Optimal Scheduling Method for Heating Furnaces in Steel Enterprises Based on Multi-agent Simulation

Abstract

A method for optimizing the scheduling of heating furnace groups in steel enterprises based on multi-agent simulation belongs to the technical field of intelligent optimization scheduling of production workshops. It includes: initialization, data reading, calculation of the ratio of cold-loaded slabs and hot-loaded slabs in each rolling unit, determination of the furnace loading mode of each rolling unit according to the ratio of cold and hot slabs and the position of cold and hot slabs in the rolling process, and the furnace loading sequence and heating furnace number of each rolling process as input, and starting the multi-agent simulation model of the heating furnace group to perform heating furnace group scheduling simulation; according to the simulation results, the evaluation index of each rolling process is calculated, and each rolling process that does not meet the index requirements is obtained in sequence, and iterative optimization and adjustment are performed, and finally the simulation scheduling strategy is output. The advantage is that it solves the scheduling and production scheduling problem of heating furnaces under the dedicated furnace mode and the smooth transition scheduling problem of different heating furnace scheduling modes between multiple rolling processes, which is more in line with the actual situation on site.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent optimization scheduling of production workshops, and in particular provides an optimization scheduling method for heating furnace groups of steel enterprises based on multi-agent simulation to solve the combined optimization problem under different heating furnace modes between multiple rolling processes. Background Art

[0002] In the production of plate and strip steel, the heating furnace is the intermediate link between continuous casting and rolling line, and undertakes the heating function of the billet. The hot delivery and hot charging process of the billet has obvious energy saving and consumption reduction and economic and social benefits. The hot delivery and hot charging process is generally divided into three types according to the furnace loading type: DHCR (direct hot charging), HCR (hot charging) and CCR (cold charging). Affected by market orders, mismatch of casting and rolling capacity, inventory and equipment maintenance, the casting and rolling room generally adopts the mixed loading mode of cold and hot slabs. In the mixed loading mode, the scheduling methods of the heating furnace include the dedicated furnace scheduling mode and the mixed loading scheduling mode. The dedicated furnace scheduling mode is that the cold billet and the hot billet are respectively loaded into different heating furnaces for heating. This production mode can improve the thermal efficiency of the heating furnace, reduce the heating quality problems caused by overburning of some steel billets, improve the yield rate, significantly reduce the heating time of the billet, and have a good energy saving effect; while the mixed loading scheduling mode does not distinguish between cold and hot billets, and the cold billets and hot billets are alternately put into the furnace for production according to the rolling order. This mode can basically ensure the requirement of continuous supply of billets by the hot rolling mill, but it cannot ensure the shortest residence time of the billet in the heating furnace. Because the heating methods of hot billets and cold billets are different in the special furnace production mode, and the production capacity of the special furnace is significantly greater than that of the ordinary furnace, it is difficult to keep the order of entering the heating furnace group consistent with the rolling order. When the order of entering the furnace and the order of leaving the furnace are inconsistent, the logistics control and scheduling organization will be very complicated. The purpose of optimizing the scheduling of the heating furnace is to reasonably arrange the order and rhythm of entering the heating furnace while meeting the rolling rhythm, so as to improve the efficiency of the heating furnace, reduce energy consumption, and reduce oxidation and burning losses. Summary of the invention

[0003] The purpose of the present invention is to provide a heating furnace group optimization scheduling method for steel enterprises, which adopts a mode combining heuristic rules and simulation to solve the scheduling and production scheduling problem of heating furnaces under the dedicated furnace mode and the smooth transition scheduling problem of different heating scheduling modes between multiple rolling processes. It enables the combination optimization under different heating furnace scheduling modes between multiple rolling processes; increases the scheduling flexibility between the casting and rolling interfaces, and is more in line with the actual situation on site.

[0004] The present invention comprises the following steps (such as Figure 1 shown):

[0005] Step 1: Initialize and read data, including the billet cutting plan, unexecuted rolling process unit R = {r1, r2…, r x}, inventory information, current rolling progress.

[0006] Step 2: Calculate the ratio of cold-charged slabs to hot-charged slabs in each rolling unit.

[0007] Step 3: Determine the furnace charging mode (dedicated furnace mode, non-dedicated furnace mode) of each rolling unit according to the ratio of hot and cold billets and the position of hot and cold billets in the rolling process. The furnace charging mode and initial time of each rolling unit are determined as follows:

[0008] (1) Special furnace mode: The combination mode of the special furnace is determined according to the number and location of the hot and cold slabs in the rolling unit (the combination mode of the three heating furnaces includes one cold and two hot, one mixed and two hot, and two mixed and one hot). Then, the order of the slabs entering the furnace and the number of the heating furnace to be entered are determined according to the combination mode. In order to meet the rolling rhythm, the cold slabs will enter the furnace in advance according to the rolling rhythm and the furnace dwelling time (initial furnace entry time = slab rolling time minus slab furnace dwelling time). After the hot slabs are taken out, they can be smoothly taken out of the furnace according to the rolling time on the premise of meeting the furnace dwelling time. Finally, the start furnace loading time of this rolling process is calculated according to the roll changing time and maintenance time of the rolling process. The initial furnace loading time = the start rolling time of this rolling process minus the minimum furnace dwelling time of the first slab according to the furnace entry order.

[0009] (2) Non-dedicated furnace mode: The slabs are put into the furnace in alternating order. The furnace entry sequence and the heating furnace number of the slabs in this rolling process are determined according to the alternating furnace entry sequence. Then, the initial furnace loading time of this rolling process is calculated according to the roll changing time and maintenance time of the rolling process. The initial furnace loading time = the start time of this rolling process minus the minimum furnace residence time of the first slab in the furnace entry sequence.

[0010] Step 4: Take the furnace entry sequence of each rolling process and the heating furnace number as input, and start the heating furnace group multi-agent simulation model to perform heating furnace group scheduling simulation.

[0011] Step 5: According to the simulation results, the evaluation index of each rolling process is calculated, including the rolling rhythm of each rolling process and the furnace dwelling time. If the rolling rhythm of each rolling process meets the preset rolling rhythm and the furnace dwelling time meets the process control standard, then execute step 7, otherwise adjust the furnace entry order of each rolling process of the special furnace according to step 6, and perform iterative simulation. If it meets the index requirements within the maximum number of iterations N2, the simulation results are output, otherwise the special furnace rolling processes that do not meet the index requirements are adjusted in turn, and the furnace entry mode is changed to the non-special furnace mode, and step 4 is re-executed to perform the heating furnace group scheduling simulation.

[0012] Step 6: Figure 2As shown, the rolling processes that do not meet the index requirements are obtained in sequence, the number of rolling processes is N3, the iteration variable j=1 is set, and the slab furnace order adjustment logic is started. If the rolling rhythm is not met, the cold billet furnace time t=t-Δt is adjusted. If the slab exceeds the maximum furnace dwell time and overburning occurs, the cold billet furnace time t=t+Δt is adjusted, where t is the last simulated slab furnace time, and Δt is the slab furnace adjustment time window for each iteration. According to the adjusted cold billet furnace time, the furnace order of the slabs in each rolling process is re-determined. When the furnace order of all rolling processes is adjusted, jump to step 4 and re-perform the heating furnace group scheduling simulation.

[0013] Step 7: Output the simulation scheduling strategy, including the furnace entry sequence and heating furnace number of each rolling process.

[0014] The multi-agent heating furnace group simulation model in step 4 of the present invention includes a heating furnace group management agent, a slab agent, a heating furnace agent, and a stepping agent. The heating furnace group management agent describes the optimal selection and task allocation process of slabs for multiple heating furnaces, as well as the management of slabs entering and leaving the furnace; the heating furnace agent describes the operation process of the heating furnace controlling the entry and exit of slabs from the perspective of the heating furnace; the slab agent describes the data and parameter information of each slab in the simulation; and the stepping agent describes the process of the heating furnace stepping function.

[0015] The above-mentioned intelligent agent is specifically described as follows:

[0016] (I) Heating furnace group management intelligent agent: The logical structure diagram of the heating furnace group management intelligent agent is as follows Figure 3 As shown. The static knowledge base is converted into the static attributes of the heating furnace group management agent, which include the heating furnace equipment, overall scheduling strategy, and slab matching strategy. Then the dynamic knowledge base is converted into the dynamic attributes of the heating furnace group management agent, which include the slab furnace queue, slab furnace queue, previous billet discharge equipment, previous billet input equipment, next billet discharge equipment, and next billet input equipment. The agent makes inference decisions based on the static and dynamic knowledge bases and rule knowledge bases. The functions of the main logic calculation modules are as follows:

[0017] (1) Slab matching calculation,

[0018] (2) Management of slab queues entering the furnace,

[0019] (3) Management of the steel-making rhythm of the heating furnace.

[0020] The intelligent agent manages the entire model through the above functions, sends the furnace entry and exit signals to the heating furnace intelligent agent, sends the updated slab information to the slab intelligent agent, and simultaneously outputs the calculated slab number, slab queue number, slab heating furnace number, latest furnace entry time, and recommended furnace entry time.

[0021] (II) Heating furnace agent: The heating furnace agent model is as follows Figure 4 As shown. The current process rules are converted into the heating furnace intelligent agent rule knowledge base. The rule knowledge base includes the heating furnace entry and exit conditions and the heating system of the heating furnace. Then the static and dynamic knowledge bases are converted into the static and dynamic attributes of the heating furnace intelligent agent. The static attributes include the heating furnace length, heating furnace width, heating furnace name, slab spacing, and the minimum distance from the exit. The dynamic attributes include the slab sequence in the furnace. The intelligent agent makes reasoning decisions based on the static and dynamic knowledge bases and the rule knowledge base. The main logical steps are as follows:

[0022] (1) Determination of slab entry conditions: When the next slab arrives at the steel loading side, it is determined whether the next slab is allowed to enter the furnace based on the size of the next slab and the distance between the slabs. If the conditions are met, the slab is notified to enter the furnace.

[0023] (2) Determination of slab discharge conditions: When the slab reaches the furnace outlet on the steel-discharging side, the slab residence time is met, and the discharge signal is received, the slab is allowed to discharge.

[0024] The intelligent agent completes the control of the heating furnace through the above functions, sends the stepping signal to the slab stepping intelligent agent, and simultaneously outputs the slab number, slab entry time, slab exit time, and slab residence time of the slabs entering and leaving the furnace.

[0025] (III) Stepping Agent: The stepping agent model is as follows Figure 5 As shown. The process rules are converted into a stepping agent rule knowledge base, which includes the start stepping judgment criteria, the stop stepping judgment criteria, and the time and temperature constraints of each heating section. Then the static and dynamic knowledge bases are converted into static and dynamic attributes of the heating furnace agent. The static attributes include the length of each heating section, the stepping cycle, and the stepping speed. The dynamic attributes include the slab position and the slab temperature in each heating section. When the in-and-out operation notification from the heating furnace agent is read, the stepping state transition will begin and the position information of the slab in the furnace will be updated. The main logical steps are as follows:

[0026] (1) Receive the in-furnace and out-furnace messages from the heating furnace intelligent agent.

[0027] (2) Change the stepping state according to the stepping and stop stepping conditions.

[0028] (3) Update the slab position and record the position information in real time.

[0029] The state transition process of the stepping agent is as follows: Figure 6 As shown, after starting to step, the rising, forward, descending, and retreating operations will be performed at the set time intervals, and then the signal from the heating furnace intelligent body and the judgment criteria will be combined to determine whether to step. If the stepping continues, it will return to the rising stage. If it stops, it will enter the waiting stage until the stepping signal is received.

[0030] The advantages of the present invention are that it solves the problem of scheduling and production scheduling of heating furnaces under the dedicated furnace mode and the problem of smooth transition scheduling of different heating furnace scheduling modes between multiple rolling processes. It is more in line with the actual situation on site and is of great significance to improving the efficiency of heating furnaces, reducing energy consumption and reducing oxidation and burning losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 Flowchart of the optimal scheduling method for heating furnace groups in steel enterprises based on multi-agent simulation.

[0032] Figure 2 Adjust the flow chart of the furnace entry sequence for special furnace rolling process that does not meet the index requirements.

[0033] Figure 3 This is the structure diagram of the heating furnace group intelligent agent.

[0034] Figure 4 This is the structural diagram of the heating furnace intelligent agent.

[0035] Figure 5 Slab stepping agent structure diagram.

[0036] Figure 6 Slab stepping process. DETAILED DESCRIPTION

[0037] The implementation of the model includes the following steps: 1. Simulation layout of heating furnace group 2. Simulation parameter setting of heating furnace group 3. Data reception 4. Model calculation and result output.

[0038] Step 1: Simulation layout of heating furnace group: simulate the plane layout according to the actual plane layout and the size and relative position of the heating furnace.

[0039] Step 2: Setting simulation parameters of heating furnace group: The simulation parameters of heating furnace group include the number of heating furnaces, heating furnace length, stepping cycle, stepping speed, residence time and process range of different steel grades, residence time and process range of different steel grades in different heating sections.

[0040] Step 3: Data reception: including planned data and real-time data. Planned data includes casting plan, furnace plan, cutting plan, hot rolling plan, and maintenance plan. Real-time data includes real-time information of slab warehouse and information of slab in heating furnace.

[0041] The key data items required for planning data are as follows:

[0042] Pouring schedule: Pouring number, pouring sequence number, continuous pouring mark, planned pouring start time

[0043] Heat plan: pouring number, furnace number, heat order, smelting steel type, tundish change mark, continuous pouring mark, weight, casting speed

[0044] Cutting plan: furnace number, slab number, slab width, slab length, slab thickness, slab weight

[0045] Hot rolling plan: hot rolling code, virtual slab number, rolling sequence, planned rolling start time

[0046] Maintenance plan: production line name, maintenance start time, maintenance end time

[0047] The key data items required for real-time data are as follows:

[0048] Real-time information of slab warehouse: including slab number and slab location information (area, stacking position, number of layers) in the warehouse

[0049] Information about slabs in the heating furnace: including heating furnace number, slab number, order of entering the furnace, time of entering the furnace, temperature of entering the furnace, estimated time of leaving the furnace, current coordinates in the furnace, current heating furnace section, and time of entering the section.

[0050] Step 4: Model calculation and result output: Follow the program flow chart (such as Figure 1 As shown in the figure, read the planned information and real-time information, start the model operation, and output the results after the model operation is completed. The output results include: rolling process code, virtual slab number, pre-matched slab number, heating furnace number, furnace entry sequence, and rolling sequence.

Claims

1. A method for optimizing the scheduling of heating furnaces in a steel enterprise based on multi-agent simulation, characterized in that: The steps include: (1) Initialization, read data, including the billet cutting plan, unexecuted rolling process unit R = {r1, r2…, r x }, inventory information, current rolling progress; (2) Calculating the ratio of cold-charged slabs to hot-charged slabs in each rolling unit; (3) According to the ratio of hot and cold billets and the position of hot and cold billets in the rolling process, the furnace loading mode of each rolling unit is determined: dedicated furnace mode, non-dedicated furnace mode. The furnace loading mode and initial time of each rolling unit are determined as follows: 1) Select the special furnace mode for the rolling process: Determine the combination mode of the special furnace according to the number and location of the cold and hot slabs in the rolling unit: that is, the combination mode of three heating furnaces, including one cold and two hot, one mixed and two hot, and two mixed and one hot; then determine the order of the slabs entering the furnace and the number of the heating furnace to be entered according to the combination mode. In order to meet the rolling rhythm, the cold slabs will enter the furnace in advance according to the rolling rhythm and the furnace dwelling time. The initial furnace entry time = slab rolling time minus slab furnace dwelling time; ensure that the hot slabs are smoothly discharged from the furnace according to the rolling time after the furnace dwelling time is met. Finally, calculate the start furnace loading time of this rolling process according to the roll changing time and maintenance time of the rolling process. The initial furnace loading time = the start rolling time of this rolling process minus the minimum furnace dwelling time of the first slab according to the furnace entry order; 2) Non-dedicated furnace mode: The slab adopts the alternate furnace entry mode. The furnace entry sequence and the heating furnace number of the slab in this rolling process are determined according to the alternate furnace entry sequence. Then, the initial furnace loading time of this rolling process is calculated according to the roll changing time and maintenance time of the rolling process. The initial furnace loading time = the start time of this rolling process minus the minimum furnace residence time of the first slab in the furnace entry sequence; (4) The order of each rolling process entering the furnace and the heating furnace number are used as input to start the heating furnace group multi-agent simulation model to perform heating furnace group scheduling simulation; (5) According to the simulation results, the evaluation index of each rolling process is calculated, including the rolling rhythm of each rolling process and the furnace dwelling time; when the rolling rhythm of each rolling process meets the preset rolling rhythm and the furnace dwelling time meets the process control standard, step (7) is executed; otherwise, the furnace entry sequence of each rolling process of the special furnace is adjusted according to step (6), and iterative simulation is performed. When the index requirements are met within the maximum number of iterations N2, the simulation results are output; otherwise, the special furnace rolling processes that do not meet the index requirements are adjusted in turn, and the furnace entry mode is changed to the non-special furnace mode, and step (4) is re-executed to perform the heating furnace group scheduling simulation; (6) Obtain each rolling process that does not meet the index requirements in sequence, the number of rolling processes is N3, set the iteration variable j=1, start the slab furnace order adjustment logic, when the rolling rhythm is not met, adjust the cold billet furnace time t=t-Δt, when the slab exceeds the maximum furnace dwell time and overburning occurs, adjust the cold billet furnace time t=t+Δt, where t is the last simulated slab furnace time, Δt is the slab furnace adjustment time window for each iteration; according to the adjusted cold billet furnace time, re-determine the slab furnace order of each rolling process, when the furnace order of all rolling processes is adjusted, jump to step (4) and re-perform the heating furnace group scheduling simulation; (7) Output simulation scheduling strategy, including the furnace entry sequence and heating furnace number of each rolling process.

2. The method for optimizing the scheduling of heating furnaces in a steel enterprise based on multi-agent simulation according to claim 1 is characterized in that: The multi-agent heating furnace group simulation model in step (4) includes a heating furnace group management agent, a slab agent, a heating furnace agent, and a stepping agent; the heating furnace group management agent describes the optimal selection of slabs for multiple heating furnaces and the task allocation process, as well as the management of slabs entering and exiting the furnace; the heating furnace agent describes, from the perspective of the heating furnace, the process of the heating furnace accepting the tasks of entering and exiting the furnace and completing the corresponding tasks; the slab agent describes the data and parameter information of each slab in the simulation; the stepping agent describes the process of the heating furnace stepping function.

3. The method for optimizing the scheduling of heating furnaces in a steel enterprise based on multi-agent simulation according to claim 2 is characterized in that: Logical structure of heating furnace group management agent: The static knowledge base is transformed into the static attributes of the heating furnace group management agent, which include the heating furnace equipment, overall scheduling strategy, and slab matching strategy. Then the dynamic knowledge base is transformed into the dynamic attributes of the heating furnace group management agent, which include the slab entering the furnace queue, the slab in the furnace queue, the previous billet discharge equipment, the previous billet feeding equipment, the next billet discharge equipment, and the next billet feeding equipment. The agent makes inference decisions based on the static and dynamic knowledge bases and the rule knowledge base. The functions of the logic calculation module are as follows: 1) Slab matching calculation, 2) Management of slab queues entering the furnace, 3) Heating furnace steel tapping management, The intelligent agent manages the entire model through the above functions, sends the furnace entry and exit signals to the heating furnace intelligent agent, sends the updated slab information to the slab intelligent agent, and simultaneously outputs the calculated slab number, slab queue number, slab heating furnace number, latest furnace entry time, and recommended furnace entry time.

4. The method for optimizing the scheduling of heating furnaces in a steel enterprise based on multi-agent simulation according to claim 2 is characterized in that: Heating furnace agent model: The current process rules are converted into a heating furnace agent rule knowledge base. The rule knowledge base includes the heating furnace entry and exit conditions and the heating system of the heating furnace. Then the static and dynamic knowledge bases are converted into static and dynamic attributes of the heating furnace agent. The static attributes include the heating furnace length, heating furnace width, heating furnace name, slab spacing, and the minimum distance from the exit. The dynamic attributes include the slab sequence in the furnace. The agent makes reasoning decisions based on the static and dynamic knowledge bases and the rule knowledge base. The logical steps are as follows: 1) Determination of slab furnace entry conditions: When the next slab arrives at the steel loading side, it is determined whether the next slab is allowed to enter the furnace based on the size of the next slab and the distance between the slabs. If the conditions are met, the slab is notified to enter the furnace; 2) Determination of slab discharge conditions: When the slab reaches the furnace outlet on the steel-discharging side, the slab residence time is met, and the discharge signal is received, the slab is allowed to be discharged; The intelligent agent completes the control of the heating furnace through the above functions, sends the stepping signal to the slab stepping intelligent agent, and simultaneously outputs the slab number, slab entry time, slab exit time, and slab residence time of the slabs entering and leaving the furnace.

5. The method for optimizing the scheduling of heating furnaces in a steel enterprise based on multi-agent simulation according to claim 2 is characterized in that: Stepping agent model: The process rules are converted into a stepping agent rule knowledge base, which includes the start stepping judgment criteria, the stop stepping judgment criteria, and the time and temperature constraints of each heating section. Then, the static and dynamic knowledge bases are converted into static and dynamic attributes of the heating furnace agent. The static attributes include the length of each heating section, the stepping cycle, and the stepping speed. The dynamic attributes include the slab position and the slab temperature in each heating section. When the in-and-out operation notification from the heating furnace agent is read, the stepping state transition will start and the position information of the slab in the furnace will be updated. The logical steps are as follows: 1) Receive the in-furnace and out-furnace messages from the heating furnace intelligent agent; 2) Change the stepping state according to the stepping and stopping stepping conditions; 3) Update the slab position and record the position information in real time.

Images