Pickling unit automatic speed control method and system based on multi-factor constraint

Abstract

The application provides a pickling unit automatic speed control method and system based on multi-factor constraints, and relates to the technical field of cold-rolled strip steel production. In view of the defects that the existing control relies on manual operation or primary automation and is difficult to cope with multiple constraints and instantaneous working conditions, the method comprises the following steps: constructing a multi-factor constraint model covering process, equipment, logistics, acid liquid and instantaneous working conditions; collecting corresponding data in real time; dynamically calculating a global optimal speed setting value through multi-layer decision logic, calculating a basic speed first and then superimposing an instantaneous working condition constraint, and directly triggering emergency disposal through high-priority alarm; issuing an instruction for execution and closed-loop feedback, and combining machine learning for self-learning optimization. The application realizes automatic intelligent control of the unit speed, guarantees quality and equipment safety, improves production efficiency and continuity, and reduces energy consumption cost.

Description

Technical Field

[0001] This invention relates to the field of automation control technology in the metallurgical industry, specifically to an automatic speed control method and system for pickling units based on multi-factor constraints. Background Technology

[0002] In the cold-rolled strip steel production process, the pickling unit, as the core pretreatment process for removing iron oxide scale from the strip surface, directly determines the product quality and production efficiency of subsequent processes such as rolling, annealing, and finishing. It is a key link in ensuring the performance of cold-rolled strip steel products (such as surface finish and dimensional accuracy) and production economy. Among them, the unit's operating speed is a core process parameter throughout the entire pickling process, and its control effect is deeply coupled with multiple key indicators: on the one hand, excessive speed can easily lead to incomplete removal of iron oxide scale (under-pickling), forming surface defects, affecting the shape control of subsequent rolling processes, and even causing production accidents such as roll sticking; on the other hand, excessive speed will cause excessive consumption of acid (over-pickling), increasing acid treatment costs and environmental pressure, while reducing output per unit time, restricting the release of production line capacity, and also aggravating ineffective wear of equipment, increasing energy consumption and maintenance costs. Therefore, achieving precise and dynamic control of the pickling unit speed is a core technical challenge in balancing production efficiency, pickling quality, equipment wear and tear, and energy consumption.

[0003] Currently, speed control in most pickling units relies primarily on manual settings and experience-based adjustments by operators. Operators set a relatively stable operating speed based on parameters such as the steel grade and specifications (thickness and width) of the strip, the surface condition of the incoming material, and the concentration and temperature of the acid solution. However, this control method, dependent on manual experience, is inherently flawed due to human factors and the complexity of the pickling process. Specifically, it exhibits the following shortcomings: 1. Conservative control strategies and underutilization of capacity potential: Insufficient pickling directly leads to surface defects in the strip steel, increasing rework costs in subsequent processes and even causing product scrap. To mitigate this risk, operators typically adopt a conservative speed setting strategy, operating at 10%-30% below the theoretical optimal speed. While this conservative control can ensure pickling quality to some extent, it significantly reduces unit capacity, resulting in insufficient equipment utilization. Especially in large-scale production scenarios, the reduced output per unit time directly weakens the company's market competitiveness.

[0004] 2. Significant response lag, leading to large fluctuations in quality and cost: During the pickling process, production conditions are constantly changing, such as continuous switching of strip steel specifications (e.g., from thick to thin plates), batch differences in the state of incoming iron oxide scale, gradual decrease in acid concentration due to continuous reaction, and deviation of acid temperature from the set value due to fluctuations in the heat exchange system. Manual observation, judgment, and adjustment involve significant time delays, typically requiring 30 seconds to several minutes to complete parameter corrections. During this period, under-pickling or over-pickling problems are highly likely. Under-pickled strip steel needs to be returned for re-pickling, increasing additional acid consumption and labor costs; over-pickling leads to excessive corrosion of the strip steel surface, reducing its mechanical properties, and also causes ineffective acid waste, increasing environmental protection pressure.

[0005] 3. Poor consistency in control effectiveness and insufficient quality stability: The effectiveness of manual control relies heavily on the experience and sense of responsibility of operators, and there are significant differences in control strategies between different shifts and different operators. This individual difference directly leads to batch-to-batch fluctuations in the pickling quality of strip steel of the same specification and steel grade, making it difficult to meet the stringent requirements for surface quality consistency in high-end cold-rolled strip steel (such as automotive steel and electronic component steel).

[0006] 4. Weak ability to handle multiple coupled factors, making global optimization difficult: Pickling is a typical complex industrial process characterized by multiple variables, strong coupling, and large time lags. Factors affecting speed control include not only continuously changing process parameters (such as acid concentration, temperature, and strip speed) but also discrete events (such as strip joint passage, acid tank switching, and equipment failure warnings). These factors are interrelated and mutually restrictive. For example, when the strip thickness increases, the speed needs to be reduced to ensure pickling time, but at the same time, the acid temperature needs to be adjusted to balance pickling efficiency. It is difficult to quickly solve for the globally optimal speed parameters under multiple constraints manually, and often only local adjustments can be made for a single factor, resulting in poor overall control performance.

[0007] 5. Insufficient response to transient conditions hinders the stability of continuous production: Numerous transient disturbances occur during the production of cold-rolled strip steel, such as impacts at strip joints, localized accumulation of iron oxide scale in incoming materials, and instantaneous pressure fluctuations in the acid circulation pump. These transient conditions are characterized by their suddenness and uncertainty, requiring the control system to quickly anticipate and respond. However, manual control relies on post-event observation and intervention, making advance prediction impossible. Adjustments are often made only after the transient disturbance has caused quality problems or equipment malfunctions, easily leading to production interruptions and affecting the continuous and efficient operation of the production line.

[0008] To address the shortcomings of manual control, the industry has gradually developed some automated control solutions. However, existing automation technologies are still in their early stages and struggle to solve global dynamic optimization problems under multiple constraints. Specifically, existing automation solutions mainly fall into two categories: one is closed-loop control for a single process parameter, such as a PID control scheme based on acid temperature. This scheme adjusts the heating device power by detecting acid temperature deviations, indirectly affecting pickling efficiency. However, this type of scheme only focuses on a single variable and does not consider other key factors such as strip specifications and incoming material conditions, making it impossible to achieve global optimization of speed parameters. The other type is simple speed chain coordination control. This scheme achieves synchronous operation of each unit by setting the speed ratio relationship for each process (such as uncoiling, pickling, straightening, and winding). However, this type of scheme uses fixed speed ratio coefficients, lacks adaptability to dynamically changing production conditions (such as strip joints and acid concentration decay), and cannot handle the coupling problem between discrete events and continuous process parameters, making it difficult to meet the dynamic speed adjustment needs in complex production scenarios.

[0009] Therefore, there is an urgent need in the market for an automatic speed control method and system for pickling units based on multi-factor constraints that can automatically, in real time, and intelligently coordinate multiple constraints to achieve global optimization control of unit speed. Summary of the Invention

[0010] To address the shortcomings of existing technologies, the purpose of this invention is to provide an automatic speed control method and system for pickling units based on multi-factor constraints.

[0011] An automatic speed control method for pickling units based on multi-factor constraints, provided by the present invention, includes: Step S1: Collect data in real time from sensors, PLC and process control system to construct the multi-factor constraint model; the data includes strip tracking signal, looper position signal, acid concentration and temperature, equipment operating status, weld detection signal and deviation detection signal; Step S2: Identify and quantify the key constraints affecting the speed of the pickling unit, and construct a multi-factor constraint model for the pickling unit; the key constraints include process constraints, equipment capacity constraints, logistics synchronization constraints, acid and energy constraints, and instantaneous equipment operating condition constraints; Step S3: Based on the multi-factor constraint model and real-time collected data, the globally optimal speed setpoint V_set is dynamically calculated using multi-layer decision logic; Step S4: Send the speed setpoint V_set to the main drive control system of the unit to control the motors of each component to accelerate and decelerate smoothly in coordination; Step S5: Monitor pickling quality using a surface inspection instrument, and combine equipment operating status and production efficiency data. Use reinforcement learning algorithm to self-learn and optimize the pickling kinetic model parameters and constraint weights, and iteratively update the speed control strategy.

[0012] Preferably, the process constraints in step S2 are quantified and calculated using an acid pickling kinetics model, including: Step S2.1: Calculate the theoretical minimum process time T_process_min based on the strip steel grade, specifications, oxide layer thickness, and acid parameters, using the following formula: T_process_min=(δ_scale^η) / (A·[HCl]^α·exp(-E_a / (R*T))·k_m·C_steel) Where T_process_min represents the theoretical minimum process time, δ_scale represents the total thickness of the oxide layer, A,α,η represent empirical constants, [HCl] represents the effective concentration of acid solution, E_a represents the apparent activation energy of the pickling reaction, R represents the ideal gas constant, T represents the acid solution temperature, k_m represents the mass transfer coefficient, and C_steel represents the steel grade coefficient. Step S2.2: Derive the maximum process speed V_process_max based on the effective length L_process of the pickling process section, as shown in the following formula: V_process_max=L_process / T_process_min.

[0013] Preferably, the multi-layer decision logic in step S3 specifically includes: The first layer calculates the base velocity using the following formula: V_base=Min(V_process_max,V_equipment_max,V_sync)·K_chem·K_stray In the formula, V_process_max is the maximum process speed corresponding to the process constraint, V_equipment_max is the maximum allowable working speed corresponding to the equipment capacity constraint, V_sync is the synchronization speed corresponding to the logistics synchronization constraint, K_chem is the speed reduction compensation coefficient corresponding to the acid and energy constraints, and K_stray is the speed reduction coefficient corresponding to the strip deviation in the instantaneous operating condition constraints of the equipment. The second layer performs instantaneous condition coverage to obtain the final speed, as shown in the following formula: V_set=Min(V_base,V_weld,V_welder,V_tlv,V_shear,V_loading,V_unloading) In the formula, V_weld is the weld seam tracking constraint speed, V_welder is the welding machine working state constraint speed, V_tlv is the tension leveler working state constraint speed, V_shear is the disc shear working state constraint speed, V_loading is the inlet coil loading preparation constraint speed, and V_unloading is the outlet unloading preparation constraint speed. Among them, edge wire escape and weld failure are the highest priority alarms. When triggered, they directly override all constraints and execute emergency deceleration or shutdown.

[0014] Preferably, the instantaneous operating condition constraints of the equipment in step S2 include inlet coiling preparation constraints, welding machine working status constraints, strip steel deviation constraints, tension leveling machine operating condition constraints, disc shear operating condition constraints, and outlet uncoiling preparation constraints. The logistics synchronization constraints include looper storage constraints and weld seam tracking constraints, which can realize differentiated speed coordination control between the inlet section and the process section.

[0015] Preferably, in step S5, the Actor-Critic reinforcement learning framework is used to achieve self-learning optimization; Using strip steel specifications, acid parameters, pickling quality, and equipment operating conditions as states, adjusting pickling kinetic model parameters and constraint weights as actions, and constructing a reward function based on comprehensive indicators of quality, efficiency, cost, and safety, production efficiency and operating costs are optimized while ensuring pickling quality.

[0016] According to the present invention, an automatic speed control system for pickling units based on multi-factor constraints is provided for implementing the method, comprising a data acquisition module, a multi-factor constraint modeling module, an adaptive speed decision module, a main drive control module, and a self-learning optimization module; The data acquisition module includes various sensors, instruments, and PLCs, used to acquire process and equipment status data in real time; The multi-factor constraint modeling module is used to construct five types of velocity constraint models; The adaptive speed decision module calculates the globally optimal speed setting value according to a priority arbitration mechanism; The main drive control module is used to drive the motors of each component to coordinate speed regulation. The self-learning optimization module is used to iteratively optimize control parameters based on reinforcement learning.

[0017] Preferably, the adaptive speed decision module has a built-in hierarchical priority arbitration logic: Level 0 is the priority of unconditional emergency shutdown, Level 1 is the priority of core safety and process interlocking, Level 2 is the priority of key instantaneous operating condition constraints, and Level 3 is the priority of conventional process, equipment and logistics constraints. Higher priority constraints directly override the calculation results of lower priority constraints.

[0018] Preferably, the system supports independent speed regulation of the inlet section and the process section; Under the conditions of inlet roll-up and welding machine stitching, the inlet section operates at the welding / threading safety speed, and the process section operates continuously at the looper synchronous speed, so as to achieve material switching without production interruption.

[0019] Preferably, it also includes a system-level safety interlock module; The safety interlock module is deeply interlocked with the acid circulation, correction, brush roller, disc shear, and media supply subsystems, and automatically limits the unit speed to a safe range when process conditions are not met.

[0020] Preferably, the self-learning optimization module adopts an offline pre-training and online real-time learning mode; Historical production data is used to complete model pre-training, and online surface quality inspection data is combined to continuously iterate and automatically optimize the parameters and constraint weights of the pickling kinetics model.

[0021] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention enables automatic setting and dynamic adjustment of unit speed, reducing reliance on manual experience. By standardizing control logic, it eliminates operational differences between different shifts and personnel, ensuring uniform pickling conditions for strip steel of the same specification and grade throughout the entire production cycle. This effectively reduces quality fluctuations and meets the stringent requirements for surface quality consistency of high-end cold-rolled strip steel.

[0022] 2. This invention, through multi-level decision-making logic, can verify the matching between the unit speed and the equipment's load-bearing capacity and process safety limits in real time, ensuring that the speed is always within a safe range; at the same time, it has intelligent response capabilities for sudden working conditions. In response to disturbances such as strip deviation, edge wire escape, and equipment abnormalities, it can quickly trigger automatic speed adjustment or emergency stop mechanisms, avoiding equipment overload damage, under-pickling / over-pickling and other quality accidents from the source, significantly improving the safety of production line operation.

[0023] 3. This invention can capture the dynamic changes of production conditions such as strip steel specifications, acid parameters, and incoming material status in real time. Through multi-constraint coupling analysis, it can accurately calculate the upper limit of safe speed under the current working conditions, speed up the unit to the optimal value, and fully tap the equipment's production capacity potential. At the same time, it can smoothly handle instantaneous working conditions such as strip steel joint passage, upper and lower coil switching, and welding. Through collaborative control, it can reduce the number and duration of production interruptions, and significantly improve the unit's production capacity and continuous operation efficiency.

[0024] 4. By precisely matching the speed parameters with process requirements, this invention can effectively avoid waste caused by under-acid pickling and rework due to excessive speed, as well as excessive consumption of acid and ineffective loss of steam energy caused by excessive speed. At the same time, it reduces ineffective wear and tear on equipment, lowers maintenance costs and environmental protection pressure, and significantly improves the economic efficiency and greenness of the production process.

[0025] 5. This invention has a built-in self-learning mechanism that can continuously iterate and optimize the control model and parameters as production data accumulates, gradually improving the accuracy and adaptability of speed control, realizing a virtuous cycle of "production-optimization-improvement", ensuring the advanced nature of control effect in the long term, adapting to the production needs of different steel grades and different specifications of strip steel, and enhancing the flexible production capacity of the production line. Attached Figure Description

[0026] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the overall flow of the automatic speed control method of the present invention; Figure 2 This is a schematic diagram of the decision logic for calculating the optimal speed setpoint in the multi-factor constraint model of this invention. Figure 3 This is a schematic diagram of the structural composition of the system of the present invention; Figure 4 This is a schematic diagram of the architecture of the multi-factor constraint model in this invention. Detailed Implementation

[0027] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0028] Example 1 The present invention provides an automatic speed control method for pickling units based on multi-factor constraints, such as... Figure 1 As shown, it includes: Step S1: Through sensors, PLC and process control system, collect data in real time corresponding to the construction of multi-factor constraint model, including strip tracking signal, looper position signal, acid concentration and temperature detection value, equipment operation status signal (such as welding machine, tension leveler, disc shear status), weld detection signal and deviation detection signal. These data are the iterative optimization parameters of multi-factor constraint model.

[0029] Step S2: Identify and quantify the key constraints affecting the pickling unit speed, and construct a multi-factor constraint model. For example... Figure 4 As shown, the multi-factor constraint model includes an input layer, a constraint layer, a decision layer, and a feedback layer. The input layer receives the data collected in step S1 and directly maps it to the constraint layer, providing raw data support for the calculation of various constraint conditions. The constraint layer is the core logic layer of the model. Based on the input layer parameters, it is divided into multiple independent constraint calculation units according to the constraint type, forming a set of constraint conditions for the entire scenario. The key constraint factors include process constraints, equipment capacity constraints, logistics synchronization constraints, acid and energy constraints, and instantaneous equipment operating condition constraints. The quantification process of the process constraints is as follows: based on the steel grade, specification parameters, and target pickling quality of the strip, the theoretical minimum process time (T_process_min) is calculated based on the pickling kinetic model, and the maximum allowable process speed (V_process_max) of the strip is derived from this. The calculation formulas are as follows: T_process_min=(δ_scale^η) / (A·[HCl]^α·exp(-E_a / (R*T))·k_m·C_steel) V_process_max=L_process / T_process_min Among them, T_process_min represents the theoretical minimum process time (unit: seconds, s), which is the core output of the model. δ_scale represents the total oxide layer thickness (unit: micrometers, μm), which is the most critical input parameter, obtained through an inlet thickness gauge or process data from the previous process (hot rolling). Typically, the oxide layer consists of FeO, Fe3O4, and Fe2O3 from the inside out; the model may consider its composite thickness or weighted average thickness. [HCl] represents the effective concentration of the acid solution (unit: weight percentage, wt% or mol / L), referring to the actual concentration of acid solution in the pickling tank in contact with the strip. T represents the acid solution temperature (unit: Kelvin, K), usually referring to the process setting temperature of the pickling tank. E_a represents the apparent activation energy of the pickling reaction (unit: kJ / mol), a key material-related parameter; this value varies for different steel grades. R represents the ideal gas constant (8.314 J / (mol·K)). A, α, and η represent empirical constants, obtained through regression analysis of a large amount of historical production data, characterizing the kinetic behavior under specific equipment and process conditions in the factory. k_m represents the mass transfer coefficient, which is related to the acid flow rate, turbulence intensity, and strip surface condition, and is used to correct for the influence of diffusion on the reaction rate. C_steel represents the steel grade coefficient, which is the core optimization mechanism for adapting the model to different steel grades. L_process represents the effective length (m, i.e., the total length of several pickling tanks, rinsing tanks, etc.) of the pickling process section.

[0030] The specifications include strip thickness and width. The quantification process for the equipment capacity constraints includes obtaining the maximum permissible operating speed (V_equipment_max) and maximum acceleration / deceleration of the key equipment in the unit. The key equipment includes the uncoiler, straightener, acid pump, squeeze roll, main drive system, looper winch, throwing roll, and coiler. The material flow synchronization constraints ensure continuous and stable material flow at the beginning and end of the production line, including looper capacity constraints and weld seam tracking constraints. The quantification process for the looper capacity constraints is as follows: based on the current capacity L_loop of the inlet and outlet loopers, calculate the synchronization speed V_sync required to maintain stable production. The quantification process for the weld seam tracking constraints is as follows: when a weld seam is detected about to pass through a key piece of equipment, a temporary deceleration constraint V_weld is applied based on the equipment characteristics. The key equipment includes the squeeze roll and acid tank. The quantification process of acid and energy constraints is as follows: real-time monitoring of acid concentration and temperature; when the acid concentration or temperature is lower than the optimal process range, a corresponding speed reduction compensation coefficient K_chem is generated; and the maximum speed of the unit is limited by K_chem to ensure the pickling effect.

[0031] The instantaneous operating condition constraints of the equipment are used to identify the instantaneous state of key workstations in the unit and to quantify the process of quantifying their constraints on speed, including: 1. Inlet Coil Preparation Constraint: When the inlet saddle is preparing a new steel coil and threading the tape, an inlet section speed limit V_loading is generated and affects the global speed through the logistics model; 2. Welding machine operating status constraints: When the welding machine is working, the speed at the inlet section is reduced to the welding speed V_welder; when a weld failure alarm is triggered, an emergency speed reduction or shutdown is activated. 3. Strip misalignment constraint: The misalignment amount is monitored in real time by the correction equipment. When the misalignment amount exceeds the safety threshold, a speed reduction coefficient K_stray or a safety speed V_stray is generated. 4. Straightening machine operating condition constraints: When the straightening machine is not straightening, the oxide layer is not removed, so reduce the pickling speed to V_tlv; 5. Circular shear operating condition constraints: When the circular shear switches widths, the process section is locked to a safe low speed V_shear; when there is an alarm for edge wire escape or blockage, the highest priority speed reduction V_shear_jam or shutdown is triggered. 6. Export unloading preparation constraint: When the export saddle is preparing for unloading and threading, the export section speed limit V_unloading is generated and affects the global speed through the logistics model.

[0032] The decision layer is the core of the model's execution decision-making. It receives all speed constraint thresholds output by the constraint layer; executes the global minimum value filtering rule, that is, takes the minimum value (V_globalmin) among all constraint speeds, which is the feasible speed under all constraints; and finally outputs the actual operating speed of the unit as the core instruction for production execution.

[0033] The feedback layer receives production data feedback (such as actual production efficiency, product quality, equipment load, etc.) corresponding to the actual operating speed of the unit output by the decision layer; it feeds the production data back to the iterative optimization parameters of the input layer to realize the dynamic updating of parameters, so that subsequent constraint calculations and decisions can adapt to changes in actual production conditions.

[0034] Step S3: Based on the constructed multi-factor constraint model and the real-time data collected in Step S1, the globally optimal speed setpoint V_set of the multi-level decision logic dynamic computer group at each moment is determined. For example... Figure 2 As shown, the multi-layer decision logic includes The formula for calculating the base velocity of the first layer is as follows: V_base=Min(V_process_max,V_equipment_max,V_sync)·K_chem·K_stray In the formula, V_process_max is the maximum process speed corresponding to the process constraint, V_equipment_max is the maximum allowable operating speed corresponding to the equipment capacity constraint, V_sync is the synchronization speed corresponding to the logistics synchronization constraint, K_chem is the speed reduction compensation coefficient corresponding to the acid and energy constraints, and K_stray is the speed reduction coefficient corresponding to the strip deviation in the instantaneous operating condition constraints of the equipment.

[0035] The second layer of instantaneous load case coverage involves first calculating the base optimal speed V_base without considering high-priority instantaneous load cases, then comparing it with the constraint speeds of all instantaneous load cases, and taking the minimum value as the final speed setpoint, as shown in the following formula: V_set=Min(V_base,V_weld,V_welder,V_tlv,V_shear,V_loading,V_unloading) In the formula, V_weld is the weld seam tracking constraint speed, V_welder is the welding machine working state constraint speed, V_tlv is the tension leveler working state constraint speed, V_shear is the disc shear working state constraint speed, V_loading is the inlet winding preparation constraint speed, and V_unloading is the outlet unloading preparation constraint speed. Furthermore, when the highest priority alarm is triggered, it directly overrides other constraint parameters, triggering emergency deceleration or shutdown. In addition to the V_set calculation based on the instantaneous operating conditions of the equipment and the direct overriding of high-priority alarms, the production line also forms a system-level, proactive safety protection network through deep interlocking with multiple subsystems such as acid circulation, web guiding, brush rollers, drying, and media supply. This mechanism ensures that when any critical process conditions are not met, the speed can be automatically adjusted to a safe range, fundamentally preventing quality accidents, equipment damage, and safety risks.

[0036] In the multi-layered decision-making logic, the priority of each constraint parameter, from high to low, is as follows: highest priority alarm constraint (wire escape, weld failure), instantaneous equipment operating condition constraint, process constraint / equipment capacity constraint / logistics synchronization constraint, and acid and energy constraint; high-priority constraints can directly override the calculation results of low-priority constraints. The calculation of the speed setpoint (V_set) is based on a hierarchical priority system. Level 0 is the highest priority, triggering an unconditional emergency shutdown; Level 1 is for core safety and process interlocks, and its constraint speed takes precedence over all regular operating condition constraints; Level 2 is for critical instantaneous operating condition constraints, and its minimum value is taken as the common constraint for this level when concurrent. The system uses an arbitration logic module to integrate all effective constraints according to the principle of 'higher-level constraints covering lower-level constraints, and the strictest constraint within the same level,' and applies necessary time window filtering and rate of change limits to finally output a stable and safe speed setpoint command. All decision-making processes are recorded in real time for traceability.

[0037] Step S4: The globally optimal speed setpoint V_set is sent to the main drive control system of the unit to control the motors of each section (uncoiler, process section, coiler, etc.) to accelerate and decelerate in a coordinated manner, smoothly transitioning to the target speed. In step S4, the main drive control system includes a PLC or DCS system; during the coordinated acceleration and deceleration process, the acceleration / deceleration of each section motor does not exceed the maximum acceleration / deceleration allowed by the equipment, ensuring that the strip runs smoothly without surges or deviations.

[0038] Step S5: Continuously monitor the actual pickling quality (e.g., via surface inspection), equipment operating status, and production efficiency, and compare them with the expected targets; based on historical data, use machine learning algorithms to fine-tune the parameters or constraint weights of the pickling kinetic model, perform self-learning optimization, and achieve continuous improvement of the control strategy. The machine learning algorithm in S5 is a reinforcement learning algorithm; the self-learning optimization process is as follows: obtain actual pickling quality data through a surface inspection instrument, combine it with equipment operating status data and production efficiency data, construct a reward function, and fine-tune the reaction rate constant and constraint weights in the pickling kinetic model to adapt the speed control strategy to changes in different production conditions. Specifically, the process is modeled as a sequential decision problem, and an Actor-Critic reinforcement learning framework is used for continuous optimization through trial and error and feedback. The aim is to automatically find the optimal speed control strategy that can improve production efficiency and reduce costs while ensuring that the pickling quality is qualified. Its status (input) includes real-time process parameters (strip steel specifications, acid concentration and temperature), quality feedback (surface inspection instrument data), equipment operating conditions, and production efficiency indicators. The actions (outputs) involve fine-tuning key parameters of the pickling kinetics model (such as the reaction rate constant) and the weights of the speed constraints, rather than directly setting the speed. The reward (evaluation criterion) is a comprehensive reward function that balances quality, efficiency, and cost. The quality reward is positively correlated with the surface inspection pass rate. The efficiency reward is positively correlated with the average production speed (but with an upper limit to prevent blind speed increases). The cost penalty is negatively correlated with acid and energy consumption. Accident penalties result in significant point deductions for triggering alarms or generating scrap. The learning and training process includes offline pre-training using historical production data and continuous learning using online real-time running data. Based on the current state, the system's policy network (Actor) provides parameter adjustment suggestions; the adjusted parameters are applied to run a production cycle (e.g., processing a coil of steel), and the reward is calculated based on actual quality and consumption after completion; the value network (Critic) evaluates the long-term value of this decision and uses this to guide the policy network's updates, making it more inclined to take actions that yield higher cumulative rewards. This process is repeated, and the strategy continuously evolves. Ultimately, maximizing the long-term cumulative reward discount is essentially seeking the optimal balance between efficiency and cost under quality constraints. Convergence is considered achieved when strategy performance (such as average reward) stabilizes and no longer significantly improves over multiple consecutive periods. Safety measures include setting strict physical limits on parameter adjustments. Manual oversight is maintained, allowing for intervention or rollback to a stable version at any time.

[0039] This invention aims to achieve automated and intelligent control of unit speed by establishing a multi-factor constraint model covering process, equipment, logistics, and instantaneous operating conditions, and dynamically calculating the optimal speed setpoint based on real-time data. This maximizes production efficiency and continuity while ensuring product quality and equipment safety.

[0040] Furthermore, the automatic speed control method for pickling units based on multi-factor constraints of the present invention is described in detail below with reference to the accompanying drawings and application scenarios under different operating conditions: I. Optimized Operating Conditions for Normal Production Speed When a new steel coil (Q95, 3.0mm thick, 1250mm wide, 1200m long, with a total pickling tank length of 240 meters) enters the production line, the production requirement is to ensure that the iron oxide scale is completely removed and that there are no defects such as under-pickling or over-pickling.

[0041] The control system invokes the pickling kinetics model, inputting the pickling reaction rate constant for Q95 steel, the strip thickness of 3.0 mm, and the target pickling quality requirements. The theoretical minimum process time T_process_min required for complete removal of iron oxide scale from this strip is calculated to be 72 seconds. Based on the correlation formula between process speed, process time, and pickling tank length: V_process_max=L_acid_tank / T_process_min Where L_acid_tank is the total length of the acid tank, and by substituting the data, V_process_max = 200m / min is obtained.

[0042] By collecting real-time operating status data of key equipment on the production line, the maximum allowable operating speed of each piece of equipment is determined. The maximum allowable speed of core equipment such as the uncoiler, coiler, and main drive system is not less than 250m / min, so V_equipment_max = 250m / min is taken. At the same time, the maximum acceleration / deceleration of each piece of equipment meets 2m / s², which can ensure smooth speed adjustment.

[0043] The position signals of the inlet and outlet loops are collected in real time. It is calculated that the current inlet and outlet loops are both in a high position (loop storage L_loop≥80% of rated storage). There is no need to replenish the loops by adjusting the process section speed. In order to ensure production efficiency, the synchronous speed V_sync=240m / min required to maintain stable production is calculated.

[0044] The acid concentration sensor detects the acid concentration in the acid tank in real time as 18% (within the optimal process range of 15%-20%), and the acid temperature sensor detects the temperature as 85℃ (within the optimal process range of 80-90℃). Therefore, the resulting speed reduction compensation coefficient K_chem=1.0, with no speed limit.

[0045] Based on the detection of sensor and equipment status signals at each workstation, there are currently no instantaneous working conditions disturbances. Specifically: there is no new steel coil preparation or threading operation at the inlet (V_loading is not activated), the welding machine is in standby mode (V_welder is not activated), the strip deviation detection value is 2mm (below the safety threshold of 5mm, K_stray=1.0), the tension leveler is working normally (the oxide layer can be effectively removed, V_tlv has no speed reduction constraint), the disc shear is in a stable shearing state (no width switching or edge wire escape / blockage, V_shear is not activated), and there is no unloading preparation operation at the outlet (V_unloading is not activated). The constraint speed for each instantaneous working condition is set to 300m / min (higher than the current constraint speeds).

[0046] Based on the decision logic, the first-level base speed and the second-level instantaneous operating condition coverage are calculated as follows: V_base=Min(200,250,240)·1.0·1.0=200m / min V_set=Min(200,300,300,...)=200m / min The control system sends the speed setpoint V_set=200m / min to the main drive system of the unit. The main drive system, operating at the maximum allowable acceleration of 2m / s², coordinates the acceleration of the motors in the uncoiler, process section, and coiler to smoothly transition to the target speed of 200m / min. During production, the real-time data acquisition module continuously monitors acid parameters, strip condition, and equipment operating data. The closed-loop self-learning module records the speed parameters and pickling quality data under the current operating conditions, accumulating data for subsequent optimization.

[0047] II. Circular edge shear wire escape condition Under the optimized production conditions described above, the unit is operating stably at an optimal speed of 200 m / min. The current strip is Q95 high-strength steel, 3.0 mm thick and 1250 mm wide. The disc shear is in normal shearing mode, and the edge wire conveyor is conveying edge wire normally. Suddenly, the edge wire blockage sensor next to the disc shear emits an "edge wire escape" alarm signal (the highest priority alarm). This signal is transmitted in real time to the control system of this invention via the PLC; simultaneously, the HMI interface receives the alarm information and triggers an audible and visual alert. The control system immediately activates the instantaneous operating condition constraint response mechanism, calls the disc shear operating condition constraint model, and determines the safe low speed corresponding to the alarm, V_shear_jam = 60 m / min (calibrated by the equipment manufacturer; this speed can avoid impact damage to the equipment caused by edge wire escape, and also facilitates operator troubleshooting).

[0048] At this time, the original base speed V_base is still 200m / min, but due to the triggering of the highest priority disk edge shear wire escape alarm, according to the priority rules of the multi-level decision logic, the high priority instantaneous working condition constraint can directly cover other constraint parameters. Therefore, the speed setpoint is recalculated: V_set=Min(200,60,300,...)=60m / min.

[0049] The control system immediately issued an emergency deceleration command of 60 m / min to the main drive system, while simultaneously setting the deceleration acceleration to -2 m / s² (the maximum allowable deceleration of the equipment) to ensure rapid and smooth deceleration of the unit and prevent the strip from experiencing surges or deviations due to sudden stops. The HMI interface simultaneously displayed alarm details (alarm type: disc shear edge wire escape; current speed: 200 m / min → 60 m / min; handling suggestion: check the edge wire conveyor and disc shear clearance), and activated the confirmation mechanism to prompt operators to address the fault promptly. During the deceleration process, the system continuously monitored parameters such as strip deviation and equipment operating status to ensure a stable deceleration process without triggering secondary faults.

[0050] The operator completes the troubleshooting (clearing the edge wire blockage and adjusting the edge wire conveyor speed) and confirms the fault is resolved through the HMI interface. After the alarm is cleared, the control system automatically re-collects all constraint parameters. At this time, all constraint parameters are consistent with the optimized operating conditions of normal production speed. Therefore, a speed command of V_set=200m / min is regenerated, and the main drive system controls the unit to smoothly accelerate to the optimal speed at an acceleration of 2m / s², restoring normal production.

[0051] III. Inlet roll-up, welding machine and process section in coordination The unit's current process section is producing a certain specification of strip steel at a speed of 200 m / min. The original steel coils at the inlet section are about to run out, requiring new steel coils to be loaded, threaded, and welded. Real-time monitoring shows that the inlet looper reserve is at the middle level (L_loop = 50% of rated reserve), and the outlet looper reserve is at the high level (L_loop = 80% of rated reserve). The inlet section is preparing to thread new steel coils (V_loading activated), and the welding machine is simultaneously starting to prepare for welding (V_welder activated). Looper reserve constraints are being considered, and the required synchronization speed to maintain stable production is being calculated (V_sync activated).

[0052] Specifically, the entry saddle begins preparations for winding and threading a new steel coil (Q235 steel, 2.0mm thickness, 1200mm width), triggering the entry winding preparation constraint, with a corresponding constraint speed of V_loading = 30m / min (the safe speed for threading and winding). Simultaneously, the welding machine starts preparations to stitch the tail of the original steel coil to the head of the new steel coil, triggering the welding machine's working state constraint, with a corresponding welding speed of V_welder = 30m / min (the optimal welding speed calibrated by the welding machine).

[0053] Because the speed needs to be reduced to 30m / min at the entrance section for winding and welding, in order to ensure the continuous operation of the process section, the system calculates the matching relationship between the looper reserve and the speed through the logistics model. The current median reserve of the entrance looper can support the continued operation of the process section. Therefore, the synchronization speed V_sync=130m / min is adjusted (this speed can ensure that the entrance looper reserve is slowly consumed during the speed reduction period at the entrance section, with a safety reserve of not less than 20%).

[0054] The acid concentration and temperature are normal (K_chem=1.0), there is no deviation (K_stray=1.0), and the equipment such as the disc shear and tension leveler are normal. The corresponding instantaneous working condition constraint speed is 300m / min.

[0055] The system prioritizes consuming the inlet loop storage, based on the decision-making logic: First-layer base speed calculation: V_base=Min(V_process_max=180m / min (new steel coil process constraint speed),V_equipment_max=250m / min,V_sync=130m / min)×1.0×1.0=130m / min.

[0056] The second layer of instantaneous operating condition coverage: V_set=Min(130,30,30,300,...)=130m / min (Note: The inlet roll preparation constraint V_loading=30m / min and the welding machine constraint V_welder=30m / min only apply to the inlet section. The process section speed runs at V_sync=130m / min to achieve speed coordination between the inlet section and the process section).

[0057] The control system issues a speed command of 30 m / min to the inlet section equipment (uncoiler, welder) and a speed command of 130 m / min to the process section and outlet section equipment, achieving differentiated speed control between the inlet and process sections. The inlet section completes the new steel coil threading and weld seam joining operations at 30 m / min. During this process, the inlet looper's reserve capacity slowly decreases from 50% to 30%. The process section operates continuously at 130 m / min without interruption. After the weld seam is completed, the operator confirms the weld is qualified via HMI, and the inlet coil preparation constraints and welding machine operating status constraints are released. The control system immediately recalculates the synchronous speed V_sync=240m / min, the base speed V_base=Min(180,250,240)=180m / min, and finally V_set=180m / min. An acceleration command is issued to the inlet section, which accelerates to 180m / min at an acceleration of 2m / s². At the same time, the inlet looper storage is replenished to 50% of the median level. The process section maintains a stable operation at 180m / min.

[0058] By coordinating the speed control of the entry section and the process section, the process section operates without interruption throughout the entire coiling and welding process, with zero production interruption time. Compared with the traditional manual control mode of "process section stopping to wait for entry operation", the production time of a single coil is shortened while the weld quality is stable, and there are no defects such as weld deviation or incomplete penetration caused by speed mismatch. This achieves the coordinated optimization of material switching and continuous production.

[0059] In summary, this invention can achieve intelligent, automatic, safe, and efficient control of the pickling unit speed for different production conditions, effectively solving the defects of traditional control methods such as conservatism, lag, and weak multi-factor coupling processing capability, and significantly improving the production efficiency, quality stability, and operational safety of the pickling production line.

[0060] The present invention also provides an automatic speed control system for pickling units based on multi-factor constraints. The automatic speed control system for pickling units based on multi-factor constraints can be implemented by executing the process steps of the automatic speed control method for pickling units based on multi-factor constraints. That is, those skilled in the art can understand the automatic speed control method for pickling units based on multi-factor constraints as a preferred embodiment of the automatic speed control system for pickling units based on multi-factor constraints.

[0061] An automatic speed control system for pickling units based on multi-factor constraints is provided by the present invention, such as... Figure 3 As shown, it includes: Data acquisition layer: Composed of various sensors, instruments and PLCs, used to collect multi-dimensional on-site production data. That is, the data acquisition module includes various sensors, instruments and PLCs to acquire process and equipment status data in real time.

[0062] The process monitoring layer includes a process computer and a database, used to run multi-factor constraint models and speed optimization algorithms. Specifically, the process monitoring layer includes a multi-factor constraint modeling module for constructing five types of speed constraint models, an adaptive speed decision module for calculating the globally optimal speed setpoint using a priority arbitration mechanism, a main drive control module for driving the motors in each section to coordinate speed regulation, and a self-learning optimization module for iteratively optimizing control parameters based on reinforcement learning. The multi-factor constraint model is constructed by identifying and quantifying key constraint factors. The speed optimization algorithm includes calculating the globally optimal speed setpoint V_set based on the multi-factor constraint model and collected real-time data, through a preset multi-layer decision logic. This speed optimization algorithm also includes a built-in reinforcement learning algorithm, which combines historical production data to optimize the pickling kinetic model parameters and constraint weights. The self-learning optimization module adopts an offline pre-training and online real-time learning mode; it uses historical production data to complete model pre-training, and continuously iterates using online surface quality inspection data to automatically optimize the pickling kinetic model parameters and constraint weights. The adaptive speed decision module has a built-in hierarchical priority arbitration logic: Level 0 is the priority of unconditional emergency shutdown, Level 1 is the priority of core safety and process interlocking, Level 2 is the priority of critical instantaneous operating condition constraints, and Level 3 is the priority of conventional process, equipment and logistics constraints; higher priority constraints directly cover the calculation results of lower priority constraints.

[0063] The system supports independent speed regulation of the inlet section and the process section. During inlet winding and welding / sewing operations, the inlet section operates at a safe welding / threading speed, while the process section operates continuously at the looper synchronous speed, achieving uninterrupted production during material switching. The system also includes a system-level safety interlock module; this module is deeply interlocked with the acid circulation, correction, brush roller, disc shear, and media supply subsystems, automatically limiting the unit speed to a safe range when process conditions are not met.

[0064] Execution control layer: This includes the main drive PLC of the unit, which communicates with the main drive control system, issues speed control commands, and monitors the execution process.

[0065] Human-Machine Interface (HMI): Used to display the current speed, the limits of various constraints, alarm information, and to allow operators to monitor and intervene as necessary.

[0066] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0067] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. An automatic speed control method for pickling units based on multi-factor constraints, characterized in that, include: Step S1: Collect data in real time from sensors, PLC and process control system to construct the multi-factor constraint model; the data includes strip tracking signal, looper position signal, acid concentration and temperature, equipment operating status, weld detection signal and deviation detection signal; Step S2: Identify and quantify the key constraints affecting the speed of the pickling unit, and construct a multi-factor constraint model for the pickling unit; the key constraints include process constraints, equipment capacity constraints, logistics synchronization constraints, acid and energy constraints, and instantaneous equipment operating condition constraints; Step S3: Based on the multi-factor constraint model and real-time collected data, the globally optimal speed setpoint V_set is dynamically calculated using multi-layer decision logic; Step S4: Send the speed setpoint V_set to the main drive control system of the unit to control the motors of each component to accelerate and decelerate smoothly in coordination; Step S5: Monitor pickling quality using a surface inspection instrument, and combine equipment operating status and production efficiency data. Use reinforcement learning algorithm to self-learn and optimize the pickling kinetic model parameters and constraint weights, and iteratively update the speed control strategy.

2. The automatic speed control method for pickling units based on multi-factor constraints according to claim 1, characterized in that, The process constraints mentioned in step S2 are quantified and calculated using an acid pickling kinetics model, including: Step S2.1: Calculate the theoretical minimum process time T_process_min based on the strip steel grade, specifications, oxide layer thickness, and acid parameters, using the following formula: T_process_min=(δ_scale^η) / (A·[HCl]^α·exp(-E_a / (R*T))·k_m·C_steel) Where T_process_min represents the theoretical minimum process time, δ_scale represents the total thickness of the oxide layer, A,α,η represent empirical constants, [HCl] represents the effective concentration of acid solution, E_a represents the apparent activation energy of the pickling reaction, R represents the ideal gas constant, T represents the acid solution temperature, k_m represents the mass transfer coefficient, and C_steel represents the steel grade coefficient. Step S2.2: Derive the maximum process speed V_process_max based on the effective length L_process of the pickling process section, as shown in the following formula: V_process_max=L_process / T_process_min.

3. The automatic speed control method for pickling units based on multi-factor constraints according to claim 1, characterized in that, The multi-layered decision logic mentioned in step S3 is specifically as follows: The first layer calculates the base velocity using the following formula: V_base=Min(V_process_max,V_equipment_max,V_sync)·K_chem·K_stray In the formula, V_process_max is the maximum process speed corresponding to the process constraint, V_equipment_max is the maximum allowable working speed corresponding to the equipment capacity constraint, V_sync is the synchronization speed corresponding to the logistics synchronization constraint, K_chem is the speed reduction compensation coefficient corresponding to the acid and energy constraints, and K_stray is the speed reduction coefficient corresponding to the strip deviation in the instantaneous operating condition constraints of the equipment. The second layer performs instantaneous condition coverage to obtain the final speed, as shown in the following formula: V_set=Min(V_base,V_weld,V_welder,V_tlv,V_shear,V_loading,V_unloading) In the formula, V_weld is the weld seam tracking constraint speed, V_welder is the welding machine working state constraint speed, V_tlv is the tension leveler working state constraint speed, V_shear is the disc shear working state constraint speed, V_loading is the inlet coil loading preparation constraint speed, and V_unloading is the outlet unloading preparation constraint speed. Among them, edge wire escape and weld failure are the highest priority alarms. When triggered, they directly override all constraints and execute emergency deceleration or shutdown.

4. The automatic speed control method for pickling units based on multi-factor constraints according to claim 1, characterized in that, The instantaneous operating conditions constraints of the equipment mentioned in step S2 include inlet coiling preparation constraints, welding machine working status constraints, strip deviation constraints, tension leveler operating conditions constraints, disc shear operating conditions constraints, and outlet uncoiling preparation constraints. The logistics synchronization constraints include looper storage constraints and weld seam tracking constraints, which can realize differentiated speed coordination control between the inlet section and the process section.

5. The automatic speed control method for pickling units based on multi-factor constraints according to claim 1, characterized in that, In step S5, the Actor-Critic reinforcement learning framework is used to achieve self-learning optimization; Using strip steel specifications, acid parameters, pickling quality, and equipment operating conditions as states, adjusting pickling kinetic model parameters and constraint weights as actions, and constructing a reward function based on comprehensive indicators of quality, efficiency, cost, and safety, production efficiency and operating costs are optimized while ensuring pickling quality.

6. An automatic speed control system for a pickling unit based on multi-factor constraints, characterized in that, The method for implementing any one of claims 1-5 includes a data acquisition module, a multi-factor constraint modeling module, an adaptive speed decision module, a main drive control module, and a self-learning optimization module; The data acquisition module includes various sensors, instruments, and PLCs, used to acquire process and equipment status data in real time; The multi-factor constraint modeling module is used to construct five types of velocity constraint models; The adaptive speed decision module calculates the globally optimal speed setting value according to a priority arbitration mechanism; The main drive control module is used to drive the motors of each component to coordinate speed regulation. The self-learning optimization module is used to iteratively optimize control parameters based on reinforcement learning.

7. The automatic speed control system for pickling units based on multi-factor constraints according to claim 6, characterized in that, The adaptive speed decision module has a built-in hierarchical priority arbitration logic: Level 0 is the priority of unconditional emergency shutdown, Level 1 is the priority of core safety and process interlocking, Level 2 is the priority of key instantaneous operating condition constraints, and Level 3 is the priority of conventional process, equipment and logistics constraints. Higher priority constraints directly override the calculation results of lower priority constraints.

8. The automatic speed control system for pickling units based on multi-factor constraints according to claim 6, characterized in that, The system supports independent speed adjustment of the inlet section and the process section; Under the conditions of inlet roll-up and welding machine stitching, the inlet section operates at the welding / threading safety speed, and the process section operates continuously at the looper synchronous speed, so as to achieve material switching without production interruption.

9. The automatic speed control system for pickling units based on multi-factor constraints according to claim 6, characterized in that, It also includes a system-level safety interlock module; The safety interlock module is deeply interlocked with the acid circulation, correction, brush roller, disc shear, and media supply subsystems, and automatically limits the unit speed to a safe range when process conditions are not met.

10. The automatic speed control system for pickling units based on multi-factor constraints according to claim 6, characterized in that, The self-learning optimization module adopts an offline pre-training and online real-time learning mode; Historical production data is used to complete model pre-training, and online surface quality inspection data is combined to continuously iterate and automatically optimize the parameters and constraint weights of the pickling kinetics model.

Images