A steel plate production-energy collaborative optimization method

Abstract

The present application belongs to the field of industrial energy optimization, and discloses a steel plate production-energy collaborative optimization method.The steps of the method are as follows: firstly, various constraint conditions and optimization targets in the cold-rolled steel coil production process are determined.Secondly, a rule-driven rolling mill distribution model is established as a first-layer optimization model, and the contract requirements and the production capacity of different rolling mills are matched, so that the contract is distributed to each rolling mill.Finally, a lower-layer time scheduling model is solved, the demand electricity fee is calculated in a rolling manner, and the contract is arranged in production with the minimization of the demand electricity fee as the target.The present application can well adapt to the actual working conditions of the cold-rolling mill while optimizing the maximum demand of the plant electricity.

Description

Technical Field

[0001] This invention belongs to the field of energy optimization scheduling in the iron and steel industry, and relates to a steel plate production-energy collaborative optimization method. Specifically, it relates to a scheduling optimization method that combines the cold rolling production process with multiple rolling mills operating in parallel under complex constraints with energy optimization. Background Technology

[0002] Cold rolling is an important step in steel plate production, and its production process includes multiple procedures. The hardening stage after pickling is the main energy-consuming point for cold rolling mills. With increasing market competition and diversified orders, factories face the dual challenge of ensuring delivery time and reducing production costs.

[0003] In the field of steel production scheduling, most research focuses on algorithm innovation to address the complex constraints and multi-objective optimization requirements of the production process. For example, Cohen et al. (Weiss Cohen, M., Foxx, H. & Ben Alul, S. Adecision support flexible scheduling system for continuous galvanizationlines using genetic algorithm. Prod. Eng. Res. Devel. 13, 43–52 (2019). https: / / doi.org / 10.1007 / s11740-018-0856-6) analogized the CGL scheduling problem to the traveling salesman problem and then systematically improved and innovated the genetic algorithm. Through experiments and optimization of a series of key parameters such as encoding method, fitness function, and mutation operator, they improved the efficiency and quality of the algorithm's solution. Dong et al. (Dong, Z., Wang, X., & Tang, L. (2021). Color-coating scheduling with a multiobjective evolutionary algorithm based on decomposition and dynamic local search. IEEE Transactions on Automation Science and Engineering. https: / / doi.org / 10.1109 / TASE.2020.3011428) established a three-objective optimization model for the color-coating production line scheduling problem, aiming to minimize the number of transition coils, thickness jump penalties, and roll switching times. They proposed a multiobjective evolutionary algorithm based on decomposition and dynamic local search (MOEA / D-DLS) to solve the model. This algorithm dynamically allocates computational resources and performs local search on subproblems with slower evolution speeds, exhibiting good convergence.Li et al. (Li, T., Meng, Y., & Tang, L. (2023). Scheduling of continuous annealing with a multi-objective differential evolution algorithm based on deep reinforcement learning. IEEE Transactions on Automation Science and Engineering. https: / / doi.org / 10.1109 / TASE.2023.3244331) established a multi-objective integer programming model for the continuous annealing scheduling problem, aiming to minimize preparation cost, advance cost, and delay cost. They proposed an adaptive multi-objective differential evolution algorithm (AMODE-DRL) based on deep reinforcement learning. This algorithm embeds deep reinforcement learning (DRL) as a controller into the multi-objective differential evolution framework, adaptively selects mutation operators through DDQN, and dynamically adjusts key parameters using DDPG, achieving intelligent decision-making on operators and parameters during the search process. Furthermore, they designed a continuous encoding method and repair strategy suitable for discrete scheduling problems to ensure the feasibility of the solution.

[0004] Under the "two-part tariff" system, demand-based electricity charges are collected based on the user's maximum average power over a certain period, which significantly affects the energy expenditure of enterprises. Moreover, demand is no longer a simple basis for electricity billing, but a core variable for industrial users to participate in the modernization of the power system, achieve cost reduction and efficiency improvement, and green and low-carbon operation. Effective management and scheduling optimization of demand are becoming key measures for industrial enterprises to enhance their competitiveness in the energy revolution. Liu Hang et al. (Liu Hang, Shen Hao, Ji Ling, et al. Dual-layer optimization peak-shaving strategy for short-process steel enterprises based on improved RTN model considering maximum demand [J]. China Electric Power, 2025, 58(8):118-129.) proposed a dual-layer optimization peak-shaving strategy considering maximum demand for hot rolling and its upstream processes. This study improved the resource task network model by introducing time window nodes, put the process constraints under multi-order production into the model, and constructed a dual-layer optimization model of grid-enterprise collaboration to achieve load smoothing and reduce electricity costs. Li Hui et al. (Li Hui, Tang Mintai, Tan Mao, et al. Flexible load optimization scheduling for the steel industry under the new power reform to meet real-time maximum demand [J]. Manufacturing Automation, 2020, 42(12):45-52.) proposed a flexible load optimization scheduling method for the steel industry to meet real-time maximum demand. This study analyzes the load characteristics of each stage in the continuous casting process, takes the ladle furnace (LF) as the key flexible load resource, and constructs a ladle furnace load model based on the resource task network (RTN) that considers process constraints, power level adjustment and scheduling costs. It establishes an optimization scheduling framework aimed at reducing real-time maximum demand, which can effectively reduce peak demand and reduce electricity costs. However, its scheduling scope is still concentrated on the ladle furnace in the refining stage and has not been extended to subsequent high-energy-consuming links such as cold rolling. The working characteristics of subsequent links have not been fully considered. Summary of the Invention

[0005] To fill the gap in existing steel production scheduling methods in the field of cold rolling, especially hard rolling, and to address the lack of consideration for maximum demand in the field of cold rolling production scheduling, this invention proposes a steel plate production-energy synergistic optimization method.

[0006] This invention addresses the complex constraints and maximum demand optimization requirements of cold rolling processes, proposing a steel plate production-energy co-optimization method. This method employs a two-layer optimization structure, separating contract allocation and maximum demand reduction tasks: the upper layer is a mill allocation model, which fully considers the rolling capacity and volume of different mills, as well as the width, thickness, and output requirements in the contract, allocating contracts to various mills with the goals of contract completion, minimizing contract splitting, and balanced allocation; the lower layer is a scheduling optimization layer, which aims to output a reasonable production scheduling Gantt chart and minimize demand electricity costs, fully considering production constraints in the hard rolling process and outputting an optimized demand curve. Both the upper and lower layers of this invention incorporate actual production experience, enabling the output of effective scheduling schemes and demand optimization based on scheduling. Compared with existing optimization scheduling methods, this invention focuses on the parallel scheduling of multiple cold rolling mills under maximum demand considerations. Experimental verification shows that while optimizing scheduling, it can reduce maximum demand, lower cold rolling mill production costs, and improve economic efficiency.

[0007] The technical solution of this invention:

[0008] A method for co-optimizing steel plate production and energy, comprising the following steps:

[0009] S1: Data acquisition and preprocessing;

[0010] For the cold rolling production scheduling system, raw data is collected, including historical production contract data, rolling mill production capacity, historical electricity consumption data of each rolling mill, historical peak demand, and the method for calculating demand electricity costs. Among them, historical production contract data includes the width, thickness, output, and grade of cold-rolled steel coils required by the contract; rolling mill production capacity includes the maximum rolling width, minimum rolling thickness, and maximum daily capacity; and historical electricity consumption data of the rolling mills is the electricity consumption of each rolling mill in a certain month.

[0011] The collected raw data is preprocessed, including abnormal contract data screening and power curve extraction. Abnormal contract data screening is used to remove overlapping contracts due to errors by data entry personnel. Power curve extraction is used to extract the power consumption curves of different rolling mills when rolling steel coils by summarizing the historical production power consumption data of each rolling mill.

[0012] The preprocessed historical production contract data is organized into structured input data and stored in the cold rolling production scheduling system. Simultaneously, based on production process requirements, equipment safety regulations, and practical operating experience, constraints such as mill rolling capacity, processing integrity, rest soft constraints, power superposition constraints, and demand calculation constraints are organized and set as constraints for the two-stage production scheduling optimization model. Economic performance indicators for the two-stage production scheduling optimization model are set to evaluate the scheduling effect. These economic performance indicators include minimizing demand electricity costs and balancing mill load. Among these, the mill rolling capacity constraints include process feasibility constraints, equipment capacity constraints, logical association constraints, time non-overlap constraints, and time boundary constraints.

[0013] S2: Construct a two-stage production scheduling optimization model;

[0014] To address the feasibility and economic objectives of cold rolling production scheduling, and taking into full account constraints such as mill rolling capacity, processing integrity, rest soft constraints, power superposition constraints, and demand calculation constraints, a two-stage production scheduling optimization model is established. The upper level is the mill allocation model, and the lower level is the time scheduling model. By solving this two-stage production scheduling optimization model, under the premise of satisfying all constraints, a production scheduling scheme that minimizes the electricity cost of demand in the next day is calculated.

[0015] S2.1: Construct a rolling mill allocation model;

[0016] The mill allocation model aims to minimize the number of contracts allocated and balance the mill load. The decision variable is the number of times each contract is processed on different mills. The objective function is expressed as:

[0017]

[0018] in, This is a constant used to set the weight of the number of times a contract is allocated on a certain rolling mill; A binary variable representing a contract. Is it in the rolling mill? Upward processing; To balance the weighting coefficients; M represents the set of all rolling mills; C represents the set of all contracts to be produced;

[0019] For rolling mill The total load is defined as:

[0020]

[0021] in, An integer variable representing a contract. at the rolling mill Number of times allocated;

[0022] The average load is defined as:

[0023]

[0024] in, For the contract The required number of rolling passes is calculated using the following formula:

[0025]

[0026] For the contract The required weight of the rolled steel coil is specified in tons. This represents the maximum rolling capacity of the mill in a single pass, expressed in tons per pass.

[0027] The constraints of the objective function include:

[0028] 1) Technological feasibility constraints: Contract allocation must conform to the width and thickness rolling capacity of the rolling mill; the set of feasible rolling mills for contract c is defined as follows:

[0029]

[0030] in, The dividing thickness is the thickness that determines the rolling thickness capability of different types of rolling mills. The dividing width for the wide plate rolling capacity of different types of rolling mills; and Contracts Requirements for the thickness and width of the rolled steel coil, This refers to a collection of rolling mills that produce steel coils with a thickness less than the dividing line thickness. This refers to the set of rolling mills capable of rolling steel coils with a width greater than the dividing width, denoted by the symbol […]. The symbol represents logical OR. This indicates a logical AND operation; for rolling mills that do not meet the process requirements, the pass number is forcibly assigned to 0.

[0031]

[0032] 2) Processing integrity constraint: All rolling passes required for each contract must be allocated, and the rolling requirements for each contract must be fully met.

[0033]

[0034] 3) Equipment capacity constraints: The total number of rolling operations on a single rolling mill shall not exceed the safety limit, and the total number of rolling operations for each rolling mill must be controlled within a safe range.

[0035]

[0036] In the formula, For rolling mill The maximum number of safe lanes;

[0037] 4) Logical relationship constraints: Contract at the rolling mill Number of times of upper distribution Device usage state variables The following logical relationships exist between them:

[0038]

[0039] In the formula, For the contract at the rolling mill Maximum number of assignments:

[0040]

[0041] in, For rolling mill The upper limit of production capacity, in tons;

[0042] In addition, integer variables It must satisfy the integer and non-negativity constraints. It is a set of non-negative integers; It is a binary variable, and its value can be 0 or 1:

[0043]

[0044] S2.2: Construct a time scheduling model;

[0045] Based on the allocation results, the time scheduling model aims to minimize the demand electricity cost and the total number of startups, with the start time of each processing task as the decision variable; the objective function is expressed as:

[0046]

[0047] in, This represents the average demand over a maximum of 15 minutes. The binary decision variable represents the contract. at the rolling mill At the moment Has any processing task started? The degree of violation of the rest and adjustment constraints indicates the rolling mill In the Does the time window violate the rest and recuperation requirements? This is the penalty coefficient;

[0048] The constraints of the objective function include:

[0049] 1) Non-overlapping time constraint: Only one task can be executed on the same rolling mill at any given time; define the rolling mill's operating state variables:

[0050]

[0051] in, The number of time steps required for a single rolling pass; The total time step; non-overlapping constraint expression:

[0052]

[0053] Substituting into equation (14), we obtain the equivalent startup time constraint:

[0054]

[0055] 2) Power superposition constraint: The total power is the real-time superposition of the power curves of each rolling mill; assuming... For the normalized rolling power curve, This indicates the time offset from the start of rolling, and the rated power of the rolling mill is... The formula for calculating total power is:

[0056]

[0057] in, Indicates time Total power demand The duration of a single task on the rolling mill;

[0058] 3) Demand Calculation Constraints: Calculate the 15-minute average demand using a sliding window; define the 15-minute sliding window average power:

[0059]

[0060] in, minutes / step; maximum demand satisfy:

[0061]

[0062] Final demand electricity pricing is based on Calculation, where This is the historical peak demand.

[0063] 4) Soft Rest Constraint: A rest period must be scheduled after every N consecutive processing cycles. Violation is permitted but penalties will be imposed; the number of rest periods is [number missing]. The total window length and time step are For rolling mills and the start time of each possible window Define the default variable for the rest period. The constraints are:

[0064]

[0065] in, For a sufficiently large positive number, such that the constraint is... The requirement is automatically met, thus establishing the rest requirement as a soft constraint.

[0066] 5) Time Boundary Constraints: Processing tasks must be completed within the feasible time window; to ensure that processing tasks do not span multiple days, the latest start time step is set to... The constraints are:

[0067]

[0068] In addition, based on the rolling mill allocation results All rolling passes for each contract must be arranged:

[0069]

[0070] S3: Model solution strategy;

[0071] To address the characteristics of large-scale mixed-integer linear programming in mill allocation and time scheduling models, a hybrid solution strategy combining heuristic initial solution generation and mathematical programming is adopted.

[0072] S3.1 Mathematical Programming Solution and Parameter Configuration;

[0073] Both the mill allocation model and the time scheduling model are solved using a solver. First, the mill allocation model is solved, allocating contracts to each mill according to constraints. Then, the time scheduling model is solved. To accelerate the solution process, a heuristic algorithm is used to generate an initial solution. Based on this initial solution, the time scheduling model is optimized using a solver. Key solution parameters, including time constraints, optimality gaps, solution focus, and parallel computing, are configured.

[0074] S3.2 Heuristic initial solution generation;

[0075] To accelerate the optimization process, a greedy heuristic algorithm is designed to generate feasible initial solutions. The core principles of the greedy heuristic algorithm include:

[0076] Contract sorting: Sorted in descending order based on the number of rolling passes required by the contract, with priority given to tasks with large workloads;

[0077] Time slot allocation: The "earliest available time" principle is adopted to find the earliest idle time period on the selected mill that can accommodate a continuous sufficient number of time steps for each contract rolling pass.

[0078] The beneficial effects of this invention are:

[0079] 1. Compared with traditional single optimization frameworks, this invention adopts a two-stage optimization architecture (upper-level mill allocation model and lower-level time scheduling model), achieving effective separation and coordination between optimized equipment resource allocation and optimized scheduling. The mill allocation model focuses on the rational allocation of mill resources, aiming to minimize the number of contract allocations and balance the load; the lower-level scheduling model focuses on the optimal arrangement of electricity costs, aiming to minimize demand electricity costs and the number of start-ups. The structure is clear and facilitates phased design and solution.

[0080] 2. This invention incorporates various complex constraints in the actual production process during the optimization process, including factors that affect actual production scheduling such as rolling mill process constraints, equipment capacity constraints, and roll change time reservations. It also adopts soft constraints to improve the feasibility of the model and ensure that the optimization scheme can be implemented.

[0081] 3. This invention integrates demand into the cold rolling production process. Through a 15-minute sliding window demand calculation method, it accurately reflects the demand billing method in actual motor settlement. It combines the optimization of demand electricity costs with production scheduling, thereby optimizing demand electricity costs without affecting actual production, and thus optimizing the factory's electricity costs. Attached Figure Description

[0082] Figure 1 This is a flowchart illustrating the application of the present invention.

[0083] Figure 2 The following are the scheduling results when the present invention is applied to a cold rolling mill with six rolling mills: (a) is the demand curve before optimization, and (b) is the demand curve after optimization. Detailed Implementation

[0084] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0085] Example

[0086] Taking the scheduling problem of a cold rolling production line with multiple rolling mills as an example, this production line needs to fulfill 29 production contracts of different specifications, and its operational objective is to minimize electricity costs while meeting contract delivery requirements. Cold rolling production lines typically face complex process constraints and electricity cost structures, requiring comprehensive consideration of equipment capacity, process limitations, and electricity market mechanisms.

[0087] A method for co-optimizing steel plate production and energy, the application process of which is as follows: Figure 1 As shown, the specific steps include:

[0088] S1: Data acquisition and preprocessing;

[0089] The data for the cold rolling production line includes: rolling mill production capacity (maximum rolling width, minimum rolling thickness, maximum daily capacity), historical production electricity consumption data for each rolling mill, historical peak demand and the calculation method for demand electricity costs, process data (rolling time, rest requirements), and on-site production record sheets.

[0090] Data Acquisition: Historical data from May 1, 2025 to May 30, 2025 for the cold rolling production line were collected to form a production database. This database was used to extract various data required for optimized scheduling, and the data from the last week was used for comparison and verification of the method's effectiveness. The data sampling time interval was set according to the actual dynamic characteristics of production. In this embodiment, the mill allocation model used batch data, the time scheduling model used 5-minute granular time data, and the cost calculation used 1-minute granular power data.

[0091] Data preprocessing: The collected raw data are preprocessed as follows:

[0092] 1) Outlier detection and removal: Remove records from production logs that contain very small tonnage values ​​and are for cut rolls;

[0093] 2) Power curve processing: Extract the conventional power consumption curve for single-coil rolling from the power consumption curve of the rolling mill;

[0094] 3) Organize equipment parameters and actual operating constraints: Based on the process requirements and equipment parameters of the cold rolling production line, define the following key constraints:

[0095] Rolling capacity constraints: Contracts with a thickness of 0.22mm or less can only be processed on rolling mills 3, 4, 5, and 6. Contracts with a width greater than or equal to 1100mm can only be processed on rolling mills No. 2 and No. 6, i.e. The safe limit for the total number of rolling operations per day for a single rolling mill is 30; the duration of a single rolling task is 40 minutes (corresponding to 8 five-minute time steps); the time window for each day is 0:00-23:20, corresponding to time steps 0-280.

[0096] Processing integrity constraint: All rolling passes required for each contract must be allocated.

[0097] Power superposition constraint: The total power is the real-time superposition of the power curves of each rolling mill.

[0098] Demand calculation constraint: The average power is calculated every 15 minutes using a sliding window, and the maximum value between the calculated value and the historical peak demand is taken as the maximum demand.

[0099] Rest and soft constraint requirements: A 20-minute rest period is required after every 5 coils are rolled out continuously, for the purpose of changing the rolls;

[0100] Electricity cost parameters: historical peak demand .

[0101] S2: Construct a two-stage production scheduling optimization model;

[0102] S2.1: Construct the mill allocation model. The mill allocation model aims to minimize the number of contracts allocated and balance the mill load. The decision variable is the number of times each contract is processed on different mills.

[0103] S2.2: Based on the allocation results, the time scheduling model aims to minimize the demand electricity cost and the total number of startups, with the decision variable being the start time of each processing task.

[0104] S3: Model solution strategy;

[0105] A hybrid solution strategy combining heuristic initial solution generation and mathematical programming is adopted. First, a feasible initial scheduling scheme is generated based on the allocation results, and then the Gurobi optimization solver is used for solving. The key parameter configurations are shown in Table 1.

[0106] S3.1 Mathematical Programming Solution and Parameter Configuration;

[0107] Both the mill allocation model and the time scheduling model are solved using solvers; key solution parameters, including maximum solution time, solution focus, and parallel computing configuration, are configured.

[0108] Table 1: Parameter Configuration of the Two-Stage Optimization Model

[0109]

[0110] S3.2 Heuristic initial solution generation;

[0111] To accelerate the optimization process, a greedy heuristic algorithm is designed to generate feasible initial solutions. The solutions are sorted in descending order according to the contract rolling pass demand, with priority given to tasks with large workloads and the "earliest available time" principle is adopted. For each contract rolling pass, the earliest available time period that can accommodate a sufficient number of consecutive time steps is found on the selected mill.

[0112] Optimization results extraction and cost calculation:

[0113] Scheduling scheme extraction. After solving the time scheduling model, the start and end times of each rolling task are extracted to generate a complete Gantt chart scheduling scheme. The scheduling results are as follows: Figure 2 As shown, (b) is the optimized demand curve, and (a) is the unoptimized demand curve, used to compare the optimization effect. In this embodiment, 29 contracts are arranged with a total of 168 rolling tasks, covering the entire 24 hours.

[0114] Accurate electricity cost calculation. Based on the dispatch scheme, the 5-minute granularity plan is increased to 1-minute granularity to accurately calculate the demand curve. Average power is calculated through a 15-minute sliding window, combined with... Determine the maximum demand.

[0115] Implementation effect analysis:

[0116] To verify the effectiveness of this method, it was compared with traditional experience-based scheduling methods. One week's production records of the cold rolling mill were extracted and the plan was revised using this method. The maximum demand after scheduling was calculated and compared with the demand before scheduling. The test results are shown in Table 2. Compared with manual scheduling methods relying on individual experience, this method can more comprehensively consider various influencing factors and optimize the maximum demand, thereby reducing the plant's production costs.

[0117] Table 2 Comparison of Maximum Demand Before and After Optimization

[0118]

[0119] In summary, this embodiment applies a steel plate production-energy co-optimization method to cold rolling production scheduling, achieving coordinated optimization of mill resource allocation and demand-based electricity costs. The upper-level mill allocation model ensures that process constraints and equipment capabilities are met, while the lower-level time scheduling model focuses on optimizing electricity costs and feasible scheduling. Case study results demonstrate that this method effectively reduces electricity costs and improves production efficiency while maintaining production process requirements, exhibiting high computational efficiency and promising practical application prospects. This method can provide scientific and efficient decision support for cold rolling production scheduling, possessing significant value for widespread application.

[0120] The above-described embodiments are merely illustrative of the implementation of the present invention, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A method for co-optimizing steel plate production and energy, characterized in that, The steps are as follows: S1: Data acquisition and preprocessing; S2: Construct a two-stage production scheduling optimization model; To address the feasibility and economic objectives of cold rolling production scheduling, and fully considering constraints such as mill rolling capacity, processing integrity, rest soft constraints, power superposition constraints, and demand calculation constraints, a two-stage production scheduling optimization model is established. The upper level is a mill allocation model, and the lower level is a time scheduling model. The mill allocation model aims to minimize the number of contracts allocated and balance the mill load, with the decision variable being the number of times each contract is processed on different mills. Based on the allocation results, the time scheduling model aims to minimize the demand electricity cost and the total number of starts, with the decision variable being the start time of each processing task. By solving this two-stage production scheduling optimization model, under the premise of satisfying all constraints, a production scheduling scheme that minimizes the demand electricity cost within the next day is calculated. S3: Model solution strategy; A hybrid solution strategy combining heuristic initial solution generation and mathematical programming is adopted; firstly, a feasible initial scheduling scheme is generated based on the allocation results, and then the Gurobi optimization solver is used for solving. S3.1 Mathematical Programming Solution and Parameter Configuration; Both the mill allocation model and the time scheduling model were solved using a solver; S3.2 Heuristic initial solution generation; A greedy heuristic algorithm is used to generate feasible initial solutions; the solutions are sorted in descending order according to the contract rolling pass demand, and the "earliest available time" principle is adopted to find the earliest idle time period on the selected mill that can accommodate a continuous sufficient number of time steps for each contract rolling pass; After solving the time scheduling model, the start time and end time of each rolling task are extracted to generate a complete scheduling scheme.

2. The steel plate production-energy synergistic optimization method according to claim 1, characterized in that, The specific implementation process of step S1 is as follows: For the cold rolling production scheduling system, raw data is collected, including historical production contract data, rolling mill production capacity, historical electricity consumption data of each rolling mill, historical peak demand, and the method for calculating demand electricity costs. Among them, historical production contract data includes the width, thickness, output, and grade of cold-rolled steel coils required by the contract; rolling mill production capacity includes the maximum rolling width, minimum rolling thickness, and maximum daily capacity; and historical electricity consumption data of the rolling mills is the electricity consumption of each rolling mill in a certain month. The collected raw data is preprocessed, including abnormal contract data screening and power curve extraction. Abnormal contract data screening is used to remove overlapping contracts due to errors by data entry personnel. Power curve extraction is used to extract the power consumption curves of different rolling mills when rolling steel coils by summarizing the historical production power consumption data of each rolling mill. The preprocessed historical production contract data is organized into structured input data and stored in the cold rolling production scheduling system. Simultaneously, based on production process requirements, equipment safety regulations, and practical operating experience, constraints such as mill rolling capacity, processing integrity, rest soft constraints, power superposition constraints, and demand calculation constraints are organized and set as constraints for the two-stage production scheduling optimization model. Economic performance indicators for the two-stage production scheduling optimization model are set to evaluate the scheduling effect. These economic performance indicators include minimizing demand electricity costs and balancing mill load. Among these, the mill rolling capacity constraints include process feasibility constraints, equipment capacity constraints, logical association constraints, time non-overlap constraints, and time boundary constraints.

3. The steel plate production-energy synergistic optimization method according to claim 2, characterized in that, The specific implementation process of step S2 is as follows: S2.1: Construct a rolling mill allocation model; The mill allocation model aims to minimize the number of contracts allocated and balance the mill load. The decision variable is the number of times each contract is processed on different mills. The objective function is expressed as: in, This is a constant used to set the weight of the number of times a contract is allocated on a certain rolling mill; A binary variable representing a contract. Is it in the rolling mill? Upward processing; To balance the weighting coefficients; M represents the set of all rolling mills; C represents the set of all contracts to be produced; For rolling mill The total load is defined as: in, An integer variable representing a contract. at the rolling mill Number of times allocated; The average load is defined as: in, For the contract The required number of rolling passes is calculated using the following formula: For the contract The required weight of the rolled steel coil is specified in tons. This represents the maximum rolling capacity of the mill in a single pass, expressed in tons per pass. The constraints of the objective function include: 1) Technological feasibility constraints: Contract allocation must conform to the width and thickness rolling capacity of the rolling mill; the set of feasible rolling mills for contract c is defined as follows: in, The dividing thickness is the thickness that determines the rolling thickness capability of different types of rolling mills. The dividing width for the wide plate rolling capacity of different types of rolling mills; and Contracts Requirements for the thickness and width of the rolled steel coil, This refers to a collection of rolling mills that produce steel coils with a thickness less than the dividing line thickness. This refers to the set of rolling mills capable of rolling steel coils with a width greater than the dividing width, denoted by the symbol […]. The symbol represents logical OR. This indicates a logical AND operation; for rolling mills that do not meet the process requirements, the pass number is forcibly assigned to 0. 2) Processing integrity constraint: All rolling passes required for each contract must be allocated, and the rolling requirements for each contract must be fully met. 3) Equipment capacity constraints: The total number of rolling operations on a single rolling mill shall not exceed the safety limit, and the total number of rolling operations for each rolling mill must be controlled within a safe range. In the formula, For rolling mill The maximum number of safe lanes; 4) Logical relationship constraints: Contract at the rolling mill Number of times of upper distribution Device usage state variables The following logical relationships exist between them: In the formula, For the contract at the rolling mill Maximum number of assignments: in, For rolling mill The upper limit of production capacity, in tons; In addition, integer variables It must satisfy the integer and non-negativity constraints. It is a set of non-negative integers; It is a binary variable, and its value can be 0 or 1: S2.2: Construct a time scheduling model; Based on the allocation results, the time scheduling model aims to minimize the demand electricity cost and the total number of startups, with the start time of each processing task as the decision variable; the objective function is expressed as: in, This represents the average demand over a maximum of 15 minutes. The binary decision variable represents the contract. at the rolling mill At the moment Has any processing task started? The degree of violation of the rest and adjustment constraints indicates the rolling mill In the Does the time window violate the rest and recuperation requirements? This is the penalty coefficient; The constraints of the objective function include: 1) Non-overlapping time constraint: Only one task can be executed on the same rolling mill at any given time; define the rolling mill's operating state variables: in, The number of time steps required for a single rolling pass; The total time step; non-overlapping constraint expression: Substituting into the following formula yields the equivalent startup time constraint: 2) Power superposition constraint: The total power is the real-time superposition of the power curves of each rolling mill; assuming... For the normalized rolling power curve, This indicates the time offset from the start of rolling, and the rated power of the rolling mill is... The formula for calculating total power is: in, Indicates time Total power demand The duration of a single task on the rolling mill; 3) Demand Calculation Constraints: Calculate the 15-minute average demand using a sliding window; define the 15-minute sliding window average power: in, Minutes / step; Maximum demand satisfy: Final demand electricity pricing is based on Calculation, where This is the historical peak demand. 4) Soft Rest Constraint: A rest period must be scheduled after every N consecutive processing cycles. Violation is permitted but penalties will be imposed; the number of rest periods is [number missing]. The total window length and time step are For rolling mills and the start time of each possible window Define the default variable for the rest period. The constraints are: in, For a sufficiently large positive number, such that the constraint is... The requirement is automatically met, thus establishing the rest requirement as a soft constraint. 5) Time Boundary Constraints: Processing tasks must be completed within the feasible time window; to ensure that processing tasks do not span multiple days, the latest start time step is set to... The constraints are: In addition, based on the rolling mill allocation results All rolling passes for each contract must be arranged: 。

Images