A constant-tension constant-velocity cooperative regulation-based ultra-fine tungsten wire drawing control system and method

Abstract

This invention discloses a control system and method for ultra-fine tungsten wire drawing based on constant tension and constant speed coordinated adjustment, relating to the field of metal wire preparation technology. The method collects tension change data, speed change data, wire diameter change data at the unwinding end, wire diameter change data at the take-up end, short-range temperature change data before and after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data from the adjacent previous pass during the final drawing stage of the target ultra-fine tungsten wire, forming an input dataset Inp for the final drawing stage. The central processing system inputs the input dataset Inp into the final drawing implicit stability reconstruction model Mdl to reconstruct the decay trajectory of the continuous synchronous steady state during the final drawing stage, forming the implicit stability state result Hid for the tail section. Then, based on the implicit stability state result Hid, a release stability switching result Rel and a tension speed relocking result Lck are generated, and short-time release stability switching and tension speed relocking control are executed during the final drawing stage.

Description

Technical Field

[0001] This invention relates to the field of metal wire preparation technology, specifically to a control system and method for drawing ultra-fine tungsten wire based on the coordinated adjustment of constant tension and constant speed. Background Technology

[0002] With the increasing demand for high-strength, fine-diameter, and highly consistent wires in fields such as diamond wire cutting, precision electronic devices, and micro-conductor processing, the continuous multi-pass drawing technology for ultra-fine tungsten wires has gradually evolved from traditional single-machine speed regulation and single-point tension control to a collaborative control technology that integrates tension acquisition, speed acquisition, roll diameter detection, temperature detection, and online diameter measurement.

[0003] Currently, existing technologies for controlling the final drawing stage of ultra-fine tungsten wires are still not sufficiently aligned with actual working conditions. On the one hand, the release of excess coil at the unwinding end, the accumulation of coil at the take-up end, the short-range thermal state misalignment before and after the final drawing die, and the transmission of residual fluctuations from the previous pass are often not uniformly identified as a single scenario, resulting in a lack of dedicated data entry points for the final drawing stage. On the other hand, most existing solutions focus on real-time corrections based on the current tension, current speed, or current wire diameter, making it difficult to reconstruct the continuous synchronous steady-state attenuation process before surface parameters exceed limits. Furthermore, it is difficult to further transform the identification results into sequential control actions such as short-term release stabilization switching and tension speed relocking. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a control system and method for drawing ultra-fine tungsten wire based on constant tension and constant speed coordinated adjustment, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution, comprising the following steps: S1. Collect the tension change data, speed change data, wire diameter change data at the unwinding end, wire diameter change data at the take-up end, short-range temperature change data before the final drawing die, short-range temperature change data after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous pass of the target ultra-fine tungsten wire during the final drawing stage, and send them to the central processing system to form the input dataset Inp for the final drawing stage. S2. The central processing system inputs the final pull-off stage input dataset Inp into the pre-constructed and pre-trained final pull-off hidden stability reconstruction model Mdl. The final pull-off hidden stability reconstruction model Mdl reconstructs the decay trajectory of the continuous synchronous steady state in the final pull-off stage, and extracts the short-term release stabilization start point, short-term release stabilization duration interval and tension velocity relocking start condition from the decay trajectory to form the tail segment hidden stability state result Hid. S3. Based on the tail section hidden stability result Hid, at the short-term release stability start point, the central processing system first releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit in the order of the take-up drive unit and the release drive unit, and then releases the tension pull lock relationship between the release drive unit and the final pull traction unit, forming the release stability switching result Rel; when the tension speed relock start condition is met, the central processing system restores the speed lock relationship and tension lock relationship in the order of the final pull traction unit, the take-up drive unit and the release drive unit, forming the tension speed relock result Lck. S4. The central processing system outputs linkage control commands based on the release-stabilization switching result Rel and the tension speed relocking result Lck, and executes short-term release-stabilization switching and tension speed relocking control in the final tensioning stage to complete the coordinated adjustment of the final tensioning stage of the ultra-fine tungsten wire with the control objective of maintaining constant tension and constant speed.

[0006] Preferably, in S1, the central processing system performs final pull-off stage identification on the collected data; When the data on the change in the winding diameter at the unwinding end indicates that the unwinding end is in the state of releasing excess winding, and the data on the change in the winding diameter at the take-up end indicates that the take-up end is in the state of accumulating winding, and the data on the slight drift of the end wire diameter indicates that the end wire diameter maintains a slight deviation during the continuous sampling period, and when there is a staggered change relationship between the short-range temperature change data before the final drawing die and the short-range temperature change data after the final drawing die, and when the residual fluctuation data of the adjacent previous pass continues into the final drawing pass, the central processing system determines that the target ultra-fine tungsten wire has entered the final drawing and finishing stage. The system retains the tension change data, speed change data, wire diameter change data at the pay-off end, wire diameter change data at the take-up end, short-range temperature change data before and after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous draw, forming the input dataset Inp for the final drawing stage.

[0007] Preferably, in S2, the obtained final pull-off stage input dataset Inp is input into the final pull-off hidden stability reconstruction model Mdl; the final pull-off hidden stability reconstruction model Mdl is constructed through a gated temporal coding and decoding reconstruction framework; The final pull-in hidden stability reconstruction model Mdl includes a scene slice input layer, a roll diameter evolution coding layer, a short-range thermal state hysteresis coding layer, a tension velocity micro-drift coding layer, a previous pass residual wave propagation coding layer, a hidden stability coupling fusion layer, and an attenuation trajectory reconstruction layer. The scene slice input layer receives the input dataset Inp from the final pull-off stage. The roll diameter coupling evolution characteristic of the output feed end of the roll diameter evolution coding layer being in the state of excess roll release and the take-up end being in the state of roll accumulation; The misalignment characteristics between the short-range temperature change data before and after final die pulling in the output of the short-range thermal state hysteresis coding layer. The linkage offset characteristics between the tension change data, velocity change data and end wire diameter micro-drift data output by the tension micro-drift coding layer; The previous residual wave propagation coding layer outputs the propagation characteristics of the adjacent previous residual wave data to the final pull. The implicit stable coupling fusion layer performs alignment and fusion on the roll diameter coupling evolution characteristics, front and rear misalignment characteristics, linkage offset characteristics and transmission characteristics according to the same time sequence position of the final pull-off stage, forming the implicit stable coupling characteristics of the final pull-off stage. Based on the implicit stability coupling characteristics of the final pull stage, the attenuation trajectory reconstruction layer reconstructs the continuous synchronous steady-state attenuation trajectory of the final pull stage and forms the implicit stability state result Hid of the tail segment.

[0008] Preferably, in S2, the construction of the final pull-in hidden stability reconstruction model Mdl includes a pre-training phase and a calibration training phase; During the pre-training phase, based on historical operating data from a continuous multi-pass drawing production line for ultra-fine tungsten wire, steady-state maintenance segments, latent steady-state decay segments, and overt instability segments were segmented. Among them, the preceding running segment before any result of end wire diameter jump, wire breakage, or speed reduction control switching occurs within the subsequent preset time period is determined as the implicit stable attenuation segment; The running segment that has already shown any of the following results: end wire diameter exceeding tolerance, wire breakage, or control loss of synchronization is identified as an explicit unstable segment; The running segment that does not exhibit abnormal control switching and whose end wire diameter remains stable is identified as the steady-state maintenance segment. The steady-state maintenance segment, the implicitly stable decay segment, and the explicit unstable segment are used to perform self-supervised pre-training on the gated temporal coding and decoding reconstruction framework, so that the representation of the roll diameter evolution relationship, the representation of the preceding and following misalignment relationship, the representation of the linkage offset relationship, and the representation of the transmission relationship form a temporal correlation. During the calibration and training phase, based on the results of end wire diameter jumps, wire breakage, speed reduction control switching, and control step loss within a subsequent preset time period, the results of historical operation segments are back-calibrated to form steady-state maintenance segments, implicit stability decay segments, and explicit instability segments. Based on the steady-state maintenance segments, implicit stability decay segments, and explicit instability segments, the final tension implicit stability reconstruction model Mdl is calibrated and trained so that the output of the final tension implicit stability reconstruction model Mdl corresponds to the tail-segment implicit stability state result Hid of the control switching at the end of the final tension stage.

[0009] Preferably, in S2, the tail-end hidden stability result Hid includes at least the short-term release stability start time, the short-term release stability duration interval, the tension velocity relock start condition, and the continuous synchronous steady-state decay level. Among them, the short-term release stabilization start point corresponds to the decay trajectory of the latent stability coupling characteristics in the final pull-off stage, at the time of entering the release stabilization trigger zone; The short-term stabilization duration interval corresponds to the decay trajectory being within the stabilization maintenance region; The initial condition for Zhang Su relocking corresponds to the condition for the decay trajectory to exit the stabilization maintenance region and enter the relockable region. The continuous synchronous steady-state decay level indicates the degree of decay of the continuous synchronous steady state during the final pull phase.

[0010] Preferably, in S3, the central processing system generates the destabilization switching result Rel based on the tail-end hidden stability state result Hid; The release stabilization switching result Rel includes the take-up drive unit speed unlock command, the release drive unit tension unlock command, and the release rhythm of the short-term release stabilization duration range; The central processing system first sends a speed unlock command to the take-up drive unit, so that the take-up drive unit releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit. The central processing system then sends a tension unlocking command to the wire release drive unit, causing the wire release drive unit to release the tension locking relationship between the wire release drive unit and the final pull traction unit. During the short-term stabilization period, the central processing system maintains the short-term stabilization control mode according to the release rhythm of the short-term stabilization period.

[0011] Preferably, in S3, after the short-term destabilization period ends, the tension rate relocking result Lck is generated based on the tension rate relocking initiation condition; The tension relock result Lck includes the reference traction speed establishment command, the take-up drive unit speed relock command, the pay-off drive unit tension relock command, and the preset relock step size.

[0012] The central processing system first sends a reference traction speed establishment command to the final traction unit, so that the final traction unit can establish the reference traction speed; The central processing system then sends a speed relock command to the take-up drive unit, so that the take-up drive unit can restore the speed lock relationship between the take-up drive unit and the final pull traction unit. Finally, the central processing system issues a tension relock command to the wire release drive unit, so that the wire release drive unit restores the tension lock relationship between the wire release drive unit and the final pull traction unit. The central processing system completes the tension speed relock step by step according to the preset relock step size.

[0013] Preferably, in S4, after executing the release-stabilization switching result Rel and the tension speed relocking result Lck, the central processing system continues to collect the tension change data, the velocity change data, and the end wire diameter change data after execution. The central processing system generates a steady-state recovery result Rcv based on the tension change data, speed change data, and end wire diameter change data after execution. The steady-state recovery result Rcv includes at least the speed lock recovery state, the tension lock recovery state, and the end wire diameter stabilization state. When the steady-state recovery result Rcv does not meet the preset recovery conditions, the central processing system re-calls the destabilization switching result Rel and the tension speed relocking result Lck corresponding to the tail section hidden steady state result Hid, and continues to execute the short-term destabilization switching control and tension speed relocking control in the final tensioning stage.

[0014] Preferably, in S4, the central processing system pre-establishes the correspondence between the continuous synchronous steady-state decay level and the release rhythm of the short-term release stability duration interval, the reference traction speed establishment command, and the preset relock step size. Based on the continuous synchronous steady-state decay level, the central processing system calls the short-term release stabilization duration interval release rhythm, the reference traction speed establishment command, and the preset relock step size corresponding to the current continuous synchronous steady-state decay level in the corresponding relationship. The system then writes the short-term release stabilization duration interval release rhythm after the call into the stabilization switching result Rel, and writes the reference traction speed establishment command and the preset relock step size after the call into the tension speed relock result Lck.

[0015] A control system for drawing ultrafine tungsten wire based on constant tension and constant speed coordinated adjustment includes a finishing data acquisition module, a final drawing modeling module, a tension speed relocking module, and a coordinated adjustment module. The finishing data acquisition module collects data on tension changes, speed changes, wire diameter changes at the unwinding end and the take-up end of the target ultra-fine tungsten wire during the final drawing stage. It also collects short-range temperature changes before and after the final drawing die, micro-drift data of the wire diameter at the end, and residual fluctuation data of the adjacent previous pass. The data is then sent to the central processing system to form the final drawing stage input dataset Inp. The final pull modeling module inputs the final pull end stage input dataset Inp into the pre-built and pre-trained final pull hidden stability reconstruction model Mdl through the central processing system. The final pull hidden stability reconstruction model Mdl reconstructs the decay trajectory of the continuous synchronous steady state in the final pull end stage, and extracts the short-term release stabilization start point, short-term release stabilization duration interval and tension velocity relocking start condition from the decay trajectory to form the tail segment hidden stability state result Hid. The tension speed relocking module, based on the tail section hidden stability result Hid by the central processing system, at the short-term release stabilization start point, first releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit in the order of the take-up drive unit and the release drive unit, and then releases the tension counter-lock relationship between the release drive unit and the final pull traction unit, forming the release stabilization switching result Rel; when the tension speed relocking start condition is met, the central processing system restores the speed lock relationship and tension lock relationship in the order of the final pull traction unit, the take-up drive unit, and the release drive unit, forming the tension speed relocking result Lck. The coordinated adjustment module outputs linkage control commands through the central processing system based on the release stabilization switching result Rel and the tension speed relocking result Lck. It executes short-term release stabilization switching and tension speed relocking control in the final tensioning stage, and completes the coordinated adjustment of the final tensioning stage of the ultra-fine tungsten wire with the control objective of maintaining constant tension and constant speed.

[0016] This invention provides a control system and method for drawing ultra-fine tungsten wire based on the coordinated adjustment of constant tension and constant speed. It has the following beneficial effects: (1) This method introduces the latent stability reconstruction model Mdl in the final tension stage, and jointly reconstructs the changes in roll diameter, short-range temperature, tension speed, end wire diameter micro-drift and residual fluctuations of the previous pass in the input dataset Inp in the final tension stage to form the latent stability result Hid of the tail section. This solves the problem that the existing technology can only process the latent instability in the final tension stage after the tension, speed or wire diameter has become obviously abnormal, and it is difficult to identify the latent instability in the final tension stage in a timely manner.

[0017] (2) This method generates the release stabilization switching result Rel and the tension speed relocking result Lck based on the tail section hidden stability state result Hid, and releases the locking relationship in the order of take-up drive unit and release drive unit, and then restores the locking relationship in the order of final pull traction unit, take-up drive unit and release drive unit. This solves the problem that existing technologies often use synchronous adjustment or unified restoration methods, which are prone to secondary impact, tension rebound and speed disturbance in the final pull stage.

[0018] (3) This method directly uses the tail-end hidden stability state result Hid output by the final tension hidden stability reconstruction model Mdl to generate the release stability switching result Rel and the tension speed relock result Lck, and after execution, the steady-state recovery result Rcv is generated and called again. This solves the problems of disconnect between state recognition and control execution, lack of recovery confirmation after control, and insufficient stability maintenance capability in the final tension stage in the existing technology, thereby improving the wire breakage suppression capability, recovery efficiency and finished product consistency in the final tension stage of ultra-fine tungsten wire. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention; Figure 2This is a schematic diagram of the module of the present invention; Figure 3 This is a schematic diagram of the Mdl structure of the final tensile hidden stability reconstruction model of the present invention; Figure 4 This is a control chain reconfiguration diagram of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0021] Example 1 Please see Figure 1 , Figure 3 and Figure 4 This invention provides a method for controlling the drawing of extremely fine tungsten wire based on the coordinated adjustment of constant tension and constant speed. To achieve the above objectives, this invention is implemented through the following technical solution, including the following steps: S1. Collect the tension change data, speed change data, wire diameter change data at the unwinding end, wire diameter change data at the take-up end, short-range temperature change data before the final drawing die, short-range temperature change data after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous pass of the target ultra-fine tungsten wire during the final drawing stage, and send them to the central processing system to form the input dataset Inp for the final drawing stage. S2. The central processing system inputs the final pull-off stage input dataset Inp into the pre-constructed and pre-trained final pull-off hidden stability reconstruction model Mdl. The final pull-off hidden stability reconstruction model Mdl reconstructs the decay trajectory of the continuous synchronous steady state in the final pull-off stage, and extracts the short-term release stabilization start point, short-term release stabilization duration interval and tension velocity relocking start condition from the decay trajectory to form the tail segment hidden stability state result Hid. S3. Based on the tail section hidden stability result Hid, at the short-term release stability start point, the central processing system first releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit in the order of the take-up drive unit and the release drive unit, and then releases the tension pull lock relationship between the release drive unit and the final pull traction unit, forming the release stability switching result Rel; when the tension speed relock start condition is met, the central processing system restores the speed lock relationship and tension lock relationship in the order of the final pull traction unit, the take-up drive unit and the release drive unit, forming the tension speed relock result Lck. S4. The central processing system outputs linkage control commands based on the release-stabilization switching result Rel and the tension speed relocking result Lck, and executes short-term release-stabilization switching and tension speed relocking control in the final tensioning stage to complete the coordinated adjustment of the final tensioning stage of the ultra-fine tungsten wire with the control objective of maintaining constant tension and constant speed.

[0022] In this embodiment, the continuous multi-pass wire drawing production line for ultra-fine tungsten wires used in electroplated diamond wire busbars is applied. The control object is limited to the final drawing stage after the last wire drawing. The working condition corresponding to this embodiment is not the general wire drawing scenario of ordinary ultra-fine metal wires, but the control scenario of ultra-fine tungsten wires that have completed the previous multi-pass diameter reduction and are about to enter the final stage of finished product winding.

[0023] In this scenario, the target ultra-fine tungsten wire is directly used for electroplating diamond wire busbars. It has higher requirements for the consistency of the wire diameter at the end, the stability of the tension speed in the final stage, and the continuity of the winding stage. Compared with ordinary fine wires, ultra-fine tungsten wires are more likely to experience the superposition of residual coil release, full coil accumulation, short-range thermal state misalignment, and residual fluctuations from the previous stage in the final drawing and winding stage.

[0024] While surface tension and surface velocity values ​​may remain within acceptable limits in the short term, the continuous synchronous steady state has already begun to decay, which can easily induce end wire diameter shift, abnormal speed reduction, or sudden wire breakage. Based on the above scenario, this embodiment uses the final tension implicit stability reconstruction model Mdl to perform reconstruction processing on the input dataset Inp for the final tension stage, and then uses the tail section implicit stability state result Hid to generate the release stability switching result Rel and the tension speed relocking result Lck to complete the coordinated adjustment of the final tension stage.

[0025] In this embodiment, a nine-pass continuous wire drawing production line is selected. The first eight passes complete the pre-diameter reduction, and the ninth pass is the final drawing pass. The average wire diameter of the tungsten wire before entering the ninth pass is controlled to be 46.8μm to 47.3μm, and the target finished wire diameter after the ninth pass is controlled to be 42.0μm to 42.4μm.

[0026] The pay-off drive unit adopts an active pay-off mechanism, the take-up drive unit adopts a high-full-wind take-up mechanism, and the final pull traction unit is located at the ninth pass exit position. The central processing system is connected to the tension collector, speed collector, pay-off end roll diameter collector, take-up end roll diameter collector, temperature collector before final pull die, temperature collector after final pull die, and end wire diameter collector.

[0027] The sampling frequency for tension and velocity variation data was set to 20kHz, the sampling frequency for pay-off and take-up diameter variation data was set to 10kHz, the sampling frequency for short-range temperature variation data before and after final die drawing was set to 10kHz, and the sampling frequency for end wire diameter micro-drift data was set to 100Hz.

[0028] The central processing system performs timing alignment on all data streams using a unified time base of 20ms. During normal operation of the final pull, the traction speed for the ninth pass is set to 820m / min to 860m / min, and the tension setting range for the final pull zone is set to 0.72N to 0.86N. The short-range temperature before the final pull die is typically distributed between 71℃ and 79℃, and the short-range temperature after the final pull die is typically distributed between 76℃ and 84℃. When the remaining coil length at the unwinding end enters the range of 18m to 6m, the coil diameter at the take-up end enters the range of 198mm to 206mm, and the full winding rate enters the range of 92% to 96%, the central processing system begins to focus on monitoring the final pull stage.

[0029] Before implementation, historical operating data is imported into the central processing system to pre-build and pre-train the final tension hidden stability reconstruction model Mdl. The input window length of the final tension hidden stability reconstruction model Mdl is set to 1.2s, and the rolling update step size is set to 40ms. After the current running round enters the final tension stage, the central processing system continuously collects tension change data, speed change data, wire diameter change data at the pay-off end, wire diameter change data at the take-up end, short-range temperature change data before final tension, short-range temperature change data after final tension, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous round.

[0030] The central processing system first performs time window slicing on the raw data, then performs final pull position alignment, and removes obviously distorted abnormal sampling points; after processing, it retains the corresponding data of the final pull stage to form the input dataset Inp for the final pull stage.

[0031] Within each 40ms update cycle, the central processing system sends the input dataset Inp from the most recent 1.2s of the final pull stage to the final pull hidden stability reconstruction model Mdl. The final pull hidden stability reconstruction model Mdl first reads the roll diameter change data at the release end and the take-up end to identify the roll diameter evolution relationship between the remaining roll release state and the roll accumulation state. Then, it reads the short-range temperature change data before and after the final pull die to identify the misalignment relationship of the short-range thermal state. Subsequently, it reads the tension change data, velocity change data, and end-point roll diameter micro-drift data to identify the tension-velocity linkage offset relationship. Finally, it reads the residual fluctuation data from the adjacent previous pass to identify the transmission relationship of residual fluctuations to the final pull pass. The final pull hidden stability reconstruction model Mdl performs simultaneous stage coupling fusion on the above relationships and reconstructs the attenuation trajectory of the continuous synchronous steady state during the final pull stage.

[0032] The central processing system extracts the short-term stabilization start point, short-term stabilization duration interval, and tension speed relock start conditions from the decay trajectory to form the tail-end hidden stabilization state result Hid. The tail-end hidden stabilization state result Hid is not an ordinary risk alarm value, but the basis for the action required for the reorganization of the control chain in the final pull stage.

[0033] In this embodiment, when the tail segment hidden stability result Hid determines that the current window has entered the destabilization trigger zone, the central processing system provides a short-term destabilization start point; When the Hid result of the tail segment determines that the current window remains in the stabilization maintenance region, the central processing system provides a short-term stabilization duration range of 0.62s to 0.88s. When the Hid result of the tail section's hidden stability state determines that the attenuation trajectory has exited the stabilization maintenance region and entered the recoverable control region, the central processing system provides the tension speed relock initiation condition.

[0034] The central processing system generates the destabilization switching result Rel based on the tail-end hidden stability state result Hid.

[0035] After reaching the short-term stabilization start point, the central processing system first sends a speed unlock command to the take-up drive unit, so that the take-up drive unit releases the speed synchronization lock relationship with the final pull traction unit, and the execution time interval is controlled to be 90ms to 120ms.

[0036] At this point, the take-up drive unit no longer strictly follows the synchronous speed of the final pull traction unit, but performs limited following within a preset release rhythm, allowing a control deviation of 0.35% to 0.60% between the take-up end speed and the final pull traction speed.

[0037] After completing the above actions, the central processing system sends a tension unlocking command to the pay-off drive unit, causing the pay-off drive unit to release the tension locking relationship with the final pull traction unit. The execution time interval is controlled to be 70ms to 100ms. At this time, the pay-off drive unit retains the pay-off stabilization function, but the front tension no longer forms a rigid tension with the final pull traction speed. During the short-term stabilization period, the central processing system does not reduce the overall speed, nor does it cancel the reference traction function of the final pull traction unit. Instead, it maintains a short-term stabilization control mode of "stabilization at the end first, stabilization at the front end later, and traction reference maintained". This is to first weaken the speed synchronization constraint generated by the accumulation of the final pull end coil, and then weaken the tension locking constraint generated by the release of the remaining coil at the front end.

[0038] Once the central processing system determines that the initial conditions for tension speed relocking are met, it begins to generate the tension speed relocking result Lck. During the tension speed relocking phase, the central processing system first issues a reference traction speed establishment command to the final traction unit, enabling the final traction unit to establish a new traction speed reference within the current operating window. The new traction speed reference does not directly return to the normal steady-state setpoint, but is called between 802 m / min and 834 m / min based on the current operating condition corresponding to the tail section implicit steady-state result Hid.

[0039] After the baseline traction speed is established, the central processing system sends a speed relock command to the take-up drive unit, so that the take-up drive unit re-establishes the speed lock relationship with the final pull traction unit. The speed relock process adopts a three-stage step control, with the first stage step set to 0.08%, the second stage step set to 0.06%, and the third stage step set to 0.04%, so that the take-up end speed gradually approaches the final pull traction baseline.

[0040] After the take-up speed lock relationship stabilizes, the central processing system sends a tension relock command to the pay-off drive unit, causing the pay-off drive unit to re-establish the tension lock relationship with the final pull traction unit. The tension relock also adopts a step-by-step regression method, with the first recovery amount set to 0.03N, the second recovery amount set to 0.02N, and the third recovery amount set to 0.01N, thereby avoiding a new transient impact caused by a one-time tension recovery at the front end. Through the above sequential control, the final pull traction unit first establishes the benchmark, the take-up drive unit then restores the speed lock, and the pay-off drive unit finally restores the tension lock, thus completing the tension speed relock in the final pull stage.

[0041] After executing the destabilization switching result Rel and the tension speed relocking result Lck, the central processing system continues to collect post-execution tension change data, post-execution velocity change data, and post-execution end wire diameter change data to form the steady-state recovery result Rcv. In this embodiment, the central processing system uses a 1.5s window after relocking is completed as a confirmation window to determine whether the velocity lock recovery state, tension lock recovery state, and end wire diameter stabilization state simultaneously meet the preset recovery conditions.

[0042] The preset recovery conditions include: the tension in the final pull zone returns to the range of 0.74N to 0.84N, the synchronization deviation between the take-up end speed and the final pull speed returns to within ±0.18%, and the micro-drift amplitude of the end wire diameter returns to within ±0.28μm.

[0043] When all three conditions above are met, the central processing system maintains the collaborative control relationship after the current tension speed relock. If any condition is not met, the central processing system does not exit directly, but instead calls up the input dataset Inp of the final pull-off stage in the latest time window, inputs the final pull-off hidden stability reconstruction model Mdl again, and regenerates the new tail segment hidden stability state result Hid, and based on this, regenerates the destabilization switching result Rel and the tension velocity relocking result Lck.

[0044] Example 2 Please refer to Figure 1 Specifically: In S1, the central processing system performs final pull-off stage identification on the collected data; When the data on the change in the winding diameter at the unwinding end indicates that the unwinding end is in the state of releasing excess winding, and the data on the change in the winding diameter at the take-up end indicates that the take-up end is in the state of accumulating winding, and the data on the slight drift of the end wire diameter indicates that the end wire diameter maintains a slight deviation during the continuous sampling period, and when there is a staggered change relationship between the short-range temperature change data before the final drawing die and the short-range temperature change data after the final drawing die, and when the residual fluctuation data of the adjacent previous pass continues into the final drawing pass, the central processing system determines that the target ultra-fine tungsten wire has entered the final drawing and finishing stage. The system retains the tension change data, speed change data, wire diameter change data at the pay-off end, wire diameter change data at the take-up end, short-range temperature change data before and after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous draw, forming the input dataset Inp for the final drawing stage.

[0045] In this embodiment, the application scenario remains the same as in Embodiment 1, which is still a continuous multi-pass wire drawing production line for ultra-fine tungsten wires used in electroplated diamond wire busbars.

[0046] Tension variation data is acquired by tension acquisition devices located before and after the final pull unit; speed variation data is acquired by encoders located on the main shaft of the final pull unit and the main shaft of the take-up drive unit; wire diameter variation data at the pay-off end and take-up end are acquired by laser rangefinders located at the pay-off end and take-up end, respectively; short-range temperature variation data before and after the final pull die are acquired by infrared temperature acquisition devices located 50mm before and 50mm after the final pull die, respectively; micro-drift data of the end wire diameter is acquired by an online laser diameter gauge located in front of the take-up drive unit; and residual fluctuation data of the adjacent previous pass are synchronously acquired by the tension acquisition device and speed acquisition device at the eighth pass exit.

[0047] Each data collector sends its corresponding data to the central processing system, which then completes the receiving, buffering, and alignment processes according to a unified time reference.

[0048] The central processing system performs final stage identification on the collected data.

[0049] During identification, a 1.2s identification window and a 40ms rolling update step are used to simultaneously determine the data on changes in wire diameter at the pay-off end, changes in wire diameter at the take-up end, micro-drift data of the wire diameter at the end, short-range temperature changes before and after final die pulling, and residual fluctuation data from adjacent previous passes. For the wire diameter change data at the pay-off end, the central processing system extracts six consecutive sampled values. When adjacent sampled values ​​decrease sequentially over time, with the difference between the first and last values ​​between 0.8mm and 3.5mm, and the corresponding remaining roll length between 18m and 6m, the unwinding end is determined to be in a state of remaining roll release. For the roll diameter change data at the take-up end, the central processing system extracts 6 consecutive sampled values; when adjacent sampled values ​​increase sequentially over time, with the difference between the first and last values ​​between 0.6mm and 2.8mm, and the take-up full roll rate between 92% and 96%, the take-up end is determined to be in a state of roll accumulation.

[0050] For the micro-drift data of the end wire diameter, the central processing system extracts 12 consecutive sample values; when at least 9 sample values ​​have the same offset direction relative to the target wire diameter reference value, the absolute value of the offset is in the range of 0.10μm to 0.45μm, and the jump between adjacent sample values ​​is not greater than 0.08μm, it is determined that the end wire diameter maintains a slight offset during the continuous sampling period.

[0051] For the short-range temperature change data before and after the final die pulling, the central processing system extracts the temperature change sequence within the corresponding identification window. When the main change directions of the two sets of sequences are consistent, and the turning point of the short-range temperature change sequence before the final die pulling occurs 100ms to 300ms earlier than the turning point of the short-range temperature change sequence after the final die pulling, it is determined that there is a staggered change relationship between the two.

[0052] For the residual fluctuation data of the adjacent previous pass, the central processing system performs time-shift matching between the residual fluctuation sequence of the eighth pass exit and the tension and velocity fluctuation sequence of the ninth pass inlet; When the normalized correlation is not less than 0.72, the difference in the main fluctuation period is not greater than 12%, and the time shift is in the range of 60ms to 180ms, it is determined that the residual fluctuation data of the adjacent previous track will continue into the final track.

[0053] When the unwinding end is in the state of releasing excess coil, the take-up end is in the state of accumulating coil, the end wire diameter maintains a slight deviation during the continuous sampling period, and there is a staggered change relationship between the short-range temperature change data before the final drawing die and the short-range temperature change data after the final drawing die, and the residual fluctuation data of the adjacent previous pass continues into the final drawing pass, the central processing system determines that the target ultrafine tungsten wire has entered the final drawing and finishing stage.

[0054] To reduce misjudgments caused by occasional fluctuations, the central processing system further requires that the above five conditions remain true within two consecutive recognition windows. After completing the final pull-off stage recognition, the central processing system only retains the tension change data, speed change data, wire diameter change data at the pay-off end, wire diameter change data at the take-up end, short-range temperature change data before and after the final pull-off die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous pass for the corresponding final pull-off stage, and organizes them into the final pull-off stage input dataset Inp. The final pull-off stage input dataset Inp serves as the dedicated input data for the subsequent final pull-off hidden stability reconstruction model Mdl, used to characterize the scene state of the ultra-fine tungsten wire used in the final pull-off stage of electroplated diamond wire busbar.

[0055] Example 3 Please refer to Figure 1 and Figure 3 Specifically: In S2, the obtained final pull-off stage input dataset Inp is input into the final pull-off hidden stability reconstruction model Mdl; the final pull-off hidden stability reconstruction model Mdl is constructed through a gated temporal coding and decoding reconstruction framework; The final pull-in hidden stability reconstruction model Mdl includes a scene slice input layer, a roll diameter evolution coding layer, a short-range thermal state hysteresis coding layer, a tension velocity micro-drift coding layer, a previous pass residual wave propagation coding layer, a hidden stability coupling fusion layer, and an attenuation trajectory reconstruction layer. The scene slice input layer receives the input dataset Inp from the final pull-off stage. The roll diameter coupling evolution characteristic of the output feed end of the roll diameter evolution coding layer being in the state of excess roll release and the take-up end being in the state of roll accumulation; The misalignment characteristics between the short-range temperature change data before and after final die pulling in the output of the short-range thermal state hysteresis coding layer. The linkage offset characteristics between the tension change data, velocity change data and end wire diameter micro-drift data output by the tension micro-drift coding layer; The previous residual wave propagation coding layer outputs the propagation characteristics of the adjacent previous residual wave data to the final pull. The implicit stable coupling fusion layer performs alignment and fusion on the roll diameter coupling evolution characteristics, front and rear misalignment characteristics, linkage offset characteristics and transmission characteristics according to the same time sequence position of the final pull-off stage, forming the implicit stable coupling characteristics of the final pull-off stage. Based on the implicit stability coupling characteristics of the final pull stage, the attenuation trajectory reconstruction layer reconstructs the continuous synchronous steady-state attenuation trajectory of the final pull stage and forms the implicit stability state result Hid of the tail segment.

[0056] In S2, the construction of the final hidden stability reconstruction model Mdl includes a pre-training phase and a calibration training phase. During the pre-training phase, based on historical operating data from a continuous multi-pass drawing production line for ultra-fine tungsten wire, steady-state maintenance segments, latent steady-state decay segments, and overt instability segments were segmented. Among them, the preceding running segment before any result of end wire diameter jump, wire breakage, or speed reduction control switching occurs within the subsequent preset time period is determined as the implicit stable attenuation segment; The running segment that has already shown any of the following results: end wire diameter exceeding tolerance, wire breakage, or control loss of synchronization is identified as an explicit unstable segment; The running segment that does not exhibit abnormal control switching and whose end wire diameter remains stable is identified as the steady-state maintenance segment. The steady-state maintenance segment, the implicitly stable decay segment, and the explicit unstable segment are used to perform self-supervised pre-training on the gated temporal coding and decoding reconstruction framework, so that the representation of the roll diameter evolution relationship, the representation of the preceding and following misalignment relationship, the representation of the linkage offset relationship, and the representation of the transmission relationship form a temporal correlation. During the calibration and training phase, based on the results of end wire diameter jumps, wire breakage, speed reduction control switching, and control step loss within a subsequent preset time period, the results of historical operation segments are back-calibrated to form steady-state maintenance segments, implicit stability decay segments, and explicit instability segments. Based on the steady-state maintenance segments, implicit stability decay segments, and explicit instability segments, the final tension implicit stability reconstruction model Mdl is calibrated and trained so that the output of the final tension implicit stability reconstruction model Mdl corresponds to the tail-segment implicit stability state result Hid of the control switching at the end of the final tension stage.

[0057] In S2, the tail-end hidden stability result Hid includes at least the short-term release stability start time, the short-term release stability duration interval, the tension velocity relock start condition, and the continuous synchronous steady-state decay level. Among them, the short-term release stabilization start point corresponds to the decay trajectory of the latent stability coupling characteristics in the final pull-off stage, at the time of entering the release stabilization trigger zone; The short-term stabilization duration interval corresponds to the decay trajectory being within the stabilization maintenance region; The initial condition for Zhang Su relocking corresponds to the condition for the decay trajectory to exit the stabilization maintenance region and enter the relockable region. The continuous synchronous steady-state decay level indicates the degree of decay of the continuous synchronous steady state during the final pull phase.

[0058] In this embodiment, the same electroplated diamond wire busbar ultra-fine tungsten wire continuous multi-pass wire drawing production line as in Embodiments 1 and 2 is used. After the central processing system forms the final drawing stage input dataset Inp in S1, it inputs the most recent 1.2s of the final drawing stage input dataset Inp as a scene segment into the final drawing hidden stability reconstruction model Mdl. The rolling update step size is set to 40ms, so each scene segment corresponds to 60 unified time series points.

[0059] The final pull hidden stability reconstruction model Mdl is constructed based on a gated temporal encoding and decoding reconstruction framework. In this embodiment, a two-layer gated loop structure is used as the basic framework for encoding and decoding. The scene slice input layer is responsible for receiving the input dataset Inp of the final pull stage, and organizing the tension change data, velocity change data, wire diameter change data at the release end, wire diameter change data at the take-up end, short-range temperature change data before final pull, short-range temperature change data after final pull, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous pass into temporal vectors of uniform length.

[0060] The convolutional diameter evolution coding layer, the short-range thermal state hysteresis coding layer, the tension velocity micro-drift coding layer, and the previous pass residual wave propagation coding layer encode the corresponding inputs respectively, and each coding layer outputs a 32-dimensional relational representation. The implicit stable coupling fusion layer fuses the four relational representations according to the same final pull-off stage temporal position to form a 96-dimensional implicit stable coupling feature of the final pull-off stage. The attenuation trajectory reconstruction layer reconstructs the continuous synchronous steady-state attenuation trajectory based on the implicit stable coupling feature of the final pull-off stage and outputs 60 steady-state holding values ​​Stb. The steady-state holding value Stb is set to a range of 0 to 1. The closer the value is to 1, the stronger the continuous synchronous steady-state holding ability. The closer the value is to 0, the weaker the continuous synchronous steady-state holding ability.

[0061] The central processing system then extracts the short-term stabilization start point, short-term stabilization duration interval, tension velocity relock start condition, and continuous synchronous steady-state decay level based on the steady-state hold value Stb sequence, forming the tail-end hidden steady state result Hid.

[0062] The training of the final-stretch hidden stability reconstruction model Mdl was completed using a combination of historical data playback and result backtracking calibration. The historical data came from the final-stretch records of 420 reels of ultra-fine tungsten wire from the same production line, of which 336 reels were used as training data, 42 reels as validation data, and 42 reels as test data.

[0063] The central processing system first uses the identification rules for the final pull-off stage in Example 2 to segment the steady-state maintenance segment, the implicit steady-state decay segment, and the explicit instability segment from the historical data. The preceding running segment before any result of end wire diameter jump, wire breakage, or switching speed reduction control within the next 1.2 seconds is labeled as the implicit steady-state decay segment. The running segment that has already resulted in any result of end wire diameter exceeding tolerance, wire breakage, or control loss of synchronization is labeled as the explicit instability segment. The running segment that has not experienced abnormal control switching and whose end wire diameter remains stable is labeled as the steady-state maintenance segment.

[0064] After segmentation, the training data yielded 1864 steady-state maintenance segments, 612 latent steady-state decay segments, and 337 overtly unstable segments. During the pre-training phase, the central processing system performed self-supervised pre-training on the final latent steady-state reconstruction model Mdl. The pre-training tasks included occluded segment reconstruction, prediction of the next steady-state maintenance value Stb, and temporal sequence consistency verification. The number of training rounds was set to 60, the batch size to 64, and the initial learning rate to 0.001.

[0065] During the calibration training phase, the central processing system continues to perform calibration training on the final pull hidden stability reconstruction model Mdl based on the three types of segment labels obtained from the result backtracking calibration. The number of calibration training rounds is set to 30, the batch size is set to 32, and the learning rate is adjusted to 0.0003, so that the tail segment hidden stability state result Hid output by the final pull hidden stability reconstruction model Mdl can correspond to the control switching in the final pull closing stage.

[0066] In this embodiment, the central processing system sets the short-term stabilization start point as the first time point when the steady-state value Stb is below 0.72 for three consecutive time points, sets the short-term stabilization duration range as the continuous range where the steady-state value Stb is between 0.72 and 0.54, sets the tension rate relocking start condition as when the steady-state value Stb is above 0.66 again for five consecutive time points and the end wire diameter micro-drift amplitude returns to within ±0.20 μm, and divides the continuous synchronous steady-state attenuation level into three levels: The minimum steady-state retention value Stb corresponding to the first level is in the range of 0.66 to 0.72; The second level corresponds to a minimum steady-state retention value Stb in the range of 0.54 to 0.66; The third level corresponds to a minimum steady-state retention value Stb of less than 0.54.

[0067] After this processing, the tail-end hidden stability result Hid is no longer a single risk value, but directly corresponds to the action basis required for the subsequent destabilization switching result Rel and the tension relocking result Lck.

[0068] To demonstrate the effectiveness of the solution in this embodiment, three sets of comparisons were set up in the test data.

[0069] Example group 3.1 adopts the method of this example; Comparative Example 1 uses a multi-parameter threshold alarm method, which only outputs an abnormal signal after the tension fluctuation or the end wire diameter deviation exceeds the set threshold; Comparative Example 2 uses a common single-branch time-series prediction model, which only uses tension and velocity sequences to output abnormal trend values. It does not construct the roll diameter evolution relationship, short-range thermal state hysteresis relationship, tension velocity micro-drift relationship, and the residual wave transmission relationship of the previous pass, nor does it output the complete tail-end hidden steady state result Hid.

[0070] During the test, the production line, raw material specifications, final pull speed setting range and sampling conditions used in Example 3.1 and Comparative Example 3.1 were kept consistent with those used in Comparative Example 3.1, with only the state reconstruction method in S2 being changed.

[0071] The data listed in the table below are example data from historical data playback combined with semi-physical control verification, used to illustrate the logic and beneficial effects of this embodiment; the specific content is shown in Table 1.

[0072] Table 1: Comparison of Validation Results of Different Control Schemes in the Final Pull-off Stage As can be seen from the data in Table 1, the accuracy rate of identifying the cryptic attenuation segment in Example 3.1 was 92.4%, which was significantly higher than that of Comparative Example 1 and Comparative Example 2.

[0073] The results show that the terminal pull hidden steady state reconstruction model Mdl does not perform ordinary trend prediction on a single sequence. Instead, it extracts four types of relationships through the convolutional diameter evolution coding layer, the short-range thermal state hysteresis coding layer, the tension velocity micro-drift coding layer, and the previous residual wave propagation coding layer, respectively. Then, the hidden steady state result Hid of the tail segment is output by the hidden steady state coupling fusion layer and the decay trajectory reconstruction layer. Therefore, it can more completely reconstruct the decay trajectory of the continuous synchronous steady state in the terminal pull closing stage.

[0074] Comparative Example 1 lacks a time-series reconstruction chain and can only trigger an alarm after the parameter exceeds the threshold. Although Comparative Example 2 uses a time-series model, it only processes the tension and velocity sequences and does not incorporate the roll diameter evolution, short-range thermal state hysteresis, and the transmission of residual fluctuations from the previous pass into the same reconstruction chain, resulting in significantly lower recognition accuracy.

[0075] Example 3.1 showed average identification deviations of 58ms and 71ms for the short-term release stabilization initiation point and the tension velocity relock initiation condition, respectively, which were significantly better than those of Comparative Example 1 and Comparative Example 2, indicating that the time point, interval and condition output by Hid in the tail-end hidden stabilization state result have high availability.

[0076] Looking further at the control effect, in Example 3.1, the number of wire breakages during the final tension stage was only 2, the standard deviation of the end wire diameter was 0.63μm, the average recovery time for switching from destabilization to tension speed relock was 5.9s, and the finished product qualification rate was 98.4%.

[0077] This indicates that the tail-end hidden stable state result Hid in this embodiment is not simply a model output, but a state result that can directly support the subsequent control chain actions.

[0078] Based on the above data, this embodiment, through the structural constraints of the final pull-hidden stability reconstruction model Mdl, the constraints of the construction of the training chain, and the constraints of the result content of the tail-end hidden stability state result Hid, fully connects the four aspects of "how to build the model, how to train it, what it outputs, and why it can directly serve the control chain switching". Therefore, compared with the existing threshold alarm method and the ordinary single-branch prediction method, it demonstrates stronger scenario adaptability, state reconstruction capability and control support capability.

[0079] Example 4 Please refer to Figure 1 and Figure 4 Specifically: In S3, the central processing system generates the destabilization switching result Rel based on the tail-end hidden stability state result Hid; The release stabilization switching result Rel includes the take-up drive unit speed unlock command, the release drive unit tension unlock command, and the release rhythm of the short-term release stabilization duration range; The central processing system first sends a speed unlock command to the take-up drive unit, so that the take-up drive unit releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit. The central processing system then sends a tension unlocking command to the wire release drive unit, causing the wire release drive unit to release the tension locking relationship between the wire release drive unit and the final pull traction unit. During the short-term stabilization period, the central processing system maintains the short-term stabilization control mode according to the release rhythm of the short-term stabilization period.

[0080] In S3, after the short-term destabilization period ends, the tension rate relocking result Lck is generated based on the tension rate relocking initiation condition. The tension relock result Lck includes the reference traction speed establishment command, the take-up drive unit speed relock command, the pay-off drive unit tension relock command, and the preset relock step size.

[0081] The central processing system first sends a reference traction speed establishment command to the final traction unit, so that the final traction unit can establish the reference traction speed; The central processing system then sends a speed relock command to the take-up drive unit, so that the take-up drive unit can restore the speed lock relationship between the take-up drive unit and the final pull traction unit. Finally, the central processing system issues a tension relock command to the wire release drive unit, so that the wire release drive unit restores the tension lock relationship between the wire release drive unit and the final pull traction unit. The central processing system completes the tension speed relock step by step according to the preset relock step size.

[0082] In this embodiment, the application scenario remains consistent with that of Embodiments 1, 2 and 3, and is still selected as the final stage after the ninth final pull zone. The central processing system does not directly use a unified deceleration method to process the tail section hidden stability result Hid, but first generates the destabilization switching result Rel, and then generates the tension speed relocking result Lck after the tension speed relocking start condition is met, thereby converting the model output into an executable control chain action.

[0083] In this embodiment, after obtaining the tail section hidden stability state result Hid, the central processing system first reads the short-term release stability start time, short-term release stability duration interval and continuous synchronous steady-state decay level from the tail section hidden stability state result Hid, and then combines it with the current final pull speed, current take-up speed, current pay-off tension and current take-up full roll rate to generate the release stability switching result Rel.

[0084] The specific generation process is as follows: The central processing system first uses the short-term destabilization start point as the execution starting point of the destabilization switching result Rel; Then, the short-term stabilization duration interval is used as the maintenance interval of the stabilization switching result Rel; Subsequently, based on the continuous synchronous steady-state decay level, the take-up drive unit speed unlocking delay, the release drive unit tension unlocking delay, the upper limit of the allowable deviation of the take-up speed, the allowable release amount of the release tension, and the segmented release time corresponding to the current continuous synchronous steady-state decay level are called from the pre-established release rhythm parameter set. The central processing system then writes the called parameters into the same control result structure in the order of "take-up drive unit speed unlock command, pay-off drive unit tension unlock command, and then short-term release stabilization continuous interval release rhythm", forming the release stabilization switching result Rel.

[0085] More specifically: Three sets of release rhythm parameters are pre-established. The first set of release rhythm parameters corresponds to the first level of continuous synchronous steady-state decay, with the take-up drive unit speed unlock delay set to 95ms, the pay-off drive unit tension unlock delay set to 80ms, the upper limit of the allowable deviation of the take-up speed set to 0.35%, and the allowable release amount of the pay-off tension set to 0.03N. The second set of release rhythm parameters corresponds to the second level of continuous synchronous steady-state decay, with the take-up drive unit speed unlock delay set to 105ms, the pay-off drive unit tension unlock delay set to 88ms, the upper limit of the allowable deviation of the take-up speed set to 0.45%, and the allowable release amount of the pay-off tension set to 0. 05N; The third set of release rhythm parameters corresponds to the three-level continuous synchronous steady-state decay level. The speed unlocking delay of the take-up drive unit is set to 115ms, the tension unlocking delay of the pay-off drive unit is set to 95ms, the upper limit of the allowable deviation of the take-up speed is set to 0.55%, and the allowable release amount of the pay-off tension is set to 0.06N. The central processing system calls the corresponding set of release rhythm parameters according to the current continuous synchronous steady-state decay level, and combines it with the length of the short-term stabilization duration interval to further allocate the called parameters to each release stage within the short-term stabilization duration interval, and finally forms the stabilization switching result Rel.

[0086] Therefore, the stabilization switching result Rel not only includes the short-term stabilization start time and the short-term stabilization duration range, but also includes the execution time of the speed unlock of the take-up drive unit, the execution time of the tension unlock of the release drive unit, the duration of each release stage within the short-term stabilization duration range, and the upper limit of the speed allowable deviation and the allowable tension release amount corresponding to each release stage.

[0087] The release stabilization switching result Rel includes the take-up drive unit speed unlock command, the take-up drive unit tension unlock command, and the short-term release stabilization duration range release rhythm.

[0088] After the central processing system reaches the short-term stabilization start point, it first sends a speed unlock command to the take-up drive unit, so that the take-up drive unit releases the speed synchronization lock relationship with the final pull traction unit. In this embodiment, the execution delay of the speed unlock command of the take-up drive unit is set to 95ms to 115ms. After unlocking, the take-up drive unit is allowed to have a restricted speed deviation of 0.35% to 0.60% relative to the final pull traction unit.

[0089] After the take-up drive unit speed is unlocked, the central processing system sends a tension unlocking command to the pay-off drive unit, causing the pay-off drive unit to release the tension locking relationship with the final pull traction unit. In this embodiment, the execution delay of the tension unlocking command of the wire feeding drive unit is set to 75ms to 95ms. After unlocking, the tension at the wire feeding end is allowed to be released within the range of 0.03N to 0.06N under limited conditions.

[0090] The short-term release stabilization interval adopts a three-stage release method: the first release segment lasts from 0.18s to 0.24s, the second release segment lasts from 0.20s to 0.28s, and the third release segment lasts from 0.16s to 0.24s. During the three release segments, the upper limit of the speed deviation at the take-up end is 0.35%, 0.45%, and 0.55% respectively, and the tension release amount at the release end is 0.03N, 0.05N, and 0.06N respectively.

[0091] With the above settings, the central processing system first releases the end velocity synchronization constraint, and then releases the front tension constraint, so that the final tension zone enters the short-term stability control mode while maintaining the reference traction effect.

[0092] The reason for adopting this order is that the final coiling accumulation state at the end of the final drawing stage makes it easier to amplify the speed synchronization constraint to the final drawing zone. If the front tension lock is released first while the end speed lock is still maintained, it is easy to cause a sudden change in the tension in the final drawing zone. Conversely, releasing the end speed lock first and then releasing the front tension lock is more conducive to weakening the locking amplification effect under the implicit stability decay state.

[0093] Once the central processing system determines that the tension speed relocking initiation condition is met based on the tail section hidden stability state result Hid, it begins to generate the tension speed relocking result Lck.

[0094] The tension relock result Lck includes the reference traction speed establishment command, the take-up drive unit speed relock command, the pay-off drive unit tension relock command, and the preset relock step size.

[0095] In this embodiment, after the central processing system meets the initial conditions for speed relocking, it first reads the continuous synchronous steady-state attenuation level in the tail section hidden steady state result Hid, and then reads the final pull speed, take-up speed, let-out tension, and end wire diameter micro-drift amplitude at the end of the short-term release stabilization. Based on the continuous synchronous steady-state attenuation level, the final pull speed, take-up speed, let-out tension, and end wire diameter micro-drift amplitude at the end of the short-term release stabilization, the central processing system generates the speed relocking result Lck.

[0096] The central processing system first calls the corresponding reference traction speed range based on the continuous synchronous steady-state decay level, and then determines the reference traction speed establishment command by combining the actual final traction speed at the end of the short-term release stabilization.

[0097] In this embodiment, when the continuous synchronization steady-state attenuation level is Level 1, the call range corresponding to the reference traction speed establishment command is 826 m / min to 832 m / min; when the continuous synchronization steady-state attenuation level is Level 2, the call range corresponding to the reference traction speed establishment command is 814 m / min to 825 m / min; and when the continuous synchronization steady-state attenuation level is Level 3, the call range corresponding to the reference traction speed establishment command is 804 m / min to 813 m / min.

[0098] The central processing system then uses the speed difference between the take-up speed and the reference traction speed at the end of the short-term release stabilization to call the take-up drive unit speed relock command in conjunction with the continuous synchronous steady-state decay level, and writes the preset relock step size into the take-up drive unit speed relock command.

[0099] In this embodiment, the preset relock step size is set to three segments: 0.08%, 0.06%, and 0.04%.

[0100] The central processing system then calls the tension relock instruction of the pay-off drive unit based on the tension difference between the pay-off tension at the end of the short-term release and the target tension range, combined with the continuous synchronous steady-state decay level, and writes the tension recovery amount into the tension relock instruction of the pay-off drive unit.

[0101] In this embodiment, the three tension recovery values ​​corresponding to the tension relock command of the wire feeding drive unit are set to 0.03N, 0.02N, and 0.01N.

[0102] After the tension speed relock result Lck is generated, the central processing system first sends a reference traction speed establishment command to the final pull traction unit, so that the final pull traction unit establishes the reference traction speed; then it sends a take-up drive unit speed relock command to the take-up drive unit, so that the take-up drive unit gradually restores the speed lock relationship with the final pull traction unit; finally, it sends a release drive unit tension relock command to the release drive unit, so that the release drive unit gradually restores the tension lock relationship with the final pull traction unit.

[0103] The reason for adopting the relocking sequence of "final pull traction unit, take-up drive unit and then release drive unit" is that after the final pull traction unit establishes the speed reference of the final pull zone, the take-up drive unit restores the speed locking relationship so that the end winding can be controlled again. Finally, the release drive unit restores the tension locking relationship, which can prevent the tension pull-back on the front end of the remaining roll from being prematurely transmitted to the final pull zone.

[0104] After executing the tension relock result Lck, the central processing system continues to collect post-execution tension change data, post-execution velocity change data, and post-execution end wire diameter change data, and uses 1.5s as the recovery confirmation window to form the steady-state recovery result Rcv; When the speed lock recovery state, tension lock recovery state, and end wire diameter stabilization state simultaneously meet the preset recovery conditions, the current coordinated control relationship is maintained. When any state does not meet the preset recovery conditions, the destabilization switching result Rel and the tension speed relocking result Lck corresponding to the tail section hidden stability state result Hid are called again to continue the control.

[0105] To verify the effectiveness of the action chain in this embodiment, 12 disc samples were used as the embodiment group, and two comparative groups were set up.

[0106] Example group 4.1 uses the method of the present invention; Comparative Example 1.2 adopts a synchronous unlocking and synchronous relocking method, that is, when entering the short-term stabilization stage, the speed synchronization locking relationship and the tension counter-locking relationship are released simultaneously, and the two types of locking relationships are restored simultaneously during the recovery stage; Comparative Example 2.2 adopts a unified deceleration and recovery method, that is, without distinguishing the order of speed locking relationship and tension locking relationship, the central processing system simultaneously issues deceleration and recovery commands to the final pull traction unit, take-up drive unit and release drive unit; the raw material specifications, final pull speed range, environmental conditions and sampling conditions used in the three sets of tests are consistent, as shown in Table 2.

[0107] Table 2: Comparison of Control Effects in the Final Tensioning Stage under Different Stabilization Switching and Tension Speed ​​Relocking Control Methods As can be seen from the results in Table 2, Example 4.1 is superior to Comparative Example 1.2 and Comparative Example 2.2 in terms of indicators such as number of wire breakages, tension fluctuation during the release stabilization stage, speed deviation during the release stabilization stage, tension overshoot during the relocking stage, wire diameter rebound during the relocking stage, and average recovery time.

[0108] In Example 4.1, the number of wire breaks was 1, while in Comparative Example 1.2 it was 4 and in Comparative Example 2.2 it was 3. This shows that the central processing system first releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit, and then releases the tension lock relationship between the pay-off drive unit and the final pull traction unit. This can preferentially release the speed lock amplification effect under the cumulative state of the end winding, and reduce sudden instability in the final pull zone.

[0109] In Example 4.1, the maximum tension fluctuation during the stabilization phase was 0.19 N, which was significantly lower than 0.34 N in Comparative Example 1.2 and 0.28 N in Comparative Example 2.2. This indicates that the graded release method of the release rhythm in the short-term stabilization duration range can release the synchronization constraint more smoothly, rather than removing all the locking relationships at once.

[0110] In Example 4.1, the maximum tension overshoot during the relocking stage was 0.11N and the maximum wire diameter rebound was 0.29μm, both significantly lower than those in the comparative example group. This indicates that the tension-speed relocking action chain of "first establishing the reference traction speed, then restoring the speed locking relationship, and finally restoring the tension locking relationship" can effectively reduce the impact at the moment of relocking.

[0111] In Example 4.1, the average recovery time from the start of stabilization to the completion of tension relock was 5.6 seconds, which was significantly shorter than that of Comparative Examples 1.2 and 2.2. This indicates that the recovery was not simply delayed, but achieved faster and more stable recovery under a more reasonable control chain sequence.

[0112] Ultimately, the finished product qualification rate of Example Group 4.1 reached 98.8%, which was higher than that of the two comparative groups. This proves that the release-stabilization switching result Rel and the tension speed relocking result Lck of this example not only constitute a clear control chain reorganization logic at the action level, but also produce a significant stabilization effect in the final pulling stage of the electroplated diamond wire busbar using ultra-fine tungsten wire.

[0113] Example 5 Please refer to Figure 1 Specifically: In S4, after executing the release stabilization switching result Rel and the tension speed relocking result Lck, the central processing system continues to collect tension change data, velocity change data, and end wire diameter change data after execution. The central processing system generates a steady-state recovery result Rcv based on the tension change data, speed change data, and end wire diameter change data after execution. The steady-state recovery result Rcv includes at least the speed lock recovery state, the tension lock recovery state, and the end wire diameter stabilization state. When the steady-state recovery result Rcv does not meet the preset recovery conditions, the central processing system re-calls the destabilization switching result Rel and the tension speed relocking result Lck corresponding to the tail section hidden steady state result Hid, and continues to execute the short-term destabilization switching control and tension speed relocking control in the final tensioning stage.

[0114] In S4, the central processing system pre-establishes the correspondence between the continuous synchronous steady-state decay level and the release rhythm of the short-term release stability duration interval, the reference traction speed establishment command, and the preset relock step size. Based on the continuous synchronous steady-state decay level, the central processing system calls the short-term release stabilization duration interval release rhythm, the reference traction speed establishment command, and the preset relock step size corresponding to the current continuous synchronous steady-state decay level in the corresponding relationship. The system then writes the short-term release stabilization duration interval release rhythm after the call into the stabilization switching result Rel, and writes the reference traction speed establishment command and the preset relock step size after the call into the tension speed relock result Lck.

[0115] In this embodiment, the application scenario remains the same as in Embodiment 1, which is still the ninth final drawing stage of the continuous multi-pass wire drawing production line for electroplated diamond wire busbar ultra-fine tungsten wire.

[0116] Before S4 is implemented, the central processing system pre-establishes the correspondence between the continuous synchronous steady-state decay level and the release rhythm of the short-term release stabilization continuous interval, the reference traction speed establishment command, and the preset relock step size.

[0117] Specifically, when the continuous synchronous steady-state decay level is level one, the first set of execution parameters is called. The first set of execution parameters includes a short-term stabilization duration range of 0.62s to 0.70s, a speed unlocking delay of 95ms for the take-up drive unit, a tension unlocking delay of 80ms for the pay-off drive unit, a reference traction speed establishment range of 826m / min to 832m / min, and preset speed relocking step sizes of 0.08%, 0.06%, and 0.04%. When the continuous synchronous steady-state decay level is level two, the second set of execution parameters is called accordingly. The second set of execution parameters includes a short-term release stabilization duration range of 0.71s to 0.82s, a speed unlocking delay of 105ms for the take-up drive unit, a tension unlocking delay of 88ms for the pay-off drive unit, a reference traction speed establishment range of 814m / min to 825m / min, and preset speed relocking step sizes of 0.07%, 0.05%, and 0.03%. When the continuous synchronous steady-state decay level is three, the third set of execution parameters is called accordingly. The third set of execution parameters includes a short-term release stabilization duration range of 0.83s to 0.92s, a speed unlocking delay of 115ms for the take-up drive unit, a tension unlocking delay of 95ms for the pay-off drive unit, a reference traction speed establishment range of 804m / min to 813m / min, and preset speed relocking step sizes of 0.06%, 0.04%, and 0.02%.

[0118] After obtaining the tail-end hidden steady-state result Hid, the central processing system first reads the continuous synchronous steady-state decay level within it, and then calls a set of execution parameters that match the current continuous synchronous steady-state decay level from the aforementioned correspondence. The release rhythm of the short-term release stabilization duration interval after the call is written into the release stabilization switching result Rel, and the reference traction speed establishment command and preset relocking step size after the call are written into the tension speed relocking result Lck. In this way, the state level output by the model no longer stays at the recognition layer, but directly enters the execution layer, becoming the source of parameters for subsequent coordinated adjustment.

[0119] After the central processing system completes the calls to the destabilization switching result Rel and the tension speed relocking result Lck, it executes short-term destabilization switching and tension speed relocking control in the aforementioned order.

[0120] During the short-term stabilization phase, the central processing system first controls the take-up drive unit to release the speed synchronization lock relationship with the final pull traction unit, then controls the release drive unit to release the tension pull lock relationship with the final pull traction unit, and maintains the release rhythm within the short-term stabilization duration after the call.

[0121] During the tension relocking phase, the central processing system first controls the final pull traction unit to establish the reference traction speed after the call, then controls the take-up drive unit to restore the speed locking relationship with the final pull traction unit according to the preset speed relocking step size after the call, and finally controls the release drive unit to restore the tension locking relationship with the final pull traction unit according to the corresponding tension recovery amount.

[0122] The purpose of the above implementation is to make the continuous synchronous steady-state decay level directly determine the execution intensity and execution rhythm, thereby avoiding over-release under mild steady-state decay conditions and under-release under severe steady-state decay conditions.

[0123] After the release-stabilization switching result Rel and the tension-speed relocking result Lck are executed, the central processing system does not terminate control. Instead, within a 1.5s recovery confirmation window, it continues to collect post-execution tension change data, post-execution velocity change data, and post-execution end wire diameter change data, and generates a steady-state recovery result Rcv. The steady-state recovery result Rcv includes velocity lock recovery state, tension lock recovery state, and end wire diameter stabilization state.

[0124] The speed lock recovery status is determined as follows: the central processing system continuously extracts 15 sets of corresponding sampled values ​​of the take-up speed and the final pull-out speed within the recovery confirmation window. When the speed deviation between the two is continuously within ±0.18%, it is determined that the speed lock recovery status meets the preset recovery conditions.

[0125] The method for determining the tension lock recovery state is as follows: The central processing system continuously extracts 15 sets of wire tension sampling values ​​and final tension sampling values ​​in the recovery confirmation window. When the tension in the final tension zone is continuously within the range of 0.74N to 0.84N, and the tension fluctuation amplitude between adjacent sampling values ​​does not exceed 0.03N, it is determined that the tension lock recovery state meets the preset recovery conditions.

[0126] The determination method for the end wire diameter stabilization state is as follows: The central processing system continuously extracts 30 sets of end wire diameter micro-drift sampling values ​​within the recovery confirmation window. When the offset of the end wire diameter relative to the target wire diameter reference value is continuously within ±0.20μm, and the offset direction between consecutive sampling values ​​no longer maintains a unidirectional continuity, the end wire diameter stabilization state is determined to meet the preset recovery conditions.

[0127] Only when the speed lock recovery state, tension lock recovery state, and end wire diameter stabilization state simultaneously meet the preset recovery conditions will the central processing system determine that the steady-state recovery result Rcv meets the requirements and maintain the current cooperative control relationship.

[0128] When the steady-state recovery result Rcv does not meet the preset recovery conditions, the central processing system does not directly return to the conventional constant tension and constant speed control mode. Instead, it calls back the release stabilization switching result Rel and the tension speed relocking result Lck corresponding to the latest tail section hidden steady state result Hid, and continues to execute the short-term release stabilization switching control and tension speed relocking control in the final tensioning stage.

[0129] When the process is restarted, the central processing system first reads the post-execution tension change data, post-execution speed change data, and post-execution end wire diameter change data in the current recovery confirmation window. Then, it merges these data with the final tension stage input dataset in the latest rolling window and sends them to the final tension implicit stability reconstruction model Mdl to update the tail section implicit stability state result Hid. Subsequently, the central processing system restarts the corresponding release rhythm, benchmark traction speed establishment command, and preset relock step size based on the updated continuous synchronous steady-state decay level, and then forms a new release stability switching result Rel and tension speed relock result Lck.

[0130] In this way, the execution effect of the previous round of actions will directly affect the result of the parameter call in the next round of actions. The reason for adopting this implementation method is that the degree of implicit stability decay in the final pull-off stage may continue to change in a short period of time. If the destabilization switching and tension speed relock are only executed once, it is easy to have insufficient control or too fast recovery. By confirming the steady-state recovery result Rcv for the second time, the central processing system can decide whether to continue to perform coordinated adjustment based on the actual recovery state, so that the whole scheme can be truly implemented.

[0131] In this embodiment, the coordinated adjustment of S4 is not simply superimposing the release-stabilization switching result Rel and the tension-speed relocking result Lck, but rather taking the continuous synchronous steady-state decay level as the core, and incorporating the release rhythm, the reference traction speed establishment command and the preset relocking step size into the same call chain.

[0132] The closer the continuous synchronization steady-state decay level is to level one, the shorter the corresponding release rhythm, the closer the benchmark traction speed establishment interval is to the normal target speed interval, and the closer the preset relock step size is to the normal recovery value; the closer the continuous synchronization steady-state decay level is to level three, the longer the corresponding release rhythm, the further the benchmark traction speed establishment interval deviates from the normal target speed interval, and the more subdivided the preset relock step size.

[0133] Through this correspondence, S4 can form differentiated execution logic based on the state level output by the model, verify the execution effect through the steady-state recovery result Rcv, and recall the control result when necessary, ultimately achieving coordinated adjustment with constant tension and constant speed as the control target during the final pulling and finishing stage of the ultra-fine tungsten wire.

[0134] Example 6 Please see Figure 2A control system for drawing ultrafine tungsten wire based on constant tension and constant speed coordinated adjustment, comprising a tail data acquisition module, a final drawing modeling module, a tension speed relocking module, and a coordinated adjustment module; The finishing data acquisition module collects data on tension changes, speed changes, wire diameter changes at the unwinding end and the take-up end of the target ultra-fine tungsten wire during the final drawing stage. It also collects short-range temperature changes before and after the final drawing die, micro-drift data of the wire diameter at the end, and residual fluctuation data of the adjacent previous pass. The data is then sent to the central processing system to form the final drawing stage input dataset Inp. The final pull modeling module inputs the final pull end stage input dataset Inp into the pre-built and pre-trained final pull hidden stability reconstruction model Mdl through the central processing system. The final pull hidden stability reconstruction model Mdl reconstructs the decay trajectory of the continuous synchronous steady state in the final pull end stage, and extracts the short-term release stabilization start point, short-term release stabilization duration interval and tension velocity relocking start condition from the decay trajectory to form the tail segment hidden stability state result Hid. The tension speed relocking module, based on the tail section hidden stability result Hid by the central processing system, at the short-term release stabilization start point, first releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit in the order of the take-up drive unit and the release drive unit, and then releases the tension counter-lock relationship between the release drive unit and the final pull traction unit, forming the release stabilization switching result Rel; when the tension speed relocking start condition is met, the central processing system restores the speed lock relationship and tension lock relationship in the order of the final pull traction unit, the take-up drive unit, and the release drive unit, forming the tension speed relocking result Lck. The coordinated adjustment module outputs linkage control commands through the central processing system based on the release stabilization switching result Rel and the tension speed relocking result Lck. It executes short-term release stabilization switching and tension speed relocking control in the final tensioning stage, and completes the coordinated adjustment of the final tensioning stage of the ultra-fine tungsten wire with the control objective of maintaining constant tension and constant speed.

[0135] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A superfine tungsten wire drawing control method based on constant tension constant speed cooperative regulation, characterized in that: Includes the following steps: S1. Collect the tension change data, speed change data, wire diameter change data at the unwinding end, wire diameter change data at the take-up end, short-range temperature change data before the final drawing die, short-range temperature change data after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous pass of the target ultra-fine tungsten wire during the final drawing stage, and send them to the central processing system to form the input dataset Inp for the final drawing stage. S2. The central processing system inputs the final pull-off stage input dataset Inp into the pre-constructed and pre-trained final pull-off hidden stability reconstruction model Mdl. The final pull-off hidden stability reconstruction model Mdl reconstructs the decay trajectory of the continuous synchronous steady state in the final pull-off stage, and extracts the short-term release stabilization start point, short-term release stabilization duration interval and tension velocity relocking start condition from the decay trajectory to form the tail segment hidden stability state result Hid. S3. Based on the tail section hidden stability result Hid, at the short-term release stability start point, the central processing system first releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit in the order of the take-up drive unit and the release drive unit, and then releases the tension pull lock relationship between the release drive unit and the final pull traction unit, forming the release stability switching result Rel; when the tension speed relock start condition is met, the central processing system restores the speed lock relationship and tension lock relationship in the order of the final pull traction unit, the take-up drive unit and the release drive unit, forming the tension speed relock result Lck. S4. The central processing system outputs linkage control commands based on the release-stabilization switching result Rel and the tension speed relocking result Lck, and executes short-term release-stabilization switching and tension speed relocking control in the final tensioning stage to complete the coordinated adjustment of the final tensioning stage of the ultra-fine tungsten wire with the control objective of maintaining constant tension and constant speed.

2. The constant tension and constant speed cooperative regulation based ultra-fine tungsten wire drawing control method according to claim 1, characterized in that: In S1, the central processing system performs final pull-off stage identification on the collected data; When the data on the change in the winding diameter at the unwinding end indicates that the unwinding end is in the state of releasing excess winding, and the data on the change in the winding diameter at the take-up end indicates that the take-up end is in the state of accumulating winding, and the data on the slight drift of the end wire diameter indicates that the end wire diameter maintains a slight deviation during the continuous sampling period, and when there is a staggered change relationship between the short-range temperature change data before the final drawing die and the short-range temperature change data after the final drawing die, and when the residual fluctuation data of the adjacent previous pass continues into the final drawing pass, the central processing system determines that the target ultra-fine tungsten wire has entered the final drawing and finishing stage. The system retains the tension change data, speed change data, wire diameter change data at the pay-off end, wire diameter change data at the take-up end, short-range temperature change data before and after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous draw, forming the input dataset Inp for the final drawing stage.

3. The constant tension and constant speed cooperative regulation based ultra-fine tungsten wire drawing control method according to claim 2, characterized in that: In S2, the obtained final pull-off stage input dataset Inp is input into the final pull-off hidden stability reconstruction model Mdl; the final pull-off hidden stability reconstruction model Mdl is constructed through a gated temporal coding and decoding reconstruction framework; The final pull-in hidden stability reconstruction model Mdl includes a scene slice input layer, a roll diameter evolution coding layer, a short-range thermal state hysteresis coding layer, a tension velocity micro-drift coding layer, a previous pass residual wave propagation coding layer, a hidden stability coupling fusion layer, and an attenuation trajectory reconstruction layer. The scene slice input layer receives the input dataset Inp from the final pull-off stage; The output of the roll diameter evolution coding layer is characterized by the roll diameter coupling evolution between the unwinding end and the roll accumulation end. The short-range thermal state hysteresis coding layer outputs a misalignment feature between the short-range temperature change data before final die drawing and the short-range temperature change data after final die drawing. The tension micro-drift coding layer outputs the linkage offset characteristics between tension change data, velocity change data, and end wire diameter micro-drift data; The previous residual wave propagation coding layer outputs the propagation characteristics of the adjacent previous residual wave data to the final pull stage. The implicit stable coupling fusion layer performs alignment and fusion on the roll diameter coupling evolution features, front and rear misalignment features, linkage offset features and transmission features according to the same final pull-off stage temporal position to form the implicit stable coupling features of the final pull-off stage. The attenuation trajectory reconstruction layer reconstructs the continuous synchronous steady-state attenuation trajectory of the final pull-off stage based on the implicit stable coupling characteristics of the final pull-off stage, and forms the implicit stable state result Hid of the tail segment.

4. The constant tension and constant speed cooperative regulation based ultra-fine tungsten wire drawing control method according to claim 3, characterized in that: In S2, the construction of the final hidden stability reconstruction model Mdl includes a pre-training phase and a calibration training phase. The pre-training phase uses historical operating data from a continuous multi-pass drawing production line for ultra-fine tungsten wire to segment into steady-state maintenance segments, latent steady-state decay segments, and overt unstable segments. Among them, the preceding running segment before any result of end wire diameter jump, wire breakage, or speed reduction control switching occurs within the subsequent preset time period is determined as the implicit stable attenuation segment; The running segment that has already shown any of the following results: end wire diameter exceeding tolerance, wire breakage, or control loss of synchronization is identified as an explicit unstable segment; The running segment that does not exhibit abnormal control switching and whose end wire diameter remains stable is identified as the steady-state maintenance segment. The steady-state maintenance segment, the implicitly stable decay segment, and the explicit unstable segment are used to perform self-supervised pre-training on the gated temporal coding and decoding reconstruction framework, so that the representation of the roll diameter evolution relationship, the representation of the preceding and following misalignment relationship, the representation of the linkage offset relationship, and the representation of the transmission relationship form a temporal correlation. The calibration training phase involves back-calibrating historical running segments based on the results of end-wire diameter jumps, wire breakage, speed reduction control switching, and control step loss within a subsequent preset time period. This results in the formation of steady-state maintenance segments, implicit stability decay segments, and explicit instability segments. Based on these segments, the final pull implicit stability reconstruction model Mdl is calibrated and trained, enabling the output of the final pull implicit stability reconstruction model Mdl to correspond to the tail-end implicit stability state result Hid of the control switching during the final pull stage.

5. The constant tension and constant speed cooperative regulation based ultra-fine tungsten wire drawing control method according to claim 4, characterized in that: In S2, the tail-end hidden stability result Hid includes at least the short-term release stability start time, the short-term release stability duration interval, the tension velocity relock start condition, and the continuous synchronous steady-state decay level. Among them, the short-term release stabilization start point corresponds to the decay trajectory of the latent stability coupling characteristics in the final pull-off stage, at the time of entering the release stabilization trigger zone; The short-term stabilization duration interval corresponds to the decay trajectory being within the stabilization maintenance region; The initial condition for Zhang Su relocking corresponds to the condition for the decay trajectory to exit the stabilization maintenance region and enter the relockable region. The continuous synchronous steady-state attenuation level indicates the degree of attenuation of the continuous synchronous steady state during the final pull-off stage.

6. The method for controlling the drawing of ultra-fine tungsten wire based on the coordinated adjustment of constant tension and constant speed according to claim 5, characterized in that: In S3, the central processing system generates the destabilization switching result Rel based on the tail-end hidden stability state result Hid; The release stabilization switching result Rel includes the take-up drive unit speed unlock command, the release drive unit tension unlock command, and the short-term release stabilization duration interval release rhythm. The central processing system first sends a speed unlock command to the take-up drive unit, so that the take-up drive unit releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit. The central processing system then sends a tension unlocking command to the wire release drive unit, causing the wire release drive unit to release the tension locking relationship between the wire release drive unit and the final pull traction unit. During the short-term stabilization period, the central processing system maintains the short-term stabilization control mode according to the release rhythm of the short-term stabilization period.

7. The method for controlling the drawing of ultra-fine tungsten wire based on constant tension and constant speed coordinated adjustment according to claim 6, characterized in that: In S3, after the short-term destabilization period ends, the tension rate relocking result Lck is generated based on the tension rate relocking initiation condition. The tension relocking result Lck includes the reference traction speed establishment command, the take-up drive unit speed relocking command, the pay-off drive unit tension relocking command, and the preset relocking step size. The central processing system first sends a reference traction speed establishment command to the final traction unit, so that the final traction unit can establish the reference traction speed; The central processing system then sends a speed relock command to the take-up drive unit, so that the take-up drive unit can restore the speed lock relationship between the take-up drive unit and the final pull traction unit. Finally, the central processing system issues a tension relock command to the wire release drive unit, so that the wire release drive unit restores the tension lock relationship between the wire release drive unit and the final pull traction unit. The central processing system completes the tension speed relock step by step according to the preset relock step size.

8. The method for controlling the drawing of ultra-fine tungsten wire based on constant tension and constant speed coordinated adjustment according to claim 7, characterized in that: In S4, after executing the destabilization switching result Rel and the tension speed relocking result Lck, the central processing system continues to collect tension change data, speed change data, and end wire diameter change data after execution. The central processing system generates a steady-state recovery result Rcv based on the post-execution tension change data, the post-execution speed change data, and the post-execution end wire diameter change data; the steady-state recovery result Rcv includes at least the speed lock recovery state, the tension lock recovery state, and the end wire diameter stabilization state; When the steady-state recovery result Rcv does not meet the preset recovery conditions, the central processing system re-calls the destabilization switching result Rel and the tension speed relocking result Lck corresponding to the tail section hidden steady state result Hid, and continues to execute the short-term destabilization switching control and tension speed relocking control in the final tensioning stage.

9. The method for controlling the drawing of ultra-fine tungsten wire based on constant tension and constant speed coordinated adjustment according to claim 8, characterized in that: In S4, the central processing system pre-establishes the correspondence between the continuous synchronous steady-state decay level and the short-term release rhythm of the stabilization duration interval, the reference traction speed establishment command, and the preset relock step size. The central processing system, based on the continuous synchronous steady-state decay level, calls the short-term release stabilization duration interval release rhythm, the reference traction speed establishment command, and the preset relocking step size corresponding to the current continuous synchronous steady-state decay level in the corresponding relationship, and writes the called short-term release stabilization duration interval release rhythm into the release stabilization switching result Rel, and writes the called reference traction speed establishment command and the preset relocking step size into the tension speed relocking result Lck.

10. A control system for drawing ultra-fine tungsten wire based on constant tension and constant speed coordinated adjustment, applied to the ultra-fine tungsten wire drawing control method based on constant tension and constant speed coordinated adjustment as described in any one of claims 1-9, characterized in that: It includes a data acquisition module for the final tension, a modeling module for the final tension, a tension speed relocking module, and a coordinated adjustment module; The finishing data acquisition module collects tension change data, speed change data, wire diameter change data at the unwinding end, wire diameter change data at the take-up end, short-range temperature change data before and after the final drawing die, micro-drift data of the end wire diameter, and residual fluctuation data of the adjacent previous pass of the target ultra-fine tungsten wire during the final drawing stage, and sends them to the central processing system to form the final drawing stage input dataset Inp. The final pull modeling module inputs the final pull end stage input dataset Inp into the pre-built and pre-trained final pull hidden stability reconstruction model Mdl through the central processing system. The final pull hidden stability reconstruction model Mdl reconstructs the decay trajectory of the continuous synchronous steady state in the final pull end stage, and extracts the short-time release stabilization start point, short-time release stabilization duration interval and tension velocity relocking start condition from the decay trajectory to form the tail segment hidden stability state result Hid. The tension speed relocking module, based on the tail section hidden stability result Hid by the central processing system, at the short-term release stabilization start point, first releases the speed synchronization lock relationship between the take-up drive unit and the final pull traction unit in the order of the take-up drive unit and the release drive unit, and then releases the tension pull lock relationship between the release drive unit and the final pull traction unit, forming the release stabilization switching result Rel; when the tension speed relocking start condition is met, the central processing system restores the speed lock relationship and tension lock relationship in the order of the final pull traction unit, the take-up drive unit and the release drive unit, forming the tension speed relocking result Lck. The coordinated adjustment module outputs linkage control commands through the central processing system based on the release stabilization switching result Rel and the tension speed relocking result Lck, and executes short-term release stabilization switching and tension speed relocking control in the final tensioning stage to complete the coordinated adjustment of the final tensioning stage of the ultra-fine tungsten wire with the control objective of maintaining constant tension and constant speed.

Images