Energy storage market risk hedging scheduling method, system and device based on multi-scene diffusion diagram prediction and storage medium

Abstract

The invention discloses an energy storage market risk hedging scheduling method, system and device based on multi-scene diffusion diagram prediction, and a storage medium, and relates to the technical field of power market scheduling, and the method comprises the steps: constructing a market space-time association initial diagram, generating a next-day price scene set and probability weight through employing a diffusion diagram space-time predictor, and carrying out the prediction of the next-day price scene set and probability weight; embedding market rule consistency and physical feasibility screening in the generation process to obtain a price scene set and a probability weight, jointly generating a health portrait set on the price scene set and the probability weight, constructing a robust scheduling input set, optimizing a hedging scheduling plan on the robust scheduling input set, and generating an execution instruction set according to the hedging scheduling plan; and performing rolling execution according to the execution instruction set and monitoring price and health consumption. According to the method, the operation conflict and the excessive circulation risk are reduced through the worst income constraint and the dynamic protection closed loop, and the executability and the robustness of scheduling are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power market dispatching, in particular to an energy storage market risk hedging dispatching method, system and device based on multi-scenario diffusion graph prediction and a storage medium. BACKGROUND

[0002] Energy storage participates in the coordinated trading and dispatch of multiple power markets (day-ahead, intra-day, ancillary services), and the price fluctuates significantly in the unified settlement period due to the influence of load, wind and light output, temperature, maintenance and cross-region interconnection. The limited line also forms a congestion-sensitive spatial price difference linkage between adjacent nodes. There is a leading / lagging coupling between different markets, and if the dispatch lacks thresholds and energy / power reservations, high-frequency small cycles and excessive degradation are easy to occur. At the same time, actual landing needs to map the strategy to the bid and EMS control limit, and has the ability to trigger and shrink in abnormal market conditions.

[0003] The existing technology mainly includes three types: one is price prediction and scenario generation, commonly using ARIMA, LSTM, Transformer or Copula / Monte Carlo to model each market independently, a small amount of work does cross-market time alignment, but lacks explicit embedding of market rule boundaries and network congestion physical feasibility in the generation stage, and is mostly trimmed or smoothed with upper and lower limits after the fact; the probability weight is mostly based on historical frequency or experience scoring. Two is robust / random dispatch, using CVaR, chance constraint or quantile programming to optimize energy storage charging and discharging and bidding under experience / hypothesis distribution, but usually separates the price uncertainty, health constraints and ancillary service reservations: the price is driven by external scenarios, the health uses static cycle / throughput upper limit, and the service reservation sets a hard constraint separately, which is difficult to form a unified, time-aligned and mapable constraint expression. Three is health management and protection, mostly using equivalent full cycle, rain flow counting or fixed threshold flow limiting offline to set a budget, and online to trigger power reduction / lock SOC according to the threshold, lacking a mechanism to convert price driving strength and congestion strength into time-varying marginal decay cost; the transaction side and the control side often have two sets of strategies, and the EMS and the bid boundary are inconsistent; a few studies introduce SoC reservation or price difference threshold, but do not unify the worst income and health budget under the robust distribution offset, and the ancillary service duration / response time is mostly an independent constraint, and the execution coordination relies on experience rules. SUMMARY

[0004] In view of the above problems, the present application is proposed.

[0005] Therefore, the technical problem solved by this invention is that existing collaborative trading and scheduling methods have the following problems: the scenario generation does not implement the consistency of rules and physical feasibility of cross-market coupling and spatial congestion during the generation stage, resulting in poor price path executability; robust scheduling does not unify price uncertainty, health constraints and ancillary service reservations into time-aligned opportunity windows and limits, which can easily trigger high-frequency small cycles in the neutral range; the health budget is static and shadow price is missing, and it is impossible to endogenize the lifetime loss cost as an operational threshold and link it with the threshold and window; and the bidding and EMS boundaries are inconsistent, lacking linkage protection for observation deviation and health over-consumption, making it difficult to balance the lower limit of returns and the degradation budget under unfavorable market conditions.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a method for generating price scenario sets and probability weights, comprising generating a price scenario set and probability weights for the next day using a diffusion graph spatiotemporal predictor based on an initial graph; embedding market rule consistency and physical feasibility screening during the generation process, and outputting the price scenario set and probability weights.

[0007] As a preferred embodiment of the price scenario set and probability weight generation method described in this invention, the step of embedding market rule consistency and physical feasibility screening during the generation process includes: using the initial graph as a conditional input, guiding the generation results of the diffusion graph spatiotemporal predictor in each settlement cycle with edge weights; using the parallel price time series of the same node as a carrier, synchronously correcting spatial adjacency and cross-market relationships with a comprehensive edge weight sequence; and outputting the guided price vector as follows: , in, For node indexing, This is a time index for the settlement period. A scenario index for price paths. In the settlement cycle Nodes generated and guided by the diffusion graph spatiotemporal predictor The price vector, in market order, consists of day-ahead energy price, intraday energy price, and ancillary service settlement price. , , These are the day-ahead energy price vector, the intraday energy price vector, and the ancillary service settlement price vector, respectively. , , This represents the portion generated before the guidance of the day-ahead energy price, intraday energy price, and ancillary service settlement price. As a spatial adjacency indicator, when a node With nodes If the adjacent element is within a range where congestion coupling exists, take 1; otherwise, take 0. In the settlement cycle Pair of nodes and The combined edge weights are derived from the edge weight sequence of the initial graph. In the settlement cycle The spatial linkage guidance coefficient is a dimensionless coefficient used to adjust the magnitude of price vector correction caused by the comprehensive edge weights. In the settlement cycle The cross-market coupling guidance coefficient is a dimensionless coefficient used to adjust the magnitude of cross-market spread correction. This is a cross-market association indicator; when performing cross-market coupling correction within the same node, it takes one value. This is a price difference correction item between the previous day and the intraday price, expressed in price. This is a price difference correction item between the intraday and previous day prices, expressed in price. For ancillary services and intraday coupling correction items, the unit is price, where This is a proportional coefficient used to establish the coupling relationship between ancillary service settlement prices and intraday energy prices.

[0008] As a preferred embodiment of the price scenario set and probability weight generation method described in this invention, the step of embedding market rule consistency and physical feasibility screening during the generation process, and outputting the price scenario set and probability weights, includes projecting the guided price vector onto the set of executable constraints defined by market rule consistency and physical feasibility screening in each settlement period, and outputting a price vector that satisfies the rules and physical constraints as follows: , in, In the settlement cycle The set of executable constraints is defined by both market rule consistency and physical feasibility screening. In the settlement cycle nodes The price vector, in market order, consists of day-ahead energy price, intraday energy price, and ancillary service settlement price. These represent the lower and upper limits of the daytime energy price, respectively, in units of price; These represent the lower and upper limits of intraday energy prices, respectively, in price units. These represent the lower and upper limits of the settlement price for ancillary services, respectively, in units of price. This is the maximum allowable spread between markets, used to limit the price difference between the previous day and the intraday price, expressed in price. This is the maximum allowed cross-market coupling limit, used to restrict the coupling price difference between ancillary services and intraday services, expressed in price. This is the scaling factor for spatial price differences, expressed in price, used to integrate edge weights. Transformed into spatial price difference constraints, For sets The projection operator is used to map the guided price vector into a price vector that satisfies market rules and physical constraints. For projection onto the scene The set of price vector components is ultimately used to construct the price scenario set for the next day; After generating and projecting all nodes and all settlement cycles, the next day's price scenario set is obtained, and probability weights are assigned to each price path to reflect the degree to which the rules and constraints are satisfied during the generation stage and the feasibility of the path. The probability weights are calculated as follows: , in, For the scene probability weights, and These are the weighting coefficients. For the scene Historical occurrence ratio score, For the scene Constraint satisfaction rating The number of scenarios in the price scenario set for the next day. The cardinality of the node set. The base of the settlement cycle index set, i.e., the number of settlement cycles on that day; For the settlement period index set, For a set of nodes, This is a set member indicator function that takes one value if the price vector belongs to the set of executable constraints, and zero value otherwise.

[0009] As a preferred embodiment of the price scenario set and probability weight generation method described in this invention, the step of embedding market rule consistency and physical feasibility screening during the generation process, and outputting the price scenario set and probability weights, includes the following: through joint calculation guided by edge weights and projection of feasible sets, after generating the price vector at each time step, the diffuse graph spatiotemporal predictor first absorbs the spatial congestion and cross-market coupling information of the initial graph with a comprehensive edge weight sequence, and then projects the executable constraint set defined by market rule consistency and physical feasibility screening, so that the generated next-day price scenario set has the rationality of cross-market time alignment and spatial linkage, and highlights the executable price path under the rules and constraints with probability weights.

[0010] As a preferred embodiment of the energy storage market risk hedging scheduling method based on multi-scenario diffusion graph prediction described in this invention, the method comprises: constructing an initial spatiotemporal correlation graph of the market, using multiple market prices, exogenous factors, node locations, and congestion relationships as inputs to form an initial graph for scenario generation; based on the initial graph, a price scenario set and probability weight generation method outputs a price scenario set and probability weights; jointly generating a health profile set on the price scenario set and probability weights to obtain a time-varying marginal decay cost sequence and daily degradation budget, used to constrain the tolerable cycle intensity, and outputting the health profile set; and constructing a robust scheduling input set based on the price scenario set and health profile set, which incorporates the ambiguity set of scenario distribution and the scheduling constraint set. The retention requirements for ancillary services are uniformly expressed as executable constraints, and a robust scheduling input set is output. The hedging scheduling plan is optimized on the robust scheduling input set to obtain the SoC retention trajectory, charge / discharge window, power limit, and price threshold. The marginal decay cost is used as the shadow price correction to ensure that the worst-case return is met and the degradation does not exceed the budget, and the hedging scheduling plan is output. An execution instruction set is generated based on the hedging scheduling plan, mapping the SoC retention trajectory and charge / discharge window to market bidding and energy management control limits, and the execution instruction set is output. The execution instruction set is executed on a rolling basis, and the price and health consumption are monitored. When the observation deviation exceeds the price scenario set range, protection is triggered to lock the SoC retention, tighten the charge / discharge window, and lower the power limit, and the execution result for the day is output.

[0011] As a preferred embodiment of the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction described in this invention, the health profile set includes a time series set for scheduling constraints, including a marginal decay cost sequence and a daily degradation budget, and organized with a unified settlement period as the time index.

[0012] As a preferred embodiment of the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction described in this invention, the following steps are taken: A robust scheduling input set is constructed based on the price scenario set and the health profile set. The ambiguous set of scenario distribution, the scheduling constraint set, and the ancillary service retention requirements are uniformly expressed as executable constraints. The output robust scheduling input set includes: using the price scenario set and probability weights as input, calculating the price quantile band in each settlement cycle to obtain the boundaries between the low and high quantiles, and marking the opportunity charging zone and opportunity discharging zone of the settlement cycle accordingly. This is used to limit the allowed operation direction and time window, avoiding high-frequency small cycles in the neutral interval of the quantile band; mapping the marginal decay cost sequence in the health profile set to a minimum price difference threshold, and only when the expected price difference formed by the opportunity charging zone and the opportunity discharging zone exceeds the minimum price difference threshold in any settlement cycle. Feasible cycles are allowed, and the daily degradation budget is converted into the equivalent throughput quota allowed for the day. The throughput quota is allocated to different time periods according to the opportunity intensity of each settlement period to obtain the health constraint upper limit for each settlement period. Based on the price percentile and the health constraint upper limit, an ambiguity set of scenario distribution is constructed around the empirical distribution of price scenario set and probability weight. The ambiguity set is further defined by setting the allowable offset range and the coverage ratio of extreme paths. The ancillary service retention requirements are transformed into energy reservation and power reservation trajectories for each service period, and time alignment and conflict resolution are performed with the opportunity charging zone, opportunity discharging zone and health constraint upper limit to obtain the available power upper limit and available energy range for each settlement period. The opportunity window, minimum price difference threshold, health constraint upper limit, energy reservation and power reservation are uniformly expressed as executable constraints on the time axis to form a robust scheduling input set.

[0013] Another objective of this invention is to provide a risk hedging and scheduling system for the energy storage market based on multi-scenario diffusion graph prediction. This system can generate a set of price scenarios and probability weights for the next day by using a diffusion graph spatiotemporal predictor based on an initial graph. In the generation process, market rule consistency and physical feasibility screening are embedded to obtain the set of price scenarios and probability weights. This solves the problem that current collaborative trading and scheduling methods contain robust scheduling that does not address price uncertainty.

[0014] As a preferred embodiment of the energy storage market risk hedging scheduling system based on multi-scenario diffusion graph prediction described in this invention, the system includes: a multi-scenario diffusion graph prediction modeling module, a health constraint and robust scheduling generation module, and an integrated execution and dynamic contraction control module. The multi-scenario diffusion graph prediction modeling module is used to construct a spatiotemporal price prediction model for the energy storage market. This is achieved by unifying multiple market prices, spatial node congestion relationships, meteorological and load characteristics into a spatiotemporal graph structure, and by using the aforementioned price scenario set and probability weight generation method to generate market price evolution scenarios. The health constraint and robust scheduling generation module is used to combine battery life cost, energy constraints, and market uncertainty to generate an executable set of scheduling constraints. The integrated execution and dynamic contraction control module is used to map the scheduling scheme to the market trading and EMS system and execute closed-loop control.

[0015] Another objective of this invention is to provide a storage market risk hedging scheduling device based on multi-scenario diffusion map prediction, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the storage market risk hedging scheduling method based on multi-scenario diffusion map prediction.

[0016] Another object of the present invention is to provide a storage medium for energy storage market risk hedging scheduling based on multi-scenario diffusion map prediction, wherein a computer program is stored thereon, and when the computer program is executed by a processor, the steps of the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction are implemented.

[0017] The beneficial effects of this invention are:

[0018] The energy storage market risk hedging scheduling method based on multi-scenario diffusion graph prediction provided by this invention takes the initial spatiotemporal correlation graph of the market as a condition, unifies the parallel time series of multi-market prices, spatial congestion, and cross-market correlation into a comprehensive edge weight sequence; in the generation stage, "edge weight guidance + executable constraint projection" is introduced, and the physical feasibility constraints of rule upper and lower limits, cross-market price difference and spatial price difference are directly executed on the price vector of each time step, and the probability weight is formed by combining the historical occurrence ratio and the degree of constraint satisfaction.

[0019] After improvement, the next-day price scenario set maintains consistency with network constraints in cross-market time alignment and spatial linkage, enhancing executability. The paths entering subsequent scheduling are closer to the real constraint boundaries, reducing the injection of scenarios that do not comply with bidding and network restrictions. Unlike generation methods that only use a single market time series or post-correction, this method incorporates biological rules and physical constraints during the generation phase, avoiding a disconnect between the generated results and executable conditions. It jointly generates marginal decay cost sequences and daily degradation budgets based on price scenarios and probability weights, and maps them to minimum price difference thresholds and current allocable energy limits according to time allocation weights derived from comprehensive edge weights. This further constructs offset contraction conditions for opportunity charging zones, opportunity discharging zones, and ambiguous sets.

[0020] This method internalizes health damage costs as a trigger threshold, suppressing high-frequency small cycles within the neutral price difference range. Even during periods of distribution offset and congestion sensitivity, it maintains service energy and power reserves to prevent encroachment, ensuring degradation does not exceed the budget. Unlike scheduling algorithms that use fixed degradation coefficients or ex-post limits, this method dynamically adjusts shadow prices and quotas based on price differences, congestion, and temperature. It also uses an ambiguity set to cover both normal and extreme market conditions, unifying health constraints and price uncertainty at the same constraint level, thus improving the consistency of robust scheduling implementation.

[0021] The SoC retention trajectory, charge / discharge window, power limit, and price threshold are optimized on the robust scheduling input, and the marginal decay cost is used as the shadow price correction endogenously in the threshold and window. Then, the scheduling parameters are integrated and mapped to the market bidding and energy management control limits, and a protection trigger is set when the observed price exceeds the scenario envelope or the health consumption approaches the daily degradation budget, and the SoC retention is locked, the window is tightened, and the power is reduced.

[0022] The holistic approach ensures that transactions and controls share a unified boundary, guaranteeing that ancillary service fulfillment and risk buffers are not compromised, and maintaining a minimum return without exceeding the degradation budget under the worst-case price path. Unlike bidding and control processes based solely on expected prices or fragmented approaches, this method reduces operational conflicts and excessive loop risks through worst-case return constraints and dynamic protection loops, thereby improving the executability and robustness of scheduling. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is an overall flowchart of a risk hedging scheduling method for the energy storage market based on multi-scenario diffusion map prediction, provided in Embodiment 1 of the present invention.

[0025] Figure 2 This diagram compares the traditional strategy of the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction provided in Embodiment 2 of the present invention with that of the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0027] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for hedging and scheduling risks in the energy storage market based on multi-scenario diffusion map prediction is provided, comprising:

[0028] S1. Construct an initial graph of market spatiotemporal correlation, using multi-market prices, exogenous factors, node locations and congestion relationships as inputs to form an initial graph for scenario generation.

[0029] Furthermore, under a unified settlement cycle time index, an initial graph is constructed, with multi-market prices as node attributes, node location and congestion relationships as edge constraints, and exogenous factors as time-series drivers. In the initial graph, the same node carries the parallel time series of day-ahead energy prices, intraday energy prices, and ancillary service settlement prices. The edge set simultaneously includes cross-market related edges and spatial congestion edges; spatial congestion edges are used to express the impact of restricted lines on price linkages of adjacent nodes, while cross-market related edges are used to express the leading and lagging relationships between different markets. The exogenous factors include load forecasts, wind power forecasts, photovoltaic power forecasts, meteorological temperatures, planned maintenance information, inter-regional mutual assistance plans, and ancillary service bidding periods and durations, all organized with a unified time index as time-series attributes to drive the price linkage intensity and constraint changes of the initial graph in different time periods. The node location is a market settlement node or regional identifier, used to determine the spatial adjacency range, and together with congestion relationships, it defines the edge set and edge constraints of the initial graph. The multi-market prices include day-ahead energy prices, intraday energy prices, and ancillary service settlement prices, organized in parallel time series with the settlement cycle as the time step. Price time series from different markets are time-aligned and cross-referenced within the initial graph, enabling subsequent scenario generation to directly support risk hedging scheduling in setting charge / discharge windows, reservation states, and power limits. To ensure the initial graph remains consistent with market rules in both time and space dimensions, this implementation explicitly calculates edge weights during initial graph construction, unifying congestion relationships and cross-market price differences into a single weight framework for use as input conditions in subsequent scenario generation.

[0030] The core calculation lies in the integrated edge weight sequence of the initial graph. For each settlement period, the unified weights of spatial congestion edges and cross-market related edges are calculated, as follows: , Where i is the node index, used to indicate the market settlement node or region identifier; j is the node index, used together with i to indicate spatially adjacent node pairs; t is the time index of the settlement cycle, used to indicate the specific time period within the unified settlement cycle; This is the combined edge weight sequence for node pairs i and j at settlement period t, used to construct the initial graph. For load forecasting in settlement period t, it is an exogenous factor used to characterize the driving force of load level on congestion. The wind power generation forecast for the settlement period t is an exogenous factor used to characterize the driving force of wind power output on congestion. The photovoltaic power generation forecast for settlement period t is an exogenous factor used to characterize the driving force of photovoltaic output on congestion. The meteorological temperature during the settlement period t is an exogenous factor used to characterize the impact of temperature conditions on congestion. The planned maintenance information in settlement period t is an exogenous factor used to characterize the impact of restricted lines and maintenance status on congestion. The cross-regional mutual aid plan during settlement period t is an exogenous factor used to characterize the impact of cross-regional support on congestion. This is a spatial adjacency indicator determined by node location and congestion relationship. It is set to one when node i and node j are within the adjacency range where congestion coupling exists, and otherwise to zero. The day-ahead energy price at node i in settlement period t is a multi-market price. The price of energy at node i in settlement period t is the intraday price, which belongs to the multi-market price. The settlement price for auxiliary services at node i in settlement period t is a multi-market price. This is a cross-market association edge indicator. It takes one when i equals j, and is used to include cross-market association edges of the same node in the weight calculation; otherwise, it takes zero. is the scaling coefficient of the exogenous factor, used to map exogenous factors of different dimensions into weighted dimensionless contributions; , which is the scaling factor for cross-market price differences and price changes, used to map price differences and price changes of different dimensions into weighted dimensionless contributions; This is a proportional coefficient used to measure the relative strength of ancillary service settlement prices and intraday energy prices when calculating cross-market correlation edge weights.

[0031] It should be noted that the above-mentioned comprehensive edge weights Unify the expression of congestion relationship drivers and cross-market correlation drivers within the same formula: when When the weight is set to zero, only the weight of the spatially congested edge is retained, which is determined by the exogenous factor and spatial adjacency, reflecting the influence of node location and congestion relationship; when In one approach, the weights of cross-market related edges at the same node are superimposed on the weights of spatial congestion edges, allowing price time series from different markets to directly participate in weight calculations within the same settlement period through price differences and price change terms. Based on... The numerical values ​​are used to form an edge set and edge weight sequence in each settlement cycle, and together with the multi-market prices and exogenous factors in the node attributes, they are organized into an initial graph, which serves as the conditional input for scenario generation. Through this calculation point, the initial graph explicitly incorporates the coupling information of congestion relationships and cross-market price differences during the construction phase, providing a premise profile that is consistent with market rules and executable for the subsequent generation of price scenario sets and probability weights.

[0032] S2. Based on the initial graph, a diffusion graph spatiotemporal predictor is used to generate the price scenario set and probability weights for the next day. Market rule consistency and physical feasibility screening are embedded in the generation process to obtain the price scenario set and probability weights.

[0033] Furthermore, in this embodiment, the initial graph obtained in step S1 is used as the input condition. The initial graph includes the day-ahead energy price, intraday energy price, and ancillary service settlement price in the node attributes, and is organized as a parallel time sequence with the settlement cycle as the time step. It also includes a set of edges jointly defined by spatial adjacency indicators and a comprehensive edge weight sequence. The diffusion graph spatiotemporal predictor progressively generates the price vector of each node on the time index of the unified settlement cycle, and embeds market rule consistency and physical feasibility screening in the generation process, so that the output is the price scenario set and probability weights for the next day. To ensure that the generated price path reflects both cross-market coupling and congestion-sensitive spatial linkage, and can be directly used in subsequent scheduling, this embodiment adopts a joint calculation of "edge weight guidance and feasible set projection" in the generation stage. The comprehensive edge weight sequence of the initial graph is used as the generation guide, and the set of executable constraints is used as the boundary for progressive projection.

[0034] First, the generated results of the diffusion graph spatiotemporal predictor in each settlement period are guided by edge weights. Using the parallel price time series of the same node as the carrier, the spatial adjacency and cross-market relationship are synchronously corrected by the comprehensive edge weight sequence to obtain the guided price vector: ,

[0035] in, For node indexes, used to indicate market settlement nodes or region identifiers; It serves as a time index for the settlement cycle, used to indicate specific time periods within a unified settlement cycle; This serves as a scenario index for price paths, used to distinguish different price paths within the next day's price scenario set. In the settlement cycle Nodes generated and guided by the diffusion graph spatiotemporal predictor The price vector, in market order, consists of day-ahead energy price, intraday energy price, and ancillary service settlement price. , , These are the three components of the price vector mentioned above; , , This is the unguided generated component, i.e., the raw output of the diffuse graph spatiotemporal predictor at this step; As a spatial adjacency indicator, when a node With nodes If the adjacent range is within a congested coupling range, take 1; otherwise, take 0. In the settlement cycle Pair of nodes and The combined edge weights are derived from the edge weight sequence of the initial graph; In the settlement cycle The spatial linkage guidance coefficient is a dimensionless coefficient used to adjust the price vector correction amplitude caused by the comprehensive edge weight; In the settlement cycle The cross-market coupling guidance coefficient is a dimensionless coefficient used to adjust the magnitude of cross-market spread correction. This is a cross-market association edge indicator, and is set to one when performing cross-market coupling correction within the same node. This is a price difference correction item between the previous day and the intraday price, expressed in price.

[0036] This is a price difference correction item between the intraday and previous day prices, expressed in price. For ancillary services and intraday coupling correction items, the unit is price, where This is a proportional coefficient used to establish the coupling relationship between ancillary service settlement prices and intraday energy prices.

[0037] It should be noted that the guided price vector is projected onto the set of executable constraints defined by market rule consistency and physical feasibility in each settlement period to obtain a price vector that satisfies both the rules and physical constraints: , in, In the settlement cycle The set of executable constraints is defined by both market rule consistency and physical feasibility screening. In the settlement cycle nodes The price vector, in market order, consists of day-ahead energy price, intraday energy price, and ancillary service settlement price. These represent the lower and upper limits of the daytime energy price, respectively, in units of price.

[0038] It should also be noted that, These represent the lower and upper limits of intraday energy prices, respectively, in units of price; These represent the lower and upper limits of the settlement price for ancillary services, respectively, in units of price; This is the maximum allowable spread between markets, used to limit the price difference between the previous day and the intraday price; the unit is price. This is the maximum allowed amount for cross-market coupling, used to limit the coupling price difference between ancillary services and intraday services, expressed in price. This is the scaling factor for spatial price differences, expressed in price, used to integrate edge weights. Transformed into spatial price difference constraints; For sets The projection operator is used to map the guided price vector into a price vector that satisfies market rules and physical constraints; For projection onto the scene The set of price vector components is ultimately used to construct the price scenario set for the next day.

[0039] After generating and projecting all nodes and all settlement cycles, the next day's price scenario set is obtained, and probability weights are assigned to each price path to reflect the degree to which the rules and constraints are satisfied during the generation phase and the feasibility of the path. The probability weights are calculated as follows: , in, For the scene The probability weight is a dimensionless value used to reflect the importance of the price path in subsequent scheduling. and , is a weighting coefficient, a dimensionless value, used to balance the proportion of historical occurrences with the degree of constraint satisfaction; For the scene The historical occurrence ratio score is a dimensionless value and belongs to the internal score of the diffusion graph spatiotemporal predictor. It is used to reflect the consistency between the scene and the historical distribution. For the scene The constraint satisfaction score is a dimensionless value, obtained by statistical analysis based on the node set and time index. The number of scenarios in the price scenario set for the next day, which is a positive integer; The cardinality of the node set, i.e., the number of nodes; The base of the settlement cycle index set, i.e., the number of settlement cycles on that day; Set of settlement cycle indexes; A set of nodes; This is a set member indicator function that takes one value if the price vector belongs to the set of executable constraints, and zero value otherwise.

[0040] Through the joint calculation of "edge weight guidance and feasible set projection" mentioned above, after generating the price vector at each time step, the diffusion graph spatiotemporal predictor first absorbs the spatial congestion and cross-market coupling information of the initial graph with a comprehensive edge weight sequence, and then projects the executable constraint set defined by the consistency of market rules and physical feasibility. This makes the generated next-day price scenario set reasonable in terms of cross-market time alignment and spatial linkage, and highlights the executable price path under the rules and constraints with probability weights. This provides a directly usable scenario profile and weight basis for the robust scheduling input set and hedging scheduling plan of the next step.

[0041] S3. Generate a set of health profiles by jointly generating a set of price scenarios and probability weights, and obtain the marginal decay cost sequence and daily degradation budget that change over time. These are used to constrain the tolerable cycle intensity and output the set of health profiles.

[0042] Furthermore, using the next-day price scenario set and probability weights output in step S2 as input, and the comprehensive edge weights in the initial graph constructed in step S1 as spatiotemporal indicators of congestion and cross-market coupling, a health profile set is jointly formed. This health profile set includes a marginal decay cost sequence and a daily degradation budget, organized using a unified settlement cycle as the time index. The core idea is to estimate the tolerable cyclic intensity tendency based on scenario-weighted price fluctuations and congestion intensity in each settlement cycle, and map this tendency to the marginal cost of lifetime loss corresponding to unit charge / discharge. Then, a daily degradation budget consistent with the asset safety threshold is formed on a daily basis, ensuring that subsequent hedging scheduling is executed with health constraints as the threshold.

[0043] First, a marginal decay cost sequence is constructed, and the scenario-weighted price-driven intensity and congestion intensity are combined into the current health damage shadow price, as calculated in the following way: , in, In the settlement cycle The marginal decay cost, expressed as the lifetime loss cost per unit of electricity, is used as a shadow price of health damage in subsequent scheduling. This is the conversion factor between price-driven intensity and congestion intensity to marginal cost of life loss, with the unit being the cost of life loss per unit price. This is a scaling factor for the effect of temperature, expressed as a factor per unit of temperature, used to convert meteorological temperature into an additive factor for lifespan loss. In the settlement cycle The meteorological temperature is an exogenous factor used to characterize the impact of temperature conditions on lifespan loss. The cardinality of the node set, i.e., the number of nodes; A set of nodes; The number of scenarios in the price scenario set for the next day, which is a positive integer; For the scene The probability weight is a dimensionless value used to reflect the importance of the price path in subsequent scheduling. In the settlement cycle nodes In the scene The following is the intraday energy price, in units of price; In the settlement cycle nodes In the scene The following is the intraday energy price, in units of price; In the settlement cycle nodes In the scene The following are the day-ahead energy prices, in price units; In the settlement cycle nodes In the scene The settlement price for ancillary services is expressed in price units. This is a proportionality coefficient used to couple the ancillary service settlement price with the intraday energy price; it is a dimensionless coefficient. It is a spatial adjacency indicator, which is dimensionless, when a node With nodes If the adjacent range is within a congested coupling range, take 1; otherwise, take 0. In the settlement cycle Pair of nodes and The overall edge weight is dimensionless and is used to represent the combined strength of congestion relationship and cross-market association; This is the scaling factor for spatial price differences, expressed in price, used to integrate edge weights. This translates into the equivalent strength of spatial price differences; These are the scaling coefficients for the price and congestion terms, both of which are dimensionless and are used to adjust the contribution ratio of different driving terms to the marginal decay cost.

[0044] It should be noted that after obtaining the marginal decay cost sequence, a daily degradation budget is formed to ensure that the total daily life loss does not exceed the asset safety threshold. Quotas are then allocated on the time axis according to the opportunity intensity weighted by the scenario, as calculated below: , in, This is the daily degradation budget, expressed as the total lifespan loss, used to limit the maximum operational intensity for the day. This is the asset safety threshold, expressed in terms of total lifespan loss, used to limit the maximum permissible lifespan loss for the day. Set of settlement cycle indexes; In the settlement cycle The marginal degradation cost, expressed as the lifetime loss cost per unit of electricity; In the settlement cycle The time allocation weight is a dimensionless coefficient, obtained from the normalized result of the scenario-weighted price-driven intensity and congestion intensity, so that the sum of the weights of each settlement cycle is one. This represents the daily weighted energy throughput limit, expressed in electricity, and is used to indicate the weighted energy throughput quota that can be tolerated on a daily basis.

[0045] Through the above two calculations, the health profile set outputs the marginal decay cost in each settlement cycle and the daily degradation budget on a daily basis. The marginal decay cost sequence is driven by scenario-weighted price fluctuations and congestion intensity, with temperature as an additive factor, so that the health damage cost changes dynamically with operational tendencies and environmental conditions. The daily degradation budget is capped by the asset safety threshold and allocated in a unified settlement cycle with time allocation weights, so that the health profile set can be directly called in the subsequent robust scheduling input set and hedging scheduling plan, constraining the tolerable cycle intensity, avoiding excessive cycles during periods of high uncertainty and congestion sensitivity, and ensuring that the minimum revenue and necessary service capacity are not affected by health damage under the background of price uncertainty.

[0046] S4. Based on the price scenario set and health profile set, construct a robust scheduling input set, unify the ambiguous set of scenario distribution, scheduling constraint set and auxiliary service retention requirements into executable constraints, and output the robust scheduling input set.

[0047] Furthermore, using the next-day price scenario set and probability weights output in step S2 as input, and the marginal decay cost sequence and daily degradation budget in the health profile set output in step S3 as the source of health constraints, combined with the congestion and cross-market coupling strength reflected in the comprehensive edge weight sequence in the initial graph of step S1, a robust scheduling input set is constructed on the time index of a unified settlement cycle. The robust scheduling input set unifies the ambiguous set of scenario distribution, the scheduling constraint set, and the ancillary service retention requirements into executable constraints, which are used to directly drive the optimization of the hedging scheduling plan in the next step. The core computation point is to simultaneously characterize the triggering conditions of the opportunity charging zone and the opportunity discharging zone in each settlement cycle, and to transform the constraints of the health profile set into the current period's allocable energy limit, so that price uncertainty and health constraints are directly coupled at the executable constraint level.

[0048] It should be noted that a robust scheduling input component is constructed for each settlement period, combining the opportunity charging zone indication, the opportunity discharging zone indication, and the current allocatable energy limit into a constraint triplet, as follows: , in, In the settlement cycle The robust scheduling input component is a triplet of executable constraints, corresponding to the opportunity charging zone indicator, the opportunity discharging zone indicator, and the current allocable energy limit in sequence. This is a set member indicator function. It takes one when the condition in parentheses is true and zero otherwise. It is used to form the trigger indication of the opportunity charging zone and opportunity discharging zone in the executable constraint. In the settlement cycle The intraday energy price is calculated by weighting the scene set and spatial average, with the unit being price. It is obtained by weighting the node set and the scene set, and its weighting factor is a probability weight, which is used to characterize the intraday price center level under the price scene set. The price difference offset of the ambiguous set of the scene distribution, in units of price, is determined by a combination of historical prediction error and risk tolerance. It is used to robustly shrink the trigger criteria in the opportunity charging zone and opportunity discharging zone, so as to maintain executability under the distribution offset. In the settlement cycle The marginal decay cost, which is the life loss cost per unit of electricity, is derived from the marginal decay cost sequence of the health profile set and is used as the shadow price threshold for health damage. In the settlement cycle The time allocation weight is a dimensionless coefficient, obtained by normalizing the scenario-weighted price-driven intensity and congestion intensity, so that the sum of the weights of each settlement cycle is one. This is the daily weighted energy throughput limit, expressed in electricity, and is derived from a set of health profiles. It is used to limit the energy throughput quota that can be tolerated on a daily basis. In the settlement cycle The energy reservation corresponding to the ancillary service retention requirements is expressed in electricity. It is converted on the time axis based on the service period, minimum duration and retention capacity parameters. It is used to ensure that the energy used to meet the ancillary services is not encroached when the opportunity charging zone and opportunity discharging zone are triggered. The daily degradation budget, in units of total lifespan loss, is derived from the health profile set and is used to limit the upper limit of the conversion between allocable energy and health damage on a daily basis. It serves as a time index for the settlement cycle, used to indicate specific time periods within a unified settlement cycle.

[0049] It should also be noted that the above-mentioned constraint triplet The first component is the Opportunity Charging Zone Indicator. It determines whether charging is allowed based on whether the difference between the scenario-weighted intraday price and the price difference offset of the scenario distribution ambiguity set in adjacent settlement periods exceeds the marginal decay cost. This criterion uses health damage cost as a trigger threshold, ensuring that entry into the Opportunity Charging Zone only occurs when the price difference signal is significant and the reward for health damage is sufficient. The second component is the Opportunity Discharge Zone Indicator, judged in the same way under the opposite price difference in adjacent settlement periods. It is used to trigger discharge when the price drops significantly and exceeds the health threshold. The third component is the current allocable energy limit, calculated from the daily weighted energy throughput limit in the health profile set and the daily degradation budget. It is allocated according to time allocation weights in each settlement period, while deducting the energy reservation for ancillary service retention needs. This ensures that health constraints and service retention form a unified executable constraint at the energy allocation level.

[0050] After completing all settlement cycles After calculation, the constraint triples for each settlement period are organized into a robust scheduling input set according to the time index of a unified settlement period. This robust scheduling input set explicitly introduces an ambiguous set of scenario distributions into the triggering logic of the opportunity charging and opportunity discharging zones, robustly contracting the triggering conditions with the price difference offset magnitude. Simultaneously, it converts the current allocable energy limit into a health constraint using marginal decay cost and daily degradation budget, and deducts energy reservation based on ancillary service retention requirements, unifying price uncertainty, health constraints, and service retention within the same constraint expression. Since the time allocation weights have already absorbed the congestion and cross-market coupling strength reflected in the comprehensive edge weight sequence in step S3, the robust scheduling input set maintains the same temporal structure as the initial one in terms of opportunity strength and energy allocation. Figure 1 The spatiotemporal characteristics of the system provide directly executable constraint inputs for the next step to set the SoC reserved trajectory, charge / discharge window, and power limit without introducing high-frequency small cycles.

[0051] S5. Optimize the hedging scheduling plan on the robust scheduling input set to obtain the SoC reserved trajectory, charge / discharge window, power limit and price threshold, and use the marginal decay cost as the shadow price correction to ensure that the worst-case return meets the target and the degradation does not exceed the budget, and output the hedging scheduling plan.

[0052] Furthermore, using the robust scheduling input set constructed in step S4 as the sole input, the robust scheduling input set provides opportunity charging zone indication, opportunity discharging zone indication, and current allocable energy limit on a unified settlement period time index, and has transformed the marginal decay cost sequence and daily degradation budget in the health profile set into executable constraints. Based on this robust scheduling input set, a hedging scheduling plan is optimized, which includes the SoC reserved trajectory, charge / discharge window, power limit, and price threshold; wherein the marginal decay cost sequence is used as a shadow price correction in this step, acting on the trigger conditions of the price threshold and charge / discharge window to ensure that the net profit is not lower than the lower limit of profit under the most unfavorable price path and that the daily lifetime loss does not exceed the daily degradation budget.

[0053] It should be noted that the constraint triples of each settlement period in the robust scheduling input set are mapped to executable windows and quotas. Opportunity charging zone indicators and opportunity discharging zone indicators are used as trigger conditions for the charging and discharging windows, respectively. The current allocable energy limit is used as the energy quota for each settlement period, forming the basis for subsequent threshold setting and power limit allocation. To incorporate the marginal decay cost sequence as a shadow price correction into the price threshold, the buy and sell thresholds are set in each settlement period as follows: , , in, In the settlement cycle The buy threshold, expressed in price, is used to trigger a buy within the charging window; In the settlement cycle The sell threshold, expressed in price, is used to trigger a sell within the discharge window. This represents the lower limit of the intraday energy price, expressed in price, derived from the market rule consistency constraint in step S2. This represents the upper limit of the intraday energy price, expressed in price units, derived from the market rule consistency constraint in step S2. In the settlement cycle The intraday energy price, weighted by the scenario and spatially averaged, is expressed in price and comes from the robust scheduling input set of step S4. The price difference offset of the ambiguous set of scene distribution, in units of price, comes from the robust setting in step S4; In the settlement cycle The marginal degradation cost, expressed as the lifetime loss cost per unit of electricity, is derived from the health profile set in step S3.

[0054] It should also be noted that, under the aforementioned price threshold settings, the executable quantities for defining the charge / discharge window and energy quota are... In the settlement cycle The constraint triplets are, in order, the opportunity charging zone indicator, the opportunity discharging zone indicator, and the current allocable energy limit; let For opportunity charging zone indication, For the indication of the opportunity discharge zone, let This represents the upper limit of allocable energy for the current period, of which , and Corresponding to The first, second, and third components all originate from the robust scheduling input set of step S4. Based on this, the core guarantee condition for forming the hedging scheduling plan is: the net profit is not lower than the lower limit of profit under the most unfavorable price path, specifically: , in, The net profit under the most unfavorable price path within the scenario set, expressed in profit, is used to guarantee a minimum profit level. The scenario index for the price path is used to distinguish different price paths in the price scenario set of the next day, and comes from step S2; The number of scenarios in the price scenario set for the next day is a positive integer, derived from step S2; This is a set of settlement cycle indexes, derived from the preceding steps; In the settlement cycle The opportunity discharge zone indicator is a dimensionless indicator quantity; a value of one indicates that discharge is permitted. In the settlement cycle The current period's allocable energy limit, in units of electricity, comes from step S4; In the settlement cycle The selling threshold, in units of price; In the settlement cycle The buy threshold, in price; In the settlement cycle The marginal degradation cost, expressed as the lifetime loss cost per unit of electricity; This is the lower limit threshold for returns, expressed in units of returns. It is used to set the lower limit for achieving the worst-case return in the hedging scheduling plan.

[0055] The above-mentioned guarantee conditions are directly based on the current allocable energy limit. With marginal decay cost Deductions are created to form health constraints, and price thresholds are used. and A threshold is applied to discharge revenue to maintain consistency between the revenue floor and the healthy budget, even with shifts in the ambiguous set of scenario distributions. Consequently, the output hedging scheduling plan is given on the timeline: the SoC retention trajectory represents the lower limit of the state range maintained in each settlement cycle, ensuring it is not lower than the energy reservation corresponding to the ancillary service retention requirements; the charge / discharge window is determined by... and The allowable range marked on the timeline is determined; the power limit is based on the current allocable energy limit within each settlement period. To align quota sourcing with ancillary service retention demand, an executable maximum charging power and maximum discharging power are determined; the price threshold is determined by... and The above outputs are provided in each settlement period. All of the above outputs are organized in a unified settlement period, and the marginal decay cost is used as the shadow price for correction, with the health damage cost endogenously used as the execution threshold. When the above guarantee conditions are found to be insufficient net income under the worst price path, the charge and discharge window is automatically tightened or the price threshold is increased within the constraint boundary of the robust scheduling input set until the worst income target is met and the daily degradation budget is not exceeded.

[0056] S6. Generate an execution instruction set based on the hedging scheduling plan, map the SoC reserved trajectory and charge / discharge window to market bidding and energy management control limits, and output the execution instruction set.

[0057] Furthermore, using the hedging scheduling plan output in step S5 as the sole source, and following the time index of a unified settlement cycle, the SoC retention trajectory, charge / discharge window, power limit, and price threshold in the hedging scheduling plan are mapped to market bidding instructions on the trading side and energy management control limit instructions on the control side, respectively. These are then aggregated into an execution instruction set, ensuring that trading and control are implemented within the same boundary and avoiding conflicting operations when price uncertainty and health constraints coexist. Specifically, during the settlement cycle... Inside, when the opportunity charging area is indicated When permitted, an energy purchase quote is generated, with its price limit directly using the purchase threshold given in the hedging scheduling plan. The quoted electricity volume is recorded as And based on the current allocable energy limit This is the upper limit of the quota, while not encroaching on the energy reservation corresponding to the demand for ancillary services. Tracking preserved with SoC When the opportunity discharge zone is indicated When permitted, an energy sell quote is generated, with its price limit set at a sell threshold. The quoted electricity volume is recorded as And based on the current allocable energy limit This is the upper limit of the quota, while ensuring that the state at the end of the settlement period is not lower than the SoC reserved trajectory. For ancillary service capacity pricing, the maximum charging power is set within the service period, using the power cap given by the offsetting scheduling plan as the boundary. With maximum discharge power This is used as the upper limit of response capacity to generate the corresponding service's quoted power, denoted as... The minimum duration and response time of the quoted price are consistent with the charge / discharge window in the hedging scheduling plan, ensuring strict time alignment between service fulfillment and energy operation. Energy management control limit instructions on the control side are issued during the settlement cycle. The lower limit of the internal setting state range is Maximum charging power is The maximum discharge power is The permitted charging period and permitted discharging period are indicated by the opportunity charging zone. With opportunity discharge zone indication Trigger, the trigger price threshold is set according to the hedging scheduling plan. and The set of market bidding orders on the trading side is denoted as... The set of energy management control limit instructions on the control side is denoted as Both during the settlement cycle The organizational order is completely consistent and aligned with the market rules output from step S2 on the timeline and the set of executable constraints defined by the physical feasibility screening.

[0058] Ultimately, and This is summarized into an instruction set for execution, denoted as... This is used to execute directly on the next day according to the settlement cycle, so that energy pricing, ancillary service capacity pricing and on-site control limits share the same hedging boundary and health threshold, avoiding high-frequency small cycles in the neutral price range, while ensuring that energy and power reserves during the service period are not encroached upon.

[0059] It should be noted that, It serves as a time index for the settlement cycle, used to indicate specific time periods within a unified settlement cycle; In the settlement cycle The opportunity charging zone indicator is a dimensionless indicator quantity; it enters the charging window when permission is granted. In the settlement cycle The opportunity discharge zone indicator is a dimensionless indicator quantity, and the discharge window is entered when permission is granted. In the settlement cycle The current allocable energy limit, in units of electricity, comes from the robust scheduling input set of step S4; In the settlement cycle The buy threshold, in price, comes from the hedging schedule in step S5; In the settlement cycle The sell threshold, in price, comes from the hedging scheduling plan in step S5; In the settlement cycle The energy reservation corresponding to the ancillary service retention demand is expressed in electricity and comes from the robust scheduling input set of step S4. In the settlement cycle The lower limit of the state range of the SoC reserved trajectory is in the form of dimensionless or equivalent energy, used to ensure that service and risk buffers are not compromised, from the hedging scheduling plan in step S5. In the settlement cycle The maximum charging power, in units of power, is derived from the power limit in step S5; In the settlement cycle The maximum discharge power, in units of power, is derived from the power limit in step S5; In the settlement cycle The energy purchase quotation is in electricity volume and is a market bidding order. In the settlement cycle The quoted electricity price for selling energy is in the form of electricity volume and is a market bidding order. In the settlement cycle The quoted power for ancillary service capacity is in power and is a market bidding instruction. It is a set of market bidding orders, and organizes energy quotations and ancillary service capacity quotations according to the settlement cycle; It is a set of energy management control limit instructions, organized according to the settlement cycle, including the lower limit of the status range, the maximum charging and discharging power, the allowed charging and discharging period, and the price threshold. To execute the instruction set, for and The merged set.

[0060] S7. Executes the instruction set in a rolling manner and monitors the price and health consumption. When the observed deviation exceeds the price scenario set range, it triggers protection, locks the SoC reservation, tightens the charge and discharge window, and lowers the power limit, and outputs the execution results for the day.

[0061] Furthermore, the execution instruction set output in step S6 The execution instruction set serves as the sole basis for execution. From the market bidding instruction set Energy management control limit instruction set It is composed of components and executed on a rolling basis over a unified settlement cycle time index. Within the settlement cycle... Internally, the control side follows the energy management control limit instruction set. Set the lower limit of the state range to the SoC reserved trajectory. Set the maximum charging power to With the maximum discharge power And according to the opportunity charging area instructions With opportunity discharge zone indication Open or close the deposit / discharge window; the trading side follows the market bidding order set. Use buy threshold Submit an energy buy quote, using the sell threshold. Submit an energy sale offer, provided the offered energy volume does not exceed the current period's allocable energy limit. And it does not encroach on the energy reservation corresponding to the ancillary service retention needs. During execution, the system monitors and observes the price vector in parallel. The observed price vector is related to the cumulative amount of health consumption. In the settlement cycle Nodes returned by market data source The actual price vector is used to compare with the price envelope range of the next day's price scenario set obtained in step S2; the accumulated health consumption is denoted as... Based on the current transaction volume and marginal decay cost The superposition estimation was obtained and compared with the daily degradation budget on the time axis. Compare them.

[0062] It should be noted that when the observation bias exceeds the range of the price scenario set, i.e. during the settlement period... Internal observation price vector If the price does not fall within the price envelope range formed by the next day's price scenario set during the settlement period, protection is immediately triggered: First, the SoC reserved trajectory is locked, specifically by increasing the lower limit of the state range in the current and several subsequent settlement periods. And maintain the current level; then tighten the charging / discharging window, specifically by adjusting the opportunity charging zone indicator. With opportunity discharge zone indication Set to off in the neutral range, retaining only those meeting the price threshold. and Furthermore, it will not encroach on the energy reserve window; at the same time, the power limit will be lowered, specifically by proportionally reducing the maximum charging power in the current and subsequent settlement cycles. With maximum discharge power And maintain consistency with the responsiveness of ancillary services. When accumulated health consumption approaches the daily degradation budget, i.e. near At the same time, perform protection contraction of equal intensity to prioritize energy reservation corresponding to SoC retention trajectory and auxiliary service retention requirements. To prevent encroachment and prioritize reducing unnecessary opportunity openings, the trading side will simultaneously shrink the market bid order set upon triggering the aforementioned protection. The quoted electricity volume and effective window are used to simultaneously tighten the set of energy management control limit instructions on the control side. The charging and discharging limits ensure that transactions and controls are coordinated within the same boundary.

[0063] It should also be noted that at the end of the day, the execution results for that day are output and denoted as... The daily execution results include the actual transaction volume, actual revenue, cumulative lifespan loss, and ancillary service performance summarized by settlement cycle, and provide the SoC retention trajectory. Opportunity charging zone indicator Opportunity discharge zone indicator Maximum charging power Maximum discharge power Buy threshold With the sell threshold The daily time sequence record is used to review the consistency of inputs and parameters from steps S1 to S5 during subsequent evaluation and retraining.

[0064] To execute the instruction set, the market bidding instruction set Energy management control limit instruction set composition; For the time index of the settlement period; In the settlement cycle The lower limit of the state range of the SoC-retained trajectory; In the settlement cycle Maximum charging power; In the settlement cycle Maximum discharge power; In the settlement cycle Opportunity charging area indicator; In the settlement cycle Opportunity discharge zone indication; In the settlement cycle The purchase threshold; In the settlement cycle The selling threshold; In the settlement cycle The current allocable energy limit; In the settlement cycle The energy reservation corresponding to the ancillary service retention needs; In the settlement cycle nodes The observed price vector; In the settlement cycle The marginal decay cost; The daily degradation budget for that day; In the settlement cycle The cumulative amount of health expenditure; This is the set of execution results for the day.

[0065] Example 2, refer to Figure 2 As an embodiment of the present invention, a method for risk hedging and scheduling in the energy storage market based on multi-scenario diffusion map prediction is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0066] First, based on Figure 2 As can be seen, this technical solution is The same day The core advantages are: the right-side strategy is based on cross-market price paths generated by multi-scenario diffusion maps, and embeds market rules and congestion-sensitive spatiotemporal constraints during the generation stage; the scheduling side constructs robust opportunity windows through quantile bands and ambiguous sets, superimposed with a healthy shadow price threshold, significantly suppressing high-frequency small cycles under neutral price differences. SoC retention trajectories and power limits are mapped to bidding and EMS in an integrated manner to ensure that ancillary services are not encroached upon; health profiles (marginal decay costs, daily degradation budgets) are transformed into thresholds and throughput quotas, so that cycles are only executed when covering lifetime attrition costs, and health consumption is controlled and remains within the budget. At the same time, even in the worst-case scenario, the "worst-case return" target is still optimized, with the right-side return curve crossing the lower limit while the left-side curve does not; when market conditions deviate, protection is automatically triggered by locking the SoC, tightening the window, and reducing power, forming a robust closed loop of return-lifetime-fulfillment.

[0067] Example 3, an embodiment of the present invention, provides a risk hedging scheduling system for the energy storage market based on multi-scenario diffusion map prediction, including a multi-scenario diffusion map prediction modeling module, a health constraint and robust scheduling generation module, and an integrated execution and dynamic contraction control module.

[0068] The multi-scenario diffusion graph prediction modeling module is used to construct a spatiotemporal price prediction model for the energy storage market. It unifies multiple market prices, spatial node congestion relationships, meteorological and load characteristics into a spatiotemporal graph structure, and uses a price scenario set and probability weight generation method to generate market price evolution scenarios. The health constraint and robust scheduling generation module is used to combine battery life cost, energy constraints and market uncertainty to generate an executable set of scheduling constraints. The integrated execution and dynamic contraction control module is used to map the scheduling scheme to the market transaction and EMS system and execute closed-loop control.

[0069] This embodiment also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction proposed in the above embodiment.

[0070] This embodiment also provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction proposed in the above embodiment.

[0071] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0072] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0073] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0074] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0075] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for generating price scenario sets and probability weights, characterized in that, include: Based on the initial graph, a diffusion graph spatiotemporal predictor is used to generate the next day's price scenario set and probability weights; The generation process incorporates market rule consistency and physical feasibility screening, and outputs a set of price scenarios and probability weights.

2. The price scenario set and probability weight generation method as described in claim 1, characterized in that: The process of embedding market rule consistency and physical feasibility screening during generation includes, Using the initial graph as input, the generated results of the diffusion graph spatiotemporal predictor in each settlement period are guided by edge weights. Using parallel price time series of the same node as a carrier, a comprehensive edge weight sequence is used to synchronously correct spatial adjacency and cross-market relationships. The output guided price vector is represented as follows: ， in, For node indexing, This is a time index for the settlement period. A scenario index for price paths. In the settlement cycle Nodes generated and guided by the diffusion graph spatiotemporal predictor The price vector, in market order, consists of day-ahead energy price, intraday energy price, and ancillary service settlement price. , , These are the day-ahead energy price vector, the intraday energy price vector, and the ancillary service settlement price vector, respectively. , , This represents the portion generated before the guidance of the day-ahead energy price, intraday energy price, and ancillary service settlement price. As a spatial adjacency indicator, when a node With nodes If the adjacent element is within a range where congestion coupling exists, take 1; otherwise, take 0. In the settlement cycle Pair of nodes and The combined edge weights are derived from the edge weight sequence of the initial graph. In the settlement cycle The spatial linkage guidance coefficient is a dimensionless coefficient used to adjust the magnitude of price vector correction caused by the comprehensive edge weights. In the settlement cycle The cross-market coupling guidance coefficient is a dimensionless coefficient used to adjust the magnitude of cross-market spread correction. This is a cross-market association indicator; when performing cross-market coupling correction within the same node, it takes one value. This is a price difference correction item between the previous day and the intraday price, expressed in price. This is a price difference correction item between the intraday and previous day prices, expressed in price. For ancillary services and intraday coupling correction items, the unit is price, where This is a proportional coefficient used to establish the coupling relationship between ancillary service settlement prices and intraday energy prices.

3. The price scenario set and probability weight generation method as described in claim 2, characterized in that: The process of embedding market rule consistency and physical feasibility screening during generation, and the output price scenario set and probability weights include, The guided price vector is projected onto the set of executable constraints defined by market rule consistency and physical feasibility in each settlement period, and the output price vector that satisfies the rules and physical constraints is represented as follows: ， in, In the settlement cycle The set of executable constraints is defined by both market rule consistency and physical feasibility screening. In the settlement cycle nodes The price vector, in market order, consists of day-ahead energy price, intraday energy price, and ancillary service settlement price. These represent the lower and upper limits of the daytime energy price, respectively, in units of price; These represent the lower and upper limits of intraday energy prices, respectively, in price units. These represent the lower and upper limits of the settlement price for ancillary services, respectively, in units of price. This is the maximum allowable spread between markets, used to limit the price difference between the previous day and the intraday price, expressed in price. This is the maximum allowed cross-market coupling limit, used to restrict the coupling price difference between ancillary services and intraday services, expressed in price. This is the scaling factor for spatial price differences, expressed in price, used to integrate edge weights. Transformed into spatial price difference constraints, For sets The projection operator is used to map the guided price vector into a price vector that satisfies market rules and physical constraints. For projection onto the scene The set of price vector components is ultimately used to construct the price scenario set for the next day; After generating and projecting all nodes and all settlement cycles, the next day's price scenario set is obtained, and probability weights are assigned to each price path to reflect the degree to which the rules and constraints are satisfied during the generation stage and the feasibility of the path. The probability weights are calculated as follows: ， in, For the scene probability weights, and These are the weighting coefficients. For the scene Historical occurrence ratio score, For the scene Constraint satisfaction rating The number of scenarios in the price scenario set for the next day. The cardinality of the node set. The base of the settlement cycle index set, i.e., the number of settlement cycles on that day; For the settlement period index set, For a set of nodes, This is a set member indicator function that takes one value if the price vector belongs to the set of executable constraints, and zero value otherwise.

4. The price scenario set and probability weight generation method as described in any one of claims 1 to 3, characterized in that: The process of embedding market rule consistency and physical feasibility screening during generation, and the output price scenario set and probability weights include, By combining edge weight guidance and feasible set projection, the diffusion graph spatiotemporal predictor first absorbs the spatial congestion and cross-market coupling information of the initial graph with a comprehensive edge weight sequence after generating the price vector at each time step. Then, it projects the executable constraint set defined by the consistency of market rules and physical feasibility, so that the generated next day's price scenario set has the rationality of cross-market time alignment and spatial linkage, and highlights the executable price path under the rules and constraints with probability weights.

5. A method for risk hedging and scheduling in the energy storage market based on multi-scenario diffusion map prediction, characterized in that, include, Construct an initial graph of market spatiotemporal correlation, using multi-market prices, exogenous factors, node locations, and congestion relationships as inputs to form an initial graph for scenario generation; Based on the initial graph, the price scenario set and probability weight are output using a price scenario set and probability weight generation method as described in any one of claims 1 to 4. A health profile set is jointly generated based on the price scenario set and probability weights to obtain the marginal decay cost sequence and daily degradation budget that change over time. These are used to constrain the tolerable cycle intensity and output the health profile set. Based on the price scenario set and health profile set, a robust scheduling input set is constructed. The ambiguous set of scenario distribution, the scheduling constraint set and the auxiliary service retention requirements are uniformly expressed as executable constraints, and the robust scheduling input set is output. Optimize the hedging scheduling plan on the robust scheduling input set to obtain the SoC retained trajectory, charge / discharge window, power limit and price threshold, and use the marginal decay cost as the shadow price correction to ensure that the worst-case return is met and the degradation does not exceed the budget, and output the hedging scheduling plan. The execution instruction set is generated based on the hedging scheduling plan, and the SoC reserved trajectory and charge / discharge window are mapped to market bidding and energy management control limits, and the execution instruction set is output. The system executes the instruction set in a rolling manner and monitors price and health consumption. When the observed deviation exceeds the price scenario set range, protection is triggered, which locks the SoC reservation, tightens the charge and discharge window, and lowers the power limit. The execution results for the day are then output.

6. The energy storage market risk hedging and scheduling method based on multi-scenario diffusion map prediction as described in claim 5, characterized in that: The health profile set includes, The time series set used for scheduling constraints includes marginal decay cost sequences and daily degradation budgets, and is organized with a unified settlement period as the time index.

7. The energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction as described in claim 5 or 6, characterized in that: Based on the price scenario set and health profile set, a robust scheduling input set is constructed. This unifies the ambiguity set of scenario distribution, the scheduling constraint set, and the auxiliary service retention requirements into executable constraints. The output robust scheduling input set includes... Using price scenario set and probability weight as input, the price percentile is calculated in each settlement cycle to obtain the boundaries between the low percentile and the high percentile. Based on this, the opportunity charging zone and opportunity discharging zone of the settlement cycle are marked to limit the allowed operation direction and time window, and to avoid generating high-frequency small cycles in the neutral interval of the percentile. The marginal decay cost sequence in the health profile set is mapped to the minimum price difference threshold. In any settlement period, a feasible cycle is only allowed when the expected price difference between the opportunity charging zone and the opportunity discharging zone exceeds the minimum price difference threshold. At the same time, the daily degradation budget is converted into the equivalent throughput quota allowed on that day, and the throughput quota is allocated to the time period according to the opportunity intensity of each settlement period to obtain the upper limit of health constraints for each settlement period. Based on price quantiles and health constraint upper limits, an ambiguity set of scenario distribution is constructed around the empirical distribution of price scenario set and probability weights. The ambiguity set is determined by setting the allowable offset range and the coverage ratio of extreme paths. The ancillary service retention requirements are transformed into energy and power reservation trajectories for each service period, and time alignment and conflict resolution are performed with the opportunity charging zone, opportunity discharging zone, and health constraint upper limit to obtain the available power upper limit and available energy range for each settlement period. Opportunity window, minimum price difference threshold, health constraint upper limit, energy reservation and power reservation are uniformly expressed as executable constraints on the time axis, forming a robust scheduling input set.

8. A risk hedging and scheduling system for the energy storage market based on multi-scenario diffusion map prediction, employing the risk hedging and scheduling method for the energy storage market based on multi-scenario diffusion map prediction as described in any one of claims 5 to 7, characterized in that: It includes a multi-scenario diffusion map prediction modeling module, a health constraint and robust scheduling generation module, and an integrated execution and dynamic contraction control module; The multi-scenario diffusion map prediction modeling module is used to construct a spatiotemporal price prediction model for the energy storage market. It unifies multiple market prices, spatial node congestion relationships, and meteorological and load characteristics into a spatiotemporal map structure, and generates market price evolution scenarios using a price scenario set and probability weight generation method as described in any one of claims 1 to 4. The health constraint and robust scheduling generation module is used to combine battery life cost, energy constraints and market uncertainty to generate an executable set of scheduling constraints; The integrated execution and dynamic contraction control module is used to map the scheduling scheme to the market trading and EMS system and execute closed-loop control.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction as described in any one of claims 5 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the energy storage market risk hedging scheduling method based on multi-scenario diffusion map prediction as described in any one of claims 5 to 7.

Images