Risk mode selection switching multi-cell short process steelmaking collaborative scheduling method

Abstract

This invention discloses a multi-electrolyzer-short-process steelmaking collaborative scheduling method for risk mode selection and switching, belonging to the fields of energy system engineering and iron and steel metallurgy. It includes: generating multiple random photovoltaic output scenarios based on photovoltaic output characteristics under different weather conditions; using an inverse cumulative distribution function to transform historical data into power boundaries at different confidence levels, generating dynamic safety envelopes for three risk modes: aggressive, robust, and risk-averse; introducing binary variables to represent the selected risk mode at each moment and imposing penalties for mode switching; employing McCormick's soft boundary envelope linearization technique for the bilinear term and introducing relaxation variables to construct flexible linearization constraints; establishing a tiered carbon trading mechanism to divide excess emissions into multiple price tiers; and establishing a multi-entity collaborative operation model with the goal of minimizing total operating costs, coordinating energy storage, steelmaking, and hydrogen production subsystems including alkaline, proton exchange membrane, and solid oxide electrolyzers to determine the optimal operating strategy. This invention solves the steel production safety risk problem caused by the limitation of a single electrolyzer and the uncertainty of renewable energy in the large-scale application of electrolyzers.

Description

Technical Field

[0001] This invention relates to the fields of energy system engineering and iron and steel metallurgy, and in particular to a multi-electrolytic cell-short-process steelmaking collaborative scheduling method for risk mode selection and switching. Background Technology

[0002] In steel production, the traditional blast furnace-converter long process uses coke as the main reducing agent, resulting in high carbon emission intensity. Hydrogen metallurgy technology, which replaces coke with hydrogen, is one of the important technological paths to achieve deep decarbonization in the steel production process. Meanwhile, with the continuous expansion of installed capacity of renewable energy sources such as photovoltaics and wind power, using renewable energy to electrolyze water to produce hydrogen can not only improve the absorption capacity of renewable energy but also provide a low-carbon hydrogen source for steel production. However, constructing a coupled system of renewable energy water electrolysis for hydrogen production and steel production faces multiple practical challenges, including the random fluctuations in renewable energy output, significant differences in the operating characteristics of various types of electrolyzers, and the rigid constraints of short-process steelmaking.

[0003] Currently, there is some research on the coupled scheduling of hydrogen energy systems and steel production. On the hydrogen production side, existing work mainly focuses on the modeling and optimization of single-type electrolyzers (such as alkaline electrolyzers (AWE), proton exchange membrane electrolyzers (PEMWE), or solid oxide electrolyzers (SOE). At the system scheduling level, existing methods typically characterize the volatility of renewable energy based on deterministic scenarios or simple uncertainty handling methods (such as typical scenario methods or robust boundary methods). On the steel production side, short-process steelmaking involves multiple steps such as electric arc furnaces, refining furnaces, and continuous casting, and the strong coupling characteristics of its material flow and energy flow have attracted the attention of some studies. In addition, carbon market mechanisms are also beginning to be introduced into industrial operation optimization, guiding emission reduction behavior by setting carbon emission prices or constraints.

[0004] Nevertheless, existing technologies still have the following three shortcomings:

[0005] (1) Hydrogen production equipment is limited to a single technology route: a single type of electrolyzer makes it difficult to balance economy and flexibility. For example, alkaline electrolyzers (AWE) have low cost but slow dynamic response, proton exchange membrane electrolyzers (PEMWE) have high responsiveness but high equipment cost, and solid oxide electrolyzers (SOE) have high efficiency but demanding operating temperatures and difficult start-up and shutdown. Existing solutions lack the ability to optimize and dynamically coordinate the combination of the above three or more heterogeneous electrolyzers, and cannot fully leverage the complementary advantages of different electrolyzers in terms of efficiency, flexibility and cost.

[0006] (2) Lack of dynamic risk perception capability at the dispatch decision level: Existing methods mostly adopt static or fixed uncertainty boundary dispatch models, which are relatively crude in handling the volatility of renewable energy and make it difficult to achieve a dynamic optimal balance between system economy and operational reliability. In particular, there is a lack of a mechanism to flexibly adjust dispatch strategies according to different risk preferences (such as aggressive consumption mode or conservative risk avoidance mode), which cannot effectively avoid the power supply security risks of steel production caused by the drastic fluctuations of renewable energy.

[0007] (3) The system coupling level does not fully consider the refined constraints of steel production: Existing coupling studies do not adequately characterize the strong coupling characteristics of material flow and energy flow in short-process steelmaking, and the scheduling scheme is difficult to achieve an effective balance among the coordination of heterogeneous hydrogen production equipment, dynamic matching of renewable energy and load, and avoidance of operational risks. This leads to prominent problems in actual operation, such as frequent equipment start-ups and shutdowns, difficulty in handling nonlinear power distribution, and difficulty in coordinating carbon emission costs and power supply reliability. Summary of the Invention

[0008] The technical problem this invention aims to solve is to provide a multi-electrolyzer-short-process steelmaking collaborative scheduling method with risk mode selection switching. This method fully leverages the complementary advantages of the three electrolyzer types in terms of efficiency, flexibility, and cost by constructing a heterogeneous hydrogen production system model that couples alkaline electrolyzers, proton exchange membrane electrolyzers, and solid oxide electrolyzers. Simultaneously, considering the volatility of renewable energy, a risk perception framework based on an inverse cumulative distribution function is designed. Chance constraints are used to transform the stochastic problem into a deterministic dynamic boundary under different risk preferences, and the McCormick soft boundary envelope linearization technique is introduced to handle the nonlinearity in power allocation. Finally, a risk-level controllable, economically optimal collaborative scheduling strategy is constructed.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0010] A collaborative scheduling method for multi-electrolytic cell-short-process steelmaking with risk mode selection switching includes the following steps:

[0011] S1. Random scene generation based on historical data statistical characteristics: According to the characteristics of photovoltaic power output under different weather types, different predicted mean and fluctuation coefficient are assigned to each typical day to generate random photovoltaic power output scenes that reflect various weather scenarios.

[0012] S2. Risk Mode Selection and Dynamic Safety Envelope Construction: The inverse cumulative distribution function is used to transform the statistical characteristics of historical data into power boundaries at different confidence levels, generating dynamic safety envelopes corresponding to three risk modes: aggressive, robust, and risk-averse. Binary variables are introduced to represent the selected risk mode at each time step, and a penalty cost is imposed on mode switching. For the bilinear term coupled with the photovoltaic allocation ratio and the total power consumption, the McCormick soft boundary envelope linearization technique is used to construct flexible linear constraints by introducing slack variables.

[0013] S3. Construction of a tiered carbon trading model: Establish a tiered carbon trading mechanism, divide excess emissions into multiple price tiers, and introduce carbon emission constraints into the hydrogen production-steelmaking collaborative system;

[0014] S4. Multi-entity collaborative operation optimization: Establish a multi-entity collaborative operation model with the goal of minimizing the total operating cost within the scheduling cycle of the hydrogen production-steelmaking collaborative system. By coordinating the energy storage system, the steelmaking subsystem, and the hydrogen production subsystem, which includes alkaline electrolyzers, proton exchange membrane electrolyzers, and solid oxide electrolyzers, optimize the configuration and determine the optimal operating strategy for each device.

[0015] A further improvement to the technical solution of this invention lies in the following: In step S1, the different weather types include sunny days, cloudy days, and rainy days; the photovoltaic output is modeled as a superposition of a deterministic scenario mean trend and a random fluctuation component, wherein the intensity of the random fluctuation component is determined by the time-varying volatility of historical data, and the standard deviation of the fluctuation is forcibly reduced to zero during the nighttime period without light; the random scenario of photovoltaic output based on historical data is constructed as shown in the following formula:

[0016]

[0017] in, This is the predicted mean based on historical data; The preset fluctuation intensity coefficient reflects the proportion of fluctuation in photovoltaic output relative to the predicted value; The minimum threshold for determining the state of no light.

[0018] A further improvement to the technical solution of the present invention is that step S2 specifically includes:

[0019] S21. Risk boundary transformation based on the inverse cumulative distribution function:

[0020] Opportunity-constrained programming is a mathematical method that transforms uncertainty into a deterministic decision boundary. Unlike traditional methods that presuppose that the error follows a specific distribution, this invention directly utilizes the statistical characteristics of historical data to calculate the power boundary at different confidence levels using the inverse cumulative distribution function.

[0021] For a given violation probability The opportunity constraints of photovoltaic power output are transformed into a set of deterministic dynamic envelopes. , :

[0022]

[0023]

[0024] In the formula, The standardized inverse cumulative distribution function, also known as the quantile function, is used to determine the safety margin at the corresponding confidence level; this formula incorporates the historical forecast mean. With proportional fluctuation Calculate the scheduling boundaries under different risk modes;

[0025] S22. Risk Mode Selection and Switching:

[0026] Introducing binary variables This indicates that the risk mode is selected at time t. It includes three modes: aggressive, moderate, and risk-averse, corresponding to high, medium, and low risk levels. The system must and can only select one mode at any given time, and mode switching is determined by variables. Capture:

[0027]

[0028]

[0029] The system performs dynamic trade-offs at every moment, and when the photovoltaic prediction deviation... When the risk level is relatively high, electricity and carbon prices are high, and the marginal economic benefits can cover the risk penalty costs of the high-risk mode, switching to the high-risk mode can significantly relax the safety lower bound. The freed-up scheduling margin allows the system to absorb more photovoltaic power, thereby reducing economic costs.

[0030] S23, McCormick soft-boundary envelope linearization:

[0031] Traditional McCormick methods typically construct strict hard-constraint envelopes, which are effective when dealing with boundaries where physical limits are defined. However, within the risk-aware framework of this invention, power boundaries... and It is a probabilistic statistical boundary derived from chance constraints, rather than an absolute physical rigid boundary.

[0032] Since opportunity constraints allow for default risk at a certain confidence level, the corresponding power allocation relationship should possess mathematical flexibility. Therefore, this invention proposes a flexible McCormick envelope mechanism. This mechanism constructs a dual buffer system at both the probabilistic and geometric levels by introducing slack variables.

[0033] First, the model must satisfy the power security domain constraint based on the current risk pattern, i.e., the total photovoltaic power absorbed. It must not exceed the lower bound of the corresponding confidence level:

[0034]

[0035] Secondly, regarding the photovoltaic allocation ratio With total consumption Bilinear terms generated by coupling Through McCormick slack variables A flexible linearization constraint is constructed; the variable assignment relationship is allowed to deviate slightly near the statistical boundary, serving as a mapping of probabilistic default risk in the opportunity constraint to the algebraic constraint level; the specific flexible envelope constraint is constructed as follows, with... For example, during activation:

[0036]

[0037] In the formula, For photovoltaic allocation ratio, slack variable It can quantify the allowable default degree of the linearized envelope boundary; due to the randomness of photovoltaic power output, its risk boundary can lead to an extremely narrow constraint space or even no solution under extreme fluctuations. It provides a numerical buffer for the linearization process of bilinear terms.

[0038] A further improvement of the technical solution of the present invention is that: in step S3, the tiered carbon trading model includes at least three price tiers, each tier corresponding to a different carbon emission range and carbon trading unit price, and the carbon trading unit price increases stepwise with the increase of excess emissions; the carbon trading cost is the sum of the products of emissions in each tier and the corresponding carbon trading unit price.

[0039] To accurately simulate the carbon market, excess emissions will be... Assigned to In the price tiers, the width of each tier is... The tiered carbon trading model is represented as follows:

[0040]

[0041]

[0042]

[0043] In the formula, is the activation variable for the carbon emission range of the i-th tier, representing the actual excess emissions allocated within that tier; g is the carbon price tier growth rate. For each carbon price tier; The benchmark carbon price; carbon prices at each tier. Typically based on benchmark price The growth rate was calculated in advance. .

[0044] A further improvement of the technical solution of the present invention is that: in step S4, the alkaline electrolytic cell, the proton exchange membrane electrolytic cell, and the solid oxide electrolytic cell have different efficiency characteristics and start-stop constraints: the alkaline electrolytic cell is suitable for base load operation, the proton exchange membrane electrolytic cell has fast response capability, and the solid oxide electrolytic cell adopts a high-temperature electrochemical mode and is coupled with the steelmaking waste heat recovery system.

[0045] A further improvement of the technical solution of the present invention is that: in step S4, the steelmaking subsystem is a short-process steelmaking process based on hydrogen-based direct reduction iron-electric arc furnace, and its energy demand includes electricity for electric furnace smelting and hydrogen for reduction reaction. The hydrogen produced by the electrolytic cell is given priority to meet the hydrogen demand for steelmaking, and the remaining part is incorporated into hydrogen storage or sold externally.

[0046] A further improvement of the technical solution of the present invention is that: in step S4, the energy storage system includes two forms: battery energy storage and hydrogen energy storage; wherein battery energy storage is used for short-term power smoothing, and hydrogen energy storage achieves cross-time energy transfer through hydrogen production in the electrolyzer and feedback from the fuel cell.

[0047] A further improvement to the technical solution of the present invention is that: in step S4, the objective function of the multi-entity collaborative operation model includes seven parts: electricity purchase cost, tiered carbon trading cost, risk mode operation cost, equipment start-up and shutdown and cyclic aging cost, curtailment penalty cost, risk mode switching penalty cost, and McCormick envelope relaxation penalty cost.

[0048] The objective function model is shown below:

[0049]

[0050] In the formula, To minimize the total system operating cost within the scheduling period T, the specific expressions for each cost are defined by system parameters and decision variables, reflecting a multi-dimensional trade-off between economy, environmental protection, reliability, and operability.

[0051] Specifically:

[0052] (I) The detailed mathematical description and physical meaning of each cost item are as follows:

[0053] (1) The cost of electricity purchase is the direct economic expense of the system obtaining energy from the external power grid, and it constitutes a major component of operating costs; the formula is shown below:

[0054]

[0055] In the formula, The time-of-use electricity price for the dispatch period t is expressed in yuan / MWh. This represents the power purchased during that period, in MW.

[0056] (2) The tiered carbon trading cost is the penalty cost for carbon emissions exceeding the free allowance. It is a key economic lever for the model to achieve low-carbon scheduling and guide the system to prioritize the use of green energy. This model adopts a more realistic tiered carbon price; the greater the excess emissions, the higher the unit carbon price, as shown in the formula below:

[0057]

[0058] In the formula, This represents the carbon trading price for this range, expressed in yuan / tCO2. The carbon emissions falling within the i-th price range are expressed in tCO2.

[0059] (3) The operating cost of the risk model is a virtual operating cost assigned based on different risk preferences. High-risk, aggressive scheduling strategies are penalized heavily, guiding the model to make prudent decisions between economic benefits and operational risks. The formula is shown below:

[0060]

[0061] In the formula, This is the unit operating cost of risk model k, in yuan. It is a binary variable indicating whether mode k is selected during time period t;

[0062] (4) The start-up and shutdown costs of electrolytic cells and steel equipment, as well as the cyclic aging costs of energy storage equipment, were quantified to reduce frequent equipment state switching and improve equipment lifespan and operational economy. The formulas are shown below:

[0063]

[0064] In the formula, and These are the start-up and shutdown costs for electrolytic cell i, respectively. and For start / stop action variables; and Let J represent the start-up cost and action variables for steelmaking process j. Unit cycle cost of energy storage; and This refers to the charging and discharging power of energy storage.

[0065] (5) To mitigate the cost of curtailment penalties, the incentive model maximizes the utilization rate of local green energy, serving as a crucial indicator for evaluating the greenness of the system; the formula is shown below:

[0066]

[0067] In the formula, The unit of abandoned light penalty coefficient, This refers to the power of abandoned light.

[0068] (6) The penalty cost for switching risk modes is the penalty cost imposed on frequent switching of risk modes to avoid drastic oscillations in the system's risk strategy, thus ensuring a smooth transition of the scheduling strategy; the formula is shown below:

[0069]

[0070] In the formula, Switch the penalty coefficient by unit. This is a variable for mode switching actions.

[0071] (7) The cost of the McCormick envelope relaxation penalty is used to ensure the numerical stability and solution efficiency of the McCormick linearization method, as shown in the following formula:

[0072]

[0073] In the formula, To relax the penalty coefficient, is a slack variable for the McCormick envelope.

[0074] (II) Establishing a subsystem model for the electrolytic cell cluster:

[0075] Electrolyzer clusters are the core hubs connecting the power system and the hydrogen energy system, and their flexible scheduling is crucial for absorbing fluctuating renewable energy. This model employs a heterogeneous cluster composed of electrolyzers using different technologies: AWE, PEMWE, and SOE. Its index is [index missing]. .

[0076] Total input power of the electrolytic cell It consists of two parts: directly absorbed photovoltaic power. And power obtained from the system's common bus, supplemented by the grid or energy storage. The formula is shown below:

[0077]

[0078] In the formula, The total input power of electrolytic cell i during time period t is expressed in MW. The photovoltaic power allocated to this electrolyzer is expressed in MW. The power obtained from the bus is expressed in MW.

[0079] At the same time, the total power drawn by all electrolytic cells from the busbar cannot exceed the total available power on the busbar, i.e., the power purchased by the power grid. With energy storage discharge power The sum, minus the power supplied to the steel load For the part, the formula is as follows:

[0080]

[0081] This constraint ensures the proper distribution of power flow and energy conservation within the system.

[0082] To reflect the direct impact of risk perception on equipment operation strategies, the operating boundary of the electrolyzer is set to be dynamically related to the current risk mode k of the system, as shown in the following formula:

[0083]

[0084]

[0085]

[0086] In the formula, and These are the upper and lower limits of the power of electrolyzer i in risk mode k, respectively, in MW; This is the corresponding gradeability limit, expressed in MW / h;

[0087] By using pattern selection variables Multiplication enables dynamic switching of operating boundaries. For example, in disaster-averse mode, to maintain more system reserves to cope with potential photovoltaic power shortages, the model's power ceiling is adjusted. It may be set to a lower level; while in aggressive mode, it is allowed to operate at a higher power level to maximize hydrogen production. It is a binary variable indicating whether the device is powered on.

[0088] To calculate start-up and shutdown costs and avoid frequent equipment operation, a standard start-up and shutdown logic is introduced, as shown in the following formula:

[0089]

[0090] In the formula, This represents the running state at time t. (Binary variable) and These represent the start-up and shutdown actions, respectively.

[0091] The first equation defines start and stop actions by comparing the states at adjacent moments; a change in state from 0 to 1 necessarily corresponds to a start. The second inequality ensures that within a given time period, the device cannot simultaneously perform the mutually exclusive actions of starting and stopping.

[0092] Hydrogen production rate of the electrolyzer With its input electrical power The relationship is linear, determined by the hydrogen production efficiency of this type of electrolyzer. Decide.

[0093]

[0094] In the formula, The unit is kg / h. The unit is kg / MWh.

[0095] (III) Steel Production Subsystem Model:

[0096] The steel production process is an energy-intensive process with extremely high requirements for continuity. This model provides a detailed model of the short electric arc furnace process, which starts from the electric arc furnace (EAF), passes through the ladle furnace (LF), and finally forms the steel in the continuous casting (CC) machine.

[0097] The core production objective of the model is to complete the predetermined production batches within the entire scheduling cycle. This macro-level goal was broken down into specific processes. The above is reflected in the constraint on total energy consumption; the formula is shown below:

[0098]

[0099] In the formula, It is the power consumption of process j at time t, in MW; This is the minimum total energy required to complete a single batch of products in this process, expressed in MWh;

[0100] This constraint ensures that the final output is met. The EAF, as the main melting equipment, [is involved]. Maximum value; energy consumption of LF and CC The value is relatively small.

[0101] Each steelmaking process has its designated operating power range that must be adhered to in order to ensure product quality and equipment safety, as shown in the formula below:

[0102]

[0103] In the formula, and Let these be the minimum and maximum operating power of process j, respectively. Its running state variables.

[0104] The value of this constraint varies significantly across different processes; EAF is the unit with the highest power consumption. The power requirement is much higher than other processes; LF has moderate power and is used for precise control of molten steel temperature and composition; CC has the lowest power requirement and is mainly used for driving equipment and compensating for heat dissipation.

[0105] The minimum continuous operating time constraint is a key industrial constraint in steel production modeling, directly related to production safety and cost. To prevent frequent start-ups and shutdowns from causing significant thermal shock, mechanical stress, and material loss to the equipment, once started, it must operate continuously for at least [duration missing]. Hours, the formula is as follows:

[0106]

[0107]

[0108] Define the startup action variable This indicates that if a startup action occurs at time t, then subsequent actions starting from time t... Operating status within a given time period It must always be 1.

[0109] This constraint is particularly important for CC, as any unplanned interruption could lead to catastrophic accidents such as leaks of steel. The value is the longest. For EAF and LF, this constraint is mainly for protecting refractory materials and saving preheating energy. The value is relatively short.

[0110] Steel production is a tightly coupled assembly line operation, with time allotted for transporting molten steel and other materials between processes. To ensure the physical feasibility of the material flow, this model establishes strict start-stop timing constraints.

[0111] The downstream process (j+1) starts running at time t, and must be based on the upstream process j starting at time t- The process has already been run before this point, as shown in the formula below:

[0112]

[0113] Upstream process j runs at time t, and its downstream process (j+1) must run at time t+1. This is based on the premise that there is still an operational plan after the specified time, in order to avoid producing intermediate products that cannot be processed; the formula is as follows:

[0114]

[0115] This constraint precisely describes the physical timeline of the molten steel's flow from upstream to downstream. Specifically, after the molten steel is melted at EAF, it needs to be transported within a certain timeframe. The product is internally transferred to LF for processing; after refining and passing inspection, it is then transported again during the specified time. The steel is then sent to the CC. This constraint ensures that the LF will not idle and wait before the steel is discharged from the EAF, and that the CC will not start before the LF has finished processing.

[0116] (iv) Model of Electric Energy Storage and Hydrogen Storage Subsystem

[0117] Electric energy storage and hydrogen storage tanks are key buffers to mitigate fluctuations in renewable energy and are the cornerstone for the reliable operation of hydrogen production-steelmaking synergistic systems.

[0118] The system's total hydrogen production is divided into two parts: one part is directly supplied to steel mills. Another portion is injected into the hydrogen storage tank. To prepare for unforeseen circumstances. The total hydrogen demand of steel mills... Hydrogen is supplied directly and extracted from hydrogen storage tanks. Both conditions are met; the formula is shown below:

[0119]

[0120]

[0121] Hydrogen storage tank inventory levels (kg) is determined by the previous inventory and the current net injection. It must be kept at a safe minimum inventory level. and maximum capacity The system operates between these parameters. Meanwhile, to ensure system robustness and hydrogen reserves, the model mandates that at least [amount missing] of the total daily hydrogen production must be [percentage missing]. A certain proportion must be stored in storage tanks to cope with possible prolonged insufficient photovoltaic output; the formula is as follows:

[0122]

[0123]

[0124]

[0125] Battery state of charge (MWh) also follows a similar dynamic balance, taking into account both charge and discharge efficiency. and Furthermore, batteries cannot be charged and discharged simultaneously; this physical characteristic is expressed through binary variables representing the charge and discharge states. and This is achieved by applying mutual exclusion constraints; the formula is shown below:

[0126]

[0127]

[0128]

[0129]

[0130]

[0131] Carbon emissions are an important dimension to consider in the model, and are divided into indirect emissions and direct emissions; the formula is shown below:

[0132]

[0133]

[0134]

[0135]

[0136] Calculate total emissions The emissions consist of indirect emissions from electricity purchased from the grid and direct emissions from electric arc furnaces. The emission factors are as follows: and Composition. Calculations are based on production volume and benchmark values. Free carbon allowances Calculate the excess emissions that need to be purchased. .

[0137] To accurately simulate the carbon market, excess emissions will be... Assigned to In the price tiers, the width of each tier is... The specific tiered carbon emission levels are as follows:

[0138]

[0139]

[0140]

[0141] In the formula, The activation variable for the tiered carbon emission range is the carbon price at each tier. Typically based on benchmark price The growth rate g was calculated in advance. .

[0142] A further improvement of the technical solution of the present invention is that the multi-subject collaborative operation model is solved using a mixed integer linear programming solver with a solution period of 168 hours and a time resolution of 1 hour, and the dynamic safety envelope generated in step S2 is embedded into the model hourly as an upper limit constraint of photovoltaic power consumption.

[0143] A further improvement to the technical solution of the present invention is that, in step S4, the multi-entity collaborative operation model further includes operating constraints of the hydrogen production subsystem, operating constraints of the steelmaking subsystem, operating constraints of the energy storage system, and system power balance constraints.

[0144] The operational constraints of the hydrogen production subsystem include the total input electrical power constraint of the electrolyzer and the equipment operation strategy constraint of the electrolyzer.

[0145] The operational constraints of the steelmaking subsystem include steel production power constraints and steel production operation constraints.

[0146] The constraints on the operation of the energy storage system include total hydrogen production allocation constraints, electrical energy storage constraints, and carbon emission constraints.

[0147] The technological advancements achieved by this invention due to the adoption of the above technical solutions are as follows:

[0148] 1. By flexibly switching between three risk modes—aggressive, conservative, and risk-averse—the system can automatically find the optimal balance between conservative and aggressive approaches, achieving a dynamic and optimal equilibrium.

[0149] 2. When the photovoltaic output fluctuates slightly, the system automatically switches to aggressive mode to effectively improve the photovoltaic absorption rate; when the photovoltaic output fluctuates drastically, the system switches to risk-averse mode to significantly reduce the risk of power default caused by photovoltaic uncertainty.

[0150] 3. By intelligently selecting and switching risk modes, this invention controls multiple costs, including electricity purchase costs, tiered carbon trading costs, risk mode operation costs, equipment start-up and shutdown and cyclic aging costs, curtailment penalty costs, risk mode switching penalty costs, and McCormick envelope relaxation penalty costs, to the lowest possible level.

[0151] 4. This invention achieves synergistic optimization of economic benefits (lowest operating costs), environmental benefits (reduced carbon emissions and curtailment of solar power), and system reliability (avoiding power default risks). Attached Figure Description

[0152] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0153] Figure 1 This is a flowchart of a multi-electrolytic cell-short-process steelmaking collaborative scheduling method for risk mode selection and switching provided by an embodiment of the present invention;

[0154] Figure 2 This is a physical architecture diagram of the hydrogen production-steelmaking co-production system in an embodiment of the present invention;

[0155] Figure 3 This is a scheduling framework diagram of the hydrogen production-steelmaking collaborative system in an embodiment of the present invention;

[0156] Figure 4 This is a boundary diagram of photovoltaic historical scenarios and risk perception in an embodiment of the present invention;

[0157] Figure 5 This is a scheduling mode switching diagram based on risk perception in an embodiment of the present invention;

[0158] Figure 6 This is a comparison diagram of the load conditions of heterogeneous electrolytic cell clusters in embodiments of the present invention;

[0159] Figure 7 This is a diagram showing the operating status of the hydrogen storage tank in an embodiment of the present invention;

[0160] Figure 8 This is a timing power diagram of the sub-steelmaking process in an embodiment of the present invention;

[0161] Figure 9 This is a diagram showing the operating status of the energy storage system in an embodiment of the present invention;

[0162] Figure 10 This is a power balance diagram of the hydrogen production-steelmaking co-production system in an embodiment of the present invention. Detailed Implementation

[0163] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, products or devices.

[0164] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments:

[0165] This invention selects a hydrogen production-steelmaking integrated demonstration park in a photovoltaic resource-rich area as the research object. The park aims to achieve low-carbon steelmaking through a short-process process by deploying a large-scale photovoltaic power generation system and a hybrid water electrolysis hydrogen production unit. The system scheduling cycle is set to T=168 hours, with a time step of 1 hour.

[0166] like Figure 1 , Figure 2 , Figure 3 As shown, a multi-electrolytic cell-short-process steelmaking collaborative scheduling method with risk mode selection switching includes the following steps:

[0167] S1. Random scene generation based on historical data statistical characteristics: According to the characteristics of photovoltaic power output under different weather types, different predicted mean and fluctuation coefficient are assigned to each typical day to generate random photovoltaic power output scenes that reflect various weather scenarios.

[0168] Specifically, collect such as Figure 4 The diagram shows a 168-hour historical photovoltaic scenario, generating a risk perception boundary map. Historical data is processed using a random scenario generation method based on statistical characteristics. For the optimization problem, photovoltaic output is modeled as a superposition of a deterministic scenario mean trend and a random fluctuation component, where the intensity of the random component is determined by the time-varying volatility of historical data, rather than a simple fixed distribution assumption. The model maps different weather types by assigning differentiated predicted means and volatility coefficients for different typical days, as shown in the following formula:

[0169]

[0170] In the formula, This is the predicted mean based on historical data; The preset fluctuation intensity coefficient reflects the fluctuation ratio of photovoltaic output relative to the predicted value. This is a minimum threshold for determining the state of no light. This mechanism ensures that at night... The standard deviation of the fluctuation is forced to zero, thereby constructing a boundary that conforms to physical laws and avoiding the generation of unreasonable random power during periods of no light.

[0171] S2. Risk Mode Selection and Dynamic Safety Envelope Construction: The inverse cumulative distribution function is used to transform the statistical characteristics of historical data into power boundaries at different confidence levels, generating dynamic safety envelopes corresponding to three risk modes: aggressive, robust, and risk-averse. Binary variables are introduced to represent the selected risk mode at each time step, and a penalty cost is imposed on mode switching. For the bilinear term coupled with the photovoltaic allocation ratio and the total power consumption, the McCormick soft boundary envelope linearization technique is used to construct flexible linear constraints by introducing slack variables.

[0172] Specifically, based on the risk perception boundary, the power boundary at different confidence levels is calculated using the inverse cumulative distribution function, and risk mode switching is performed. By introducing slack variables, a dual buffering system at both probabilistic and geometric levels is constructed.

[0173] For a given violation probability The opportunity constraints of photovoltaic power output are transformed into a set of deterministic dynamic envelopes. , The formula is shown below:

[0174]

[0175]

[0176] In the formula, This is the standardized inverse cumulative distribution function, also known as the quantile function, used to determine the safety margin at the corresponding confidence level. This formula incorporates the historical forecast mean. With proportional fluctuation The scheduling boundaries under different risk modes are calculated.

[0177] Introducing binary variables This indicates that the risk mode is selected at time t. It includes three modes: aggressive, moderate, and risk-averse, corresponding to high, medium, and low risk profiles. The system must and can only select one mode at any given time; mode switching is determined by variables. Capture, the formula is as follows:

[0178]

[0179]

[0180] Generate as Figure 5 Based on the risk perception mode switching graph, the system performs dynamic trade-offs at every moment, and when the photovoltaic prediction deviation... When the risk level is relatively high, electricity and carbon prices are high, and the marginal economic benefits can cover the risk penalty costs of the high-risk mode, switching to the high-risk mode can significantly relax the safety lower bound. The freed-up scheduling margin allows the system to absorb more solar power, thereby reducing economic costs.

[0181] First, the model must satisfy the power security domain constraint based on the current risk pattern, i.e., the total photovoltaic power absorbed. It must not exceed the safe lower bound at the corresponding confidence level, as shown in the formula below:

[0182]

[0183] Secondly, regarding the photovoltaic allocation ratio With total consumption Bilinear terms generated by coupling Through McCormick slack variables A flexible linearization constraint is constructed. The variables are allowed to deviate slightly from their statistical boundaries, serving as a mapping of probabilistic default risk in the opportunity constraint to the algebraic constraint level. The specific flexible envelope constraint is constructed as follows: For example, the formula for activation is as follows:

[0184]

[0185] In the formula, For photovoltaic allocation ratio, slack variables It can quantify the tolerance for default at the linearized envelope boundary. Due to the stochastic nature of photovoltaic power output, its risk boundary, under extreme fluctuations, may lead to an extremely narrow constraint space or even no solution for the model. It provides a numerical buffer for the linearization process of bilinear terms.

[0186] S3. Construction of a tiered carbon trading model: Establish a tiered carbon trading mechanism, divide excess emissions into multiple price tiers, and introduce carbon emission constraints into the hydrogen production-steelmaking collaborative system;

[0187] The tiered carbon trading model includes at least three price tiers, each corresponding to a different carbon emission range and carbon trading unit price, and the carbon trading unit price increases stepwise as the excess emissions increase; the carbon trading cost is the sum of the products of the emissions in each tier and the corresponding carbon trading unit price;

[0188] To accurately simulate the carbon market, excess emissions will be... Assigned to In the price tiers, the width of each tier is... The tiered carbon trading model is represented as follows:

[0189]

[0190]

[0191]

[0192] In the formula, is the activation variable for the carbon emission range of the i-th tier, representing the actual excess emissions allocated within that tier; g is the carbon price tier growth rate. For each carbon price tier; The benchmark carbon price; carbon prices at each tier. Typically based on benchmark price The growth rate was calculated in advance. .

[0193] S4. Multi-entity collaborative operation optimization: Establish a multi-entity collaborative operation model with the goal of minimizing the total operating cost within the scheduling cycle of the hydrogen production-steelmaking collaborative system. By coordinating the energy storage system, the steelmaking subsystem, and the hydrogen production subsystem, which includes alkaline electrolyzers, proton exchange membrane electrolyzers, and solid oxide electrolyzers, optimize the configuration and determine the optimal operating strategy for each device.

[0194] Specifically, the objective function of the multi-entity collaborative operation model is to minimize the total system operating cost within the scheduling cycle of the hydrogen production-steelmaking collaborative system, and to achieve optimal configuration by configuring the energy storage system, steelmaking subsystem, and hydrogen production subsystem.

[0195] The objective function of the multi-agent collaborative operation model is:

[0196]

[0197] The optimization objective of the model is to minimize the total system operating cost within the scheduling period T. The objective function; To minimize the total system operating cost within the scheduling period T, the specific expressions for each cost are defined by system parameters and decision variables; C1 is the electricity purchase cost, C2 is the tiered carbon trading cost, C3 is the risk mode operation cost, C4 is the equipment start-up, shutdown and cyclic aging cost, C5 is the curtailment penalty cost, C6 is the risk mode switching penalty cost, and C7 is the McCormick envelope relaxation penalty.

[0198] Furthermore, the multi-entity collaborative operation model also includes operational constraints for the hydrogen production subsystem, the steelmaking subsystem, the energy storage system, and the system power balance constraints; as detailed below:

[0199] (1) Establish operating constraints for the hydrogen production subsystem:

[0200] A. The total input power constraint of the electrolytic cell is shown in the following formula:

[0201]

[0202]

[0203] In the formula, The total input power of electrolytic cell i during time period t is expressed in MW. The photovoltaic power allocated to this electrolyzer, in MW. The power drawn from the busbar is expressed in MW. Simultaneously, the total power drawn from the busbar by all electrolyzers cannot exceed the total available power on the busbar, i.e., the power purchased by the grid. With energy storage discharge power The sum, minus the power supplied to the steel load The part.

[0204] B. Equipment operation strategy constraints for electrolytic cells;

[0205] To reflect the direct impact of risk perception on equipment operation strategies, the operating boundary of the electrolyzer is set to be dynamically related to the current risk mode k of the system, as shown in the following formula:

[0206]

[0207]

[0208]

[0209]

[0210] In the formula, and These are the upper and lower power limits (MW) of electrolyzer i in risk mode k, respectively. This corresponds to the gradeability limit (MW / h). (This is related to the mode selection variable.) Multiplication enables dynamic switching of operating boundaries. For example, in disaster-averse mode, to maintain more system reserves to cope with potential photovoltaic power shortages, the model's power ceiling is adjusted. It may be set to a lower level; while in aggressive mode, it is allowed to operate at a higher power level to maximize hydrogen production. This is a binary variable indicating whether the device is powered on. This represents the running state at time t. (Binary variable) and These represent start-up and shutdown actions, respectively. The first equation defines start-up and shutdown actions by comparing the states at adjacent moments; a change in state from 0 to 1 necessarily corresponds to a start-up. The second inequality ensures that within a given time period, the device cannot simultaneously perform the mutually exclusive actions of starting and stopping.

[0211] Hydrogen production rate of the electrolyzer With its input electrical power The relationship is linear, determined by the hydrogen production efficiency of this type of electrolyzer. The decision is made using the formula shown below:

[0212]

[0213] In the formula, The unit is kg / h. The unit is kg / MWh. Figure 6 The statistical distribution of the operating load rate of various types of electrolyzers throughout the entire scheduling cycle is shown.

[0214] (2) Establish operating constraints for the steelmaking subsystem:

[0215] A. Constraints on steel production capacity, as shown in the following formula:

[0216]

[0217]

[0218] In the formula, It is the power consumption (MW) of process j at time t. This is the minimum total energy (MWh) required to complete a single batch of products in this process. This constraint ensures that the final output is met. The EAF, as the main melting equipment, [is involved in this process]. Maximum value; energy consumption of LF and CC The value is relatively small. and Let these be the minimum and maximum operating power of process j, respectively. This is its operating state variable. The value of this constraint varies significantly for different processes; EAF is the unit with the highest power consumption. The power requirement is much higher than other processes; LF has moderate power and is used for precise control of molten steel temperature and composition; CC has the lowest power requirement and is mainly used for driving equipment and compensating for heat dissipation.

[0219] As a high-energy-consuming and rigidly constrained load, steel production exhibits a high degree of synergy between its scheduling strategies and the operational status of energy storage systems, such as... Figure 7 and Figure 8 As shown.

[0220] B. Constraints on steel production operations, as shown in the following formula:

[0221]

[0222]

[0223]

[0224]

[0225] The minimum continuous operating time constraint is a key industrial constraint in steel production modeling, directly related to production safety and cost. To prevent frequent start-ups and shutdowns from causing significant thermal shock, mechanical stress, and material loss to the equipment, once started, it must operate continuously for at least [duration missing]. Hours. A startup action variable is defined. This indicates that if a startup action occurs at time t, then subsequent actions starting from time t... Operating status within a given time period It must always be 1. This constraint is especially important for CC, as any unplanned interruption could lead to catastrophic accidents such as leaks. The value is the longest. For EAF and LF, this constraint is mainly for protecting refractory materials and saving preheating energy. The value is relatively short. Steel production is a tightly coupled assembly line operation, with time required for transporting molten steel and other materials between processes. To ensure the physical feasibility of material flow, this model establishes strict start-stop timing constraints. If a downstream process (j+1) starts running at time t, it must be preceded by its upstream process j starting at time t- This assumes that a process has already run before time t. If upstream process j runs at time t, then its downstream process (j+1) must run at time t+1. The premise is that there is still an operational plan after the time limit, in order to avoid producing intermediate products that cannot be processed.

[0226] (3) Establish operating constraints for the energy storage system:

[0227] A. Constraints on the allocation of total system hydrogen production:

[0228] The system's total hydrogen production is divided into two parts: one part is directly supplied to steel mills. Another portion is injected into the hydrogen storage tank. To prepare for unforeseen circumstances. The total hydrogen demand of steel mills... Hydrogen is supplied directly and extracted from hydrogen storage tanks. They are both satisfied, as shown in the formula below:

[0229]

[0230]

[0231]

[0232]

[0233]

[0234] Hydrogen storage tank inventory levels (kg) is determined by the previous inventory and the current net injection. It must be kept at a safe minimum inventory level. and maximum capacity The system operates between these parameters. Meanwhile, to ensure system robustness and hydrogen reserves, the model mandates that at least [amount missing] of the total daily hydrogen production must be [percentage missing]. A certain proportion must be stored in tanks to cope with possible prolonged insufficient photovoltaic output.

[0235] B. Energy storage constraints:

[0236] Battery state of charge (MWh) also follows a similar dynamic balance, taking into account both charge and discharge efficiency. and Furthermore, batteries cannot be charged and discharged simultaneously; this physical characteristic is expressed through binary variables representing the charge and discharge states. and This is achieved by applying mutual exclusion constraints. The operating strategy of the energy storage system is as follows: Figure 9 As shown, its main function is to facilitate electricity price arbitrage and smooth system power fluctuations. Energy storage charges during the daytime when photovoltaic resources are abundant and electricity prices are low, at which time its state of charge rises rapidly, as shown in the formula below:

[0237]

[0238]

[0239]

[0240]

[0241]

[0242] C. Constraints of the carbon emission model:

[0243] Carbon emissions are an important dimension to consider in the model, and are divided into indirect emissions and direct emissions, as shown in the formula below:

[0244]

[0245]

[0246]

[0247]

[0248] Calculate total emissions The emissions consist of indirect emissions from electricity purchased from the grid and direct emissions from electric arc furnaces. The emission factors are as follows: and Composition. Calculations are based on production volume and benchmark values. Free carbon allowances Calculate the excess emissions that need to be purchased. .

[0249] (4) Establish system power balance constraints:

[0250] The system power balance constraint is the foundation of system energy conservation, ensuring that at any time t, the total power supply of the system equals the total power consumption. (Generation example follows...) Figure 10 Power balance diagram of hydrogen production-steelmaking co-production system; formula shown below:

[0251]

[0252] In the formula, The photovoltaic power allocated to electrolytic cell i; This refers to the energy storage discharge power; This represents the total power of the electrolytic cell; Power output for the steelmaking process; Power for charging energy storage.

[0253] Curtailed power is defined as the difference between the average predicted photovoltaic power and the actual photovoltaic power absorbed by the system, as shown in the following formula:

[0254]

[0255] In the formula, This represents the average of the predicted photovoltaic power.

[0256] S5, Risk Mode Scenarios Comparison

[0257] To verify the ability of the risk perception switching strategy proposed in this invention to balance economy and system reliability, three representative comparative scenarios were set up. The confidence levels used in comparative scenarios b and c are strictly consistent with the parameter settings of the main scenario a under the corresponding mode. By quantitatively analyzing the system scheduling results under different risk preferences, the limitations of a single fixed mode and the superiority of the multi-risk mode optimization strategy are explored. The specific settings of each scenario are as follows:

[0258] (1) Strategy of the present invention - Scenario a: Allows CEMS to automatically and dynamically switch between three risk modes based on the real-time fluctuation of photovoltaic forecasts in order to seek the optimal solution of the system.

[0259] (2) Low-risk operation scenario - Scenario b: Simulates the traditional industrial high-reliability operation strategy. The system is forced to be in risk avoidance mode throughout the entire scheduling cycle, and the photovoltaic output is estimated using the narrowest confidence interval. Priority is given to ensuring system safety, and an extremely cautious attitude is taken towards the consumption of new energy.

[0260] (3) High-risk operation scenario - Scenario c: Simulates the operation strategy that pursues the maximization of new energy consumption. The system is forced to always be in aggressive mode throughout the entire dispatch cycle, adopting a wider confidence interval, tending to believe in high photovoltaic output, and reducing dependence on the grid at the cost of bearing high default risk.

[0261] Table 1 Comparison Results of Multi-Risk Mode Scenarios

[0262]

[0263] Other costs include risk mode operation standby fees, McCormick linearization default penalties, and slack-off penalties.

[0264] By comparing the data in Table 1, the risk mode switching optimization strategy proposed in this invention has achieved a significant advantage in total cost. The total cost of RMB 13.0161 million is reduced by approximately 25.7% in the low-risk operation scenario and by approximately 30.2% in the high-risk operation scenario. The results demonstrate that a single fixed risk preference is not the best solution to the uncertainty problem of green electricity.

[0265] In contrast to scenario b, while ensuring absolute system security and minimizing default risks, the cost is extremely low photovoltaic (PV) consumption. Because the system's estimation of PV output is overly conservative, a large amount of usable green energy is wasted, directly increasing the system's electricity purchase costs and indirect carbon costs. Furthermore, although this scenario does not incur severe default penalties, the sheer volume of wasted solar power means that its overall risk cost, including penalties, is still higher than the strategy outlined in this invention, demonstrating the resource waste resulting from excessive conservatism.

[0266] In contrast to scenario c, although it achieved an 81.92% grid integration rate and the lowest electricity purchase and carbon emission costs, its total cost reached 18.6455 million yuan. This is because the strategy overestimated the reliability of photovoltaics. When the actual photovoltaic output could not meet its aggressive dispatch plan, the system experienced a severe power deficit, leading to a significant increase in McCormick slack variables and triggering a massive McCormick linearization default penalty of 9.1496 million yuan. This demonstrates that pursuing a high grid integration rate while ignoring physical fluctuation risks, although saving on electricity costs on paper, comes at an unbearable economic cost in terms of system stability.

[0267] In summary, this invention constructs a heterogeneous hydrogen production system model coupling alkaline electrolyzers, proton exchange membrane electrolyzers, and solid oxide electrolyzers. Through optimized combination and dynamic start-up and shutdown, it fully leverages the complementary advantages of the three electrolyzer types in terms of efficiency, flexibility, and cost. Addressing the volatility of renewable energy, a risk perception framework based on the inverse cumulative distribution function is designed. This framework utilizes chance constraints to transform the stochastic problem into a deterministic dynamic boundary under different risk preferences, and introduces the McCormick soft boundary envelope linearization technique to handle nonlinear issues in power allocation. Based on this, a risk-level controllable, economically optimal collaborative scheduling strategy is constructed. This method, through the heterogeneous hydrogen production architecture, achieves an effective complementarity between low-cost baseload support and rapid response peak shaving and valley filling capabilities, thus enabling the proposed strategy to possess excellent cost control capabilities.

[0268] The dynamic strategy proposed in this invention successfully strikes an optimal balance between conservative and aggressive approaches by flexibly switching risk modes. When photovoltaic fluctuations are relatively small, the system automatically switches to an aggressive mode to improve the grid integration rate; when photovoltaic fluctuations are severe, it switches to a risk-averse mode to mitigate default risks. This mechanism enables the strategy of this invention to keep other costs to a minimum, achieving synergistic optimization of economic benefits, environmental benefits, and system reliability.

[0269] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-electrolytic cell-short-process steelmaking collaborative scheduling method for risk mode selection and switching, characterized in that, Includes the following steps: S1. Random scene generation based on historical data statistical characteristics: According to the characteristics of photovoltaic power output under different weather types, different predicted mean and fluctuation coefficient are assigned to each typical day to generate random photovoltaic power output scenes that reflect various weather scenarios. S2. Risk mode selection and dynamic safety envelope construction: The inverse cumulative distribution function is used to transform the statistical characteristics of historical data into power boundaries under different confidence levels, generating dynamic safety envelopes corresponding to three risk modes: aggressive, robust and risk-averse. Introduction Binary variables represent the selected risk mode at each time step and impose penalty costs on mode switching; for the bilinear term coupled with the photovoltaic allocation ratio and the total power consumption, the McCormick soft boundary envelope linearization technique is adopted, and flexible linearization constraints are constructed by introducing slack variables. S3. Construction of a tiered carbon trading model: Establish a tiered carbon trading mechanism, divide excess emissions into multiple price tiers, and introduce carbon emission constraints into the hydrogen production-steelmaking collaborative system; S4. Multi-entity collaborative operation optimization: Establish a multi-entity collaborative operation model with the goal of minimizing the total operating cost within the scheduling cycle of the hydrogen production-steelmaking collaborative system. By coordinating the energy storage system, the steelmaking subsystem, and the hydrogen production subsystem, which includes alkaline electrolyzers, proton exchange membrane electrolyzers, and solid oxide electrolyzers, optimize the configuration and determine the optimal operating strategy for each device.

2. The method according to claim 1, characterized in that, In step S1, the different weather types include sunny days, cloudy days, and rainy days. The photovoltaic output is modeled as a superposition of a deterministic scenario mean trend and a random fluctuation component, wherein the intensity of the random fluctuation component is determined by the time-varying volatility of historical data, and the standard deviation of the fluctuation is forcibly reduced to zero during the nighttime period without light. The random scenario of photovoltaic output based on historical data is constructed as shown in the following formula: in, This is the predicted mean based on historical data; The preset fluctuation intensity coefficient reflects the proportion of fluctuation in photovoltaic output relative to the predicted value; The minimum threshold for determining the state of no light.

3. The method according to claim 1, characterized in that, Step S2 specifically includes: S21. Risk boundary transformation based on the inverse cumulative distribution function: For a given violation probability The opportunity constraints of photovoltaic power output are transformed into a set of deterministic dynamic envelopes. , : In the formula, The standardized inverse cumulative distribution function, also known as the quantile function, is used to determine the safety margin at the corresponding confidence level; this formula incorporates the historical forecast mean. With proportional fluctuation Calculate the scheduling boundaries under different risk modes; S22. Risk Mode Selection and Switching: Introducing binary variables This indicates that the risk mode is selected at time t. It includes three modes: aggressive, moderate, and risk-averse, corresponding to high, medium, and low risk levels. The system must and can only select one mode at any given time, and mode switching is determined by variables. Capture: The system performs dynamic trade-offs at every moment, and when the photovoltaic prediction deviation... When the risk level is relatively high, electricity and carbon prices are high, and the marginal economic benefits can cover the risk penalty costs of the high-risk mode, switching to the high-risk mode can significantly relax the safety lower bound. The freed-up scheduling margin allows the system to absorb more photovoltaic power, thereby reducing economic costs. S23, McCormick soft-boundary envelope linearization First, the model must satisfy the power security domain constraint based on the current risk pattern, i.e., the total photovoltaic power absorbed. It must not exceed the lower bound of the corresponding confidence level: Secondly, regarding the photovoltaic allocation ratio With total consumption Bilinear terms generated by coupling Through McCormick slack variables A flexible linearization constraint is constructed; the variable assignment relationship is allowed to deviate slightly near the statistical boundary, serving as a mapping of probabilistic default risk in the opportunity constraint to the algebraic constraint level; the specific flexible envelope constraint is constructed as follows, with... For example, during activation: In the formula, For photovoltaic allocation ratio, slack variable It can quantify the allowable default degree of the linearized envelope boundary; due to the randomness of photovoltaic power output, its risk boundary can lead to an extremely narrow constraint space or even no solution under extreme fluctuations. It provides a numerical buffer for the linearization process of bilinear terms.

4. The method according to claim 1, characterized in that, In step S3, the tiered carbon trading model includes at least three price tiers, each corresponding to a different carbon emission range and carbon trading unit price, and the carbon trading unit price increases stepwise as the excess emissions increase; the carbon trading cost is the sum of the products of the emissions of each tier and the corresponding carbon trading unit price. To accurately simulate the carbon market, excess emissions will be... Assigned to In the price tiers, the width of each tier is... The tiered carbon trading model is represented as follows: In the formula, is the activation variable for the carbon emission range of the i-th tier, representing the actual excess emissions allocated within that tier; g is the carbon price tier growth rate. For each carbon price tier; The benchmark carbon price; carbon prices at each tier. Typically based on benchmark price The growth rate was calculated in advance. .

5. The method according to claim 1, characterized in that, In step S4, the alkaline electrolyzer, proton exchange membrane electrolyzer, and solid oxide electrolyzer have different efficiency characteristics and start-up / shutdown constraints: the alkaline electrolyzer is suitable for base load operation, the proton exchange membrane electrolyzer has fast response capability, and the solid oxide electrolyzer adopts a high-temperature electrochemical mode and is coupled with the steelmaking waste heat recovery system.

6. The method according to claim 1, characterized in that, In step S4, the steelmaking subsystem is a short-process steelmaking process based on hydrogen-based direct reduction iron-electric arc furnace. Its energy requirements include electricity for electric furnace smelting and hydrogen for reduction reaction. The hydrogen produced by the electrolytic cell is given priority to meet the hydrogen demand for steelmaking, and the remaining part is incorporated into hydrogen storage or sold externally.

7. The method according to claim 1, characterized in that, In step S4, the energy storage system includes two forms: battery energy storage and hydrogen energy storage. Battery energy storage is used for short-term power stabilization, while hydrogen energy storage achieves cross-time energy transfer through hydrogen production in an electrolyzer and feedback from a fuel cell.

8. The method according to claim 1, characterized in that, In step S4, the objective function of the multi-entity collaborative operation model includes seven parts: electricity purchase cost, tiered carbon trading cost, risk mode operation cost, equipment start-up and shutdown and cyclic aging cost, curtailment penalty cost, risk mode switching penalty cost, and McCormick envelope relaxation penalty cost. The objective function model is shown below: In the formula, To minimize the total system operating cost within the scheduling period T, the specific expressions for each cost are defined by the system parameters and decision variables.

9. The method according to claim 1, characterized in that, The multi-subject collaborative operation model is solved using a mixed integer linear programming solver with a solution period of 168 hours and a time resolution of 1 hour. The dynamic safety envelope generated in step S2 is then embedded into the model hourly as an upper limit constraint on the photovoltaic power absorption capacity.

10. The method according to claim 1, characterized in that, In step S4, the multi-entity collaborative operation model also includes operating constraints for the hydrogen production subsystem, operating constraints for the steelmaking subsystem, operating constraints for the energy storage system, and system power balance constraints. The operational constraints of the hydrogen production subsystem include the total input electrical power constraint of the electrolyzer and the equipment operation strategy constraint of the electrolyzer. The operational constraints of the steelmaking subsystem include steel production power constraints and steel production operation constraints. The constraints on the operation of the energy storage system include total hydrogen production allocation constraints, electrical energy storage constraints, and carbon emission constraints.

Images