Coupling decision optimization system and method for multi-power market

Abstract

The application discloses a multi-power market-oriented coupling decision optimization system and method, and relates to the technical field of power market transactions.The system comprises a data access module, a rolling adjustable power calculation module, an irreversible capacity state machine module, a unit physical constraint access module, a cross-market optimization solution module, and a man-machine verification and automatic declaration module.The application realizes unified modeling of heterogeneous markets through unified market objects and a four-state time sequence state machine; an irreversible capacity state machine and a pre-declaration capacity gate are constructed to realize real-time deduction and concurrent consistency control of cross-market capacity through atomic transactions; multi-source data is fused to automatically calculate rolling adjustable power; a two-stage rolling optimization combined with a subsequent value function is adopted to decouple the physical delivery layer and the financial settlement layer, and to adaptively switch the difference and price settlement mode.The application realizes a second-level online decision and automatic declaration closed loop, eliminates over-issuance risks, improves income, controls downward risks, and meets compliance audit requirements.

Description

Technical Field

[0001] This invention belongs to the field of power system automation and power market trading technology, and in particular relates to a coupled decision optimization system and method for multiple power markets. Background Technology

[0002] Currently, power generation companies mainly use the following existing technologies when making cross-market transaction decisions and providing system support: The first type is the independent support model for subsystems. The medium-to-long-term, day-ahead, and real-time markets are each managed by independent business systems or manual ledgers, with a lack of data linkage between systems. Allocable electricity in each market is usually estimated manually in spreadsheets, and capacity occupancy relationships between markets rely on manual verification.

[0003] The second type is the sequential decision-making single-market optimization model. This model establishes optimization models for each individual market according to their opening time, using the transaction results of the previous market as fixed input, optimizing only the current market, and not passing on the remaining capacity status to subsequent markets.

[0004] The third type is the fixed-ratio allocation mode. The system mechanically allocates the predicted power according to a preset fixed ratio, and the adjustment of the ratio is highly dependent on human experience.

[0005] The fourth type is a research-based approach based on a general optimization framework. Several multi-market joint bidding frameworks have been proposed in academia, such as modeling sequential markets based on multi-stage stochastic programming, approximating sequential bids using value functions based on approximate dynamic programming, and simulating bid equilibrium based on multi-agent deep reinforcement learning.

[0006] However, the aforementioned existing technologies have the following significant technical shortcomings when dealing with the complex scenarios of the electricity market: First, there is a lack of system-level irreversible state management for cross-market capacity occupancy relationships. In the independent support and sequential decision-making model of subsystems, the deduction of available capacity in subsequent markets after a transaction in one market relies on manual verification or lacks any linkage. The system lacks a state carrier that spans all markets, records locked and remaining allocable capacity, and guarantees irreversibility. This easily leads to data inconsistency issues such as over-issuance, duplicate occupancy, and capacity conflicts, and it is impossible to block illegal applications in real time at the moment of a transaction.

[0007] Secondly, heterogeneous clearing mechanisms are difficult to model and trigger uniformly within a single system. Medium- and long-term markets often involve bilateral negotiation or centralized bidding, rolling matching markets use continuous order book matching, and day-ahead and real-time markets use unified marginal clearing. These three mechanisms have completely different application time windows, transaction judgment logic, and data interfaces. Existing single-market systems cannot be compatible with these three mechanisms under a unified data structure and time-series triggering mechanism, especially lacking support for rolling matching / continuous matching markets and adjustable electricity calculation.

[0008] Secondly, the calculation of adjustable electricity volume in the rolling matching process relies on manual labor and data fusion is prone to errors. The adjustable electricity volume in the rolling matching market needs to integrate three types of multi-source heterogeneous data: power forecast, contracted electricity volume, and traded rolling adjustment volume. These data typically come from different systems and spreadsheet templates, with inconsistent field naming and time period column naming formats. Manual alignment, contract weighting and merging, and time period column parsing are highly prone to errors, and the calculation results cannot be fed back into the cross-market optimization model in real time.

[0009] Furthermore, the general optimization framework is not tied to the physical constraints and hardware links of heterogeneous markets, making direct engineering implementation difficult. Existing multi-stage stochastic programming, approximate dynamic programming, or multi-agent reinforcement learning solutions are mostly geared towards two or three types of homogeneous markets, failing to cover the five types of heterogeneous markets, including rolling matching, inter-provincial and regional mutual assistance, and their irreversible physical delivery constraints. These solutions are mostly offline simulations, failing to form an online closed loop with physical constraints such as unit ramp-up, start-up and shutdown, minimum output, and channel capacity, as well as the automatic reporting interface of the trading platform. Simultaneously, some black-box methods are uninterpretable in compliance audit scenarios, requiring retraining for rule changes, and cannot be directly used as online-running, auditable engineering systems.

[0010] Finally, existing solutions largely assume a single settlement rule, making it difficult to adapt to the switch between volume-based settlement and price-based settlement. Existing solutions generally implicitly assume that medium- and long-term contract electricity is physically deducted, while spot trading only involves the remaining volume in volume-based settlement. Under price-based settlement, however, all grid-connected electricity is cleared and settled at the spot price, and medium- and long-term contracts degenerate into financial price difference contracts for separate settlement, without occupying physical saleable capacity. Existing systems lack the ability to adaptively switch capacity definitions and revenue models according to market settlement rules, making them difficult to apply across settlement rules and regions. Furthermore, the lack of reliable human-machine verification and automated reporting loops between optimization results and execution leads to a disconnect between decision-making and execution.

[0011] Therefore, how to provide a coupled decision optimization system and method that can adapt to multiple heterogeneous market clearing mechanisms, realize irreversible state management of remaining capacity across markets, automatically calculate adjustable electricity, adaptively switch settlement rules, and have online closed-loop execution capabilities is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0012] In view of this, the present invention aims to propose a coupled decision optimization system and method for multiple electricity markets, in order to overcome the shortcomings of the existing technology, such as the lack of system-level irreversible state management of cross-market capacity occupancy leading to over-generation and capacity conflicts, the difficulty in uniformly modeling and triggering heterogeneous clearing mechanisms, the dependence of rolling matching of adjustable power calculation on manual labor and the error-prone data fusion, the difficulty of binding the general optimization framework with the physical constraints of generating units and the hardware links of the trading platform to form an online closed loop, the difficulty in adapting to the smooth switching of differential settlement and differential price settlement rules, and the lack of reliable human-machine verification and automatic reporting closed loop from optimization results to execution, which leads to the disconnect between decision-making and execution.

[0013] To achieve the above objectives, the technical solution of the present invention is implemented as follows: This invention provides a coupled decision optimization system for multiple electricity markets. The system includes a data access and market object modeling module, a rolling adjustable power calculation module, an irreversible capacity state machine module, a unit physical constraint access module, a cross-market optimization solution module, and a human-machine verification and automatic reporting module. The data access and market object modeling module is used to facilitate the participation of power generation companies. Each electricity market establishes a unified market description object, which includes market type, price formation mechanism, settlement model, decision-making time window, clearing time, delivery period set, upper and lower limits of declared electricity volume, and interface description. It is an integer not less than 4; The adjustable power calculation module is used to integrate three types of multi-source heterogeneous data: power prediction, contract power, and rolling adjustment, and eliminate format ambiguity to calculate the adjustable power of each station at each time period. The unit physical constraint access module is used to read the upper and lower limits of ramp rate, start and stop status and minimum start time, maximum and minimum output and upper limit of transmission channel capacity constraints from the generator dispatch system, and use them as the feasible domain boundary input of the cross-market optimization solution module; The irreversible capacity state machine module is used to maintain capacity state objects for each delivery day according to the delivery period, and realize declaration reservation, atomic deduction after clearing and snapshot generation; The cross-market optimization solution module is used to adaptively switch between capacity caliber and revenue model according to the settlement mode. It adopts a two-stage rolling optimization combined with the subsequent value function to solve the suggested electricity volume and suggested electricity price for the market currently in the decision window. The human-machine verification and automatic declaration module is used to automatically submit declarations through the power trading platform interface after opportunity cost verification and manual compliance review, and automatically obtain the transaction results and write them back to the irreversible capacity state machine module after clearing.

[0014] Furthermore, the adjustable power calculation module is specifically used for: Header identification and field parsing are performed on three types of Excel data: power forecast, contracted power volume, and rolling adjustment. To address ambiguity in column name prefixes for different time periods, a reverse matching method from largest to smallest is used to determine the column segment number. When multiple contracts exist for the same station at the same time period, the combined electricity volume is calculated using the following formula. and weighted average electricity price : ; ; in For the first The electricity volume of the contract, For the first The electricity price under the contract; Calculate the adjustable power using the following formula. : ; in For station Time period Power prediction of electricity consumption, For station Time period Contracted electricity volume, For station Time period Adjust the power using the roller; when When it is positive, it serves as the upper limit for eligible electricity sales; when... When the value is negative, its absolute value is used as the rigid mandatory purchase quantity to generate a separate electricity purchase declaration.

[0015] Furthermore, the irreversible capacity state machine module is specifically used for: definition For time period The declared reserve quantity awaiting clearance For the market The remaining salable quantity after the decision; During initialization, the time periods are... Set as capacity limit. Set to 0; When the market During the delivery period Number of applications submitted At that time, a capacity gate check is performed before the application is submitted to verify whether the submitted quantity does not exceed the current available sales capacity after deducting concurrent reservations and whether it meets the unit's physical constraints; if the check passes, the reservation operation is performed using atomic transactions. ; When the market During the period Clear out and write back the actual transaction volume Then, release the reserved and update the remaining marketable inventory using atomic transactions: ; and the actual transaction volume The locked battery level details are written with an irreversible flag, whereby... This represents the remaining available stock in the subsequent market. For the market During the period The actual transaction volume; Each time the capacity state object changes, a data snapshot with a millisecond-level timestamp is generated.

[0016] Furthermore, the cross-market optimization solution module is specifically used for: Price scenarios are generated based on the joint distribution of historical prices, and subsequent value functions are estimated. , Indicates time period The remaining available quantity is The expected return that can be obtained from the optimal allocation across all subsequent markets, and about It is monotonically non-decreasing and marginally decreasing. Construct the objective function for a two-stage solution: ; The constraints are: ; in For transaction functions, For all known information at the decision moment, For time period The unit's output is adjustable. For time period Channel capacity, For the market During the period The expected clearing price For the market The lower limit of the declared electricity volume, For the market The upper limit of the declared electricity volume, For time period Total adjustable power, For mathematical expectation, For the market During the period The number of applications, For time period The declared reserve quantity awaiting clearance For the market The remaining salable quantity after the decision. For the market During the period The declared electricity price; Calculate the marginal opportunity cost across markets: ;in This is the left derivative of the subsequent value function at the current remaining salable quantity; when the expected marginal revenue corresponding to the proposed electricity price is lower than... At that time, an early warning is triggered.

[0017] Furthermore, the cross-market optimization solution module decouples capacity and revenue modeling into a physical delivery layer and a financial settlement layer, and applies it according to the market. Settlement mode Field-driven branch: when When settling accounts for the difference, the electricity volume under medium- and long-term contracts is considered a rigid deduction from the physical available capacity, and the revenue is the sum of the product of the transaction volume and the transaction price in each market. when When settling price differences, all available resources are used to clear positions in the spot market, with medium- and long-term contracts serving as financial CFD positions. Settlement is done separately. The initial available sales volume is the upper limit of physical deliverables. Revenue is the sum of the revenue from the total spot electricity volume and the electricity cost of the medium- to long-term CFD. The revenue from the total spot electricity volume equals the day-ahead declared volume multiplied by the day-ahead clearing price plus the actual generation minus the day-ahead declared volume multiplied by the real-time clearing price. The electricity cost of the medium- to long-term CFD is... ,in The contract price. The spot price serves as a reference point for medium- to long-term settlement. This refers to the medium- to long-term contract for difference (CFD) number. For time period No. The position size of long-term CFDs.

[0018] Based on the same concept, this invention also provides a coupled decision optimization method for multiple electricity markets, applied to a coupled decision optimization system for multiple electricity markets, comprising the following steps: Step 1, involving power generation companies Each electricity market establishes a unified market description object and advances the four-state sequential state machine migration according to the decision-making time window and clearing time of each market. The four states include not open market, decision-making window open, declared pending clearing, and cleared locked. Step 2: For each delivery day, maintain the capacity status object according to the delivery period, perform capacity gate verification before declaration, execute the declaration reservation with atomic transactions after the verification passes, release the reservation with atomic transactions after clearing and irreversibly deduct the remaining market salable quantity, and generate a data snapshot; Step 3: Obtain the physical constraints of the generator set from the generator set dispatching system as the boundary of the feasible region for optimization; Step 4: Integrate power prediction, contracted power volume and rolling adjustment data, eliminate ambiguity in the column name prefixes for each time period, and calculate the adjustable power volume for each station in each time period; Step 5: Adaptively switch between capacity caliber and revenue model based on the settlement mode field of the market description object, use two-stage rolling optimization combined with subsequent value function to solve for the suggested declared electricity volume and suggested declared electricity price, and calculate the cross-market marginal opportunity cost for verification; Step 6: After manual compliance review and capacity gate verification, the application is automatically submitted through the power trading platform interface, and the capacity status object is written back after clearing.

[0019] Furthermore, in step 4, the adjustable power is calculated. The formula is: ;in, For station Time period Power prediction of electricity consumption, For station Time period Contracted electricity volume, For station Time period Adjust the power using the roller; when When the positive time is used as the upper limit for the electricity that can be declared for sale, When it is negative, its absolute value is used as the rigid mandatory purchase quantity to generate a separate electricity purchase declaration.

[0020] Furthermore, in step 5, when the settlement mode is price difference settlement, all available power output enters the spot market for clearing, and the medium- and long-term contracts are settled separately as financial price difference contract positions, without deducting the physical sellable capacity. The revenue includes the revenue from the total spot power output and the electricity fees from the medium- and long-term price difference contracts.

[0021] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention achieves cross-market data consistency and irreversible control, eliminating the risks of over-issuance and capacity conflicts at the system level. Through an irreversible capacity state machine and a pre-application capacity gate, this invention performs atomic deductions in real time at the moment of transaction and application, blocking over-applications. This ensures the atomicity and irreversibility of cross-market capacity management, and provides millisecond-level data snapshots to guarantee the traceability and auditability of the entire decision-making process.

[0022] 2. Achieved unified support for heterogeneous markets and fusion calculation of rolling-match data. This invention uses a unified market description object to accommodate three heterogeneous mechanisms: bilateral negotiation, continuous order book matching, and unified marginal clearing, and comprehensively covers five types of electricity markets. At the same time, through automatic parsing of multi-source data, time-period reverse disambiguation, and contract weighted merging, it realizes automated calculation of adjustable electricity volume in rolling-matching, eliminates human calculation errors, and feeds the calculation results back to the cross-market optimization model in real time.

[0023] 3. Possesses high real-time engineering online operation capability. This invention adopts a two-stage rolling optimization combined with subsequent value function approximation method, reducing the exponential complexity of multi-stage stochastic optimization to linear, meeting the requirement of second-level online solution in large-scale scenarios; at the same time, the optimization input is deeply bound to the physical constraints of the unit, and the output results are automatically reported through the trading platform interface after opportunity cost verification and human-machine compliance review, realizing an integrated closed loop of decision-making and execution.

[0024] 4. Significantly improved overall returns while achieving controllable and auditable risks. Through cross-market capacity coordination and opportunity cost quantification, this invention significantly improved the total returns of power generation companies in simulation verification; the introduction of Conditional Value at Risk (CVaR) constraints ensured a bottom line for returns under extremely unfavorable scenarios, and each decision recommendation was accompanied by a comparison of opportunity costs and scenario returns, possessing high interpretability and meeting the compliance audit requirements of power trading.

[0025] 5. Adaptive switching and cross-regional applicability of settlement rules are achieved. This invention decouples the physical delivery layer from the financial settlement layer, enabling the same system to automatically switch between capacity definitions, revenue models, and opportunity cost definitions under both differential settlement and price difference settlement rules. This design not only adapts to market rules in different regions but also allows for a smooth switch from differential settlement to price difference settlement, requiring only configuration modifications without system reconstruction. Attached Figure Description

[0026] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 A schematic diagram of the overall architecture of a coupled decision optimization system for multiple electricity markets is provided for embodiments of the present invention. Figure 2 This is a schematic diagram of a multi-market opening sequence state machine provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the irreversible remaining capacity state machine and the pre-application capacity gate process provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the flow chart of the adjustable power calculation module for rolling matching provided in an embodiment of the present invention; Figure 5 A schematic diagram of the two-stage rolling optimization and opportunity cost verification process provided in this embodiment of the invention; Figure 6 This is a schematic diagram of the online closed-loop timing of human-machine verification and automatic reporting provided in an embodiment of the present invention. Detailed Implementation

[0027] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0028] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0029] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0030] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0031] like Figures 1 to 6 As shown, this invention provides a coupled decision optimization system and method for multiple electricity markets. The coupled decision optimization system for multiple electricity markets includes a data access and market object modeling module, a rolling adjustable power calculation module, an irreversible capacity state machine module, a unit physical constraint access module, a cross-market optimization solution module, and a human-machine verification and automatic reporting module. The data access and market object modeling module is used to facilitate the participation of power generation companies. Each electricity market establishes a unified market description object, which includes market type, price formation mechanism, settlement model, decision-making time window, clearing time, delivery period set, upper and lower limits of declared electricity volume, and interface description. It is an integer not less than 4; The adjustable power calculation module is used to integrate three types of multi-source heterogeneous data: power prediction, contract power, and rolling adjustment, and eliminate format ambiguity to calculate the adjustable power of each station at each time period. The unit physical constraint access module is used to read the upper and lower limits of ramp rate, start and stop status and minimum start time, maximum and minimum output and upper limit of transmission channel capacity constraints from the generator dispatch system, and use them as the feasible domain boundary input of the cross-market optimization solution module; The irreversible capacity state machine module is used to maintain capacity state objects for each delivery day according to the delivery period, and realize declaration reservation, atomic deduction after clearing and snapshot generation; The cross-market optimization solution module is used to adaptively switch between capacity caliber and revenue model according to the settlement mode. It adopts a two-stage rolling optimization combined with the subsequent value function to solve the suggested electricity volume and suggested electricity price for the market currently in the decision window. The human-machine verification and automatic declaration module is used to automatically submit declarations through the power trading platform interface after opportunity cost verification and manual compliance review, and automatically obtain the transaction results and write them back to the irreversible capacity state machine module after clearing.

[0032] A coupled decision optimization method for multiple electricity markets, applied to a coupled decision optimization system for multiple electricity markets, includes the following steps: Step 1: Establish a unified market description object for each of the K electricity markets in which power generation companies participate, and advance the four-state sequential state machine migration according to the decision time window and clearing time of each market. The four states include not open market, decision window open, declared pending clearing, and cleared and locked. Step 2: For each delivery day, maintain the capacity status object according to the delivery period, perform capacity gate verification before declaration, execute the declaration reservation with atomic transactions after the verification passes, release the reservation with atomic transactions after clearing and irreversibly deduct the remaining market salable quantity, and generate a data snapshot; Step 3: Obtain the physical constraints of the generator set from the generator set dispatching system as the boundary of the feasible region for optimization; Step 4: Integrate power prediction, contracted power volume and rolling adjustment data, eliminate ambiguity in the column name prefixes for each time period, and calculate the adjustable power volume for each station in each time period; Step 5: Adaptively switch between capacity caliber and revenue model based on the settlement mode field of the market description object, use two-stage rolling optimization combined with subsequent value function to solve for the suggested declared electricity volume and suggested declared electricity price, and calculate the cross-market marginal opportunity cost for verification; Step 6: After manual compliance review and capacity gate verification, the application is automatically submitted through the power trading platform interface, and the capacity status object is written back after clearing.

[0033] Specifically, the embodiments of the present invention are as follows: I. System Overall Architecture and Module Functions like Figure 1 As shown, the coupled decision optimization system for multiple power markets provided by this invention adopts a four-layer architecture design, including an application layer, a computing layer, a data layer, and an external system interface layer.

[0034] The application layer, oriented towards the user interface, includes a decision calendar module, a rolling adjustable power workbench module, a remaining capacity dashboard module, an opportunity cost early warning module, and a declaration module. The decision calendar module displays the opening, closing, clearing, and settlement periods of various markets—medium- and long-term, rolling matching, day-ahead, real-time, inter-provincial, and regional mutual aid markets—in a timeline format. It automatically identifies the current decision node, locked markets, and available markets, triggering corresponding optimization calculations when a market enters the decision window. The rolling adjustable power workbench module supports batch import of three types of Excel data: power forecast, contract power, and rolling adjustment. It automatically completes header recognition, time period column disambiguation, station name mapping, and contract weighted merging. It calculates the adjustable power for each station and time period using the formula: adjustable power equals power forecast power minus contract power minus rolling adjustment power, and distinguishes buy and sell directions with positive and negative signs. The remaining capacity dashboard module displays the total power generation forecast, locked power, and remaining allocable power in each market in real time. After a transaction in any market, the irreversible capacity state machine automatically deducts the excess capacity and blocks over-declaration. The opportunity cost early warning module issues an alert when electricity is planned to be sold in a certain market at a price lower than the opportunity cost across markets, and displays a comparison of expected returns waiting for subsequent markets. The application module automatically submits applications through the trading platform interface after the optimization results have been verified by opportunity cost and manually reviewed for compliance. After clearing, it automatically obtains the transaction results and writes back the capacity status, revenue, and forecast.

[0035] The computational layer comprises eight core computational modules. The data access and market object modeling module establishes unified market description objects for each of the K electricity markets in which power generation companies participate. The unit physical constraint access module reads constraints such as ramp rate limits, start-up and shutdown status, minimum start-up time, maximum and minimum output, and transmission channel capacity limits from the generator dispatch system, and uses these as the feasible domain boundary input for the cross-market optimization solution module. The rolling adjustable power calculation module integrates three types of multi-source heterogeneous data—power forecast, contract power, and rolling adjustment—and eliminates format ambiguity to calculate the adjustable power for each station at each time period. The irreversible capacity state machine module maintains capacity state objects for each delivery day and delivery period, enabling declaration reservation, atomic deduction after clearing, and snapshot generation. The scenario generation and CVaR module generates price scenarios based on the joint distribution of historical prices and introduces conditional value-at-risk constraints to control downside risk. The two-stage optimization module and settlement mode adaptive module adaptively switch between capacity caliber and revenue model based on the settlement mode field, employing two-stage rolling optimization combined with subsequent value functions to solve for suggested declared power and suggested declared electricity prices in the current decision window market. The opportunity cost verification module calculates the cross-market marginal opportunity cost and compares it with the expected marginal revenue corresponding to the proposed electricity price. The human-machine verification and automatic submission module automatically submits the application through the power trading platform interface after opportunity cost verification and manual compliance review.

[0036] The data layer comprises five data storage units: a market object repository, a price joint distribution, a CapState snapshot, contract and rolling data, and medium- and long-term CFD positions. The external system interface layer includes interfaces for the power trading platform and the generator dispatching system, used for submitting clearing queries and obtaining physical constraints such as ramp-up, start-up, shutdown, output, and channel capacity, respectively.

[0037] II. Market Sequence State Machines and Unified Modeling like Figure 2 As shown, this invention maintains a four-state sequential state machine for each market. The state set includes: not open, decision window open, submitted but awaiting clearing, and cleared and locked. The system is for participating... Each electricity market shall establish a unified market description object. ,in Market type is an integer not less than 4. Values ​​are determined based on medium- to long-term, rolling matching, day-ahead, real-time, inter-provincial, and regional mutual assistance price formation mechanisms. Values ​​are determined through bilateral negotiation or centralized bidding, continuous order book matching, and unified marginal clearing in the market. Settlement mode The values ​​are derived from differential settlement and differential price settlement. Each market description object also includes a decision time window, clearing time, delivery period set, upper and lower limits of declared electricity volume, and interface description.

[0038] The background scheduling thread advances state transitions according to the decision time window and clearing time of each market. When a market transitions to the decision window open state, the cross-market optimization solution module is triggered to solve for that market. When a market transitions to the cleared and locked state, the irreversible capacity state machine module is triggered to execute atomic transaction updates. After the state transition is completed, the system enters the next settlement period loop, repeating the above process until all markets are settled.

[0039] III. Irreversible Capacity State Machine and Pre-Declaration Capacity Gate like Figure 3 As shown, the irreversible capacity state machine module maintains a CapState capacity state object for each delivery day, categorized by delivery period. Let the delivery day be divided into... Each delivery period For each time period The total power generation forecast of power generation enterprises is The capacity status object organizes fields by delivery period, including total generation forecasts and their distribution for each period, and market data. Remaining salable inventory after decision Time period The declared reserve quantity awaiting clearing And details of locked battery levels.

[0040] During system initialization, Set as capacity limit. Set to 0. When the market During the period Number of applications submitted At this time, the system first performs a pre-declaration capacity gate check, verifying whether the declared quantity for each delivery period covered by the market does not exceed the current available sales volume after deducting concurrent reservations, and simultaneously checking whether it meets the unit physical constraints. If any delivery period fails the check, the declaration is blocked and an alarm is triggered, returning the feasible upper limit. If the check passes, the reservation operation is performed using atomic transactions, and the capacity is allocated to the designated storage capacity. Increase The amount is temporarily locked to prevent it from being repeatedly allocated in the concurrent market, and then submitted to the trading platform to await clearing.

[0041] When the market During the period Clear out and write back the actual transaction volume Then, the system releases the reservation using an atomic transaction. minus Update the remaining available stock for all subsequent market transactions and include the actual transaction volume. The locked electricity volume details are written with an irreversible flag. Once committed, this deduction transaction cannot be reversed within the system, corresponding to the physical rule in the electricity market that traded electricity volume cannot be repurchased or cancelled. Each time the capacity status object changes, the system generates a data snapshot with a millisecond-level timestamp, recording the source of the change and the values ​​before and after the change, to support retrospective historical decision-making based on time points.

[0042] Specifically, the declaration reservation operation is represented as follows: (1); Clearing write-back and irreversible deduction operations are represented as follows: (2); in For time period The declared reserve quantity awaiting clearance For the market During the period The number of applications, For the market The remaining salable quantity after the decision. For the market During the period The actual transaction volume.

[0043] IV. Adjustable Power Calculation and Data Fusion for Rolling Tap like Figure 4 As shown, the adjustable power calculation module integrates three types of heterogeneous data: power prediction Excel, contract power Excel, and adjustable power calculation Excel. First, it performs header recognition and field parsing, automatically identifying the station name column, date column, and time period column. To address the ambiguity of the time period column name prefix, a reverse matching method from largest to smallest is used. Each column name is matched in descending order from segment 24 to segment 1, i.e., time period reverse disambiguation. The first match is the segment number of that column, ensuring that larger numbers take precedence and are not mismatched by smaller prefixes.

[0044] Following this, station mapping alignment is performed. Specifically, station names are automatically mapped, and a mapping table is maintained between external station names and internal station codes, normalizing station names from different data sources and markets into a unified code. When multiple contracts exist for the same station in the same time period, the electricity price is weighted and averaged based on the electricity volume of each contract. The combined electricity volume and combined electricity price are calculated using the following formulas: (3); (4); in For the first The electricity volume of the contract, For the first The electricity price under the contract, The total electricity consumption after merging. The system uses a weighted average electricity price. It also supports incremental consolidation, ensuring that batch arrivals are equivalent to a one-time full consolidation.

[0045] Adjustable power consumption is calculated using the following formula: (5); in For station Time period Power prediction of electricity consumption, For station Time period Contracted electricity volume, For station Time period Adjust the power using the roller. For station Time period Adjustable power consumption. When When positive, it indicates that there is a surplus of electricity available for sale during that period. This value serves as the upper limit for the electricity that the power station can declare for sale in the rollover market during that period, and is included in the initial surplus of electricity available for sale during that delivery period. When... When the value is negative, it indicates that the contract was over-signed and there is a delivery gap during that period. Its absolute value is used as a rigid purchase quantity to generate a separate electricity purchase declaration for fulfillment, and is not included in the cross-market electricity sales surplus allocation and irreversible transmission chain.

[0046] V. Two-stage rolling optimization and opportunity cost verification like Figure 5 As shown, the cross-market optimization solution module employs a two-stage rolling optimization method combined with a subsequent value function. At each decision node, all undecided markets are aggregated into a virtual subsequent market, and the problem degenerates into two stages. The first stage determines the declared electricity volume and declared electricity price for the current market, and the second stage invests the remaining capacity into the subsequent market to obtain the expected subsequent revenue.

[0047] First, enter the current market. Based on the remaining available stock and the observed actual clearing price, a conditional scenario is generated. This is achieved using Latin hypercube sampling, based on the joint distribution of historical prices. Each price scenario is analyzed. When the previous market has cleared and the actual clearing price has been observed, the system directly resamples from the conditional distribution to generate a subsequent price scenario consistent with the observed state. For the parametric case, Latin hypercube sampling is performed again based on the conditional mean and conditional covariance; for the nonparametric case, historical samples are weighted according to observation bias using kernel weights.

[0048] Subsequent value function Indicates time period The remaining available quantity is This represents the expected return that can be obtained from the optimal allocation across all subsequent markets. The estimation method is based on the generated price scenarios, and for each scenario, [the following is a partial equation / parameter] is used. The unit electricity volume is greedily allocated from high to low according to the clearing price of each subsequent market during that period. The revenue of this scenario is obtained by summing the transaction revenue of each market, and the average is taken over all scenarios. . about It is monotonically non-decreasing and marginally decreasing.

[0049] The objective function for the two-stage solution is: (6); The constraints are: (7); in This is the transaction function, which depends on the price formation mechanism. In a market with a uniform marginal clearing mechanism, all bid volumes are executed when the bid price is not higher than the clearing price; otherwise, no volume is executed. In a market with continuous order book matching, matching is based on the current order book status, and partial execution is allowed. In a bilaterally negotiated market, the bid volume is the executed volume, and the bid price is the executed price. For all known information at the decision moment, For the market During the period The declared electricity volume For the market During the period The declared electricity price For time period The unit's output is adjustable. For time period Channel capacity, For the market During the period The expected clearing price For the market The lower limit of the declared electricity volume, For the market The upper limit of the declared electricity volume, For time period Total adjustable power, For mathematical expectation, For time period The declared reserve quantity awaiting clearance For the market The remaining salable quantity after the decision.

[0050] The two-dimensional problem for each time period is solved by combining grid search with scene sampling. Each delivery period is independent and can be computed in parallel.

[0051] The opportunity cost verification module calculates the cross-market marginal opportunity cost: (8); in This is the left derivative of the subsequent value function at the current remaining salable quantity. This refers to the cross-market marginal opportunity cost. When the expected marginal revenue corresponding to the proposed electricity price is higher than... At that time, a capacity gate verification is performed before the application is submitted. If the expected marginal revenue corresponding to the proposed electricity price is lower than the threshold, the verification will be conducted. When the time comes, an alert is triggered and a comparison of the expected returns of the two strategies—selling in the current market and holding until the next market—is output under different price scenarios.

[0052] In another implementation, the optimization objective of maximizing the risk-adjusted expected total return includes a conditional value at risk constraint: (9); in This is the confidence coefficient, with a default value of 0.95. This is the preset minimum acceptable profit. Value at risk is determined by the conditions. Total returns are sorted in ascending order by scenario, with the worst return being selected. The average of proportional scenarios ensures that the total return does not fall below the bottom line under extremely unfavorable conditions.

[0053] VI. Human-machine verification and automatic reporting closed loop like Figure 6 As shown, the online closed-loop process of human-machine verification and automatic reporting involves five participants: the optimizer, human-machine verification, the capacity gate, the trading platform API, and the capacity snapshot library. The optimizer submits the optimal solution and opportunity cost to the human-machine verification team, who only perform compliance review without altering the algorithm's optimization process. After successful review, a report is requested to the capacity gate. The capacity gate performs pre-reporting capacity gate verification; upon successful verification, the report is automatically submitted via the trading platform API. After the trading platform returns the clearing result, the capacity gate performs atomic deduction and generates a snapshot in the capacity snapshot library. The capacity snapshot library writes back the capacity status, and the system moves to the next market, cycling until all markets are completed.

[0054] VII. Settlement Mode Adaptive Mechanism The cross-market optimization solution module decouples capacity and revenue modeling into a physical delivery layer and a financial settlement layer, and then optimizes it according to the market. Settlement mode Field-driven branch: when When settling accounts for the difference, the electricity volume under medium- and long-term contracts is considered a rigid deduction from the physical available capacity, and the revenue is the sum of the product of the transaction volume and the transaction price in each market. when When settling price differences, all available resources are used to clear positions in the spot market, with medium- and long-term contracts serving as financial CFD positions. Settlement is done separately. The initial available sales volume is the upper limit of physical deliverables. Revenue is the sum of the revenue from the total spot electricity volume and the electricity cost of the medium- to long-term CFD. The revenue from the total spot electricity volume equals the day-ahead declared volume multiplied by the day-ahead clearing price plus the actual generation minus the day-ahead declared volume multiplied by the real-time clearing price. The electricity cost of the medium- to long-term CFD is... ,in The contract price. The spot price serves as a reference point for medium- to long-term settlement. This refers to the medium- to long-term contract for difference (CFD) number. For time period No. The position size of long-term CFDs.

[0055] In this embodiment, the cross-market optimization solution module decouples capacity and revenue modeling into a physical delivery layer and a financial settlement layer. The physical delivery layer is always valid; the physically deliverable electricity volume at any delivery period is consistently constrained by the actual adjustable output of the generating units and the channel capacity, and physical delivery is irreversible once it occurs. The financial settlement layer is market-based. Settlement mode Field-driven branching.

[0056] Under the differential settlement branch, the electricity volume under medium- and long-term contracts is considered a rigid deduction from the physical saleable capacity. The net saleable surplus after deducting contracts is allocated to each spot market, and the irreversible surplus transmission chain across markets still holds. The profit is the sum of the products of transaction volume and transaction price in each market, and the objective function is: (10); in, For all known information at the decision moment, For the market The declared electricity volume For the market The declared electricity price For mathematical expectation, For the first Electricity declared by each market For the first The declared electricity price for each market For the first The market's clearing price For the first The transaction function for each market; Under the price difference settlement branch, all available output enters the spot market for clearing. Medium- and long-term contracts do not reduce physical sellable capacity but are settled separately as financial price difference contract positions. At this point, the initial sellable surplus is taken as the upper limit of physical deliverables, and the irreversible surplus transfer chain only affects the spot physical reporting dimension. The revenue is the sum of the revenue from total spot electricity and the electricity fees from medium- and long-term price difference contracts. The objective function is: (11); The total revenue from spot electricity is equal to the day-ahead declared electricity multiplied by the day-ahead clearing price, plus the actual electricity generated minus the day-ahead declared electricity multiplied by the real-time clearing price. For all known information at the decision moment, For the market The declared electricity volume For the market The declared electricity price For mathematical expectation, The contract price. The spot price serves as a reference point for medium- to long-term settlement. For CFD positions, For medium- to long-term CFDs, For time period No. The volume of medium- and long-term CFD positions. Medium- and long-term hedging positions under the CFD branch do not consume physical available capacity; their function is to hedge against spot price fluctuations and improve conditions for downside risk to risk-free value.

[0057] The two branches share all infrastructure, including market modeling, unit physical constraints, data parsing and merging, scenario generation and conditional value at risk, human-machine verification and reporting closed loop, etc., differing only in the meaning of capacity caliber, objective function, solution and opportunity cost according to the market. Settlement mode Switch.

[0058] VIII. Specific Implementation Method 1 (Price Difference Settlement Model) Taking a single delivery period on a certain delivery day for a new energy power plant as an example, the predicted power output for that period is 200MWh, the maximum adjustable output of the generating units is approximately 200MWh, the remaining capacity of the channel is 150MWh, the electricity volume under signed medium- and long-term contracts is 80MWh, and the adjusted volume traded in the roll-on / roll-off market is 0MWh. This power plant participates in the roll-on / roll-off market, the day-ahead market, and the real-time market. The historical average prices in each market for that period are approximately RMB 400 / MWh, RMB 520 / MWh, and RMB 350 / MWh, respectively. The maximum capacity that can be absorbed in the day-ahead market during that period is approximately 100MWh.

[0059] This embodiment adopts a differential settlement model. Under differential settlement, all physical output enters the spot market and is settled at the spot price. Therefore, the upper limit of physical deliverables is the smaller of the adjustable output of 200MWh and the channel capacity of 150MWh, i.e., 150MWh. The 80MWh of medium- and long-term contracts is not deducted from the available quantity, but is used as a financial differential contract position with a contract price of 450 yuan / MWh and an expected average spot price of 500 yuan / MWh as the settlement reference point.

[0060] The day-ahead financial benchmark position is determined within the physical deliverable cap of 150MWh. Since the day-ahead expected price of 520 yuan / MWh is higher than the real-time expected price of 350 yuan / MWh, a larger position of approximately 130MWh is preferred to be locked in day-ahead. Real-time settlement of deviations will be based on the actual power generation minus the day-ahead declared volume, with joint conditional risk-based constraints to avoid the deviation risk caused by excessive day-ahead price locking.

[0061] The expected return of a medium- to long-term hedging position of 80MWh is 80×(450-500)=-4000 yuan, which is the expected cost of hedging. However, it reduces the portfolio volatility when spot prices decline and increases the return in the worst-case scenario of 5%, thereby improving the risk-adjusted return under the conditional value at risk constraint.

[0062] Opportunity cost is assessed based on the marginal value of price risk allocation over time. Irreversible deductions, pre-declaration gates, human-machine verification, and automatic declaration loops only apply to the physical declaration dimension of spot goods. Confirmation of medium- and long-term CFD positions goes through the financial settlement layer and is not written into the physical balance transfer chain.

[0063] IX. Specific Implementation Method Two (Differential Settlement Mode) This embodiment uses the differential settlement mode, and the parameters are the same as those in Embodiment 1.

[0064] The initial saleable surplus is calculated using the adjustable power formula. The adjustable power is 200 - 80 - 0 = 120 MWh, which is greater than 0 MWh. Therefore, the initial saleable surplus is 120 MWh, and the reserve is 0 MWh. Neither the adjustable output of 200 MWh nor the channel capacity of 150 MWh constitutes a tighter upper limit; therefore, the upper limit for declaration is 120 MWh.

[0065] Based on the historical joint distribution, 100 sets of prices from the three markets were sampled. The empirical characteristics are that the rolling average price is about 400 yuan / MWh, the day-ahead average price is about 520 yuan / MWh, and the real-time average price is about 350 yuan / MWh. The day-ahead price and the real-time price are positively correlated.

[0066] The subsequent value function at the rolling allocation decision point is estimated, with the subsequent markets being day-ahead and real-time, and the capacity cap for each market during that period set at 100 MWh. Under a restricted greedy allocation, when the remaining capacity does not exceed 100 MWh, it is prioritized to be entirely invested in the higher-priced day-ahead market, and the subsequent value function is approximately equal to 520 multiplied by the remaining capacity. When the remaining capacity exceeds 100 MWh, the excess portion is transferred to the real-time market, and the marginal value decreases to approximately 350 yuan / MWh. Therefore, the subsequent value function is monotonically non-decreasing, and the marginal value decreases from 520 yuan / MWh to 350 yuan / MWh.

[0067] Of the two subsequent markets, the one with the highest day-ahead price has a maximum absorption capacity of 100MWh during that period. Therefore, of the remaining 120MWh available for sale, 100MWh should be reserved for the day-ahead market. The approximately 20MWh exceeding the day-ahead capacity, if not sold through roll-to-roll, will only fall into the lower-priced real-time market at 350 yuan / MWh. Therefore, the marginal opportunity cost of the roll-to-roll market is the derivative of the subsequent value function at 120MWh, which is 350 yuan / MWh.

[0068] The grid search process optimizes the declared electricity volume and declared electricity price. Since the expected rolling price of 400 yuan / MWh is higher than the marginal opportunity cost of 350 yuan / MWh, the model suggests selling approximately 20MWh during the rolling process and reserving 100MWh for the day-ahead period. The declared electricity price is set slightly lower than the expected clearing price of 395 yuan / MWh to ensure a successful transaction. In contrast, if a certain optimization yields a declared electricity price of 340 yuan / MWh, which is lower than the opportunity cost of 350 yuan / MWh, the opportunity cost verification module triggers an alert and displays a comparison of the revenue from selling during the rolling process versus reserving until the day-ahead period or in real-time.

[0069] When submitting a rolling settlement declaration of 20MWh, a reservation operation is performed, with the reserved amount set to 20MWh. After passing the capacity gate verification (20≤120-0), and after human-machine compliance review, the declaration is automatically submitted. After the rolling settlement clears and writes back the actual transaction volume of 20MWh, the reserved amount is released to 0MWh, and the remaining salable inventory is updated to 120-20=100MWh, marked as irreversible, and a snapshot is generated.

[0070] The remaining 100 MWh of available supply is then entered into the day-ahead market, and the above steps are repeated. This 100 MWh can be absorbed by the day-ahead market at approximately 520 yuan / MWh. After the day-ahead market is cleared, any remaining supply is rolled over to the real-time market until the entire market for that period is cleared. The same process is executed in parallel during each delivery period.

[0071] 10. Verification of Technical Effects Each technical feature of this invention can be verified independently, and the verification does not depend on real fund transactions.

[0072] Irreversible capacity consistency verification is achieved by constructing concurrent application test cases, asserting that the sum of the traded electricity volume and the reserved volume at any time during any period is no greater than the initial available sales volume, and that there are no records of traded electricity sales being offset by subsequent repurchases. When injecting over-allocation applications, it is asserted that the gate will definitely be blocked, with verification indicators being zero over-allocation times and zero capacity conflicts.

[0073] The correctness verification of the adjustable power consumption of the rolling batch uses an Excel test set containing ambiguous column names with prefixes and multiple contracts. It asserts that the reverse order disambiguation column mapping is correct, the weighted combined electricity price is equal to the sum of the products of the electricity price and the power consumption of each contract divided by the sum of the power consumption of each contract, and the results of batch import and one-time import are consistent.

[0074] The effectiveness backtesting was optimized by using the actual clearing prices of 30 historical delivery days for post-mortem simulation backtesting. The decision returns of this invention were compared daily with the fixed-ratio allocation baseline and the independent sequential single-market decision baseline. The total return of this invention is approximately 8.5% higher than the fixed-ratio allocation method and approximately 4.7% higher than the independent sequential single-market decision method, and the total return is no less than the set baseline in the worst 5% scenario under the conditional value at risk constraint.

[0075] Performance verification, conducted across 96 settlement periods, 5 markets, and 100 scenarios, yielded a single end-to-end decision time of approximately 1.2 seconds, with an approximate complexity of O(n). Moreover, each decision is made within seconds.

[0076] The closed-loop availability verification uses a simulated environment of the trading platform interface to verify that the entire chain of reservation, declaration, clearing write-back, status update, and snapshot can be executed completely, and to verify alarm and degradation behaviors under abnormal conditions such as token expiration and interface timeout.

[0077] The adaptive verification of the settlement mode sets up the differential quantity and differential price settlement modes on the same set of calculation examples. It asserts that the system automatically switches the capacity caliber, revenue model and opportunity cost definition, and constructs an equivalent scenario where the contract price equals the average spot price at the reference point. It verifies that the settlement electricity cost under the two modes tends to be consistent under this equivalent condition.

[0078] In summary, compared with the prior art, the present invention provides a coupled decision optimization system and method for multiple electricity markets, which has the following significant advantages: First, it achieves cross-market data consistency and irreversible control, completely eliminating the risks of over-generation and capacity conflicts from the system's underlying layer. This invention transforms the power market rule of "once a transaction is completed, it is physically irreversible" into an atomic transaction mechanism at the computer system level by constructing an irreversible capacity state machine and pre-application capacity gate verification. Before submitting an application in any market, the system verifies in real time whether the application quantity exceeds the current available sales capacity after deducting concurrent reservations and the physical constraints of the generating units, blocking illegal applications at the source. During clearing and write-back, the system releases the reserved and irreversibly deducts the remaining available sales capacity in subsequent markets through atomic transactions, while simultaneously generating a data snapshot with millisecond-level timestamps. This design ensures strong consistency and irreversibility in cross-market capacity management, and guarantees traceability and auditability of the entire decision-making process through end-to-end snapshots, meeting the stringent compliance requirements of power trading.

[0079] Secondly, this invention achieves unified support for heterogeneous markets and automated fusion of rolling matching data. It uses a unified market description object to accommodate three types of heterogeneous clearing mechanisms: bilateral negotiation, continuous order book matching, and unified marginal clearing. This comprehensively covers five types of electricity markets: medium- and long-term, rolling matching, day-ahead, real-time, inter-provincial, and regional mutual assistance, filling the gap in the existing single-market system's inability to provide unified modeling. Simultaneously, addressing the pain point of error-prone fusion of multi-source heterogeneous data in the rolling matching process, this invention achieves fully automated calculation of adjustable electricity quantities through intelligent meter header recognition, reverse order disambiguation of time period columns from largest to smallest, automatic mapping of station names, and weighted average merging of contracted electricity quantities. This not only compresses hours of manual data processing work to the second level, eliminating human calculation errors, but also ensures that the calculation results can be fed back to the cross-market optimization model in real time and accurately.

[0080] Third, it possesses high real-time engineering-ready online operation capabilities and is deeply bound to physical constraints. This invention employs a two-stage rolling optimization method combined with subsequent value function approximation, reducing the exponential complexity of complete multi-stage stochastic optimization to linear complexity. In large-scale simulation examples covering 96 delivery periods, 5 markets, and 100 scenarios, the end-to-end time for a single decision is only about 1.2 seconds, fully meeting the real-time requirements of second-level online solutions in the power market. Furthermore, this invention deeply binds the optimization input to physical constraints such as unit ramp-up, start-up and shutdown, maximum and minimum output, and channel capacity, ensuring that the optimization result is no longer a purely mathematical solution detached from physical boundaries, but rather an engineering solution that can be directly physically executed by the unit scheduling system.

[0081] Fourth, it achieves adaptive switching and cross-regional applicability of settlement rules. This invention innovatively decouples capacity and revenue modeling into a physical delivery layer and a financial settlement layer. The physical delivery layer maintains an irreversible kernel, while the financial settlement layer is driven by settlement mode fields. This allows the same system to automatically switch between capacity caliber, revenue model, and opportunity cost definition under both differential settlement and price difference settlement rules. This design not only adapts to the two settlement methods in the basic rules of the electricity spot market, achieving cross-regional applicability, but also allows for a smooth switch from differential settlement to price difference settlement, requiring only modification of configuration parameters without system reconstruction, demonstrating strong engineering scalability.

[0082] Fifth, it significantly improves overall returns while ensuring controllable and auditable risks. Through cross-market capacity coordination and opportunity cost quantification, this invention, in simulation verification based on historical publicly available electricity price distributions, achieves approximately 8.5% higher total returns compared to the fixed-proportion allocation method and approximately 4.7% higher than independent sequential single-market decision-making. Simultaneously, the introduction of Conditional Value at Risk (CVaR) constraints guarantees a minimum return under extremely unfavorable scenarios; the opportunity cost verification module ensures that each decision recommendation includes marginal value assessment and scenario return comparison, providing high interpretability and avoiding the unexplainable risks of black-box models in compliance audits.

[0083] Sixth, it achieves an integrated online closed loop for decision-making and execution. This invention establishes a seamless interaction across the entire chain, from the optimizer, human-machine verification, capacity gate, trading platform interface to the capacity snapshot database. After manual compliance review, the optimization results are automatically submitted for declaration through the trading platform interface. After clearing, the transaction results are automatically obtained and the capacity status is written back. This design retains the human supervision step to ensure compliance while automating the execution process, completely solving the problem of the disconnect between decision-making and execution in traditional solutions, and significantly improving trading efficiency and system reliability.

[0084] In summary, this invention not only solves the technical bottlenecks of existing technologies in cross-market capacity management, heterogeneous data fusion, online real-time solution and settlement rule adaptation, but also provides new energy power generation companies with a safe, efficient, compliant and interpretable coupled decision optimization engineering solution in complex power market environments, which has extremely high commercial application value and industry promotion prospects.

[0085] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A coupled decision optimization system for multiple electricity markets, characterized in that, It includes a data access and market object modeling module, a roll-to-roll adjustable power calculation module, an irreversible capacity state machine module, a unit physical constraint access module, a cross-market optimization solution module, and a human-machine verification and automatic reporting module. The data access and market object modeling module is used to facilitate the participation of power generation companies. Each electricity market establishes a unified market description object, which includes market type, price formation mechanism, settlement model, decision-making time window, clearing time, delivery period set, upper and lower limits of declared electricity volume, and interface description. It is an integer not less than 4; The adjustable power calculation module is used to integrate three types of multi-source heterogeneous data: power prediction, contract power, and rolling adjustment, and eliminate format ambiguity to calculate the adjustable power of each station at each time period. The unit physical constraint access module is used to read the upper and lower limits of ramp rate, start and stop status and minimum start time, maximum and minimum output and upper limit of transmission channel capacity constraints from the generator dispatch system, and use them as the feasible domain boundary input of the cross-market optimization solution module; The irreversible capacity state machine module is used to maintain capacity state objects for each delivery day according to the delivery period, and realize declaration reservation, atomic deduction after clearing and snapshot generation; The cross-market optimization solution module is used to adaptively switch between capacity caliber and revenue model according to the settlement mode. It adopts a two-stage rolling optimization combined with the subsequent value function to solve the suggested electricity volume and suggested electricity price for the market currently in the decision window. The human-machine verification and automatic declaration module is used to automatically submit declarations through the power trading platform interface after opportunity cost verification and manual compliance review, and automatically obtain the transaction results and write them back to the irreversible capacity state machine module after clearing.

2. The coupled decision optimization system for multiple electricity markets according to claim 1, characterized in that, The adjustable power calculation module is specifically used for: Header identification and field parsing are performed on three types of Excel data: power forecast, contracted power volume, and rolling adjustment. To address ambiguity in column name prefixes for different time periods, a reverse matching method from largest to smallest is used to determine the column segment number. When multiple contracts exist for the same station at the same time period, the combined electricity volume is calculated using the following formula. and weighted average electricity price : ； ； in For the first The electricity volume of the contract, For the first The electricity price under the contract; Calculate the adjustable power using the following formula. : ； in For station Time period Power prediction of electricity consumption, For station Time period The contracted electricity volume, For station Time period Adjust the power using the roller; when When it is positive, it serves as the upper limit for eligible electricity sales; when... When the value is negative, its absolute value is used as the rigid mandatory purchase quantity to generate a separate electricity purchase declaration.

3. The coupled decision optimization system for multiple electricity markets according to claim 1, characterized in that, The irreversible capacity state machine module is specifically used for: definition For time period The declared reserve quantity awaiting clearance For the market The remaining salable quantity after the decision; During initialization, the time periods are... Set as capacity limit. Set to 0; When the market During the period Number of applications submitted At that time, a capacity gate verification is performed before the application is submitted to verify whether the submitted quantity does not exceed the current saleable surplus after deducting the concurrent reservation and whether it meets the physical constraints of the unit. If the validation passes, the reserved operation will be executed using an atomic transaction: ; When the market During the period Clear out and write back the actual transaction volume Then, release the reserved and update the remaining marketable inventory using atomic transactions: ; and the actual transaction volume The locked battery level details are written with an irreversible flag, whereby... This represents the remaining available stock in the subsequent market. For the market During the period The actual transaction volume; Each time the capacity state object changes, a data snapshot with a millisecond-level timestamp is generated.

4. The coupled decision optimization system for multiple electricity markets according to claim 1, characterized in that, The cross-market optimization solution module is specifically used for: Price scenarios are generated based on the joint distribution of historical prices, and subsequent value functions are estimated. , Indicates time period The remaining available quantity is The expected return that can be obtained from the optimal allocation across all subsequent markets, and about It is monotonically non-decreasing and marginally decreasing. Construct the objective function for a two-stage solution: ； The constraints are: ； in For transaction functions, For all known information at the decision moment, For time period The unit's output is adjustable. For time period Channel capacity, For the market During the period The expected clearing price For the market The lower limit of the declared electricity volume, For the market The upper limit of the declared electricity volume, For time period Total adjustable power, For mathematical expectation, For the market During the period The number of applications, For time period The declared reserve quantity awaiting clearance For the market The remaining salable quantity after the decision. For the market During the period The declared electricity price; Calculate the marginal opportunity cost across markets: ;in This is the left derivative of the subsequent value function at the current remaining salable quantity; when the expected marginal revenue corresponding to the proposed electricity price is lower than... When this happens, an alert is triggered.

5. The coupled decision optimization system for multiple electricity markets according to claim 1, characterized in that, The cross-market optimization solution module decouples capacity and revenue modeling into a physical delivery layer and a financial settlement layer, and then optimizes it according to the market. Settlement mode Field-driven branch: when When settling accounts for the difference, the electricity volume under medium- and long-term contracts is considered a rigid deduction from the physical available capacity, and the revenue is the sum of the product of the transaction volume and the transaction price in each market. when When settling price differences, all available resources are used to clear positions in the spot market, with medium- and long-term contracts serving as financial CFD positions. Settlement is done separately. The initial available sales volume is the upper limit of physical deliverables. Revenue is the sum of the revenue from the total spot electricity volume and the electricity cost of the medium- to long-term CFD. The revenue from the total spot electricity volume equals the day-ahead declared volume multiplied by the day-ahead clearing price plus the actual generation minus the day-ahead declared volume multiplied by the real-time clearing price. The electricity cost of the medium- to long-term CFD is... ,in The contract price. The spot price serves as a reference point for medium- to long-term settlement. This refers to the medium- to long-term contract for difference (CFD) number. For time period No. The position size of long-term CFDs.

6. A coupled decision optimization method for multiple electricity markets, applied to the coupled decision optimization system for multiple electricity markets as described in any one of claims 1 to 5, characterized in that, Includes the following steps: Step 1, involving power generation companies Each electricity market establishes a unified market description object and advances the four-state sequential state machine migration according to the decision-making time window and clearing time of each market. The four states include not open market, decision-making window open, declared pending clearing, and cleared locked. Step 2: For each delivery day, maintain the capacity status object according to the delivery period, perform capacity gate verification before declaration, execute the declaration reservation with atomic transactions after the verification passes, release the reservation with atomic transactions after clearing and irreversibly deduct the remaining market salable quantity, and generate a data snapshot; Step 3: Obtain the physical constraints of the generator set from the generator set dispatching system as the boundary of the feasible region for optimization; Step 4: Integrate power prediction, contracted power volume and rolling adjustment data, eliminate ambiguity in the column name prefixes for each time period, and calculate the adjustable power volume for each station in each time period; Step 5: Adaptively switch between capacity caliber and revenue model based on the settlement mode field of the market description object, use two-stage rolling optimization combined with subsequent value function to solve for the suggested declared electricity volume and suggested declared electricity price, and calculate the cross-market marginal opportunity cost for verification; Step 6: After manual compliance review and capacity gate verification, the application is automatically submitted through the power trading platform interface, and the capacity status object is written back after clearing.

7. The coupled decision optimization method for multiple electricity markets according to claim 6, characterized in that, Calculate the adjustable power in step 4. The formula is: ;in, For station Time period Power prediction of electricity consumption, For station Time period The contracted electricity volume, For station Time period Adjust the power using the roller; when When the positive time is used as the upper limit for the electricity that can be declared for sale, When it is negative, its absolute value is used as the rigid mandatory purchase quantity to generate a separate electricity purchase declaration.

8. The coupled decision optimization method for multiple electricity markets according to claim 6, characterized in that, In step 5, when the settlement mode is price difference settlement, all available power output enters the spot market for clearing. The medium and long term contracts are settled separately as financial price difference contract positions, without deducting the physical sellable capacity. The revenue includes the revenue from the total spot power output and the electricity fees from the medium and long term price difference contracts.

Images