A ladle low-energy consumption matching optimization method under a given steelmaking production scheduling scheme
By constructing a total energy consumption optimization model for steel ladles and an adaptive large neighborhood search algorithm, the selection and matching of steel ladles in steelmaking production are optimized, solving the problem of high energy consumption in steel ladle selection under given production scheduling conditions, and achieving a reduction in energy consumption and an improvement in steel ladle utilization.
Patent Information
- Application Number
- CN202611586997.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-14
- Publication Date
- 2026-11-10
AI Technical Summary
In steelmaking production, existing technologies struggle to effectively balance various constraints such as steel grade compatibility, thermal condition, restrictions on the use of special steel grades, and time connections between furnace runs under given production schedules. This leads to problems such as long ladle waiting times, the selection of ladles in low-temperature conditions, or the activation of new ladles, making it difficult to optimize ladle selection to reduce energy consumption.
An optimization model is constructed with the goal of minimizing the total energy consumption of the ladle. An adaptive large neighborhood search algorithm that integrates simulated annealing acceptance criteria is adopted. An initial solution is constructed through rule heuristics, and the ladle selection is adjusted by combining destruction and repair operators to optimize the energy consumption of the ladle during idle, preheating and baking and start-up.
While meeting production constraints, the overall energy consumption during the ladle turnover process was reduced, unnecessary waiting and additional energy consumption from using the ladle were decreased, and the utilization level of the ladle was improved.
Smart Images

Figure CN123070653A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of steelmaking production scheduling and ladle selection optimization technology, and in particular to a low-energy-consumption ladle selection optimization method under a given steelmaking production scheduling scheme. Background Technology
[0002] The ladle is a crucial piece of equipment in the steelmaking process for holding and transferring molten steel, and its selection directly impacts overall production efficiency and energy consumption. Under given production scheduling conditions, ladle selection must simultaneously meet multiple constraints, including steel grade compatibility, thermal condition, restrictions on the use of special steel grades, frequency of use, and time coordination between heats. Existing technologies often rely on manual experience, fixed priority rules, or heat-by-heat matching for ladle allocation. As the number of heats increases and constraints become more coupled, the number of candidate ladle combinations grows rapidly. Traditional rule-based methods struggle to balance the ladle requirements of the current heat with those of subsequent heats, easily leading to problems such as excessively long ladle waiting times, the selection of ladles in low-temperature conditions, additional baking, or the activation of new ladles. The adaptive large neighborhood search algorithm can expand the solution space search range through different destruction and repair operators. Combined with a simulated annealing acceptance mechanism, it can enhance the ability to escape local optima, making it suitable for multi-heat, multi-constraint ladle selection and combination optimization. Therefore, it is necessary to establish an adaptive large neighborhood search algorithm based on the fusion simulated annealing acceptance criterion to optimize ladle selection and reduce the overall energy consumption of the ladle turnover process while meeting production constraints. Summary of the Invention
[0003] Based on the aforementioned existing problems, this invention proposes a method for optimizing the selection of low-energy-consumption steel ladles under a given steelmaking production scheduling scheme.
[0004] The technical solution of the present invention is as follows: a low-energy consumption ladle selection optimization method under a given steelmaking production scheduling scheme, constructing a ladle selection optimization model with the goal of minimizing the total energy consumption of the ladle, considering steel type, ladle material, thermal state, special steel type usage restrictions, and time constraints, constructing an initial solution based on rule heuristics, and using an adaptive large neighborhood search algorithm that integrates simulated annealing acceptance criteria to solve for the optimal feasible solution.
[0005] The total energy consumption of the steel ladle TEC Including idle energy consumption E 1. Energy consumption for preheating and baking E 2 and Activation Energy Consumption E 3; ; Subtract the actual processing time of each heat cycle from the total time the ladle is occupied to calculate the idle energy consumption generated by the ladle in the non-processing state. E 1; ; er pThis represents the baking energy consumption coefficient, expressed in kgce / min. j Indicates the first j Furnace batch; n j Indicates the total number of furnace runs; p Indicates the ladle serial number; Te j,p Indicates furnace number j Allocated to steel ladle p Upward transfer end time; P j Indicates furnace number j Available ladle collection; Tb j,p Indicates furnace number j Allocated to steel ladle p Upward transfer start time; i Indicates the first i Processing stage; n i Indicates the total number of stages; m Indicates smelting equipment; Pt j,m Indicates furnace number j In smelting equipment m The smelting time on the surface; X i,j,m Indicates if the furnace number j In the stage i Assigned to machine m If the processing is performed, the value is 1; otherwise, the value is 0. M i Indicates the first i The set of available equipment for each process; When the same ladle is used for two consecutive heat transfers, it needs to wait in the baking area for its next use. The preheating and baking energy consumption of the ladle is calculated based on the waiting time. E 2; ; T p Indicates the service life of the ladle; Wt j,j′,p This indicates the waiting time for transferring two adjacent heats of the same ladle. t Indicates the cumulative number of times the ladle has been used; Calculate the energy consumption for using the ladle based on whether it is activated. E 3; As long as the steel ladle p The activation variable of the ladle is assigned and must be responsible for at least one heat. K p The value is set to 1; U is a positive integer; ; ; eq p Indicates the energy consumption coefficient of a new or slag line coil; K p Indicates when the steel ladle p The value of 1 indicates the assigned furnace transfer number; otherwise, it is 0. Q j,p Indicates furnace number j Assigned to steel ladle p A value of 1 indicates a transfer, otherwise a value of 0.
[0006] The specific timing constraints are as follows: To ensure that the molten steel in each heat is carried and transferred by a single ladle, one and only one ladle is selected from the set of available ladles for that heat. ; ; a j,p Indicates furnace number j Steel ladles are permitted. p It is 1 if it is true, otherwise it is 0. J Indicates the set of furnace batches; The number of heat cycles each ladle can handle shall not exceed its maximum permitted number of uses; ; For any two different heats assigned to the same ladle, the transfer sequence is unique; ; Z j,j′,p Indicates if the furnace number j In the furnace j Previously assigned to steel ladles p A value of 1 indicates a successful transfer, otherwise a value of 0. Z j′,j,p Indicates if the furnace number j In the furnace j Previously assigned to steel ladles p A value of 1 indicates a successful transfer, otherwise a value of 0. ; Furnace j and furnace j' Continuous transfer from the same ladle, and heat number j Ranked by furnace j' Previously, furnace batch j' The transfer time must not start earlier than the furnace batch. j The sum of the transfer end time and the necessary preparation time for the steel ladle; Z j′,j′′,p Indicates if the furnace number jIn the furnace j Previously assigned to the ladle p A value of 1 indicates a successful transfer, otherwise a value of 0. Z j′′,j,p Indicates if the furnace number j In the furnace j Previously assigned to steel ladles p A value of 1 indicates a successful transfer, otherwise a value of 0. ; S p This indicates the minimum baking time required before using the ladle; Tb j,p Indicates furnace number j Allocated to steel ladle p Upward transfer start time; Tb j′,p Indicates furnace number j 'Assigned to steel ladle p Upward transfer start time; Q j′,p Indicates furnace number j ′ was allocated to the ladle p The value is 1 if the transfer is successful; otherwise, it is 0. The start time of ladle transfer is equal to the completion time of its steelmaking stage, and the end time of ladle transfer is equal to the time when the ladle enters the baking position. If the heat is not assigned to this ladle, the constraint is relaxed using U. ; ; Tf i,j Indicates furnace number j During the processing stage i Completion time; Δ t BE This indicates the transfer time from the baking position to the first processing stage, Δ. t CB This indicates the transfer time from the last processing stage to the baking position; IC and CC represent the die casting stage and the continuous casting stage, respectively. i =1 indicates the first processing stage; Furnace j With furnace j' They were assigned to the same ladle and in the same heat. j Located in the furnace j' Previously, the waiting time between two ladle transfer tasks was equal to the difference between the start time of the later transfer and the end time of the previous transfer. ; ; Wtj,j',p This indicates the waiting time for the transfer of the same ladle to adjacent furnaces.
[0007] Rule-based heuristic construction of initial solutions: After meeting the requirements for steel type, ladle material, thermal state, special steel type usage restrictions, and time constraints, the ladle selection enters the optimization decision-making stage. From the set of available ladles, ladles with short dwell time, no need for additional baking, and matching the current furnace process route are selected based on rule heuristics.
[0008] The rule-based heuristic construction of the initial solution is as follows: The ladle selection scheme is encoded using a ladle-heat allocation vector, and the execution sequence, time status and energy consumption of each ladle are obtained by decoding. Encoding: The ladle-heat allocation production decision problem is transformed into a serialized data structure. Ladle-heat allocation encoding: s =(p1,p2,…, p j ,…,p N ); N For the total number of heats in the steelmaking production plan, the coded number is... j Bit value p j Representing the j The ladle number assigned to each heat batch; Decoding: Based on the encoding and timing constraints, calculate the start time, end time, waiting time, and baking time for each ladle, and finally solve for the total energy consumption of the ladle; during decoding, invalid codes are removed to ensure the feasibility of the solution; Initial solution is constructed using rule-based heuristics. s 0 .
[0009] The initial solution for the first iteration s 0 Perform destruction and repair to generate new candidate solutions. s In the initialization phase of the adaptive large neighborhood search algorithm that incorporates the simulated annealing acceptance criterion, let... s = s 0 ; The destruction and repair operators are designed as follows: Design principle of the destruction operator: the initial solution for each iteration s, From the set of destructive operators Ω - Select one destruction operator D from the middle s Destruction is performed to obtain a destruction solution. s - ; Destruction operator set Ω -={D1, D2, D3}, including the random furnace failure operator D1, the marginal energy consumption worst failure operator D2, and the same steel type failure operator D3; Repair operator design principle: From the repair operator set Ω + Choose a repair operator R to repair the broken solution. s - Repairing, s ′ represents the solution obtained after repair; Repair operator set Ω + ={R1, R2, R3}, including rule-priority repair operator R1, incremental energy consumption greedy repair operator R2, and Regret-k regret value repair operator R3; Roulette wheel selection options include destruction and repair operators; ; ; P (D d ) and P(R r The numbers and represent the probabilities of the destruction operator being selected and the repair operator being selected, respectively. and Indicates the number to be calculated d The weight of the destruction operator and the weight to be calculated for the first destruction operator r N represents the weights of the repair operators, where h is the index of the summation of all operators; D N represents the total number of destruction operators. R Indicates the total number of repair operators; , Let h represent the weight of the h-th disruptive operator and the weight of the h-th repair operator in the summation process; Construct a cumulative probability interval, generate two random numbers in the range [0,1], and select the destruction operator and the repair operator based on the interval in which the random numbers fall; after accumulating a complete cycle, update the weights of the destruction operator and the repair operator. Perform destruction and repair operations to generate new candidate solutions; perform feasibility verification on the new candidate solutions to check whether they violate the restrictions on steel type, ladle material, thermal state, and use of special steel types; if the new candidate solution is feasible, proceed to simulated annealing for scoring; if it is not feasible, perform a roulette wheel selection to choose the destruction and repair operators again.
[0010] If a new candidate solution is feasible, the simulation annealing score is used to determine whether to accept it. If accepted, the new candidate solution is assigned a score. s ′ is the preferred candidate solution s ′′; Based on the Metropolis criterion, with probability P SA Accept new candidate solutions s ′′, where Δ fThis represents the change in energy consumption. T SA The current temperature gradually decreases according to the cooling function during the iteration process; when Δ f When <0, it indicates a preferred candidate solution. s '' is better than the initial solution s The preferred candidate solution will definitely be accepted; when Δ f When the value is ≥0, it means that the preferred candidate solution is inferior to the current solution. The preferred candidate solution is accepted with a certain probability, and this probability gradually decreases as the temperature decreases. ; Rate the operators in this round: ; Ψ=σ1, highest reward: optimal candidate solution s ''Breaking the historical best f ( s ′′)< f ( s The highest score will be awarded. Ψ=σ2, Medium reward: Optimal candidate solution s The solution did not break the historical best, but it is better than the current solution, so a moderate reward is given. Ψ=σ3, basic reward: the preferred candidate solution is worse than the current solution, but is accepted by the acceptance criterion of simulated annealing; Ψ=σ4, no reward: the preferred candidate solution is rejected by the simulated annealing criterion, or the generated preferred candidate solution itself is infeasible, and the operator does not score in this round; f ( s ′′) represents the preferred candidate solution s The objective function value corresponding to ′′; f ( s ) represents the historical best solution s The corresponding objective function value; f ( s () represents the objective function value corresponding to the initial solution; Determine if the operator weight update cycle has been reached. If not, directly update the simulated return temperature and iterate. If the operator weight update cycle has been reached, use an adaptive mechanism to update the operator weights. After one statistical period, the operator weights are updated after multiple iterations, and the updated operator weights are used in the next round of roulette selection. ; ; , They represent the firstd The historical weight of the first destruction operator, the first d The updated weights of the destructive operators; , , indicating the first r The historical weights of the repair operator, the first r The updated weights of the repair operator; λ is the learning rate, with a value in the range [0,1]; Indicates the number of times within the statistical period d The total score obtained by each destruction operator; Indicates the number of times within the statistical period r The total score obtained by each repair operator; Indicates the number of times within the statistical period d Number of times the destruction operator is called; Indicates the number of times within the statistical period r Number of times the repair operator is called; Update the simulated annealing temperature; set the initial temperature, final temperature, and cooling coefficient to represent the cooling iteration process; the cooling iteration process uses an exponential cooling method to adjust the number of iterations. T ′SA =α T SA α is the cooling coefficient, which takes a value in (0,1); T ′SA This indicates the new temperature after cooling. T SA This indicates the temperature before cooling down; The solution is iterated until the termination condition is met, and the optimal ladle selection scheme and total energy consumption are obtained.
[0011] The specific restrictions on the use of steel grades, ladle materials, thermal states, and special steel grades are as follows: Rule 1: Ladle material and steel grade compatibility constraints. Low-carbon steel, ultra-low-carbon steel, and steel grades sensitive to carbon increase should be given priority for low-carbon ladles. When low-carbon ladle turnover is insufficient, low-carbon ladles can be used to smelt high-carbon steel.
[0012] Rule 2: Thermal Condition and Baking Constraints for Steel Ladles. Steel ladles must meet thermal condition and baking requirements before being put into use. Ladles with a continuous usage interval of less than or equal to 2 hours are considered normal ladles; those exceeding 2 hours are considered new ladles; and new ladles between 2 and 4 hours old must be placed in a horizontal ladle baking machine for baking.
[0013] Rule 3: Ladle Constraints in Continuous Casting. A new ladle cannot be used for the first heat of a newly cast continuous casting operation.
[0014] Rule 4: Steel Grade Quality and Safety Constraints. The first trial production of a product must not use the first batch of a new ladle or the first batch of a slag line ladle. For nuclear power products, Class A products, bearing steel, high-grade structural steel, ultra-low carbon stainless steel, and some new energy and passenger vehicle steels, normal ladles should be prioritized. The first batch of a new ladle, the first batch of a slag line ladle, or a newly erected ladle cannot be used; normal turnover ladles must be prioritized.
[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) Under the given steelmaking production scheduling plan, the steel ladle allocation relationship of each heat is optimized by comprehensively considering the steel type, ladle material, thermal state, special steel type usage restrictions, maximum usage times and time connection requirements, so that the allocation plan meets the production process and ladle turnover requirements.
[0016] (2) Taking the total energy consumption of idle ladle, preheating and baking energy consumption and activation energy consumption as the optimization target, coordinating the relationship between ladle scheduling and energy consumption helps to reduce unnecessary waiting and additional energy consumption generated by activating the ladle.
[0017] (3) The initial solution is constructed through rule heuristics. The adaptive large neighborhood search algorithm is used to adjust the allocation relationship between the furnace and the ladle by combining the destruction operator and the repair operator. The operator weight is updated adaptively according to the search effect. At the same time, the simulated annealing acceptance criterion is introduced so that the search process can accept feasible solutions with a certain probability, providing a search path to escape local optima and improving the ladle selection results.
[0018] (4) Under the comparative conditions of this embodiment, compared with the manual ladle selection scheme, the scheme obtained by the present invention reduces the number of ladles used by 1 and the total energy consumption is reduced by 1.07 kgce / t, indicating that the present invention can improve the ladle utilization level and reduce the energy consumption during the use of ladle while meeting the given production schedule requirements. Attached Figure Description
[0019] Figure 1 This is a schematic diagram of the steelmaking production process in an embodiment of the present invention.
[0020] Figure 2 This is the technology roadmap adopted. Detailed Implementation
[0021] This embodiment focuses on ladle selection under a given steelmaking production scheduling plan. For example... Figure 1 As shown, steelmaking production includes smelting, refining, casting, and ingot casting stages, with each heat processed on corresponding equipment according to its process route. Given the processing equipment, process route, and processing time arrangement for each heat, the method of this invention is used to determine the ladle used for each heat and the sequence of heats that each ladle will continuously serve.
[0022] like Figure 2As shown, the low-energy consumption selection and optimization method for steel ladles in this embodiment includes the following steps.
[0023] S1. Obtain production scheduling data and ladle status data.
[0024] Obtain the furnace number, steel type, process route, processing equipment, and time and corresponding energy consumption coefficient of each processing stage in a given production schedule, and obtain the ladle usage restrictions for special steel types.
[0025] Based on factors such as steel type, ladle material, and restrictions on the use of special steel types, a set of candidate ladles is determined for each heat. The thermal state, number of uses, and available time, which change as the task is executed, are updated and assessed during subsequent decoding and feasibility verification.
[0026] S2. Establish a ladle selection optimization model.
[0027] The decision-making process focuses on the allocation relationship between heat cycles and ladles, with the goal of minimizing total ladle energy consumption. The objective function is: TEC = E 1+ E 2+ E 3; in, E 1 represents the idle energy consumption during the non-processing state of the ladle when it undertakes a heat-running task. E 2 represents the energy consumption for preheating and baking between consecutive heat cycles performed by the same ladle. E 3 represents the energy consumption for ladle activation. All energy consumption parameters are calculated based on the corresponding duration, activation status, and energy consumption coefficient. The timing range between the task's occupancy period and adjacent tasks is determined separately to avoid double counting.
[0028] The ladle selection optimization model meets the following conditions: each heat is assigned one and only one applicable ladle; the ladle allocation meets the restrictions on steel type, material, thermal state and special steel type; the number of heats undertaken by each ladle does not exceed the allowed number of uses; the tasks undertaken by the same ladle do not conflict in time, and the necessary preparation time requirements are met between adjacent tasks; the ladle transfer time is coordinated with the corresponding process time in the given production schedule.
[0029] S3. Encode and construct the initial solution. Use the ladle numbers corresponding to the heats to construct the allocation vector: s = (p1, p2, ..., p...). j , ..., p N ); where N is the total number of furnace runs, 53 furnaces, p j The ladle number assigned to the j-th heat.
[0030] An initial solution is constructed using a rule-based heuristic. Among the candidate ladles that meet the constraints, ladles with short dwell times, no need for additional baking, and that match the current heat's process route are prioritized. After each allocation, the task sequence, time status, and usage count of the corresponding ladle are updated.
[0031] Decode the allocation vector to obtain the task execution sequence, time taken, and waiting time between adjacent tasks for each ladle, and calculate the total energy consumption. Denote the obtained feasible initial solution as s0, and compare the current solution s0 with the historical best solution s0. All are initialized to s0.
[0032] S4. Perform damage and repair operations.
[0033] The system includes a set of destruction operators for random furnace destruction, worst marginal energy consumption destruction, and destruction of the same steel grade, as well as a set of repair operators for rule-based repair, incremental energy consumption greedy repair, and Regret-k regret value repair.
[0034] Random heat disruption is used to remove ladle allocation relationships for some heats; worst marginal energy consumption disruption is used to prioritize the removal of allocation relationships that contribute significantly to the energy consumption of the current scheme; same steel type disruption is used to remove the allocation relationships for some heats of the same steel type so that their ladle arrangements can be reorganized.
[0035] During the remediation process, the removed heats will be reassigned to ladles that meet the constraints. Rule-priority remediation allocates heats according to the selection priority rule; incremental energy consumption greedy remediation prioritizes feasible allocations that cause a smaller increase in total energy consumption; Regret-k regret value remediation combines the cost differences between different feasible allocations to determine the processing priority of the heats to be allocated.
[0036] Based on the current weights of each operator, a destruction operator and a repair operator are selected respectively by roulette wheel to perform destruction and repair on the current solution, and a new candidate solution s′ is obtained.
[0037] S5. Verify the candidate solutions and make an acceptance decision.
[0038] Candidate solutions are decoded to verify whether they meet the requirements for steel type compatibility, ladle material, thermal state, special steel type usage restrictions, usage frequency, and time coordination. Infeasible candidate solutions are discarded, and new candidate solutions are generated. For feasible candidate solutions, the energy consumption difference between them and the current solution is calculated: Δ f = f ( s )- f ( s ); in, f ( s )and f ( s) represent the total energy consumption of the new candidate solution and the total energy consumption of the current solution, respectively.
[0039] When Δ f When Δ ≤ 0, accept candidate solutions and update the current solution; when Δ f When >0, exp(-Δ f A candidate solution is accepted with a probability of T), where T is the current simulated annealing temperature. If a candidate solution is better than the historical best solution, the historical best solution s is updated synchronously. .
[0040] S6. Update operator weights and simulated annealing temperature.
[0041] The destruction and repair operators used in this round are scored based on whether the candidate solution refreshes the historical best, improves the current solution, and is accepted.
[0042] Once the set weight update cycle is reached, the weights of each operator are updated based on their cumulative scores and the number of times they are called within that cycle, so that operators with better search performance receive a higher probability of being selected in the future.
[0043] According to T′ SA =αT SA The simulated annealing temperature is lowered, with an initial temperature of 10000 and an ending temperature of 1, where α is set to 0.998. The process of candidate solution generation, feasibility verification, acceptance determination, and parameter updating is repeated until the set termination conditions are met.
[0044] S7. Output the optimized solutions and compare them.
[0045] Output the historical best feasible solution obtained during the search process, including the ladle number corresponding to each heat, the heat sequence undertaken by each ladle and the total energy consumption of the solution, and draw the ladle selection Gantt chart.
[0046] In comparison, the manual scheme and the method of the present invention use the same furnace task, production scheduling, initial ladle conditions, process constraints and energy consumption calculation methods to calculate the number of ladles used and the total energy consumption of each scheme.
[0047] The optimized solution improves ladle task coordination by adjusting the allocation relationship between heat cycles and ladles and the combination of consecutive ladles. Based on the comparison results of this embodiment, the optimized solution reduces the number of ladles used by one compared to the manual solution, resulting in a total energy consumption reduction of 1.07 kgce / t.
[0048] The above results show that, under the production scheduling and ladle conditions given in this embodiment, the method of the present invention can reduce the number of ladles used and reduce total energy consumption while meeting the ladle requirements of each furnace by optimizing ladle selection and turnover arrangement.
Claims
1. A method for optimizing the selection of low-energy-consumption steel ladles under a given steelmaking production scheduling scheme, characterized in that, A ladle selection optimization model is constructed with the goal of minimizing the total energy consumption of the ladle. The model considers steel type, ladle material, thermal state, restrictions on the use of special steel types, and time constraints. An initial solution is constructed based on rule heuristics, and an adaptive large neighborhood search algorithm that integrates simulated annealing acceptance criteria is used to solve for the optimal feasible solution.
2. The method for optimizing low-energy consumption ladle selection under a given steelmaking production scheduling scheme according to claim 1, characterized in that, The total energy consumption of the steel ladle TEC Including idle energy consumption E 1. Energy consumption for preheating and baking E 2 and Activation Energy Consumption E 3; ; Subtract the actual processing time of each heat cycle from the total time the ladle is occupied to calculate the idle energy consumption generated by the ladle in the non-processing state. E 1; ; er p This represents the baking energy consumption coefficient, expressed in kgce / min. j Indicates the first j Furnace batch; n j Indicates the total number of furnace runs; p Indicates the ladle serial number; Te j,p Indicates furnace number j Allocated to steel ladle p Upward transfer end time; P j Indicates furnace number j Available ladle collection; Tb j,p Indicates furnace number j Allocated to steel ladle p Upward transfer start time; i Indicates the first i Processing stage; n i Indicates the total number of stages; m Indicates smelting equipment; Pt j,m Indicates furnace number j In smelting equipment m The smelting time on the surface; X i,j,m Indicates if the furnace number j In the stage i Assigned to machine m If the processing is performed, the value is 1; otherwise, the value is 0. M i Indicates the first i The set of available equipment for each process; When the same ladle is used for two consecutive heat transfers, it needs to wait in the baking area for its next use. The preheating and baking energy consumption of the ladle is calculated based on the waiting time. E 2; ; T p Indicates the service life of the ladle; Wt j,j′,p This indicates the waiting time for two adjacent heats of the same ladle to be transferred; t Indicates the cumulative number of times the ladle has been used; Calculate the energy consumption for using the ladle based on whether it is activated. E 3; As long as the steel ladle p The activation variable of the ladle is assigned and must be responsible for at least one heat. K p The value is set to 1; U is a positive integer; ; ; eq p Indicates the energy consumption coefficient of a new or slag line coil; K p Indicates when the steel ladle p The value of 1 indicates the assigned furnace transfer number; otherwise, it is 0. Q j,p Indicates furnace number j Assigned to steel ladle p A value of 1 indicates a transfer, otherwise a value of 0.
3. The method for optimizing low-energy consumption ladle selection under a given steelmaking production scheduling scheme according to claim 1, characterized in that, The specific timing constraints are as follows: To ensure that the molten steel in each heat is carried and transferred by a single ladle, one and only one ladle is selected from the set of available ladles for that heat. ; ; a j,p Indicates furnace number j Steel ladles are permitted. p It is 1 if it is true, otherwise it is 0. J Indicates the set of furnace batches; The number of heat cycles each ladle can handle shall not exceed its maximum permitted number of uses; ; For any two different heats assigned to the same ladle, the transfer sequence is unique; ; Z j,j′,p Indicates if the furnace number j In the furnace j Previously assigned to steel ladles p A value of 1 indicates a successful transfer, otherwise a value of 0. Z j′,j,p Indicates if the furnace number j In the furnace j Previously assigned to steel ladles p A value of 1 indicates a successful transfer, otherwise a value of 0. ; Furnace j and furnace j' Continuous transfer from the same ladle, and heat number j Ranked by furnace j' Previously, furnace batch j' The transfer time must not start earlier than the furnace batch. j The sum of the transfer end time and the necessary preparation time for the steel ladle; Z j′,j′′,p Indicates if the furnace number j In the furnace j Previously assigned to the ladle p A value of 1 indicates a successful transfer, otherwise a value of 0. Z j′′,j,p Indicates if the furnace number j In the furnace j Previously assigned to steel ladles p A value of 1 indicates a successful transfer, otherwise a value of 0. ; S p This indicates the minimum baking time required before using the ladle; Tb j,p Indicates furnace number j Allocated to steel ladle p Upward transfer start time; Tb j′,p Indicates furnace number j 'Assigned to steel ladle p Upward transfer start time; Q j′,p Indicates furnace number j ′ was allocated to the ladle p The value is 1 if the transfer is successful; otherwise, it is 0. The start time of ladle transfer is equal to the completion time of its steelmaking stage, and the end time of ladle transfer is equal to the time when the ladle enters the baking position. If the heat is not assigned to this ladle, the constraint is relaxed using U. ; ; Tf i,j Indicates furnace number j During the processing stage i Completion time; Δ t BE This indicates the transfer time from the baking position to the first processing stage, Δ. t CB This indicates the transfer time from the last processing stage to the baking position; IC and CC represent the die casting stage and the continuous casting stage, respectively. i =1 indicates the first processing stage; Furnace j With furnace j' They were assigned to the same ladle and in the same heat. j Located in the furnace j' Previously, the waiting time between two ladle transfer tasks was equal to the difference between the start time of the later transfer and the end time of the previous transfer. ; ; Wt j,j',p This indicates the waiting time for the transfer of the same ladle to adjacent furnaces.
4. The method for optimizing low-energy consumption ladle selection under a given steelmaking production scheduling scheme according to claim 1, characterized in that, Rule-based heuristic construction of initial solutions: After meeting the requirements for steel type, ladle material, thermal state, special steel type usage restrictions, and time constraints, the ladle selection enters the optimization decision-making stage. From the set of available ladles, ladles with short dwell time, no need for additional baking, and matching the current furnace process route are selected based on rule heuristics.
5. The method for optimizing low-energy consumption ladle selection under a given steelmaking production scheduling scheme according to claim 1, characterized in that, The rule-based heuristic construction of the initial solution is as follows: The ladle selection scheme is encoded using a ladle-heat allocation vector, and the execution sequence, time status and energy consumption of each ladle are obtained by decoding. Encoding: The ladle-heat allocation production decision problem is transformed into a serialized data structure. Ladle-heat allocation encoding: s =(p1,p2,…, p j ,…,p N ); N For the total number of heats in the steelmaking production plan, the coded number is... j Bit value p j Representing the j The ladle number assigned to each heat batch; Decoding: Based on the encoding and timing constraints, calculate the start time, end time, waiting time, and baking time for each ladle, and finally solve for the total energy consumption of the ladle; during decoding, invalid codes are removed to ensure the feasibility of the solution; Initial solution is constructed using rule-based heuristics. s 0 .
6. The method for optimizing the selection of low-energy-consumption steel ladles under a given steelmaking production scheduling scheme according to claim 5, characterized in that, The initial solution for the first iteration s 0 Perform destruction and repair to generate new candidate solutions. s In the initialization phase of the adaptive large neighborhood search algorithm that incorporates the simulated annealing acceptance criterion, let... s = s 0 ; The destruction and repair operators are designed as follows: Design principle of the destruction operator: the initial solution for each iteration s, From the set of destructive operators Ω - Choose one of the destruction operators D pairs s Destruction is performed to obtain a destruction solution. s - ; Destruction operator set Ω - ={D1, D2, D3}, including the random furnace failure operator D1, the marginal energy consumption worst failure operator D2, and the same steel type failure operator D3; Repair operator design principle: From the repair operator set Ω + Choose a repair operator R to repair the broken solution. s - Repair. s ′ represents the solution obtained after repair; Repair operator set Ω + ={R1, R2, R3}, including rule-priority repair operator R1, incremental energy consumption greedy repair operator R2, and Regret-k regret value repair operator R3; Roulette wheel selection options include destruction and repair operators; ; ; P (D d ) and P(R r The numbers and represent the probabilities of the destruction operator being selected and the repair operator being selected, respectively. and Indicates the number to be calculated d The weight of the destruction operator and the weight to be calculated for the first destruction operator r N represents the weights of the repair operators, where h represents the summation index of all operators; D N represents the total number of destruction operators. R Indicates the total number of repair operators; , Let h represent the weight of the h-th disruptive operator and the weight of the h-th repair operator in the summation process; Construct a cumulative probability interval, generate two random numbers in the range [0,1], and select the destruction operator and the repair operator based on the interval in which the random numbers fall; After accumulating a complete cycle, update the weights of the destruction operator and the repair operator; Perform destruction and repair operations to generate new candidate solutions; perform feasibility verification on the new candidate solutions to check whether they violate the restrictions on steel type, ladle material, thermal state, and use of special steel types; if the new candidate solution is feasible, proceed to simulated annealing for scoring; if it is not feasible, perform a roulette wheel selection to choose the destruction and repair operators again.
7. The method for optimizing low-energy consumption ladle selection under a given steelmaking production scheduling scheme according to claim 6, characterized in that, If a new candidate solution is feasible, the simulation annealing score is used to determine whether to accept it. If accepted, the new candidate solution is assigned a score. s ′ is the preferred candidate solution s ′′; Based on the Metropolis criterion, with probability P SA Accept new candidate solutions s ′′, where Δ f This represents the change in energy consumption. T SA The current temperature gradually decreases according to the cooling function during the iteration process; when Δ f When <0, it indicates a preferred candidate solution. s '' is better than the initial solution s The preferred candidate solution will always be accepted; when Δ f When the value is ≥0, it means that the preferred candidate solution is inferior to the current solution. The preferred candidate solution is accepted with a certain probability, and this probability gradually decreases as the temperature decreases. ; Rate the operators in this round: ; Ψ=σ1, highest reward: optimal candidate solution s ''Breaking the historical best f ( s ′′)< f ( s The highest score will be awarded. Ψ=σ2, Medium reward: Optimal candidate solution s The solution did not break the historical best, but it is better than the current solution, so a moderate reward is given. Ψ=σ3, basic reward: the preferred candidate solution is worse than the current solution, but is accepted by the acceptance criterion of simulated annealing; Ψ=σ4, no reward: the preferred candidate solution is rejected by the simulated annealing criterion, or the generated preferred candidate solution itself is infeasible, and the operator does not score in this round; f ( s ′′) represents the preferred candidate solution s The objective function value corresponding to ′′; f ( s ) represents the historical best solution s The corresponding objective function value; f ( s () represents the objective function value corresponding to the initial solution; Determine if the operator weight update cycle has been reached. If not, directly update the simulated return temperature and iterate. If the operator weight update cycle has been reached, use an adaptive mechanism to update the operator weights. After one statistical period, the operator weights are updated after multiple iterations, and the updated operator weights are used in the next round of roulette selection. ; ; , They represent the first d The historical weight of the first destruction operator, the first d The updated weights of the destructive operators; , , indicating the first r The historical weights of the repair operator, the first r The updated weights of the repair operators; λ is the learning rate, which takes values in the range [0,1]. Indicates the number of times within the statistical period d The total score obtained by each destruction operator; Indicates the number of times within the statistical period r The total score obtained by each repair operator; Indicates the number of times within the statistical period d Number of times the destruction operator is called; Indicates the number of times within the statistical period r Number of times the repair operator is called; Update the simulated annealing temperature; The initial temperature, the final temperature, and the cooling coefficient are set to represent the cooling iteration process; The cooling iteration process uses an exponential cooling method to adjust the number of iterations. T ′SA =α T SA α is the cooling coefficient, which takes a value in (0,1); T ′SA This indicates the new temperature after cooling. T SA This indicates the temperature before cooling down; The solution is iterated until the termination condition is met, and the optimal ladle selection scheme and total energy consumption are obtained.
