High-performance copper wire drawing intelligent control method based on multi-source data

Abstract

This invention discloses a high-performance intelligent control method for copper wire drawing based on multi-source data, belonging to the field of intelligent control technology for copper wire drawing. The method includes collecting multi-source data from the wire drawing machine and preprocessing it to generate a wire drawing process dataset. This dataset is then input into a physical constraint model constructed based on the process mechanism. This model describes the dynamic balance relationship between wire diameter reduction rate, die temperature rise, and drum speed. The optimal wire diameter control benchmark and die cooling requirements are obtained through model state inversion. A dual-deviation signal is generated by comparing theoretical control values ​​with real-time measured values. A multi-objective optimization algorithm is used to solve for the optimal adjustment of parameters such as the die compression ratio, which is then converted into action commands and issued to the actuators to synchronously adjust related systems. This method can reduce control deviation, extend the die operating cycle, improve wire diameter control accuracy and drawing stability, improve copper wire surface quality and dimensional uniformity, and optimize energy consumption distribution.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology for copper wire drawing, specifically a high-performance intelligent control method for copper wire drawing based on multi-source data. Background Technology

[0002] Copper wire drawing is a crucial process in metal processing, and its product quality and production efficiency directly depend on the control precision of the drawing machine. Currently, copper wire drawing control largely employs a traditional single-parameter feedback control mode. This mode adjusts the drawing process parameters by collecting single or limited data during the drawing process and combining it with the operator's experience. This type of control relies primarily on direct feedback from real-time measurement data, failing to deeply integrate with the intrinsic mechanisms of the drawing process, and can only passively adjust for current deviations.

[0003] In existing technologies, wire diameter control and die cooling control during the wire drawing process are independent, failing to consider the dynamic correlation between wire diameter reduction rate, die temperature rise, and drum speed. This makes it impossible to accurately obtain the theoretical control benchmark for the wire drawing process, resulting in lag in wire diameter deviation adjustment. Furthermore, improper temperature control can easily lead to accelerated wear and shortened lifespan of the die. Simultaneously, existing control methods often adjust single parameters independently, failing to achieve coordinated optimization of die compression ratio, drawing fluid flow rate, and main motor torque distribution. This makes it difficult to balance wire drawing efficiency and product quality, easily leading to problems such as scratches on the copper wire surface and poor wire diameter uniformity. Summary of the Invention

[0004] This invention aims to solve at least one of the technical problems existing in the prior art; To this end, the present invention proposes a high-performance intelligent control method for copper wire drawing based on multi-source data, comprising: Collect multi-source data during the operation of the wire drawing machine, preprocess the multi-source data, and generate a wire drawing process dataset; The drawing process dataset is input into a physical constraint model based on the process mechanism, which describes the dynamic balance between wire diameter reduction rate, die temperature rise and drum speed. Using the physical constraint model, the current wire drawing process is inverted to calculate the theoretically optimal wire diameter control benchmark and mold cooling requirements. The calculated optimal wire diameter control benchmark and mold cooling requirements are compared with the real-time wire diameter measurement value and actual mold temperature value directly extracted from the multi-source data to generate wire diameter deviation signal and temperature deviation signal. Based on the wire diameter deviation signal and temperature deviation signal, a set of optimal wire drawing process parameter adjustments are obtained in a multi-objective optimization algorithm. The wire drawing process parameters include the die compression ratio, the flow rate of the wire drawing fluid, and the torque distribution of the main motor. The calculated wire drawing process parameter adjustment amount is converted into action commands that the equipment can execute and sent to the actuator of the wire drawing machine; The actuator synchronously adjusts the assembly gap of the wire drawing die, the wire drawing fluid circulation system, and the main transmission system according to the action command, so that the wire drawing machine enters a new operating condition.

[0005] Furthermore, the multi-source data is preprocessed to generate a wire drawing process dataset, including: By using sensor groups arranged at the inlet and outlet of the wire drawing machine, multi-source data during the operation of the wire drawing machine are collected synchronously. The multi-source data includes the wire diameter data of the copper wire, the temperature data of the wire drawing die, and the rotation speed data of the wire drawing drum. The multi-source data is preprocessed to remove abnormal fluctuation values ​​and to interpolate and complete missing time-slice data to generate a continuous and stable wire drawing process dataset, specifically including: A sliding time window is applied to the wire diameter data, and the moving average and standard deviation of the data within the window are calculated; When the wire diameter data of a certain time slice deviates from the moving average by more than three times the standard deviation, the wire diameter data point corresponding to the current time slice is marked as an instantaneous interference point and removed. For the missing time series after removal, a linear interpolation method based on the changing trend of adjacent data points is used to fill them; The temperature data of the mold and the rotation speed data of the drum are also processed for noise smoothing using the sliding time window method, while preserving the low-frequency trend components of the data. The wire diameter data, die temperature data, and drum rotation speed data, after anomaly removal, interpolation completion, and noise smoothing, are timestamped and integrated to generate the continuous and stable wire drawing process dataset.

[0006] Furthermore, the drawing process dataset is input into a physical constraint model constructed based on the process mechanism, including: In the physical constraint model, a differential equation is defined regarding the change in the cross-sectional area of ​​the copper wire. The input variable of the differential equation is the rotational speed data of the drum, and the output variable is the instantaneous speed of the copper wire. The temperature data of the mold is used as a boundary condition in the differential equation to correct the plastic deformation coefficient of the copper wire at high temperature. Based on the instantaneous velocity and plastic deformation coefficient, the theoretical mold compression angle for maintaining a constant copper wire cross-section under the current working conditions can be deduced. The theoretical mold compression angle is used as the core intermediate variable in the output of the physical constraint model.

[0007] Furthermore, the physical constraint model is used to perform state inversion on the current wire drawing process to calculate the theoretically optimal wire diameter control benchmark and die cooling requirements, including: Read the theoretical mold compression angle output by the physical constraint model, and combine it with the preset product specification tolerance zone to calculate the optimal wire diameter control benchmark within an allowable floating range; By retrieving the temperature data of the mold over time and analyzing its upward trend, and combining this with the thermal conductivity of the mold material, the required heat dissipation power to bring the mold temperature back to its normal operating range can be calculated. The heat dissipation power is converted into the mold cooling requirement, which is specifically reflected in the theoretical increase in the flow rate of the drawing fluid.

[0008] Furthermore, the calculated optimal wire diameter control benchmark and mold cooling requirements are compared with the real-time wire diameter measurement value and actual mold temperature value directly extracted from the multi-source data to generate wire diameter deviation signals and temperature deviation signals, including: Establish a closed-loop comparator to calculate the difference between the real-time wire diameter measurement value and the optimal wire diameter control benchmark to obtain the wire diameter error value. The wire diameter error value is subjected to a first-order low-pass filter to remove high-frequency glitches caused by mechanical vibration, thereby obtaining the wire diameter deviation signal. The actual temperature value of the mold is compared with the target temperature value corresponding to the theoretical increase in the flow rate of the drawing fluid, and the temperature error value is calculated. The temperature error value is amplified proportionally to generate the temperature deviation signal used to drive the cooling system.

[0009] Furthermore, based on the wire diameter deviation signal and the temperature deviation signal, a set of optimal wire drawing process parameter adjustments is obtained in a multi-objective optimization algorithm, including: Initialize a feasible solution space that includes the compression ratio of the mold, the flow rate of the drawing fluid, and the torque distribution of the main motor; In the multi-objective optimization algorithm, minimizing the absolute value of the wire diameter deviation signal and minimizing the absolute value of the temperature deviation signal are taken as two parallel optimization objectives. The dynamic equilibrium relationship defined by the physical constraint model is introduced as a constraint condition to prevent the solution in the solution space from causing instability in the wire drawing process; Through iterative optimization, a Pareto optimal solution is selected from the feasible solution space, and the parameter change corresponding to the Pareto optimal solution is used as the adjustment amount of the wire drawing process parameters.

[0010] Furthermore, the specific execution process of the multi-objective optimization algorithm includes: A non-dominated sorting genetic algorithm is used as the core of the multi-objective optimization algorithm to encode individuals in the feasible solution space; In each generation of evolution, non-dominated ranking is performed based on the individual’s performance in wire diameter control and temperature control, and crowding calculation is introduced to maintain population diversity. When the variation of the optimal solution in the target space for several consecutive generations is less than a set threshold, the iteration terminates and the current Pareto optimal solution is output.

[0011] Furthermore, the calculated wire drawing process parameter adjustment amounts are converted into action commands that the equipment can execute, including: Establish a device instruction mapping table, which defines the correspondence between wire drawing process parameters and actual actuator actions; The compression ratio adjustment of the mold is mapped to the number of pulse commands that drive the servo motor to move the mold base. The flow rate adjustment of the drawing fluid is mapped to the adjustment command of the inverter output frequency; The torque distribution adjustment of the main motor is mapped to the speed ratio command between multiple drive motors.

[0012] Furthermore, the process of adjusting the assembly clearance of the wire drawing die by the actuator also includes a dynamic adjustment step: Real-time monitoring of the load current feedback value of the servo motor when executing pulse number commands; If the load current feedback value exceeds the rated current threshold of the motor, it is determined that there is a foreign object stuck in the mold; Pause the current mold gap adjustment action and generate an alarm signal. At the same time, lock the mold compression ratio at the current value and wait for manual intervention.

[0013] Furthermore, after the actuator completes an adjustment, a closed-loop verification step is also included: Continuously collect and adjust the wire drawing process dataset, focusing on the adjusted wire diameter data and die temperature data; Calculate the new deviation between the adjusted wire diameter data and the control baseline, and the temperature difference between the adjusted mold temperature data and the target temperature; If both the new deviation value and the temperature difference value are within the allowable error range, then the intelligent control closed loop is considered complete. If any value exceeds the error range, the adjusted wire drawing process dataset, along with the deviation information, will be fed back to the physical constraint model to trigger a new round of parameter optimization calculations.

[0014] Compared with the prior art, the beneficial effects of the present invention are: A physical constraint model is constructed based on the wire drawing process mechanism. This model describes the dynamic balance between wire diameter reduction rate, die temperature rise, and drum speed. Using this model, the state of the wire drawing process is inverted to calculate the theoretically optimal wire diameter control benchmark and die cooling requirements. By combining the process mechanism with model construction and state inversion, the limitations of traditional control relying solely on passive adjustments based on real-time data can be overcome. This makes the determination of wire diameter control and die cooling requirements more scientific, effectively reducing control deviations caused by a lack of theoretical basis. It allows the state control of the wire drawing process to better align with actual process requirements, reduces unnecessary temperature rise losses in the die, extends the die operating cycle, and improves the accuracy of wire diameter control.

[0015] The optimal wire diameter control benchmark calculated by the physical constraint model and the mold cooling requirements are compared with the real-time wire diameter measurement value and the actual mold temperature value extracted from multi-source data to generate wire diameter deviation signals and temperature deviation signals. Based on the dual deviation signals, the optimal adjustment amounts of the mold compression ratio, drawing fluid flow rate, and main motor torque distribution are solved in a multi-objective optimization algorithm. The adjustment amounts are converted into action commands and sent to the actuators to synchronously adjust the mold assembly gap, the drawing fluid circulation system, and the main drive system. By driving the synchronous optimization and adjustment of multiple process parameters through dual deviation signals, the drawbacks of traditional independent adjustment of single parameters can be solved. This achieves coordinated linkage between various process parameters and the actuators, making the operation of each system of the wire drawing machine more coordinated, reducing equipment operation fluctuations caused by asynchronous parameter adjustments, improving the stability of copper wire drawing, improving the surface quality and dimensional uniformity of copper wire products, and optimizing energy consumption distribution in the wire drawing process. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of the high-performance copper wire drawing intelligent control method based on multi-source data described in this invention. Figure 2 A flowchart for constructing a physical constraint model and outputting core intermediate variables; Figure 3 A graph showing the temperature and deviation analysis of copper wire drawing dies; Figure 4 Analysis chart of the effect of multi-objective deviation control in copper wire drawing; Figure 5 This diagram illustrates the abnormal load current monitoring and alarm for the servo motor of the wire drawing die. Detailed Implementation

[0017] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] See Figure 1 The system collects multi-source data generated during the operation of the wire drawing machine and preprocesses this data to generate a continuous and reliable wire drawing process dataset. This processed dataset is then input into a physical constraint model based on the process mechanism. This model defines the dynamic balance between wire diameter reduction rate, die temperature rise, and drum speed. The physical constraint model is used to perform inverse calculations on the current wire drawing process state, deriving the theoretically ideal wire diameter control benchmark and the required die cooling demand. These calculated optimal wire diameter control benchmarks and die cooling demands are compared with wire diameter measurements and actual die temperature values ​​directly read from real-time multi-source data, generating wire diameter deviation signals and temperature deviation signals that reflect the difference between actual and theoretical values. Based on the generated deviation signals, a multi-objective optimization algorithm is used to solve for and find a set of optimal wire drawing process parameter adjustments. These parameters include the die compression ratio, the flow rate of the drawing fluid, and the torque distribution of the main motor. The solved parameter adjustments are then converted into action commands that can be directly recognized and executed by each actuator of the wire drawing machine according to a preset mapping relationship, and these commands are then sent to the corresponding execution units. The actuator synchronously and precisely adjusts the assembly gap of the drawing die, the drawing fluid circulation system, and the main drive system according to the received action instructions, so that the drawing machine transitions to a new and more optimized operating condition, thereby realizing closed-loop intelligent control of the entire drawing process.

[0019] In one embodiment of the invention, a sensor group is formed by deploying wire diameter measuring instruments, infrared temperature sensors, photoelectric encoders, and other devices at the wire inlet and outlet ends of the wire drawing machine. The sensor group synchronously collects raw multi-source data during the wire drawing machine's operation at millisecond-level frequencies. This raw multi-source data includes wire diameter data of the copper wire entering and exiting the die, temperature data at specific measuring points on the drawing die, and real-time rotational speed data of the drawing drum for each pass. In a specific implementation, the collected raw multi-source data is preprocessed to remove abnormal fluctuation values ​​and interpolate missing time slice data to generate a continuous and stable wire drawing process dataset. In a specific implementation, a sliding time window with a fixed length of N sampling points is applied to the wire diameter data. In the data processing server, each time a new wire diameter data point is received, the moving average and standard deviation of all N wire diameter data points within the sliding time window centered on that point are calculated. In practice, when the value of the wire diameter data point corresponding to a certain time slice deviates from the moving average of its corresponding sliding time window by more than three standard deviations, the data processing logic marks the wire diameter data point corresponding to the current time slice as an instantaneous interference point and removes it from the original data sequence. In practice, for the gaps in the wire diameter data time series caused by removing instantaneous interference points, a linear interpolation method based on the changing trends of adjacent valid data points is used to fill them, restoring the continuity of the wire diameter data. In practice, the temperature data of the die and the rotational speed data of the drum are also subjected to noise smoothing using a sliding time window method. The purpose of smoothing is to preserve the low-frequency trend components of the die temperature data and the drum rotational speed data. In practice, the wire diameter data, die temperature data, and drum rotational speed data, after anomaly removal, interpolation completion, and noise smoothing, are aligned and integrated based on millisecond-level timestamps to generate a final continuous and stable wire drawing process dataset for subsequent modeling.

[0020] In some embodiments, the length N of the sliding time window is dynamically determined based on the wire drawing speed. It is understood that the faster the wire drawing speed, the longer the length of copper wire passing through per unit time, requiring a shorter window length to capture rapid changes. It is understood that the data processing server stores a lookup table of recommended N values ​​for different wire diameters and drawing speeds, dynamically indexed based on the centralized line speed data during the wire drawing process. In some embodiments, when applying linear interpolation, if more than three consecutive instantaneous interference points are detected, the system considers this a persistent interference, pauses interpolation, and initiates a data quality alarm, awaiting operator confirmation. In specific implementations, a weighted average calculation method can be used for data processing within the sliding time window to enhance sensitivity to the current data. The weighted average calculation formula is: in: This represents the weighted moving average calculated at time point t. It is the length of the sliding time window. This represents the original wire diameter data at time point ti. It is the weight assigned to the i-th historical data point. The weight of a data point decreases exponentially as i increases, meaning that the closer a data point is to the current time point t, the greater its weight.

[0021] In one embodiment of the present invention, see [reference] Figure 2 In the physical constraint model of the control system, a differential equation is defined to describe the continuous change in cross-sectional area of ​​the copper wire during plastic deformation in the die. In practice, the input variable of the differential equation is the rotational speed data of a specific drawing drum extracted from a continuous and stable drawing process dataset. The output variable is the instantaneous velocity of the copper wire passing through the die deformation zone under the current drum rotational speed. Real-time die temperature data extracted from the continuous and stable drawing process dataset is introduced as a boundary condition into the aforementioned differential equation for simultaneous solution. The main function of the die temperature boundary condition is to dynamically correct the plastic deformation coefficient, which characterizes the flow properties of copper in the deformation zone, in the differential equation. As the die temperature increases, the plastic deformation coefficient changes accordingly. Based on the calculated instantaneous velocity of the copper wire and the corrected plastic deformation coefficient, the physical constraint model calculates, through back-calculation, the theoretical die compression angle required to maintain a stable and constant cross-sectional area during the copper wire cross-section reduction process under the combined conditions of the current drum rotational speed, current die temperature, and wire specifications. The calculated theoretical mold compression angle is a core intermediate variable output by the physical constraint model. The theoretical mold compression angle will be passed to the subsequent state inversion module for calculation.

[0022] In some embodiments, the differential equation describing the change in the cross-sectional area of ​​the copper wire adopts an engineering model based on the principle of volume invariance and the constitutive relationship of plastic flow. It is understood that the model parameters involved in the differential equation, such as the initial cross-sectional area, target cross-sectional area, and friction coefficient, are preset according to the specific grade of copper wire material and the die specifications. The relationship between the plastic deformation coefficient and the die temperature is determined through a preset coefficient-temperature lookup table or fitting function. When die temperature data is input, the physical constraint model will look up the corresponding plastic deformation coefficient value from the lookup table or calculate it through the function. In some embodiments, when calculating the instantaneous velocity of the copper wire, the physical constraint model comprehensively considers the drum rotation speed, drum diameter, and possible slip rate. The solution process of the physical constraint model is real-time. Whenever a new, continuous, and stable wire drawing process dataset is input, the model initiates a calculation and outputs the theoretical die compression angle at the corresponding moment. The back-calculation of the theoretical die compression angle relies on an implicit equation solver. The solver uses an iterative method to find a compression angle value that satisfies the force balance and volumetric flow rate equality in the deformation zone.

[0023] The differential equations of the physical constraint model can be expressed in the following form: in: It is a geometric constant related to the initial cross-sectional area of ​​the copper wire. It is the instantaneous velocity of the copper wire within the deformation zone of the mold. These are coordinates along the mold axis. Indicates instantaneous velocity along Rate of change of direction It is related to the mold temperature The relevant plastic deformation coefficient function, This is a driving term constant related to the linear velocity of the drum. In practical implementation, the formula... The specific form of the function was obtained through calibration experiments. In practical implementation, the boundary conditions, namely the current mold temperature, will be used. Substitute the actual measured value into The function can calculate the current operating conditions. The specific value is determined by combining it with the input drum speed data. The instantaneous velocity is obtained by solving the above differential equation. The distribution along the mold axis is used to calculate the theoretical mold compression angle.

[0024] In one embodiment of the present invention, the state inversion module of the control system reads the theoretical die compression angle output by the physical constraint model. Combining this with the target wire diameter and its upper and lower tolerances given in the pre-stored product specification drawings, the state inversion module converts the theoretical die compression angle into the corresponding theoretical wire diameter value through geometric relationships. Centering on the theoretical wire diameter value, the state inversion module calculates a permissible range of values ​​with respect to the tolerance band width specified in the product specification drawings. This calculated range serves as the optimal wire diameter control benchmark. The thermal state analysis module of the control system retrieves recent time-series die temperature data from a continuously stable drawing process dataset. The thermal state analysis module analyzes the upward trend of the die temperature data over time. Based on the material of the drawing die, the thermal state analysis module queries the thermal conductivity coefficient of that material from the system's material library. Combining the upward trend of the die temperature data with the thermal conductivity coefficient, the thermal state analysis module uses thermodynamic formulas to calculate the heat dissipation power required to reduce the die temperature from its current value and return it to the preset normal operating temperature range. The thermal state analysis module calculates the heat dissipation power and, based on the efficiency of the heat exchanger in the drawing fluid circulation system and the specific heat capacity of the drawing fluid, converts it into the theoretically required increase in volumetric flow rate. This volumetric flow rate is the mold cooling requirement. The closed-loop comparator module of the control system directly extracts the real-time wire diameter measurement value fed back by the online measurement sensor from the real-time data stream. The closed-loop comparator module obtains the optimal wire diameter control reference from the state inversion module. The closed-loop comparator module calculates the difference between the real-time wire diameter measurement value and the center value of the optimal wire diameter control reference to obtain the original wire diameter error value. The closed-loop comparator module performs a first-order low-pass filter on the original wire diameter error value. The first-order low-pass filter can filter out high-frequency burr signals caused by mechanical vibration of the drawing machine or measurement noise. After the first-order low-pass filter, a smooth wire diameter deviation signal that can be used for control is obtained.

[0025] In practical implementation, the temperature comparison module of the control system extracts the actual mold temperature value from the real-time data stream, and obtains the expected target temperature value corresponding to the mold cooling requirement from the thermal state analysis module. The temperature comparison module subtracts the actual mold temperature value from the expected target temperature value to calculate the original temperature error value. The temperature comparison module then multiplies the original temperature error value by a fixed proportional gain coefficient, which amplifies the temperature error value, generating a temperature deviation signal with a clear driving direction and amplitude. This temperature deviation signal is used to control the cooling system.

[0026] In some embodiments, the calculation formula for the optimal wire diameter control benchmark considers long-term drift caused by die wear. It is understood that the center value of the optimal wire diameter control benchmark will be dynamically fine-tuned as the theoretical die compression angle changes. The tolerance band width is a fixed process parameter and does not change with dynamic calculations. In some embodiments, the analysis of the die temperature rise trend uses a linear regression method, obtaining the slope of temperature change over time by fitting temperature data points within the most recent time window. The calculation of heat dissipation power is based on a simplified one-dimensional steady-state heat transfer model. The model inputs include the current die temperature, the target die temperature, the contact area between the die and the drawing fluid, and the thermal conductivity coefficient of the die material. The digital implementation formula for the first-order low-pass filter is: in: This represents the wire diameter deviation signal output after filtering at time k. This represents the wire diameter deviation signal output at time k-1. This represents the original wire diameter error value calculated at time k. The filter coefficient α is between 0 and 1, and it determines the cutoff frequency of the filter. In practice, the value of the filter coefficient α is set based on a combination of the wire drawing speed and the response frequency of the wire diameter measuring sensor; a higher wire drawing speed corresponds to a smaller filter coefficient α.

[0027] See Figure 3 This is a temperature and deviation analysis chart for a copper wire drawing die, visually reflecting the linkage between die temperature changes and control signals. The actual die temperature gradually rises from approximately 65℃ to a peak of approximately 105℃, then slowly declines, exhibiting an overall "rise then fall" trend accompanied by small, high-frequency fluctuations. The target die temperature initially stabilizes at approximately 65℃, then slowly and linearly decreases over time, reflecting the dynamic adjustment of cooling requirements. The temperature deviation signal is highly synchronized with the actual temperature, peaking at approximately 100℃, directly driving cooling actuators such as the drawing fluid flow rate. The continuous rise in die temperature from 0 to 80 minutes stems from the frictional heat generated by the plastic deformation of the copper wire, a typical heat accumulation process in wire drawing. The temperature reaches its peak and then falls back from 80 to 120 minutes, indicating that the cooling system has been activated and is effective based on the temperature deviation signal, gradually pulling the die temperature towards the target range.

[0028] In one embodiment of the present invention, before running, the optimization solution module first initializes a feasible solution space consisting of three decision variables: the compression ratio adjustment of the die, the flow rate adjustment of the drawing fluid, and the torque distribution adjustment of the main motor. In specific implementation, the value range of each decision variable in the feasible solution space is preset with upper and lower bounds based on the mechanical and physical limits, electrical safety limits, and process experience of the drawing machine. In specific implementation, the multi-objective optimization algorithm built into the optimization solution module sets two parallel optimization objectives: the first objective is to minimize the absolute value of the wire diameter deviation signal, and the second objective is to minimize the absolute value of the temperature deviation signal. The multi-objective optimization algorithm introduces the dynamic balance relationship between the wire diameter reduction rate, die temperature rise, and drum speed described by the physical constraint model as a hard constraint condition. This hard constraint condition ensures that during the optimization search process, all candidate process parameter combinations will not lead to unstable states in the drawing process such as wire breakage, die overheating, or drum slippage. Under the premise of satisfying hard constraints, the multi-objective optimization algorithm iteratively searches within the feasible solution space and finally selects a Pareto optimal solution that achieves the best trade-off between minimizing wire diameter deviation and minimizing temperature deviation.

[0029] In practical implementation, the optimization module outputs the final wire drawing process parameter adjustments as the die compression ratio adjustment, drawing fluid flow rate adjustment, and main motor torque distribution adjustment corresponding to the Pareto optimal solution. The multi-objective optimization algorithm employs a non-dominated sorting genetic algorithm as its core framework. This algorithm encodes individuals in the feasible solution space, with each individual corresponding to a specific numerical combination of die compression ratio adjustment, drawing fluid flow rate adjustment, and main motor torque distribution adjustment. During the evolution of each generation, the non-dominated sorting genetic algorithm performs non-dominated sorting on all individuals in the population based on their wire diameter control performance and temperature control performance. Wire diameter control performance is evaluated by the absolute value of the predicted wire diameter deviation signal after applying the set of adjustments, and temperature control performance is evaluated by the absolute value of the predicted temperature deviation signal. In practical implementation, the non-dominated sorting genetic algorithm introduces crowding calculation, which measures the distribution density of each individual in the target space composed of the absolute values ​​of wire diameter deviation and temperature deviation. Introducing crowding calculation helps maintain population diversity. The iteration termination condition of the non-dominated sorting genetic algorithm is set as follows: when the position change of the best individual in the frontier of the population in the target space is less than a pre-set threshold after several generations of evolution, the algorithm terminates the calculation and outputs the Pareto optimal solution of the current generation.

[0030] In some embodiments, the decision variables are encoded using real numbers, and the gene values ​​of each variable are initialized with a uniform distribution within its feasible region. It is understood that the specific calculations of wire diameter control performance and temperature control performance require substituting the candidate parameter adjustments into a simplified rapid process model for simulation and prediction. The selection of the Pareto optimal solution employs an elite selection strategy based on crowding comparison, prioritizing individuals with high non-dominant rank and high crowding. In some embodiments, the specific number of generations for "several consecutive generations" and the threshold for "variation magnitude" in the iteration termination condition are set according to the requirements for control real-time performance; higher real-time performance requirements result in fewer generations and a larger threshold, and vice versa. When evaluating individuals, the simplified model predicting the absolute values ​​of wire diameter deviation and temperature deviation will call the core computational unit of the physical constraint model for rapid evaluation. A formula for calculating crowding is as follows: in: Indicates an individual in the target space The level of congestion, This indicates the number of optimization objectives, which is relevant in specific application scenarios. , This indicates that in the target space, in the individual After that, individuals adjacent to the non-dominated hierarchy In the The value of the objective function Indicates in the individual Previously, individuals adjacent to the non-dominated hierarchy In the The value of each objective function. In practical implementation, congestion. The larger the size, the more likely the individual is to be affected. The smaller the density of surrounding solutions, the better the individual The greater the contribution, the better in maintaining population diversity.

[0031] Optionally, before optimization begins, a set of high-quality initial solutions will be loaded from the historical database to accelerate optimization convergence. Optionally, the adjustment amount of the wire drawing process parameters obtained from each optimization will be superimposed on the current value, and it will be determined whether the superposition exceeds the equipment safety limit. If it does, boundary truncation will be performed. Refer to Table 1 for the typical value range constraints of the feasible solution space on the three decision variables.

[0032] Table 1: Typical Value Range Constraints for Decision Variables See Figure 4This is a chart analyzing the multi-objective deviation control effect of copper wire drawing, visually demonstrating the convergence process of wire diameter deviation and temperature deviation under the control closed loop. From 0 to 30 seconds, the wire diameter deviation fluctuates drastically, peaking at 0.17 mm, far exceeding the allowable upper limit of 0.08 mm; the temperature deviation fluctuates synchronously, peaking at approximately 8°C, also exceeding the allowable upper limit of 3°C. This stage represents the start-up of wire drawing or a sudden change in operating conditions, and the system is in a disturbance response state. From 30 to 60 seconds, the multi-objective optimization algorithm takes effect, and the wire diameter deviation rapidly drops below 0.08 mm, while the temperature deviation simultaneously falls below 3°C. Both curves show an oscillating decay trend, indicating that the control algorithm is gradually correcting the process parameters and converging towards the target state. From 60 to 100 seconds, the wire diameter deviation stabilizes in the range of 0.02 to 0.07 mm, and the temperature deviation stabilizes in the range of 2 to 3.5°C, both remaining within the allowable thresholds; the fluctuation amplitude narrows significantly, and the system enters a stable operating state, verifying the multi-objective optimization algorithm's ability to coordinate the control of two parallel objectives.

[0033] In one embodiment of the present invention, the wire drawing process parameter adjustment amount output by the optimization solution module is sent to the instruction conversion module. The instruction conversion module pre-stores an equipment instruction mapping table, which defines in a structured form the digital mapping relationship between three types of parameters—the die compression ratio adjustment amount, the wire drawing fluid flow rate adjustment amount, and the main motor torque distribution adjustment amount—and the specific actuator actions of the wire drawing machine. In a specific implementation, the instruction conversion module, according to the equipment instruction mapping table, maps the received die compression ratio adjustment amount into a pulse quantity instruction controlling the movement of the die base servo motor. The pulse quantity instruction determines the angle the servo motor needs to rotate, thereby precisely adjusting the assembly gap of the wire drawing die. The instruction conversion module, according to the equipment instruction mapping table, maps the received wire drawing fluid flow rate adjustment amount into a frequency adjustment instruction controlling the output frequency of the wire drawing fluid circulation pump. The output frequency adjustment instruction adjusts the flow rate of the wire drawing fluid in the pipeline by changing the speed of the circulation pump motor. The instruction conversion module maps the received torque distribution adjustment amount of the main motor into speed ratio instructions issued to multiple drive main motors according to the device instruction mapping table. The speed ratio instructions coordinate the output of each main motor to achieve torque redistribution.

[0034] In practice, the generated pulse quantity command, output frequency adjustment command, and speed ratio command are synchronously sent to the servo drive, frequency converter, and multi-motor controller of the wire drawing machine. When the actuator performs the die gap adjustment action, the drive module monitors the real-time load current feedback value of the servo motor in real time. The real-time load current feedback value reflects the resistance encountered by the servo motor when driving the die base to move. If the real-time load current feedback value of the servo motor continuously exceeds the rated current threshold marked on the servo motor nameplate, the system determines that there is foreign object jamming or mechanical interference during the die movement. Once foreign object jamming is determined, the system immediately sends an emergency stop signal to the servo drive to pause the current die gap adjustment action. At the same time, an alarm signal with audible and visual prompts is generated on the operator's human-machine interface, and the die compression ratio parameter is locked at the current value in the control logic, awaiting on-site inspection and manual intervention by the operator.

[0035] After the actuator completes an adjustment, the system initiates a closed-loop verification step. This step continuously collects new wire drawing process datasets after the adjustment, focusing on the adjusted wire diameter and die temperature data. In practice, the closed-loop verification step calculates the new deviation between the adjusted wire diameter and the optimal wire diameter control benchmark provided by the state inversion module, and simultaneously calculates the new temperature difference between the adjusted die temperature and the expected target temperature in the temperature comparison module. If both the calculated new deviation and temperature difference are within the system's preset allowable error range, the logic judgment unit determines that the intelligent control closed loop is complete. If either the new deviation or temperature difference exceeds the allowable error range, the logic judgment unit packages the newly collected wire drawing process dataset along with the current new deviation and temperature difference information, sending this as a feedback data packet back to the physical constraint model's input interface. Upon receiving the feedback data packet, the physical constraint model triggers a new round of parameter optimization calculations based on the latest data.

[0036] In some embodiments, the device instruction mapping table is stored in the non-volatile memory of the control system in the form of a two-dimensional data table, supporting online editing and calibration via host computer software. It can be understood that the mapping relationship between the pulse quantity instruction and the compression ratio adjustment depends on the number of pulses per revolution of the servo motor, the lead screw, and the mechanical transmission ratio of the mold base. The mapping relationship between the output frequency adjustment instruction and the flow rate adjustment needs to be calibrated according to the flow-frequency characteristic curve of the centrifugal pump. In addition to audible and visual prompts, the alarm signal includes specific fault codes and suggested handling measures, and is sent to the workshop management server via the industrial network. In specific implementations, the allowable error range is set independently for the new deviation value of the wire diameter and the new temperature difference value, with the allowable error range for the wire diameter typically being more stringent than that for the temperature. The condition for triggering a new round of parameter optimization calculation can be set to the new deviation value or the new temperature difference value exceeding the allowable error range for three consecutive sampling periods to avoid false triggering due to noise from a single measurement.

[0037] The condition for determining whether control is complete in the closed-loop verification step can be represented by the following logical expression: in: This represents the calculated new deviation value, which is the difference between the adjusted wire diameter and the optimal wire diameter control reference center value. This represents the positive threshold value for the preset allowable error range of the wire diameter. This represents the calculated new temperature difference value, that is, the difference between the adjusted mold temperature and the desired target temperature. This represents the positive threshold of the preset temperature tolerance range, and the symbol "&" represents the logical "AND" operation. In practice, the logical expression only outputs "true" when both the absolute value of the new wire diameter deviation and the absolute value of the new temperature difference are less than the temperature tolerance range, thus determining that the current control loop is complete.

[0038] See Figure 5This is a graph showing the abnormal load current monitoring and alarm of the servo motor in a wire drawing die, intuitively demonstrating the execution effect of the servo motor load current abnormality detection and alarm logic. From 0 to 30 seconds, the load current stabilizes in the 4.3-5.8A range with no significant fluctuations, and the servo motor drives the die gap adjustment smoothly; the alarm signal remains normal, and the system is fault-free. From 30 to 40 seconds, the load current suddenly surges to 8.5A, far exceeding the rated current threshold of 7A, triggering the die foreign object jamming detection; the alarm signal immediately switches to alarm status, and the system pauses the die gap adjustment action, locking the current compression ratio; the pink shaded area in the graph represents the alarm activation window, visually marking the occurrence and duration of the abnormality. From 40 to 100 seconds, the abnormal load current drops back to the 4-6A range, gradually returning to normal levels; the alarm signal returns to normal, and the system resumes normal control flow after manual intervention; subsequent load current fluctuations are consistent with the initial stage, verifying that the fault has been eliminated.

[0039] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A high-performance intelligent control method for copper wire drawing based on multi-source data, characterized in that, The method includes: Collect multi-source data during the operation of the wire drawing machine, preprocess the multi-source data, and generate a wire drawing process dataset; The drawing process dataset is input into a physical constraint model based on the process mechanism, which describes the dynamic balance between wire diameter reduction rate, die temperature rise and drum speed. Using the physical constraint model, the current wire drawing process is inverted to calculate the theoretically optimal wire diameter control benchmark and mold cooling requirements. The calculated optimal wire diameter control benchmark and mold cooling requirements are compared with the real-time wire diameter measurement value and actual mold temperature value directly extracted from the multi-source data to generate wire diameter deviation signal and temperature deviation signal. Based on the wire diameter deviation signal and temperature deviation signal, a set of optimal wire drawing process parameter adjustments are obtained in a multi-objective optimization algorithm. The wire drawing process parameters include the die compression ratio, the flow rate of the wire drawing fluid, and the torque distribution of the main motor. The calculated wire drawing process parameter adjustment amount is converted into action commands that the equipment can execute and sent to the actuator of the wire drawing machine; The actuator synchronously adjusts the assembly gap of the wire drawing die, the wire drawing fluid circulation system, and the main transmission system according to the action command, so that the wire drawing machine enters a new operating condition.

2. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 1, characterized in that, The multi-source data is preprocessed to generate a wire drawing process dataset, including: By using sensor groups arranged at the inlet and outlet of the wire drawing machine, multi-source data during the operation of the wire drawing machine are collected synchronously. The multi-source data includes the wire diameter data of the copper wire, the temperature data of the wire drawing die, and the rotation speed data of the wire drawing drum. The multi-source data is preprocessed to remove abnormal fluctuation values ​​and to interpolate and complete missing time-slice data to generate a continuous and stable wire drawing process dataset, specifically including: A sliding time window is applied to the wire diameter data, and the moving average and standard deviation of the data within the window are calculated; When the wire diameter data of a certain time slice deviates from the moving average by more than three times the standard deviation, the wire diameter data point corresponding to the current time slice is marked as an instantaneous interference point and removed. For the missing time series after removal, a linear interpolation method based on the changing trend of adjacent data points is used to fill them; The temperature data of the mold and the rotation speed data of the drum are also processed for noise smoothing using the sliding time window method, while preserving the low-frequency trend components of the data. The wire diameter data, die temperature data, and drum rotation speed data, after anomaly removal, interpolation completion, and noise smoothing, are timestamped and integrated to generate the continuous and stable wire drawing process dataset.

3. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 2, characterized in that, The drawing process dataset is input into a physical constraint model built based on the process mechanism, including: In the physical constraint model, a differential equation is defined regarding the change in the cross-sectional area of ​​the copper wire. The input variable of the differential equation is the rotational speed data of the drum, and the output variable is the instantaneous speed of the copper wire. The temperature data of the mold is used as a boundary condition in the differential equation to correct the plastic deformation coefficient of the copper wire at high temperature. Based on the instantaneous velocity and plastic deformation coefficient, the theoretical mold compression angle for maintaining a constant copper wire cross-section under the current working conditions can be deduced. The theoretical mold compression angle is used as the core intermediate variable in the output of the physical constraint model.

4. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 3, characterized in that, Using the physical constraint model, the current wire drawing process is inverted to calculate the theoretically optimal wire diameter control benchmark and die cooling requirements, including: Read the theoretical mold compression angle output by the physical constraint model, and combine it with the preset product specification tolerance zone to calculate the optimal wire diameter control benchmark within an allowable floating range; By retrieving the temperature data of the mold over time and analyzing its upward trend, and combining this with the thermal conductivity of the mold material, the required heat dissipation power to bring the mold temperature back to its normal operating range can be calculated. The heat dissipation power is converted into the mold cooling requirement, which is specifically reflected in the theoretical increase in the flow rate of the drawing fluid.

5. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 4, characterized in that, The calculated optimal wire diameter control benchmark and mold cooling requirements are compared with the real-time wire diameter measurement value and actual mold temperature value directly extracted from the multi-source data to generate wire diameter deviation signals and temperature deviation signals, including: Establish a closed-loop comparator to calculate the difference between the real-time wire diameter measurement value and the optimal wire diameter control benchmark to obtain the wire diameter error value. The wire diameter error value is subjected to a first-order low-pass filter to remove high-frequency glitches caused by mechanical vibration, thereby obtaining the wire diameter deviation signal. The actual temperature value of the mold is compared with the target temperature value corresponding to the theoretical increase in the flow rate of the drawing fluid, and the temperature error value is calculated. The temperature error value is amplified proportionally to generate the temperature deviation signal used to drive the cooling system.

6. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 5, characterized in that, Based on the wire diameter deviation signal and temperature deviation signal, a set of optimal wire drawing process parameter adjustments is obtained in a multi-objective optimization algorithm, including: Initialize a feasible solution space that includes the compression ratio of the mold, the flow rate of the drawing fluid, and the torque distribution of the main motor; In the multi-objective optimization algorithm, minimizing the absolute value of the wire diameter deviation signal and minimizing the absolute value of the temperature deviation signal are taken as two parallel optimization objectives. The dynamic equilibrium relationship defined by the physical constraint model is introduced as a constraint condition to prevent the solution in the solution space from causing instability in the wire drawing process; Through iterative optimization, a Pareto optimal solution is selected from the feasible solution space, and the parameter change corresponding to the Pareto optimal solution is used as the adjustment amount of the wire drawing process parameters.

7. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 6, characterized in that, The specific execution process of the multi-objective optimization algorithm includes: A non-dominated sorting genetic algorithm is used as the core of the multi-objective optimization algorithm to encode individuals in the feasible solution space; In each generation of evolution, non-dominated ranking is performed based on the individual’s performance in wire diameter control and temperature control, and crowding calculation is introduced to maintain population diversity. When the variation of the optimal solution in the target space for several consecutive generations is less than a set threshold, the iteration terminates and the current Pareto optimal solution is output.

8. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 7, characterized in that, The calculated wire drawing process parameter adjustment amounts are converted into action commands that the equipment can execute, including: Establish a device instruction mapping table, which defines the correspondence between wire drawing process parameters and actual actuator actions; The compression ratio adjustment of the mold is mapped to the number of pulse commands that drive the servo motor to move the mold base. The flow rate adjustment of the drawing fluid is mapped to the adjustment command of the inverter output frequency; The torque distribution adjustment of the main motor is mapped to the speed ratio command between multiple drive motors.

9. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 8, characterized in that, The process of adjusting the assembly clearance of the wire drawing die by the actuator also includes a dynamic adjustment step: Real-time monitoring of the load current feedback value of the servo motor when executing pulse number commands; If the load current feedback value exceeds the rated current threshold of the motor, it is determined that there is a foreign object stuck in the mold; Pause the current mold gap adjustment action and generate an alarm signal. At the same time, lock the mold compression ratio at the current value and wait for manual intervention.

10. The high-performance copper wire drawing intelligent control method based on multi-source data as described in claim 9, characterized in that, After the actuator completes an adjustment, a closed-loop verification step is also included: Continuously collect and adjust the wire drawing process dataset, focusing on the adjusted wire diameter data and die temperature data; Calculate the new deviation between the adjusted wire diameter data and the control baseline, and the temperature difference between the adjusted mold temperature data and the target temperature; If both the new deviation value and the temperature difference value are within the allowable error range, then the intelligent control closed loop is considered complete. If any value exceeds the error range, the adjusted wire drawing process dataset, along with the deviation information, will be fed back to the physical constraint model to trigger a new round of parameter optimization calculations.

Images