A cable annealing control method and system based on parameter optimization

Abstract

The application discloses a cable annealing control method and system based on parameter optimization, and belongs to the technical field of cable annealing control; the method comprises the following steps: S1, deploying sensors in an annealing device to obtain original parameter data, constructing a control variable tensor and a grain uniformity index, and simultaneously constructing an energy consumption function to obtain effective annealing energy consumption; S2, constructing a harmonic target function and a tensor update amount, introducing disturbance and inertia propulsion items for updating, obtaining actual control instructions to realize cable annealing control; a multivariable control variable tensor is constructed, differential modeling of the evolution process of the grain uniformity is carried out, the lagging influence of multi-period parameters on the microstructure is sensed in real time, the system can accurately predict the uniformity of the structure based on the current control state, the grain consistency control capability is significantly improved, and the conductivity and mechanical ductility of the conductor are further enhanced.

Description

Technical Field

[0001] This invention relates to a cable annealing control method and system based on parameter optimization, which belongs to the field of cable annealing control technology. Background Technology

[0002] Cable annealing, as a crucial step in the heat treatment process of conductor materials, directly impacts the metallurgical performance indicators of the product, such as resistivity, ductility, and grain size distribution. Currently, continuous annealing production lines for cables primarily achieve control by setting constant annealing current, temperature, and traction speed. These parameters are typically set by operators based on experience or pre-set process parameters, lacking a dynamic perception and response mechanism for real-time operating conditions. Traditional control systems often process parameters such as temperature, current, and time separately, ignoring their cross-coupling relationships in the physical process. This results in control strategies that cannot accurately adapt to the heat treatment characteristics of cables of different specifications. Furthermore, the recrystallization, growth, and uniformity formation of conductor grains during annealing exhibit significant time lag characteristics. Current systems struggle to capture the residual effects of previous cycles on the current grain evolution and find it difficult to deeply embed historical information into the control strategy. On the other hand, annealing involves high energy consumption, especially during prolonged high-current operation, where energy efficiency significantly decreases. However, existing control systems lack assessment and optimization models for energy consumption per unit length, making it difficult to effectively reduce energy consumption while ensuring performance. Furthermore, existing control optimization methods are mostly based on static parameter tuning or traditional PID control. They lack the ability to dynamically identify and adjust for problems such as furnace disturbances, power grid fluctuations, or unstable traction speeds that occur during the annealing process. The control behavior is prone to oscillations or response lags, which is not conducive to the stable control of annealing quality.

[0003] Existing technologies also have technical problems such as reliance on experience for adjusting annealing process parameters, inaccurate control of grain structure consistency, high energy consumption, lag in control response, and lack of adaptability of control strategies under different cable specifications. Summary of the Invention

[0004] To address the shortcomings of the prior art, this invention provides a cable annealing control method and system based on parameter optimization.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] A cable annealing control method based on parameter optimization includes the following steps:

[0007] S1. Deploy sensors in the annealing equipment to obtain the raw parameter data of the annealing equipment. After preprocessing, construct a control variable tensor. Based on the control variable tensor, construct a grain uniformity index. At the same time, construct the energy consumption function of the annealing process to obtain the effective annealing energy consumption.

[0008] S2. Construct a harmonic objective function based on the grain uniformity index and effective annealing energy consumption. Then, construct a tensor update quantity based on the harmonic objective function, combined with the grain uniformity index and effective annealing energy consumption. Simultaneously, introduce disturbance and inertial propulsion terms to update the control variable tensor, obtain the actual control command, and realize cable annealing control.

[0009] Furthermore, the raw parameter data in S1 includes furnace surface temperature data obtained from a temperature sensor, instantaneous annealing current data obtained from a current sensor, cable tension data obtained from a tension sensor, and moving speed data obtained from a linear velocity sensor, and undergoes preprocessing such as noise reduction, synchronization, standardization, and normalization.

[0010] Furthermore, in S1, a control variable tensor is constructed, specifically by... The preprocessed parameter data for each annealing control cycle is represented by a control variable tensor. The formula is as follows:

[0011] ;

[0012] in, The current sampling period The control variable tensor; Indicates the annealing control cycle number index; This is the total number of annealing control cycles; It is the index of time-series sampling points within a single period; It is the total number of sampling points in each period; It is in the The sampling period The instantaneous annealing current data after preprocessing at each sampling point represents the instantaneous current intensity of the heated conductor; It is in the The sampling period The preprocessed conductor surface temperature data from each sampling point represents the current thermal state of the cable surface. It is in the The sampling period The preprocessed moving speed data of each sampling point represents the cable running speed during the annealing process; It is in the The sampling period The preprocessed cable tension data from each sampling point represents the tension force of the cable during the traction process.

[0013] Furthermore, the formula for the grain uniformity index in S1 is:

[0014] ;

[0015] in, The current sampling period The grain uniformity index is used to characterize the current sampling period. The degree of structural uniformity in the distribution of conductor grains; It is a historical cycle weighting factor, expressing the first The hysteresis contribution of each sampling period to the current grain uniformity is determined using the exponentially weighted moving average method, with a reference range of [value missing]. ; It is to prevent small positive numbers from having a denominator of zero, such as ; This is the tension sensitivity coefficient, which describes the sensitivity of grains to tensile stress during annealing. It is determined through statistical analysis of microscopic images of conductor grain size distribution after annealing under different tension conditions. The reference value range is [insert range here]. ; It is in the The sampling period The normalized effective heating time for each sampling point is calculated from the speed and furnace length. It is the main driving factor for grain growth, representing the driving force for metal recrystallization under the combined effect of temperature and time; It is a positive enhancement term for cable movement speed, guiding the control system to maintain operation within the allowable cable speed, avoiding annealing runaway due to being too slow or too fast; π is the mathematical constant pi. This indicates the tension suppression term, which describes the suppressive effect of tension on grain uniformity; This indicates the residual grain evolution trajectory throughout the entire annealing process.

[0016] Furthermore, after obtaining the grain uniformity index Then, using the same tensor element as input, an energy consumption function is constructed, defined as the effective annealing energy consumption in S1, with the following formula:

[0017] ;

[0018] in, The current sampling period Effective annealing energy consumption; It is the normalized resistance of the conductor material, which is determined by looking up a table based on the conductor material. This is the normalized value corresponding to the optimal annealing temperature, the physically optimal annealing temperature, determined by differential scanning calorimetry. The reference range is... ; This is the temperature deviation penalty coefficient, which controls the degree of penalty for temperature deviation. It is determined through penalty term construction techniques in multi-objective control, and the reference value range is [range missing]. ; It is an estimated value of the electric heating energy consumption of the cable, reflecting the adjustment of energy density by speed control, and is the core item of speed-heat coordinated control in the annealing process; It represents the degree of deviation between the current temperature and the optimal annealing temperature; It is an exponential correction factor for energy efficiency caused by temperature deviation, simulating the physical process that energy consumption is fully utilized when the temperature is correct, and energy efficiency decreases when the temperature deviates. It is the comprehensive energy consumption contribution value, which is an energy consumption estimate that takes into account electrical, thermal and mechanical factors.

[0019] Furthermore, the harmonic objective function in S2 is formulated as follows:

[0020] ;

[0021] in, The current sampling period The harmonic objective function represents the overall system cost under the current annealing control process; The historical best grain uniformity reference value was determined using statistical analysis based on historical parameter data obtained from existing databases. The reference value range is [range missing]. ; The energy consumption benchmark for typical operating conditions is determined using the extreme value method based on historical production data obtained from existing databases, with a reference value range of [missing value]. ; This is the performance preference index coefficient, which determines the system's emphasis on grain control. It is obtained through Bayesian optimization, and the reference value range is [value range missing]. ; This is the energy tolerance coefficient, representing the system's sensitivity to energy consumption control. It is obtained through Bayesian optimization and has a reference value range of [value missing]. ; The cost of expressing a deviation from the optimal grain structure; The expression describes the penalty for increased energy consumption, indicating that the closer the annealing energy consumption is to the baseline, the lower the energy consumption, and the higher the energy consumption, the greater the penalty value.

[0022] Furthermore, using the harmonic objective function Grain uniformity index With effective annealing energy consumption The combined gradients are used to construct the tensor update in S2, as shown in the formula:

[0023] ;

[0024] in, The current sampling period Tensor update quantity; It is the master learning rate; It is the index of the control variable; It is the gradient of the overall harmonic objective with respect to each control variable, obtained through automatic differentiation; This is the inversion adjustment coefficient, which determines whether to significantly amplify the performance gradient to counteract the shift of the energy consumption term in the dominant direction of the objective function. The reference value range is... ; It is the partial derivative of the grain performance index with respect to each control variable, expressing the sensitivity of the grain structure to changes in the variables; This indicates the relative relationship between energy consumption and grain structure; It constitutes the asymmetric update factor, reflecting the relative ratio of current energy consumption to performance; The direction of gradient descent of the harmonic objective function with respect to the control variables is described; This is an asymmetric inversion term used to offset the gradient update deviation in grain quality control that may be caused by the energy consumption penalty in the first term.

[0025] Furthermore, to avoid local optima, disturbance and inertial propulsion terms are introduced in S2 to update the control variable tensor, as shown in the formula:

[0026] ;

[0027] in, The current sampling period The updated control variable tensor; The amplitude of the disturbance is determined by performing Fast Fourier Transform spectral analysis on historical disturbance data in the existing database. The reference range is as follows: ; It is the disturbance period, obtained based on existing external disturbance modeling and prediction, with a reference value range of [value missing]. ; This is the inertial propulsion coefficient, derived using a sliding window error change rate fitting method. The reference range is... ; It is a dynamic sine term, used to introduce controllable periodic disturbances to simulate furnace atmosphere disturbances; It is the inertial term, used to enhance the physical continuity of the control direction;

[0028] Finally, from the updated control variable tensor The first section of the current control cycle is extracted as the actual control command to be executed and output to the lower computer for real-time control of current regulation, furnace temperature regulation, speed execution and tension device set point, so as to realize cable annealing control.

[0029] A cable annealing control system based on parameter optimization, the system comprising a parameter data acquisition and preprocessing module, a control variable tensor construction module, a grain uniformity index generation module, an annealing energy consumption calculation module, a harmonic target module, a tensor update module, and a control command generation module;

[0030] The parameter data acquisition and preprocessing module acquires the raw parameter data of the annealing equipment, performs preprocessing to obtain the preprocessed parameter data, and sends it to the control variable tensor construction module.

[0031] The control variable tensor construction module combines the preprocessed parameter data with the preprocessed parameter data from historical sampling periods to construct a control variable tensor, which is then sent to the grain uniformity index generation module and the annealing energy consumption calculation module.

[0032] The grain uniformity index generation module constructs a grain uniformity index based on the control variable tensor and the cumulative effect of thermal history and electro-mechanical combined action, and sends it to the harmonization target module.

[0033] The annealing energy consumption calculation module constructs the energy consumption function of the annealing process based on the control variable tensor, obtains the effective annealing energy consumption, and sends it to the harmonization target module.

[0034] The harmonic objective module, based on the grain uniformity index and effective annealing energy consumption, constructs a harmonic objective function with the dual objectives of maximizing grain uniformity and minimizing energy consumption, and sends it to the tensor update module;

[0035] The tensor update module constructs the tensor update quantity based on the harmonic objective function, combined with the grain uniformity index and effective annealing energy consumption. At the same time, it introduces disturbance and inertial propulsion terms to update the control variable tensor and sends the updated control variable tensor to the control command generation module.

[0036] The control command generation module extracts the first section of the current control cycle from the updated control variable tensor as the actual control command to be executed, and outputs it to the lower computer for real-time control of current regulation, furnace temperature regulation, speed execution and tension device setpoint, so as to realize cable annealing control.

[0037] The beneficial effects of this invention are as follows: a control variable tensor integrating multiple variables such as current, temperature, speed, and tension is constructed. By differentially modeling the evolution process of grain uniformity, the hysteretic influence of multi-period parameters on microstructure is perceived in real time. The proposed grain uniformity function expresses the comprehensive driving and inhibiting terms of grain growth through weighted and nonlinear coupling of thermal history, electrical intensity, running speed, and mechanical tension. This enables the system to accurately predict the uniformity of the microstructure based on the current control state, significantly improving the grain consistency control capability, and thus enhancing the conductivity and mechanical ductility of the conductor.

[0038] By constructing an effective annealing energy consumption function based on resistance heating theory and a moving speed coupling mechanism, the nonlinear influence of temperature deviation on energy efficiency is considered. A feedback structure of temperature-power-speed three-dimensional regulation is introduced, enabling the control system to automatically approach the minimum energy consumption per unit length while ensuring quality. Compared with traditional constant setpoint operation, this system can penalize high-energy-consumption conditions according to dynamic changes in operating conditions, guiding the control strategy to converge towards the direction of minimum energy consumption, thereby improving the overall energy efficiency of the equipment and reducing energy costs.

[0039] The system incorporates a periodic sinusoidal disturbance term. By extracting the dominant frequency from the historical disturbance spectrum using fast Fourier transform, a periodic disturbance is constructed to simulate the impact of external factors such as furnace atmosphere and power supply stability on the control system, effectively avoiding optimization from getting trapped in local optima. An inertial propulsion term is extracted using the error sliding window backtracking method to establish a dynamic inertial structure between control variables, making the tensor update path smooth and directionally continuous, resulting in a more stable overall optimization path and stronger resistance to disturbances. Attached Figure Description

[0040] Figure 1 This is a structural diagram of a cable annealing control system based on parameter optimization as described in this invention;

[0041] Figure 2 This is a flowchart of a cable annealing control method based on parameter optimization according to the present invention. Detailed Implementation

[0042] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0043] A cable annealing control method based on parameter optimization includes the following steps:

[0044] S1. Obtain the raw parameter data of the annealing equipment, and after preprocessing, construct the control variable tensor. Based on the control variable tensor, construct the grain uniformity index and at the same time construct the energy consumption function of the annealing process to obtain the effective annealing energy consumption.

[0045] Sensors are deployed in the annealing equipment to acquire raw parameter data, including conductor surface temperature data from a temperature sensor, instantaneous annealing current data from a current sensor, cable tension data from a tension sensor, and moving speed data from a linear velocity sensor. This raw parameter data, after preprocessing such as noise reduction, synchronization, standardization, and normalization, is input into the cable annealing control system in time-series form. The preprocessing procedures are well-known to those skilled in the art and will not be elaborated upon here.

[0046] Will The preprocessed parameter data for each annealing control cycle is represented by a control variable tensor. This means, specifically:

[0047] ;

[0048] in, The current sampling period The control variable tensor; Indicates the annealing control cycle number index; This is the total number of annealing control cycles; It is the index of time-series sampling points within a single period; It is the total number of sampling points in each period; It is in the The sampling period The instantaneous annealing current data after preprocessing at each sampling point represents the instantaneous current intensity of the heated conductor; It is in the The sampling period The preprocessed conductor surface temperature data from each sampling point represents the current thermal state of the cable surface. It is in the The sampling period The preprocessed moving speed data of each sampling point represents the cable running speed during the annealing process; It is in the The sampling period The preprocessed cable tension data from each sampling point represents the tension force of the cable during the traction process.

[0049] The purpose of constructing the aforementioned control variable tensor is to capture the cross-cycle correlation between current, temperature, mechanical tension, and speed during the annealing process, enabling the cable annealing control system to perceive the hysteresis effect of past parameters on the current microstructure evolution.

[0050] Furthermore, since the grain uniformity of annealing is jointly determined by the multi-cycle temperature integral effect, the current-time coupling effect, and the velocity-tension coupling effect, the control variable tensor... This is mapped onto macroscopic performance indices that can explain grain structure evolution. To express the cumulative effect based on the combined effects of thermal history and electromechanical processes, a grain uniformity index is constructed as follows:

[0051] ;

[0052] in, The current sampling period The grain uniformity index is used to characterize the current sampling period. The degree of structural uniformity in the distribution of conductor grains; It is a historical cycle weighting factor, expressing the first The hysteresis contribution of each sampling period to the current grain uniformity is determined using the exponentially weighted moving average method, with a reference range of [value missing]. ; It is to prevent small positive numbers from having a denominator of zero, such as ; This is the tension sensitivity coefficient, which describes the sensitivity of grains to tensile stress during annealing. It is determined through statistical analysis of microscopic images of conductor grain size distribution after annealing under different tension conditions. The reference value range is [insert range here]. ; It is in the The sampling period The normalized effective heating time for each sampling point is calculated from the speed and furnace length. It is the main driving factor for grain growth, representing the driving force for metal recrystallization under the combined effect of temperature and time; It is a positive enhancement term for cable movement speed, guiding the control system to maintain operation within the allowable cable speed, avoiding annealing runaway due to being too slow or too fast; π is the mathematical constant pi. This indicates the tension suppression term, which describes the suppressive effect of tension on grain uniformity; This indicates the residual grain evolution trajectory throughout the entire annealing process;

[0053] The above formula makes grain uniformity a continuously differentiable function that can be optimized. Its physical meaning is that higher temperatures, longer holding times, and lower speeds make it easier for grains to become uniform; while excessively high current can exacerbate the surface-to-internal temperature difference, thereby reducing grain uniformity.

[0054] After obtaining the grain uniformity index Then, using the same tensor element as input, an energy consumption function is constructed, defined as the effective annealing energy consumption per unit length:

[0055] ;

[0056] in, The current sampling period Effective annealing energy consumption; It is the normalized resistance of the conductor material, which is determined by looking up a table based on the conductor material. This is the normalized value corresponding to the optimal annealing temperature, the physically optimal annealing temperature, determined by differential scanning calorimetry. The reference range is... ; This is the temperature deviation penalty coefficient, which controls the degree of penalty for temperature deviation. It is determined through penalty term construction techniques in multi-objective control, and the reference value range is [range missing]. ; It is an estimated value of the electric heating energy consumption of the cable, reflecting the adjustment of energy density by speed control, and is the core item of speed-heat coordinated control in the annealing process; It represents the degree of deviation between the current temperature and the optimal annealing temperature; It is an exponential correction factor for energy efficiency caused by temperature deviation, simulating the physical process that energy consumption is fully utilized when the temperature is correct, and energy efficiency decreases when the temperature deviates. It is the comprehensive energy consumption contribution value, which is an energy consumption estimate that takes into account electrical, thermal and mechanical factors.

[0057] S2. Based on the grain uniformity index and effective annealing energy consumption, a harmonic objective function is constructed. Then, based on the harmonic objective function, combined with the grain uniformity index and effective annealing energy consumption, a tensor update quantity is constructed. At the same time, disturbance and inertial propulsion terms are introduced to update the control variable tensor, obtain the actual control command, and realize cable annealing control.

[0058] Furthermore, in order to optimize the balance between quality and energy consumption, the aforementioned two outputs... and The input is given to a harmonic objective function to generate a solvable control objective, constructed as follows:

[0059] ;

[0060] in, The current sampling period The harmonic objective function represents the overall system cost under the current annealing control process; The historical best grain uniformity reference value was determined using statistical analysis based on historical parameter data obtained from existing databases. The reference value range is [range missing]. ; The energy consumption benchmark for typical operating conditions is determined using the extreme value method based on historical production data obtained from existing databases, with a reference value range of [missing value]. ; This is the performance preference index coefficient, which determines the system's emphasis on grain control. It is obtained through Bayesian optimization, and the reference value range is [value range missing]. ; This is the energy tolerance coefficient, representing the system's sensitivity to energy consumption control. It is obtained through Bayesian optimization and has a reference value range of [value missing]. . The cost of expressing a deviation from the optimal grain structure; The expression describes the penalty for increased energy consumption, indicating that the closer the annealing energy consumption is to the baseline, the lower the energy consumption, and the higher the energy consumption, the greater the penalty value.

[0061] To ensure the harmonic objective function truly drives control input updates, gradient update variables based on asynchronous inversion analysis are constructed. This is then utilized using the harmonic objective function. Grain uniformity index With effective annealing energy consumption Combinatorial gradient construction of tensor update:

[0062] ;

[0063] in, The current sampling period Tensor update quantity; It is the master learning rate; It is the index of the control variable; It is the gradient of the overall harmonization target (grain performance + energy consumption) with respect to each control variable, obtained through automatic differentiation; This is the inversion adjustment coefficient, which determines whether to significantly amplify the performance gradient to counteract the shift of the energy consumption term in the dominant direction of the objective function. The reference value range is... ; It is the partial derivative of the grain performance index with respect to each control variable, expressing the sensitivity of the grain structure to changes in the variables; This indicates the relative relationship between energy consumption and grain structure; It constitutes the asymmetric update factor, reflecting the current relative ratio of energy consumption to performance; It describes the direction of gradient descent of the harmonic objective function with respect to the control variables; This is an asymmetric inversion term used to offset the gradient update deviation in grain quality control that may be caused by the energy consumption penalty in the first term.

[0064] Based on the above update amount, the control variable tensor To update the algorithm and avoid local optima, perturbations and inertial propulsion terms are introduced, resulting in:

[0065] ;

[0066] in, The current sampling period The updated control variable tensor; The amplitude of the disturbance is determined by performing Fast Fourier Transform spectral analysis on historical disturbance data in the existing database. The reference range is as follows: ; It is the disturbance period, obtained based on existing external disturbance modeling and prediction, with a reference value range of [value missing]. ; This is the inertial propulsion coefficient, derived using a sliding window error change rate fitting method. The reference range is... ; It is a dynamic sine term, used to introduce controllable periodic disturbances to simulate furnace atmosphere disturbances; It is the inertial term, used to enhance the physical continuity of the control direction.

[0067] Finally, from the new tensor The first section of the current control cycle is extracted as the actual control command to be executed and output to the lower computer for real-time control of current regulation, furnace temperature regulation, speed execution and tension device setpoint, so as to realize full-process, dynamic and linkage control.

[0068] A cable annealing control system based on parameter optimization, the system comprising a parameter data acquisition and preprocessing module, a control variable tensor construction module, a grain uniformity index generation module, an annealing energy consumption calculation module, a harmonic target module, a tensor update module, and a control command generation module;

[0069] The parameter data acquisition and preprocessing module acquires the raw parameter data of the annealing equipment, performs preprocessing to obtain the preprocessed parameter data, and sends it to the control variable tensor construction module.

[0070] The control variable tensor construction module combines the preprocessed parameter data with the preprocessed parameter data from historical sampling periods to construct a control variable tensor, which is then sent to the grain uniformity index generation module and the annealing energy consumption calculation module.

[0071] The grain uniformity index generation module constructs a grain uniformity index based on the control variable tensor and the cumulative effect of thermal history and electro-mechanical combined action, and sends it to the harmonization target module.

[0072] The annealing energy consumption calculation module constructs the energy consumption function of the annealing process based on the control variable tensor, obtains the effective annealing energy consumption, and sends it to the harmonization target module.

[0073] The harmonic objective module, based on the grain uniformity index and effective annealing energy consumption, constructs a harmonic objective function with the dual objectives of maximizing grain uniformity and minimizing energy consumption, and sends it to the tensor update module;

[0074] The tensor update module constructs the tensor update quantity based on the harmonic objective function, combined with the grain uniformity index and effective annealing energy consumption. At the same time, it introduces disturbance and inertial propulsion terms to update the control variable tensor and sends the updated control variable tensor to the control command generation module.

[0075] The control command generation module extracts the first section of the current control cycle from the updated control variable tensor as the actual control command to be executed, and outputs it to the lower computer for real-time control of current regulation, furnace temperature regulation, speed execution and tension device setpoint, so as to realize cable annealing control.

[0076] A control variable tensor integrating multiple variables such as current, temperature, velocity, and tension was constructed. By differentially modeling the evolution process of grain uniformity, the hysteretic effects of multi-period parameters on microstructure were perceived in real time. The proposed grain uniformity function expresses the comprehensive driving and inhibiting terms of grain growth through weighted and nonlinear coupling of thermal history, electrical intensity, running speed, and mechanical tension. This enables the system to accurately predict microstructure uniformity based on the current control state, significantly improving the grain consistency control capability, and thus enhancing the conductivity and mechanical ductility of the conductor.

[0077] By constructing an effective annealing energy consumption function based on resistance heating theory and a moving speed coupling mechanism, the nonlinear influence of temperature deviation on energy efficiency is considered. A feedback structure of temperature-power-speed three-dimensional regulation is introduced, enabling the control system to automatically approach the minimum energy consumption per unit length while ensuring quality. Compared with traditional constant setpoint operation, this system can penalize high-energy-consumption conditions according to dynamic changes in operating conditions, guiding the control strategy to converge towards the direction of minimum energy consumption, thereby improving the overall energy efficiency of the equipment and reducing energy costs.

[0078] The system incorporates a periodic sinusoidal disturbance term. By extracting the dominant frequency from the historical disturbance spectrum using fast Fourier transform, a periodic disturbance is constructed to simulate the impact of external factors such as furnace atmosphere and power supply stability on the control system, effectively avoiding optimization from getting trapped in local optima. An inertial propulsion term is extracted using the error sliding window backtracking method to establish a dynamic inertial structure between control variables, making the tensor update path smooth and directionally continuous, resulting in a more stable overall optimization path and stronger resistance to disturbances.

[0079] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cable annealing control method based on parameter optimization, characterized in that, Includes the following steps: S1. Deploy sensors in the annealing equipment to obtain the raw parameter data of the annealing equipment. After preprocessing, construct a control variable tensor. Based on the control variable tensor, construct a grain uniformity index. At the same time, construct the energy consumption function of the annealing process to obtain the effective annealing energy consumption. The construction of the control variable tensor specifically involves... The preprocessed parameter data for each annealing control cycle is represented by a control variable tensor. The formula is as follows: ； in, The current sampling period The control variable tensor; Indicates the annealing control cycle number index; This is the total number of annealing control cycles; It is the index of time-series sampling points within a single period; It is the total number of sampling points in each period; It is in the The sampling period The instantaneous annealing current data after preprocessing at each sampling point represents the instantaneous current intensity of the heated conductor; It is in the The sampling period The preprocessed conductor surface temperature data from each sampling point represents the current thermal state of the cable surface. It is in the The sampling period The preprocessed moving speed data of each sampling point represents the cable running speed during the annealing process; It is in the The sampling period The preprocessed cable tension data from each sampling point represents the tension force of the cable during the traction process; S2. Construct a harmonic objective function based on grain uniformity index and effective annealing energy consumption. Then, construct a tensor update quantity based on the harmonic objective function, combined with grain uniformity index and effective annealing energy consumption. Then, introduce disturbance and inertial propulsion terms to update the control variable tensor, obtain the actual control command, and realize cable annealing control. The introduced disturbance and inertial propulsion terms update the control variable tensor, as shown in the formula: ； in, The current sampling period The updated control variable tensor; The current sampling period Tensor update quantity; The amplitude of the disturbance is determined by performing Fast Fourier Transform spectral analysis on historical disturbance data in the existing database. The reference range is as follows: ; It is the disturbance period, obtained based on existing external disturbance modeling and prediction, with a reference value range of [value missing]. ; This is the inertial propulsion coefficient, derived using a sliding window error change rate fitting method. The reference range is... ; It is a dynamic sine term, used to introduce controllable periodic disturbances to simulate furnace atmosphere disturbances; It is the inertial term, used to enhance the physical continuity of the control direction; Finally, from the updated control variable tensor The first segment of the current control cycle is extracted as the actual control command to be executed and output to the lower-level machine for real-time control of current adjustment, furnace temperature adjustment, speed execution, and tension device setpoint, thereby realizing cable annealing control; the first segment refers to the current control cycle. Below, the control variable vector closest to the current moment.

2. The cable annealing control method based on parameter optimization according to claim 1, characterized in that: The raw parameter data in S1 includes furnace surface temperature data from a temperature sensor, instantaneous annealing current data from a current sensor, cable tension data from a tension sensor, and moving speed data from a linear velocity sensor, and has undergone preprocessing such as noise reduction, synchronization, standardization, and normalization.

3. The cable annealing control method based on parameter optimization according to claim 2, characterized in that, The formula for the grain uniformity index in S1 is: ； in, The current sampling period The grain uniformity index is used to characterize the current sampling period. The degree of structural uniformity in the distribution of conductor grains; It is a historical cycle weighting factor, expressing the first The hysteresis contribution of each sampling period to the current grain uniformity is determined using the exponentially weighted moving average method, with a reference range of [value missing]. ; It is to prevent tiny positive numbers from having a denominator of zero. ; This is the tension sensitivity coefficient, which describes the sensitivity of grains to tensile stress during annealing. It is determined through statistical analysis of microscopic images of conductor grain size distribution after annealing under different tension conditions. The reference value range is [insert range here]. ; It is in the The sampling period The normalized effective heating time for each sampling point is calculated from the speed and furnace length. It is the main driving factor for grain growth, representing the driving force for metal recrystallization under the combined effect of temperature and time; It is a positive enhancement to the cable movement speed, guiding the control system to maintain operation within the allowable cable speed and avoiding annealing runaway due to being too slow or too fast; This indicates the tension suppression term, which describes the suppressive effect of tension on grain uniformity; This indicates the residual grain evolution trajectory throughout the entire annealing process.

4. The cable annealing control method based on parameter optimization according to claim 3, characterized in that, After obtaining the grain uniformity index Then, using the same tensor element as input, an energy consumption function is constructed, defined as the effective annealing energy consumption in S1, with the following formula: ； in, The current sampling period Effective annealing energy consumption; It is the normalized resistance of the conductor material, which is determined by looking up a table based on the conductor material. This is the normalized value corresponding to the optimal annealing temperature, the physically optimal annealing temperature, determined by differential scanning calorimetry. The reference range is... ; This is the temperature deviation penalty coefficient, which controls the degree of penalty for temperature deviation. It is determined through penalty term construction techniques in multi-objective control, and the reference value range is [range missing]. ; It is an estimated value of the electric heating energy consumption of the cable, reflecting the adjustment of energy density by speed control, and is the core item of speed-heat coordinated control in the annealing process; It represents the degree of deviation between the current temperature and the optimal annealing temperature; It is an exponential correction factor for energy efficiency caused by temperature deviation, simulating the physical process that energy consumption is fully utilized when the temperature is correct, and energy efficiency decreases when the temperature deviates. It is the comprehensive energy consumption contribution value, which is an energy consumption estimate that takes into account electrical, thermal and mechanical factors.

5. The cable annealing control method based on parameter optimization according to claim 4, characterized in that, The harmonic objective function in S2 is formulated as follows: ； in, The current sampling period The harmonic objective function represents the overall system cost under the current annealing control process; The historical best grain uniformity reference value was determined using statistical analysis based on historical parameter data obtained from existing databases. The reference value range is [range missing]. ; The energy consumption benchmark for typical operating conditions is determined using the extreme value method based on historical production data obtained from existing databases, with a reference value range of [missing value]. ; This is the performance preference index coefficient, which determines the system's emphasis on grain control. It is obtained through Bayesian optimization, and the reference value range is [value range missing]. ; This is the energy tolerance coefficient, representing the system's sensitivity to energy consumption control. It is obtained through Bayesian optimization and has a reference value range of [value missing]. ; The cost of expressing a deviation from the optimal grain structure; The expression describes the penalty for increased energy consumption, indicating that the closer the annealing energy consumption is to the baseline, the lower the energy consumption, and the higher the energy consumption, the greater the penalty value.

6. The cable annealing control method based on parameter optimization according to claim 5, characterized in that, Using the harmonic objective function Grain uniformity index With effective annealing energy consumption The combined gradients are used to construct the tensor update in S2, as shown in the formula: ； in, The current sampling period Tensor update quantity; It is the master learning rate; It is the index of the control variable; It is the gradient of the overall harmonic objective with respect to each control variable, obtained through automatic differentiation; This is the inversion adjustment coefficient, which determines whether to significantly amplify the performance gradient to counteract the shift of the energy consumption term in the dominant direction of the objective function. The reference value range is... ; It is the partial derivative of the grain performance index with respect to each control variable, expressing the sensitivity of the grain structure to changes in the variables; This indicates the relative relationship between energy consumption and grain structure; It constitutes the asymmetric update factor, reflecting the relative ratio of current energy consumption to performance; The direction of gradient descent of the harmonic objective function with respect to the control variables is described; This is an asymmetric inversion term used to offset the gradient update deviation in grain quality control that may be caused by the energy consumption penalty in the first term.

7. A cable annealing control system based on parameter optimization, characterized in that, The control method described in any one of claims 1-6 is applied.

Images