voltage boundary constraints. The expansion scheme output includes expansion site identifiers, expansion capacity, access node identifiers, and boundary verification results.

[0107] Regarding the feasible region improvement constraint determination and write-back, after updating the expansion plan, this embodiment re-executes the consistency check and zero-gating generation according to the closed-loop process of Embodiment 1 to obtain the updated zero-gating matrix and the updated feasible candidate set. It also calculates the updated zero-gating percentage for the key bottleneck object set, using the same statistical caliber as described above, and applies this method to key bottleneck charging stations. Statistics, and analysis of key bottleneck power distribution nodes based on their associated charging stations. Weighted statistics are used to determine the percentage of nodes set to zero. When the size of the updated feasible candidate set is larger than the size of the original feasible candidate set and the percentage of the key bottleneck object set set to zero in the updated zero-gating matrix is ​​less than the zero-gating threshold, the expansion scheme is determined to have passed the feasible domain improvement constraint determination. The expansion scheme parameters are then written back to the spatiotemporal graph node attributes and the boundary parameters of the dual network model. The written-back content includes at least the update of charging station service capacity, the update of distribution node capacity redundancy, and the update of line margin. After the write-back, the next round of risk field inference and IEC consistency verification is triggered, thus forming a closed-loop enhancement link of evidence package-driven expansion-expansion reverse expansion of feasible domain-re-verification and re-gating.

[0108] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, characterized in that, The process includes the following steps: S1, acquiring park traffic data, charging station operation data, industrial load data, renewable energy output data, and distribution network operation boundary data, aligning them with a unified time granularity, constructing a dual-network model including the traffic network and distribution network, and setting charging stations as shared nodes; simultaneously, determining the critical line set and line margin based on the distribution network operation boundary data; S2, constructing a spatiotemporal graph based on charging stations and inferring the site-side risk field and grid-side risk field using a spatiotemporal synchronization graph neural network; S3, constructing an energy consistency checker (IEC) and generating a candidate time period set and candidate charging station set for each charging demand; calculating a consistency check value based on the site-side risk field, the grid-side risk field, vehicle arrival time window, charging station service capacity, distribution node capacity margin, and line margin, and comparing it with a consistency threshold to obtain the consistency value. S4. A zeroing gate matrix is ​​generated from the consistency verification result matrix. The assignment variables corresponding to the candidate time periods and candidate charging station combinations that do not meet the consistency threshold are zeroed to form a feasible candidate set. S5. A multi-objective mixed integer programming collaborative scheduling model is established and solved under the constraints of the feasible candidate set. The station-side triggered safety valve relaxation variables and grid-side triggered safety valve relaxation variables are introduced into the charging station capacity constraints and distribution node capacity constraints, respectively, and a bottleneck evidence package is generated. S6. The bottleneck evidence package is output. The bottleneck evidence package includes the trigger trajectory, constraint shadow price distribution and zeroing ratio distribution. The key bottleneck object set is determined based on the constraint shadow price distribution and the zeroing ratio distribution. S7. An expansion candidate set is generated based on the bottleneck evidence package, and an evidence chain expansion planning model is constructed to obtain the expansion scheme. S8, perform feasible region improvement constraint determination, and write the expansion scheme into the spatiotemporal graph and the dual network model to enter the next round of rolling collaborative optimization.

2. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The consistency verification value is jointly determined by traffic accessibility consistency, power distribution capacity consistency, and line sensitivity consistency. Traffic accessibility consistency is determined by judging the arrival time window satisfaction of vehicles corresponding to charging demand at candidate charging stations during candidate time periods. Power distribution capacity consistency is determined by judging that the capacity margin of the power distribution nodes connected to candidate charging stations during candidate time periods is not less than the capacity margin threshold. Line sensitivity consistency is determined by judging that the line margin of critical lines in the set of critical lines during candidate time periods is not less than the line margin threshold.

3. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The zeroing gating matrix is ​​obtained by binarizing the consistency check result matrix. The binarization includes marking candidate time periods and candidate charging station combinations with consistency check values ​​not less than the consistency threshold as feasible, marking candidate time periods and candidate charging station combinations with consistency check values ​​less than the consistency threshold as zero, and removing the assignment variables corresponding to candidate time periods and candidate charging station combinations that do not meet the consistency threshold from the variable domain of the multi-objective mixed integer programming collaborative scheduling model.

4. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The site-side risk field is generated by traffic flow density, betweenness centrality of key traffic nodes, and historical queue length sequence of charging stations. The grid-side risk field is generated by industrial load platform intensity sequence, distribution node load centrality, and line power flow sensitivity sequence.

5. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The conflict threshold constraint for industrial load platform periods is achieved by calculating the proportion of energy that the charging load falls into during the industrial load platform period and limiting the proportion of energy to not exceed the conflict threshold.

6. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The green electricity matching threshold constraint is achieved by calculating the charging load matching ratio during the period when the renewable energy output is greater than the first output threshold and limiting the charging load matching ratio to be no less than the matching threshold. Furthermore, the satisfaction of the charging load matching ratio is based on the premise of simultaneously satisfying the distribution node capacity constraint and the line current carrying boundary constraint.

7. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The multi-objective mixed-integer programming collaborative scheduling model includes unique assignment constraints for charging demand, charging station capacity constraints, distribution node capacity constraints, line current carrying boundary constraints, industrial load platform time-period conflict threshold constraints, and green electricity matching threshold constraints. The model allows the site-side triggered safety valve relaxation variables and the grid-side triggered safety valve relaxation variables to be positive only during periods when renewable energy output exceeds a first output threshold and the industrial load platform time-period conflict threshold constraints are met. The site-side triggered safety valve relaxation variables and the grid-side triggered safety valve relaxation variables are each assigned a relaxation upper limit threshold, and a relaxation penalty coefficient is set in the objective function of the multi-objective mixed-integer programming collaborative scheduling model. Furthermore, the relaxation penalty coefficient is not less than the penalty coefficient threshold.

8. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The constraint shadow price distribution is composed of the constraint shadow prices output during the solution process of the multi-objective mixed integer programming collaborative scheduling model. The zeroing ratio distribution is obtained by statistically analyzing the proportion of candidate time periods and candidate charging station combinations that are zeroed in the zeroing gate matrix to the total number of candidate time periods and candidate charging station combinations, and summarizing them by time period and spatial location.

9. A method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The set of key bottleneck objects includes charging stations with a trigger count greater than the trigger count threshold and distribution nodes with a constraint shadow price greater than the tightness threshold.

10. The method for coordinated optimization of charging load and industrial load in a zero-carbon industrial park oriented towards energy integration, as described in claim 1, is characterized in that... The expansion candidate set is jointly determined by a trigger count threshold, a tightness threshold, and a green electricity matching gap threshold. The green electricity matching gap threshold is determined based on the proportion of unassigned charging demand energy to total charging demand energy during periods of high renewable energy output. The expansion candidate set includes expansion candidate charging stations, expansion candidate distribution nodes, and expansion candidate regions. The expansion candidate charging stations are determined by charging stations with a trigger count greater than the trigger count threshold. The expansion candidate distribution nodes are determined by distribution nodes with a constraint shadow price greater than the tightness threshold. The expansion candidate regions are determined by the green electricity matching gap threshold and are only included in the expansion candidate set. The process involves selecting sites and determining the capacity for charging facility expansion, embedding voltage boundary constraints, line current boundary constraints, and distribution node capacity boundary constraints. The feasible region improvement constraint determination includes updating the charging station service capacity, distribution node capacity margin, and line margin based on the expansion plan, and re-executing steps S3 and S4 to obtain an updated zeroing gate matrix and an updated feasible candidate set. When the size of the updated feasible candidate set is larger than the original feasible candidate set size and the zeroing percentage of the key bottleneck object set in the updated zeroing gate matrix is ​​less than the zeroing percentage threshold, the expansion plan is determined to pass the feasible region improvement constraint determination.

Images