A copper rod manufacturing process linkage control system and method

Abstract

The present application relates to the field of metal continuous casting automation control, and discloses a copper rod processing and manufacturing process linkage control system and method, the method comprising: dividing the cast blank into virtual slices according to the pulse signal to construct a data queue; calculating the heat content characteristics of the temperature and flow signals, processing abnormal data and marking characteristic anchor points; combining the linear speed to deduce the theoretical time stamp of the slice reaching the rack; comparing the torque differential polarity and the heat content mutation direction within the theoretical time window to verify the signal, extracting the actual time stamp to correct the theoretical time stamp; generating a feedforward torque instruction according to the heat content characteristics and issuing it to the frequency converter. Through the cross-checking of thermal state and mechanical dynamics and the adaptive correction of the time stamp, the temperature measurement distortion interference is eliminated, the spatial accurate alignment is realized, the torque reserve is established before the abnormal material contacts the rack, the hysteresis of the traditional passive control is overcome, the tension shock is reduced, and the risk of belt breakage is reduced.

Description

Technical Field

[0001] This invention relates to the field of automated control for continuous casting and rolling of metals, specifically to a linkage control system and method for copper rod processing and manufacturing. Background Technology

[0002] Continuous casting and rolling of copper rods is a core process in cable manufacturing, mainly including metal smelting, continuous casting on a crystallizing wheel, and multi-stand continuous rolling. To maintain production continuity and stability, the control system typically needs to monitor production line operating parameters in real time. During the continuous rolling stage, the unit generally uses a variable frequency drive (VFD)-based speed closed-loop control system to adjust motor speed and control material tension between stands. Simultaneously, optical temperature measurement equipment is installed in the transition area between the crystallizing wheel outlet and the continuous rolling mill inlet to collect the surface temperature of the cast billet, assessing the stability of the upstream casting process and providing data reference for subsequent processing.

[0003] In the on-site operating environment, water vapor splashing from the casting area cooling system often obstructs optical temperature measurement equipment, causing fluctuations and distortions in the acquired temperature signals. Relying solely on feedback data from a single optical sensor, the system struggles to accurately distinguish between genuine localized temperature changes in the material and spurious signals generated by environmental interference. Furthermore, the continuous rolling process involves the plastic expansion of the metal material, and mechanical equipment operation is subject to transmission wear and material slippage. Conventional material tracking methods primarily rely on time integration of linear velocity to calculate displacement. This approach accumulates spatial calculation errors during long-term continuous operation, making it difficult for the control system to accurately pinpoint the actual time and spatial location of abnormal materials arriving at downstream stands.

[0004] When materials with genuine localized temperature abrupt changes enter the continuous rolling mill stand, their deformation resistance increases accordingly. Existing speed closed-loop control systems rely on passive feedback mechanisms, requiring the regulator to wait for a physical disturbance to contact the stand and cause a drop in the actual motor speed before outputting a torque compensation command. Limited by the inherent response time of the mechanical transmission chain and basic electrical control, this passive adjustment method cannot proactively intervene before abnormal materials enter the deformation zone. The delayed torque response is insufficient to effectively mitigate transmission system oscillations caused by material hard point impacts, easily leading to instability in processing tension between stands and increasing the risk of strip breakage during continuous production. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a linkage control system and method for the copper rod processing and manufacturing process. It solves the problems of increased material deformation resistance and tension oscillation in the transmission system caused by local temperature abrupt changes and hard points resulting from process fluctuations entering the continuous rolling mill during the continuous casting and rolling production of copper rods, as well as the problems of traditional temperature measurement methods being susceptible to distortion due to water vapor interference and the mechanical response lag of the basic speed closed-loop control system, which prevents active intervention.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of the present invention provides a cross-domain linkage control method for the continuous casting and rolling process of copper rods, comprising the following steps: S10, acquire the angular displacement pulse signal of the first encoder, divide the continuously generated physical billet into a virtual slice sequence of preset length according to the angular displacement pulse signal, and build a first-in-first-out data queue mapped to the physical space in the controller. S20: Simultaneously receive the surface temperature signal of the thermometer and the flow signal of the flow meter, calculate the time change gradient of the surface temperature signal, and determine that the temperature measurement signal is distorted when the time change gradient is greater than the preset water vapor blocking threshold. Use the historical enthalpy data in the first-in-first-out data queue to perform spatial polynomial interpolation to generate interpolation feature index to replace the current abnormal data, calculate the enthalpy feature index corresponding to each virtual slice, perform sliding window difference calculation on the enthalpy feature index sequence in the first-in-first-out data queue, extract feature slices with a range absolute value greater than the steady-state dead zone threshold and mark them as feature anchor points. S30 receives the linear velocity signal from the second encoder, uses the linear velocity signal as a reference to drive the first-in-first-out data queue to perform shift calculations, and combines the rack spacing to deduce the theoretical expected timestamps of each virtual slice reaching the deformation zone of each downstream rack. S40, when the virtual slice marked with feature anchor points arrives at the rack in the time window, the actual electromagnetic torque data of the strain gauge is collected, low-pass filtering and time first-order differential operation are performed on the actual electromagnetic torque data, and the polarity direction of the time first-order differential operation result is compared with the enthalpy change direction corresponding to the feature anchor point. If the polarity direction does not match, the feature anchor point is determined to be a sensor distortion artifact and the queue anchor point fuse program is triggered. If the polarity direction matches, the extreme point time where the absolute value of the differential result is greater than the noise dead zone threshold is extracted as the actual arrival timestamp. The kinematic slip coefficient is calculated using the actual arrival timestamp and the theoretical expected timestamp corresponding to the downstream rack is reconstructed. S50: When a virtual slice that has not triggered the anchor point fuse procedure enters the electromechanical response time window, a feedforward torque additional command is generated based on the thermal enthalpy characteristic index and sent to the corresponding strain gauge. After determining that the virtual slice has left the deformation zone, the feedforward torque additional command is canceled according to the attenuation slope.

[0007] Furthermore, based on the angular displacement pulse signal, the continuously generated physical billet is divided into a sequence of virtual slices of a preset length. Specifically, this includes: acquiring the pulse increment value fed back by the first encoder within the control scanning cycle; multiplying the ratio of the pulse increment value to the single-turn pulse resolution by pi and the working pitch circle diameter of the crystallizing wheel to calculate the physical displacement increment; performing an accumulation operation on the physical displacement increment using a continuous displacement accumulation register; generating a virtual slice with an incrementing number when the accumulated value reaches or exceeds the set slice length limit; subtracting the slice length limit from the accumulated value and retaining the difference as the initial base for the next round of accumulation.

[0008] Furthermore, the addressing depth of the first-in-first-out (FIFO) data queue is equal to the ratio of the total physical path length of the physical billet moving from the crystallizing wheel outlet to the downstream end stand outlet to the fixed slice length. The FIFO data queue consists of multiple data structure units arranged in address order. The data structure units are divided into an area for storing numerical numbers, a timestamp area for recording the system reference time and expected arrival time, and a feature data area for storing the enthalpy characteristic index and logical judgment flag bits.

[0009] Furthermore, spatial polynomial interpolation is performed using historical enthalpy data from the first-in-first-out (FIFO) data queue to generate interpolation feature exponents to replace current abnormal data. Specifically, this includes: when the temperature measurement signal is determined to be distorted, actively discarding the currently collected abnormal temperature data, tracing back and extracting the effective enthalpy feature data of adjacent historical nodes in the FIFO data queue; constructing a spatial polynomial using the effective enthalpy feature data, and calculating the theoretical enthalpy estimate of the current spatial location through an interpolation algorithm, and storing the theoretical enthalpy estimate as replacement data in the feature data area corresponding to the virtual slice.

[0010] Furthermore, after determining that the temperature measurement signal is distorted, the system accumulates the number of consecutively distorted slices. When the number of consecutively distorted slices exceeds the set fault tolerance limit, the polynomial interpolation operation is automatically paused, and the feature data is maintained at the most recent valid historical steady-state value until the surface temperature fed back by the sensor returns to the allowable range and the absolute value of the time change gradient is continuously lower than the water vapor blocking threshold, at which point the abnormal degradation maintenance state is exited.

[0011] Furthermore, the enthalpy characteristic index corresponding to each virtual slice is calculated. The specific process is as follows: the effective surface temperature is multiplied by the first thermodynamic weighting coefficient to obtain the first product; the instantaneous volumetric flow rate of cooling water collected by the flow meter is calculated as the ratio of the current billet linear velocity, and the ratio is multiplied by the second thermodynamic weighting coefficient to obtain the second product; the first product is subtracted from the second product to obtain the enthalpy characteristic index of the current virtual slice.

[0012] Furthermore, a sliding window difference calculation is performed on the enthalpy characteristic index sequence in the first-in-first-out data queue. Specifically, this includes: defining a sliding data window containing a specific number of virtual slices in memory, calculating the maximum and minimum values ​​of the enthalpy characteristic index within the sliding data window in real time to obtain the window range; when the window range is greater than a preset steady-state dead zone threshold, comparing the enthalpy gradient values ​​of each slice within the sliding window, selecting the core virtual slice with the steepest gradient change, and writing feature anchor point markers into its data structure unit.

[0013] Furthermore, the first-in-first-out data queue is driven by the linear velocity signal as a reference for shift calculation. Specifically, this includes: determining whether the linear velocity fed back by the second encoder is lower than the preset speed dead zone threshold. If it is lower than the speed dead zone threshold, the queue shift operation is paused to filter out minor electrical jitter. If the linear velocity is higher than the speed dead zone threshold, the linear velocity is integrated over time to obtain the instantaneous displacement. When the accumulated value of the instantaneous displacement reaches the slice length set value, a shift instruction is triggered to offset all data structure units in the queue to the next level address by one unit.

[0014] Furthermore, by combining the rack spacing, the theoretical expected timestamps of each virtual slice reaching the deformation zone of each downstream rack are deduced. The specific calculation process is as follows: the current system absolute time base value is used as the initial term; the remaining physical distance between the virtual slice and the corresponding rack is calculated, and the remaining physical distance is equal to the total length of the fixed physical path minus the product of the dynamic queue index number of the virtual slice and the fixed slice length; the remaining physical distance is divided by the material transfer speed after combining the forward slip coefficient correction to obtain the transfer time; the transfer time is added to the initial term to obtain the theoretical expected timestamp.

[0015] Furthermore, when the virtual slice with the feature anchor mark arrives at the rack's time window, the process specifically includes: using the theoretical expected timestamp minus a preset lead time as the window start time, and using the theoretical expected timestamp plus a preset lag time as the window end time, and continuously sampling actual electromagnetic torque data within this dynamic capture window.

[0016] Furthermore, the polarity direction of the first-order time derivative calculation result is compared with the enthalpy change direction corresponding to the feature anchor point. The polarity matching logic is as follows: when the enthalpy change direction is negative and the polarity direction of the first-order torque-time derivative is positive, or when the enthalpy change direction is positive and the polarity direction of the first-order torque-time derivative is negative, it is determined that the two have a physically negative correlation matching relationship; if the negative correlation matching relationship is not satisfied, it is determined that the polarity direction does not match.

[0017] Furthermore, the queue anchor point circuit breaker procedure includes: erasing the characteristic anchor point marker bit of the corresponding virtual slice in the first-in-first-out data queue, clearing its corresponding theoretical expected timestamp, and blocking the issuance process of feedforward torque additional instructions for the slice.

[0018] Furthermore, the kinematic slip coefficient is calculated and the theoretical expected timestamp corresponding to the downstream rack is reconstructed. Specifically, this includes: subtracting the theoretical expected timestamp from the actual physical timestamp to calculate the time deviation; multiplying the time deviation by the current actual linear velocity to obtain the spatial displacement compensation; dividing the spatial displacement compensation by the slice length fixed value and performing a rounding operation to obtain the discrete pointer index offset; superimposing the pointer index offset onto the current address addressing pointer of the data queue to achieve physical alignment; inputting the time deviation into the discrete proportional-integral regulator; using the calculated proportional adjustment term and integral adjustment term to sum and correct the forward slip coefficient of the previous cycle; and outputting the updated forward slip coefficient for the timestamp extrapolation of subsequent racks.

[0019] Furthermore, a feedforward torque additional instruction is generated based on the enthalpy characteristic index, specifically including: obtaining the steady-state reference enthalpy index of the production line, calculating the difference between the steady-state reference enthalpy index and the currently extracted enthalpy characteristic index; multiplying the difference by a pre-calibrated material stiffness conversion coefficient to calculate the theoretical feedforward torque compensation amount; and performing absolute amplitude limiting processing on the theoretical feedforward torque compensation amount when the absolute value of the theoretical feedforward torque compensation amount is greater than the compensation dead zone threshold.

[0020] Furthermore, when issuing the feedforward torque additional command to the frequency converter, a smooth handover mechanism for control is introduced. The smooth handover mechanism for control uses a first-order inertial filtering algorithm to smooth the theoretical feedforward torque compensation amount in the step form on the time axis, generating a smooth torque feedforward command, and superimposing the smooth torque feedforward command onto the basic torque command channel of the frequency converter.

[0021] Furthermore, after determining that the virtual slice has left the deformation zone, the feedforward torque additional command is cancelled according to the attenuation slope. Specifically, this includes: calculating the current position of the virtual slice in the absolute space coordinate system in real time; when the tail coordinate of the virtual slice crosses the exit coordinate of the frame, forcing the theoretical feedforward torque compensation amount to zero; using a first-order inertial filtering algorithm to attenuate the smooth torque feedforward command superimposed on the frequency converter to zero according to the predetermined inertial time constant, and smoothly returning the control of the main motor to the original speed closed-loop regulator.

[0022] A second aspect of the present invention provides a cross-domain linkage control system for a continuous casting and rolling process of copper rods, comprising: a controller for performing logical operations, maintaining a data queue, and issuing torque commands; upstream casting zone equipment, including a crystallizing wheel driven by a drive motor, a first encoder coaxially mounted on the main shaft of the crystallizing wheel, a temperature measuring instrument installed at the outlet end of the crystallizing wheel, and a flow meter installed in the cooling water circuit of the crystallizing wheel; transition zone equipment, including a speed measuring roller located between the crystallizing wheel and the downstream continuous rolling zone, a second encoder mounted on the shaft end of the speed measuring roller; and downstream continuous rolling zone equipment, including a first stand to an end stand arranged in series, each stand being driven by a corresponding motor, and each motor being independently controlled by a frequency converter supporting torque control mode; the controller is connected to the first encoder, the temperature measuring instrument, the flow meter, the second encoder, and the frequency converter corresponding to each stand via a communication network.

[0023] This invention provides a linkage control system and method for the copper rod processing and manufacturing process. It has the following beneficial effects: 1. This invention calculates the enthalpy characteristic index of surface temperature and flow rate signals, processes abnormal data using spatial polynomial interpolation, and collects the first-order differential polarity of the inverter torque within the extrapolated time window, cross-comparing it with the direction of enthalpy abrupt change. This mechanism utilizes the physical correspondence between thermodynamic state and mechanical dynamics feedback for bidirectional verification, eliminating optical temperature measurement distortion caused by on-site water vapor obstruction and reducing the risk of miscontrol caused by environmental interference.

[0024] 2. This invention extracts the actual arrival timestamps corresponding to the extreme points of the torque response, calculates the time deviation between these timestamps and the theoretical timestamps, and then calculates the spatial displacement compensation to correct the addressing pointers of the data queue, while simultaneously updating the forward slip coefficients between racks. This method achieves closed-loop adaptive correction of the kinematic derivation model, reduces the cumulative error of displacement integrals caused by equipment wear and physical slippage, and ensures that the control end can track abnormal material nodes on the production line in real time and accurately.

[0025] 3. This invention calculates the feedforward torque compensation based on the extracted enthalpy characteristic index before abnormal material enters the frame deformation zone, and then adds it to the torque command channel of the frequency converter after filtering and smoothing. This feedforward intervention method based on precise spatial alignment compensates for the lag defect of traditional speed closed-loop control, which requires waiting for the speed to drop before responding passively. It establishes a torque reserve before abnormal material contacts the frame, reduces tension fluctuations caused by mechanical impact, and lowers the risk of belt breakage during continuous processing. Attached Figure Description

[0026] Figure 1 This is a system framework diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention; Figure 3 This is a schematic diagram of the spatial discretization and data queue construction logic of the present invention; Figure 4 This is a schematic diagram of the thermodynamic state calibration and beacon anchoring logic of the present invention; Figure 5 This is a schematic diagram of the deduction-driven and spatiotemporal prediction solution logic of the present invention; Figure 6 This is a schematic diagram of the extreme value capture and dynamic adaptive reconstruction logic of the present invention; Figure 7 This is a schematic diagram of the feedforward torque compensation and smooth handover logic of the present invention; Figure 8 The above are comparison images of the feature decoupling extraction effect of the present invention, where (a) is the original grayscale image, (b) is the feature map output by the traditional mixed photometric evaluation algorithm, and (c) is the morphology gradient extraction map output by the feature decoupling method of the present invention. Figure 9 This is a comparison of the dynamic curves of tension response of different control systems when an abnormal cold spot suddenly enters the rolling mill stand according to the present invention.

[0027] Among them, 10 is the controller; 20 is the crystallizing wheel; 21 is the drive motor; 22 is the first encoder; 23 is the thermometer; 24 is the flow meter; 30 is the speed measuring roller; 31 is the second encoder; 40 is the continuous rolling mill; 41 is the first frame; 42 is the first motor; 43 is the first frequency converter; 44 is the nth frame; 45 is the nth motor; 46 is the nth frequency converter; and 50 is the communication network. Detailed Implementation

[0028] The technical solutions in 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.

[0029] See attached document Figure 1 , Figure 1 This is a hardware architecture diagram of the linkage control system according to the present invention. The present invention provides a linkage control system for copper rod processing and manufacturing, including a controller 10 and field devices connected to the controller 10 via a communication network 50. The field devices are distributed in the upstream casting zone, the transition zone, and the downstream continuous rolling zone.

[0030] A crystallizing wheel 20 is installed in the upstream casting zone. A drive motor 21 is connected to the crystallizing wheel 20 and drives it to rotate. A first encoder 22 is coaxially mounted on the main shaft of the crystallizing wheel 20. The first encoder 22 collects the angular displacement of the crystallizing wheel 20 and outputs an angular displacement pulse signal characterizing the length of the cast billet. A thermometer 23 is installed at the outlet end of the crystallizing wheel 20. The thermometer 23 collects the surface temperature of the cast billet after it leaves the crystallizing wheel 20. A flow meter 24 is installed in the cooling water circuit of the crystallizing wheel 20. The flow meter 24 collects the instantaneous volumetric flow rate of the cooling water circuit of the crystallizing wheel 20.

[0031] A speed measuring roller 30 is provided in the transition zone. The speed measuring roller 30 is located between the crystallizing wheel 20 and the downstream continuous rolling zone. A second encoder 31 is installed at the shaft end of the speed measuring roller 30. The second encoder 31 collects the speed of the speed measuring roller and outputs the linear velocity signal of the billet before it enters the deformation zone.

[0032] The downstream continuous rolling zone is equipped with a continuous rolling mill 40, which includes a first stand 41 to an nth stand 44 arranged in series. A first motor 42 drives the first stand 41 to operate. A first frequency converter 43 is electrically connected to and controls the first motor 42. An nth motor 45 drives the nth stand 44 to operate. An nth frequency converter 46 is electrically connected to and controls the nth motor 45. Both the first frequency converter 43 and the nth frequency converter 46 support torque control mode. The first frequency converter 43 and the nth frequency converter 46 feed back the actual value of electromagnetic torque to the controller 10 through the communication network 50 and receive the feedforward torque additional command issued by the controller 10. The controller 10 has a built-in feature authenticity backtracking verification logic. By linking and comparing the dynamic response characteristics of the first frequency converter 43 with the thermodynamic history record of the temperature measuring instrument 23, it identifies and filters optical sensor artifacts caused by on-site cooling water vapor obstruction.

[0033] See attached document Figure 2 , Figure 2 This is a flowchart of the linkage control method according to the present invention. The present invention provides a linkage control method for a copper rod processing and manufacturing process, comprising the following steps: S10, acquire the angular displacement pulse signal of the first encoder 22, divide the continuously generated physical billet into a virtual slice sequence of preset length according to the angular displacement pulse signal, and build a first-in-first-out data queue mapped to the physical space in the controller 10. S20, synchronously receive the surface temperature signal of the thermometer 23 and the flow signal of the flow meter 24, calculate the time change gradient of the surface temperature signal, and determine that the temperature measurement signal is distorted when the time change gradient is greater than the preset water vapor blocking threshold. Use the historical enthalpy data in the first-in-first-out data queue to perform spatial polynomial interpolation to generate interpolation feature index to replace the current abnormal data, calculate the enthalpy feature index corresponding to each virtual slice, perform sliding window difference calculation on the enthalpy feature index sequence in the first-in-first-out data queue, extract the feature slices with the absolute value of the range greater than the steady-state dead zone threshold and mark them as feature anchor points; S30, receives the linear velocity signal from the second encoder 31, uses the linear velocity signal as a reference to drive the first-in-first-out data queue to perform shift calculation, and combines the rack spacing to deduce the theoretical expected timestamps of each virtual slice reaching the deformation zone of each rack. S40, when the virtual slice marked with feature anchor points arrives at the time window of the first rack 41, the actual electromagnetic torque data of the first frequency converter 43 is collected, low-pass filtering and time first-order differential operation are performed on the actual electromagnetic torque data, and the polarity direction of the time first-order differential operation result is compared with the enthalpy change direction corresponding to the feature anchor point. If the polarity direction does not match, the feature anchor point is determined to be a sensor distortion artifact and the queue anchor point fuse program is triggered. If the polarity direction matches, the extreme point time where the absolute value of the differential result is greater than the noise dead zone threshold is extracted as the actual arrival timestamp. The kinematic slip coefficient is calculated using the actual arrival timestamp and the theoretical expected timestamp corresponding to the downstream rack is reconstructed. S50, when a virtual slice that has not triggered the anchor point fuse procedure enters the electromechanical response time window, a feedforward torque additional command is generated based on the thermal enthalpy characteristic index and sent to the nth frequency converter 46. After determining that the virtual slice has left the deformation zone of the nth frame 44, the feedforward torque additional command is canceled according to the attenuation slope.

[0034] See attached document Figure 3 , Figure 3 This is a schematic diagram of the spatial discretization and data queue construction logic according to the present invention. In this embodiment, to achieve state tracking of continuously produced metal materials, they need to be converted into locationable discrete data nodes within the control system. From a physical perspective, each rotation of the crystallizing wheel 20 at a certain angle will result in the output of a solid metal billet of corresponding arc length. Based on this correspondence, the controller 10 receives the angular displacement pulse signal output by the first encoder 22 in real time through its internal high-speed counting channel, and then calculates the material's formation length.

[0035] As a preferred method, the internal processor of controller 10 reads the pulse increment value of the high-speed counting channel according to the set main control cycle scan period. The main cycle scan period is typically set based on the refresh rate of the field communication bus, and its value ranges from 1 millisecond to 5 milliseconds. For the hardware anti-interference isolation and quadrature decoding processing of the encoder pulse signal, those skilled in the art can use conventional high-speed counting module hardware circuits combined with underlying firmware to implement it. Its pulse anti-interference acquisition is a well-known technology in the field and will not be elaborated upon here. Controller 10 converts the pulse increment read within a single scan cycle into the physical displacement increment of the casting billet. The formula for calculating the physical displacement increment is as follows: ; In the formula, Represents the physical displacement increment of the physical billet within a single control scan cycle; This represents the pulse increment value fed back by the first encoder 22 within the current scan cycle; This represents the single-turn pulse resolution of the first encoder 22; This represents the working pitch circle diameter of crystallizing wheel 20.

[0036] After acquiring the instantaneous length increment, the system further performs logical-level fixed-length segmentation of the continuous material. The controller 10 sets up a continuous displacement accumulation register in memory, using this register to calculate the physical displacement increment. Perform an accumulation operation. The value in the continuous shift accumulation register reaches or exceeds the set slice length. At this time, controller 10 generates a virtual slice object at the logic level. To avoid the continuous accumulation of long-cycle truncation errors, the system does not directly force the register to zero, but instead subtracts the accumulated value in the current register. The difference in computation is retained as the initial base for the next round of accumulation. Through this pulse accumulation and length threshold triggering mechanism, the continuously generated physical billets are cut into discrete virtual slice sequences at equal intervals.

[0037] Each virtual slice in the virtual slice sequence is identified by an incrementing numerical number according to the order in which it was generated. The slice length is fixed. The constants are determined based on the effective arc length of the deformation zone of the downstream continuous rolling mill 40 and the memory addressing limit of the controller 10. In typical copper rod processing scenarios, It is usually set between 0.1 meters and 0.5 meters.

[0038] To store and transmit the data of the aforementioned virtual slices, the controller 10 allocates a contiguous addressable storage area in its internal memory to construct a first-in-first-out (FIFO) data queue. The addressing depth of the FIFO data queue is equal to the total physical path length of the physical billet moving from the exit of the crystallizing wheel 20 to the exit of the downstream nth stand 44, multiplied by a fixed slice length. The ratio of .

[0039] When the ratio has a decimal part, the controller 10 performs a rounding up operation to determine the final queue length. The first-in-first-out (FIFO) data queue consists of multiple data structure units arranged in address order, each data structure unit uniquely corresponding to a virtual slice in the virtual slice sequence. When a new virtual slice object is generated, the controller 10 triggers an enqueue operation, writing the initial state data of the virtual slice into the first address storage unit of the FIFO data queue.

[0040] To accommodate multi-source state information, the controller 10 defines the data dimensions of the data structure unit. The data structure unit is configured as a composite data type in the underlying code, internally divided into an identification area, a timestamp area, and a feature data area. The identification area stores the numerical identifier of the virtual slice. The timestamp area records the system reference time when the virtual slice leaves the crystallization wheel 20 outlet, and reserves address space for storing the expected arrival time variables of each rack.

[0041] The feature data area is configured as a floating-point data structure to receive and store the enthalpy characteristic index and multiple logic judgment flags extracted in subsequent calculation steps. This first-in-first-out data queue based on composite data type maps the continuously flowing metal material on the physical production line to a queue array with independent data tags inside the controller 10, establishing a data traceability channel for cross-process domain linkage control.

[0042] See attached document Figure 4 , Figure 4 This is a schematic diagram of the thermodynamic state calibration and beacon anchoring logic according to the present invention. In this embodiment, the controller 10 needs to fuse multi-source sensor data and filter environmental interference to extract feature reference points that can characterize the true physical state of the metallic material.

[0043] Considering that the sampling periods of the thermometer 23 and the flowmeter 24 typically differ at the hardware level, the controller 10 internally constructs a unified system synchronization clock to achieve alignment of multi-source data in spatiotemporal coordinates. When a new virtual slice is stored at the first address of the first-in-first-out data queue, the controller 10 synchronously captures the surface temperature signal fed back by the thermometer 23 and the flow signal fed back by the flowmeter 24 at the current moment. For analog-to-digital conversion and low-pass filtering of analog signals, those skilled in the art can use conventional PLC analog input modules; the hardware filtering and digital conversion are well-known technologies in the field and will not be elaborated upon here.

[0044] In continuous casting and rolling mills, the drift and obstruction of cooling water vapor can easily hinder the optical measurement channel of the temperature measuring instrument 23, leading to a momentary distortion and drop in the surface temperature signal. To identify this type of physical distortion, the system is configured with water vapor obstruction gradient verification logic. The controller 10 calculates the time-varying gradient of the surface temperature signal. This gradient is calculated by subtracting the temperature value of the previous sampling cycle from the temperature value of the current sampling cycle, and then dividing by the system scan cycle time. Due to the high heat capacity of metallic materials, their internal and surface temperatures are generally unlikely to undergo drastic changes in a very short time. The system compares the absolute value of the calculated time-varying gradient with a preset water vapor obstruction threshold.

[0045] As a preferred method, the moisture blocking threshold is calibrated based on the maximum theoretical cooling rate of the metal material. In conventional copper rod processing, this threshold is typically set to 30℃ / s to 50℃ / s. When the absolute value of the time-varying gradient exceeds the aforementioned moisture blocking threshold, the controller 10 determines that the current surface temperature signal is distorted and identifies it as an optical sensor artifact caused by on-site moisture interference.

[0046] Instantaneous distortion of the temperature measurement signal can disrupt the continuity of thermodynamic states in the first-in-first-out (FIFO) data queue. Upon determining that the surface temperature signal is distorted, the controller 10 immediately triggers the spatial polynomial interpolation reconstruction mechanism under abnormal conditions. At this time, the system actively discards the currently acquired abnormal temperature data and instead traces back to extract the effective enthalpy characteristic data of adjacent historical nodes in the FIFO data queue.

[0047] The controller 10 uses these historical data points to construct a second- or third-order Lagrange polynomial, and then uses a spatial polynomial interpolation algorithm to calculate the theoretical enthalpy estimate for the current spatial location. The system generates this interpolation characteristic exponent and stores it as substitute data in the characteristic data area corresponding to the virtual slice. To prevent the interpolation calculation from diverging due to continuous and prolonged water vapor obstruction, the controller 10 is equipped with an anomaly counter.

[0048] When the number of consecutively distorted slices exceeds the set fault tolerance limit, for example, twenty consecutively distorted slices, the system automatically pauses the interpolation operation, maintaining the feature data at the most recent valid historical steady-state value. The system continuously monitors the feedback signal from the thermometer 23 until the absolute value of the surface temperature returned to the reasonable allowable range corresponding to the valid historical steady-state value, and the absolute value of the time change gradient within a consecutive preset number of sampling periods is lower than the water vapor blocking threshold. Only then does the system determine that the sensor signal has truly returned to normal and exit the abnormal degradation maintenance state. This interpolation substitution and degradation mechanism repairs the data gaps caused by short-term water vapor interference and maintains the sequence integrity of subsequent peak-finding logic.

[0049] After ensuring the input data is valid or completing interpolation repair, the controller 10 calculates the corresponding enthalpy characteristic index for each virtual slice. From a physical mechanism perspective, the actual thermodynamic state of the metal after it leaves the crystallizing wheel 20 depends not only on the surface temperature measured optically, but also on the combined effects of the heat carried away by the cooling water and the casting speed. Based on this, the controller 10 combines surface temperature, cooling water flow rate, and billet linear velocity to calculate the enthalpy characteristic index using a multivariate linear equation. The calculation formula is as follows: ; In the formula, The enthalpy characteristic index representing the current virtual slice; The effective surface temperature collected by the thermometer 23 or the effective temperature reconstructed after evaluation; The instantaneous volumetric flow rate of the cooling water circuit of crystallizer 20, collected by flow meter 24; This represents the instantaneous linear velocity of the cast billet calculated by the first encoder 22 pulses; and These are pre-calibrated thermodynamic weighting coefficients. The specific values ​​of these two coefficients are determined by fitting historical operating data based on the cooler structure and heat exchange efficiency of the production line. The value ranges from 0.8 to 1.2. The value ranges from 0.05 to 0.2.

[0050] After acquiring the enthalpy characteristic index sequence, the controller 10 performs sliding window difference calculation on the data sequence in the first-in-first-out data queue. The controller 10 defines a sliding data window in memory containing a specific number of virtual slices, typically set to 10 to 50. As the queue data shifts sequentially, the system calculates the maximum and minimum values ​​of the enthalpy characteristic index within this window in real time, thereby deriving the window range. The magnitude of this range directly reflects the degree of drastic change in the enthalpy state within the local space.

[0051] When the range exceeds the preset steady-state dead zone threshold, it indicates that a significant abrupt change in the hot and cold states of the metal material has occurred in that local area. The steady-state dead zone threshold should be set slightly larger than the amplitude of the system background thermal noise fluctuation during steady-state operation of the production line. Once the calculated range exceeds the steady-state dead zone threshold, the controller 10 further compares the enthalpy gradient values ​​of each slice within the sliding window, selects the core virtual slice with the steepest gradient change, and confirms it as a feature anchor point in the judgment flag bit of its data structure unit. This virtual slice with the feature anchor point mark will subsequently be transmitted to the downstream rack as a spatiotemporal reference beacon for dynamic reconstruction verification.

[0052] See attached document Figure 5 , Figure 5 This is a schematic diagram of the deduction-driven and spatiotemporal prediction solution logic according to the present invention. In this embodiment, the system needs to establish a purely theoretical kinematic feedforward deduction model in a physical region lacking direct sensor measurement, and then calculate the expected time points for each virtual slice to reach each downstream rolling mill stand. The related processing is divided into the following sub-steps: The physical displacement of continuously flowing metal needs to be mapped in real time to the address offset of the internal memory of controller 10 to maintain synchronization between the digital queue and the physical production line. To this end, controller 10 obtains the actual linear velocity of the metal material at the inlet of the continuous rolling mill 40 via the second encoder 31. Since there is usually a certain material free-hanging transition zone between the crystallizing wheel 20 and the continuous rolling mill 40, measuring the linear velocity using the second encoder 31 at the inlet can more accurately reflect the motion state of the material as it enters the rolling stage. Controller 10 performs time integration on this linear velocity to obtain the instantaneous displacement of the material within the transition zone. When the accumulated value of the instantaneous displacement reaches or exceeds the slice length setpoint... At that time, controller 10 triggers a shift instruction for the first-in-first-out data queue.

[0053] To ensure the absolute conservation of spatial length mapping, the system subtracts the slice length constant from the current instantaneous cumulative displacement value when a shift command is triggered. The resulting operational difference is used as the initial integration base for the next control cycle, thus avoiding the accumulation of truncation errors. Under the action of the shift instruction, the data in all data structure units in the queue will be shifted as a whole to the next memory address by one unit.

[0054] As a preferred approach, to prevent erroneous queue shifts caused by minute electrical jitter in the speed signal during temporary production line shutdowns or extremely low-speed tape threading, the controller 10 incorporates speed dead-zone detection logic before integration. When the linear velocity fed back by the second encoder 31 is lower than a preset speed dead-zone threshold, the system suspends the queue displacement accumulation and shifting operations. The speed dead-zone threshold is typically set based on the encoder's underlying noise level, for example, it can be set to 0.02 m / s to 0.05 m / s, thereby ensuring that the spatial mapping of the data does not undergo non-physical distortion.

[0055] As the queue continuously shifts sequentially, to establish a time reference for subsequent dynamic comparisons, controller 10 performs dynamic calculations of theoretically expected timestamps for virtual slices containing feature anchor point markers in the sequence. The calculation process is based on the principle of volumetric flow rate conservation in metal plastic deformation. Under ideal no-slip steady-state operation, the time required for material to traverse a specific physical distance can be inferred from the ratio of the remaining path length to the current linear velocity.

[0056] The controller 10 constructs a one-dimensional absolute spatial coordinate system with the crystallizing wheel 20 outlet as the origin at the bottom layer, and reads the internally stored equipment mechanical parameters to extract the fixed absolute physical coordinates of the deformation zone at the entrance of the first frame 41 in this coordinate system. Subsequently, the system uses the current system absolute clock and kinematic coordinate equations to calculate the theoretical expected timestamp of the virtual slice reaching the first frame 41. The formula for calculating the theoretical expected timestamp is as follows: ; In the formula, The theoretically expected timestamp representing the arrival of the virtual slice with the characteristic anchor mark at the first rack 41; This represents the system absolute time base value currently read by controller 10; This represents the total length of the fixed physical path from the outlet of crystallizing wheel 20 to the center of the deformation zone of the first frame 41; This represents the current address index of the virtual slice in the first-in-first-out data queue. The index number dynamically increments as the queue shifts. This represents a fixed slice length. This represents the current feed line speed measured by the second encoder 31.

[0057] Through the algebraic mapping of the aforementioned spatial coordinates, the numerator part calculates the actual remaining physical distance between the virtual slice and the first stand 41. For the subsequent nth stand 44 within the continuous rolling mill 40, due to the volumetric elongation and forward slip effect of the metal material in the continuous rolling deformation zone, its actual transfer speed between stands is not equal to the theoretical circumferential linear velocity of the work rolls of each stand. Therefore, the controller 10 not only uses the fixed mechanical center distance between each stand and the theoretical elongation rate of continuous rolling, but also introduces a pre-calibrated forward slip coefficient to correct the theoretical linear velocity, thereby calculating the actual material transfer speed between adjacent stands. Based on the corrected material transfer speed, the system uses the remaining physical position removal method to perform step-by-step time accumulation calculation of the transfer speed.

[0058] After calculating the theoretically expected timestamps of the virtual slice reaching subsequent racks, the controller 10 writes them into the reserved address space within the timestamp area of ​​the corresponding data structure unit. As the material continues to move and its speed changes in real time, the system continuously refreshes the theoretically expected timestamps in each control cycle until the remaining physical distance of the virtual slice approaches zero. This dynamic extrapolation mechanism establishes a benchmark point on the time axis, providing a theoretical time basis for subsequent dynamic cross-comparison using torque feedback and the opening of the spatiotemporal window.

[0059] See attached document Figure 6 , Figure 6 This is a schematic diagram of the extreme value capture and dynamic adaptive reconstruction logic according to the present invention. In this embodiment, the system needs to cross-compare the purely theoretical kinematic derivation results with the underlying actual physical dynamic feedback to correct spatial calculation deviations caused by material slippage or equipment wear. The relevant processing is divided into the following sub-steps: To avoid the enormous computational overhead of global search and the false triggering caused by fluctuations in normal processes, controller 10 opens a dynamic capture window of a specific width on the time axis based on the previously calculated theoretical expected timestamp. Controller 10 sets the start time of the window to the theoretical expected timestamp minus a preset lead time and the end time to the theoretical expected timestamp plus a preset lag time. The specific values ​​of the lead time and lag time are determined based on the maximum speed fluctuation rate of the production line's historical operation, and are typically set to 0.5 seconds to 2.0 seconds in conventional applications. After entering the dynamic capture window, controller 10 reads the torque feedback value of the corresponding stand main motor in the continuous rolling mill 40 at high speed via the field industrial bus. For the closed-loop torque observation and bus data transmission of the inverter's internal current, those skilled in the art can use conventional industrial Ethernet communication protocols in conjunction with standard driver firmware to achieve this; the torque data acquisition is a well-known technology in the field and will not be described in detail here.

[0060] When metallic materials enter the deformation zone of the continuous rolling mill 40 after experiencing a drastic change in enthalpy in a localized area, the uneven internal temperature directly alters the material's local resistance to plastic deformation. To maintain a set stable rolling speed, the main motor passively outputs a transiently abrupt electromagnetic torque to overcome this fluctuation in deformation resistance. The controller 10 stores a continuously sampled torque feedback sequence in the dynamic capture window and applies a differential filtering algorithm to eliminate periodic high-frequency torque noise generated by the meshing of mechanical transmission gears.

[0061] After filtering, the system executes torque extremum peak-finding logic, performing time first-order differential operations on the actual electromagnetic torque data to identify the specific system time corresponding to the absolute maximum value of the torque change gradient in the sequence. To avoid misjudging system background noise as feature anchor points, the controller 10 compares the identified absolute maximum value with a preset effective abrupt change threshold, i.e., a noise dead zone threshold.

[0062] After confirming the amplitude is valid, the controller 10 further executes the feature authenticity backtracking verification logic, comparing the polarity direction of the first-order derivative of torque time with the enthalpy change direction corresponding to the feature anchor point. From a physical perspective, if the material experiences a sudden temperature drop (i.e., the enthalpy change direction is negative), its deformation resistance increases, and the motor will passively output a larger sudden torque to maintain its speed, resulting in a positive polarity direction for the first-order derivative of torque, and vice versa. If the polarity directions of the two do not conform to the aforementioned negative correlation matching relationship, the system determines that the torque fluctuation is caused by other mechanical disturbances, and the original feature anchor point is actually a distortion artifact caused by moisture obstruction of the optical sensor, and immediately triggers the queue anchor point melting procedure.

[0063] In the anchor point failure procedure, the controller 10 actively erases the feature anchor point marker bit of the virtual slice in the first-in-first-out data queue, clears its corresponding theoretical expected timestamp, and blocks the issuance of subsequent feedforward torque additional commands for the slice, thereby preventing the control system from diverging due to false features. Only when the absolute maximum value exceeds the effective mutation threshold and the polarity direction is successfully matched, the controller 10 defines the time point at which this dynamic extreme value is captured as the real physical timestamp of the feature anchor point actually arriving at the corresponding rack.

[0064] The effective mutation threshold is set as a percentage of the rated torque of the main motor, typically between 5% and 10% of the rated torque. Considering that severe slippage may occur in extreme cases, leading to excessive delays in feature anchor points, the controller 10 incorporates a timeout discard mechanism. If no significant torque extreme value is detected after the window deadline, the system determines that the capture has failed, actively abandons the comparison calculation of the current feature anchor point, clears the buffer, and waits for the next feature anchor point in the processing sequence, thus preventing the underlying program from falling into an infinite waiting logical dead zone.

[0065] After successfully acquiring the actual physical timestamp, controller 10 immediately performs deviation calculation and adaptive reconstruction to correct the theoretical model. Controller 10 calculates the time deviation between the actual physical timestamp and the theoretically expected timestamp. The formula for calculating the time deviation is as follows: ; In the formula, Represents time deviation; Represents the actual physical timestamp captured by the system within the dynamic capture window; This represents the theoretically expected timestamp calculated by controller 10 during the simulation phase.

[0066] Time deviation directly reflects the degree of deviation between the feedforward kinematic model and the actual physical operating state of the production line. If the calculated time deviation is greater than zero, it indicates that the actual arrival time of the physical materials is later than theoretically expected, meaning that there is a more severe mechanical slippage phenomenon on the production line than the theoretically calculated value. Controller 10 multiplies the time deviation by the current actual linear velocity and converts it into a spatial displacement compensation amount.

[0067] Since the addressing pointer of the first-in-first-out data queue is a discrete integer index, the controller 10 divides the spatial displacement compensation amount by the slice length setpoint. The integer part is then used to calculate the discrete pointer index offset. This pointer index offset is then added to the current addressing pointer of the first-in-first-out data queue, and the spatial displacement remainder generated by the division operation is updated in the continuous displacement accumulation register. This ensures that the physical alignment of the data queue index position is achieved while avoiding the loss of microscopic displacement calculation precision.

[0068] As a preferred approach, to prevent the continuous accumulation and expansion of calculation errors in subsequent stands, the system further introduces a dynamic parameter adaptive reconfiguration mechanism. The controller 10 uses the calculated time deviation as the error input value and transmits it to the system's built-in discrete proportional-integral controller to dynamically correct the forward slip coefficient used in the previous calculation of the continuous rolling mill cascade transfer speed. Since the controller 10 executes a discrete control cycle, the forward slip coefficient correction formula adopts a discrete summation form: ; In the formula, This represents the forward sliding coefficient after the update and reconstruction; This represents the old forward slip coefficient used in the current control cycle; This represents the proportional gain coefficient, used to provide a fast response to time deviations, and its value is typically set between 0.001 and 0.01. This represents the integral gain coefficient, used to eliminate long-term steady-state slippage error, and its value is usually set between 0.0001 and 0.001. This represents the calculated time deviation.

[0069] Through the aforementioned algorithmic loop, the system utilizes the feedback of the underlying dynamic extreme value characteristics to dynamically update the kinematic feedforward parameters of the top layer online. This entire mechanism eliminates long-term integral errors caused by unmeasurable physical disturbances, enabling the digital queue within the controller to accurately anchor to the metal physical nodes on the physical production line over long periods.

[0070] See attached document Figure 7 , Figure 7 This is a schematic diagram of the feedforward torque compensation and smooth control handover logic according to the present invention. In this embodiment, the controller 10 generates a feedforward torque command in advance based on a spatially aligned first-in-first-out data queue before the metal material with abnormal temperature enters the frame deformation zone, and realizes a smooth switching between feedforward control and basic closed-loop control. The related processing is divided into the following sub-steps: The controller 10 monitors the current address index of each virtual slice in the first-in-first-out (FIFO) data queue in real time. When the physical distance between a virtual slice and the entrance of a specific stand in the continuous rolling mill 40 is less than a preset pre-reading distance threshold, the controller 10 extracts the enthalpy characteristic index inside the virtual slice. The pre-reading distance threshold is determined based on the product of the dynamic response time of the main motor and its transmission system and the current linear velocity. After extracting the enthalpy characteristic index, the controller 10 calculates the feedforward torque compensation.

[0071] According to the principles of metal plasticity mechanics, there is a direct mapping relationship between the temperature state of a material and its resistance to rolling deformation; that is, a decrease in temperature leads to an increase in deformation resistance, requiring a larger motor torque to maintain a constant rolling speed. The formula for calculating the feedforward torque compensation is as follows: ; In the formula, This represents the amount of feedforward torque compensation. The material stiffness conversion factor reflects the torque demand fluctuation caused by a unit change in enthalpy. It is usually calibrated by linear regression analysis of historical rolling force data and temperature gradient of the production line. The steady-state baseline enthalpy index represents the stable operation of the production line under rated process conditions; The enthalpy characteristic index represents the virtual slice extracted by controller 10.

[0072] When the absolute value of the calculated feedforward torque compensation exceeds the preset compensation dead zone threshold, the controller 10 determines that the metal material about to enter the frame deformation zone has a significant temperature anomaly, and then triggers the feedforward torque compensation mechanism. The compensation dead zone threshold is usually set slightly larger than the torque disturbance range that the system's conventional closed-loop control can handle. As a preferred approach, the compensation dead zone threshold is set to 2% to 5% of the rated torque of the main motor.

[0073] To prevent feedforward torque runaway caused by sensor malfunction or extreme data anomalies, controller 10 is configured with absolute amplitude limiting protection before outputting the feedforward torque compensation. The system will calculate... The value is compared with a preset compensation upper limit threshold, which is typically set at 10% to 15% of the main motor's rated torque. If... If the limit is exceeded, it will be forcibly clamped at the compensation upper limit threshold to ensure the safety of the mechanical transmission system.

[0074] To prevent the sudden increase of feedforward torque from causing mechanical shock to the transmission gearbox and leading to excessive output current of the main motor, controller 10 introduces a smooth handover mechanism when superimposing the feedforward torque compensation onto the main drive system. As a preferred approach, the system employs a first-order inertial filtering algorithm to smooth the step-like feedforward torque compensation over time. The formula for calculating the smoothed output torque feedforward command is as follows: ; In the formula, This represents the smooth torque feedforward command output to the main motor driver during the current control cycle; This represents the smooth torque feedforward command output from the previous control cycle; The control scan cycle time of controller 10; This represents the filter's inertia time constant, which determines the smoothness of the feedforward torque application and is typically set between 0.1 and 0.5 seconds. This represents the feedforward torque compensation amount after theoretical calculation and amplitude limiting.

[0075] The controller 10 directly superimposes the smooth torque feedforward command onto the basic torque command channel of the main motor driver as an additional feedforward amount for the basic torque command channel. For the torque command superposition and vector control loop inside the inverter or servo driver, those skilled in the art can use standard inverter communication messages and built-in macro configurations to achieve this. The underlying drive control is a well-known technology in the field and will not be described in detail here.

[0076] As the metal material continues to move forward, when it is detected that all virtual slices corresponding to the abnormal temperature range have completely passed through the deformation zone of the corresponding frame, the controller 10 needs to smoothly hand over the control of the main motor to the original speed closed-loop regulator.

[0077] The controller 10 continuously calculates the current position of the virtual slice in the absolute space coordinate system. When the tail coordinate of the virtual slice crosses the exit coordinate of the rack, the controller 10 compensates for the theoretical feedforward torque in the formula. Forced zeroing. At this point, the first-order inertial filtering algorithm continues to function, smoothing the torque feedforward command. The filter smoothly decays from its current value to zero according to the predetermined filter inertia time constant.

[0078] During the decay process, the system's original speed closed-loop regulator automatically adjusts its integral output according to torque changes, thus achieving a smooth transition from feedforward control to single closed-loop control. This exit mechanism avoids rolling tension fluctuations caused by the sudden withdrawal of feedforward commands, ensuring the process continuity and stability of the continuous rolling production line.

[0079] The actual operation of the system of this invention will be specifically described below using a continuous casting and rolling industrial production line used to produce 8mm high conductivity copper rods. This production line has an annual capacity of 100,000 tons, and the rated linear velocity at the outlet of the crystallizing wheel 20 is 0.25m / s.

[0080] 1) Basic parameter configuration and hardware mapping: In this embodiment, the working pitch circle diameter of the crystallizing wheel 20 To ensure high-precision spatial mapping at low speeds, the first encoder 22, with a range of 2.5m, is a high-resolution absolute encoder with a single-turn pulse resolution of [missing information]. The value is 100,000 ppr. The main control loop scan period of controller 10 is set to 10 ms.

[0081] Under steady-state operation, the crystallizing wheel rotates at 1.91 rpm, corresponding to a rated linear velocity of 0.25 m / s. Within a 10 ms scan cycle, the pulse increment value fed back by the first encoder... There are approximately 32.

[0082] Based on the physical displacement increment mapping relationship, the physical displacement increment of controller 10 within a single cycle is calculated as follows: ; In the formula, Represents the physical displacement increment within a single scan cycle; Represents the pulse increment value of the first encoder; Represents the resolution of a single pulse cycle; This represents the working pitch circle diameter of the crystallizing wheel.

[0083] The system sets a fixed value for slice length. =0.2m. When the accumulated value in the continuous displacement accumulator register reaches 0.2m, the controller generates a virtual slice.

[0084] The total length of the fixed physical path from the outlet of crystallizing wheel 20 to the center of the deformation zone of the downstream first frame 41 is known. =15m. Based on this, the basic addressing depth of the first-in-first-out data queue constructed in the main control memory is: ; In the formula, This represents the base addressing depth of the data queue; Represents the total length of a fixed physical path; This represents a fixed value for the slice length.

[0085] In other words, within the physical space between continuous casting and the first rolling mill stand, the system simultaneously maintains 75 discrete data structure units, thereby realizing equidistant digital segmentation of the continuous metal material movement state.

[0086] 2) Specific calculation process for multi-source feature calibration and dynamic reconstruction: During continuous production, two types of sudden disturbances occurred simultaneously in a certain local period: one was a sudden surge in cooling water pressure, causing a real local "cold spot" in the metal material; the other was water vapor splashing and obstructing the thermometer 23, causing a sharp drop in optical temperature measurement. To illustrate the feature decoupling and deduction calculation process of this invention, the following example calculation is performed on a "virtual slice #1024" that experienced an anomaly.

[0087] During the generation of this slice, the surface temperature reported by thermometer 23 dropped from a steady-state 850°C to 500°C within 0.2 seconds. The system calculates the time-varying gradient: ; In the formula, This represents the time-varying gradient of the temperature signal.

[0088] Since the gradient is much greater than the preset water vapor blocking threshold of 50℃ / s, the controller 10 determines it as water vapor blocking, actively discards the outlier, and performs spatial polynomial interpolation using adjacent historical data to reconstruct the surface temperature of the slice. The temperature has been revised to a reasonable 835℃.

[0089] At this time, the local instantaneous flow rate fed back by the flow meter Soaring to 120L / min, current linear velocity The speed is 0.25 m / s. The thermodynamic weighting coefficient is set. =1.0, =0.1. The enthalpy characteristic index of this slice is calculated as follows: ; In the formula, Represents the characteristic index of enthalpy; Represents the reconstructed surface temperature; Represents instantaneous flow rate; Represents the current linear velocity; and This is the thermodynamic weighting coefficient.

[0090] Given that the steady-state baseline enthalpy index of the production line is 820, and the current sliding window range is |820−787|=33, which is greater than the steady-state dead zone threshold of 15, slice #1024 was successfully marked as a "feature anchor".

[0091] As the material moves, the system updates its theoretical deductions in real time. Assume the current absolute system time... The dynamic index number of the slice in the queue. =20, Actual linear velocity at the inlet =0.24m / s. The theoretically expected timestamp for its arrival at the first rack 41 is calculated as follows: ; In the formula, Represents the theoretically expected timestamp; Represents the current system absolute time; Represents a dynamic index number; This represents the actual linear velocity at the inlet.

[0092] After the time axis enters the dynamic capture window around 245.83s, the controller performs first-order differential peak finding on the electromagnetic torque fed back by the first frequency converter 43. At an absolute time of 246.10s, the system captures a maximum point that exceeds the noise threshold.

[0093] Polarity matching verification: The enthalpy characteristic index abruptly changes to negative, decreasing from 820 to 787, while the first derivative of the torque is positive. The cold spot causes increased resistance, resulting in a sudden increase in motor torque. These two factors are physically negatively correlated, so the verification passes and the anchor point fuse procedure is not triggered. Establish the actual physical timestamp. =246.10s. Subsequently, the controller calculates the time deviation: ; In the formula, Represents time deviation; Represents the actual physical timestamp.

[0094] The fact that the actual arrival time is later than the theoretical time indicates that mechanical slippage exists in the transition zone. The system feeds this time deviation back to the PI controller, which automatically corrects the forward slip coefficient of the subsequent racks through discrete summation, and adds the displacement compensation corresponding to the error to the pointer index.

[0095] Finally, when the material is about to reach the nth frame 44, the system calculates the feedforward torque compensation, assuming a stiffness conversion factor. =0.8 N·m: .

[0096] In the formula, This represents the amount of feedforward torque compensation. Represents the material stiffness conversion factor; It represents the steady-state baseline enthalpy index.

[0097] The compensation amount is smoothed by a first-order inertial filter and then sent to the underlying driver. The motor actively applies additional load before the cold point is reached, which completely offsets the mechanical shock.

[0098] 3) Experimental verification and effect comparison: To evaluate the control effect of the present invention in a real continuous casting and rolling environment, this embodiment uses 30 consecutive days of operating data from a copper plant for comparative verification. The control group uses a traditional "constant speed closed-loop control + simple threshold alarm" system; the experimental group deploys the "cross-process domain linkage control system" of the present invention. Both groups operate under similar cooling water network fluctuations and environmental disturbances.

[0099] The experimental statistical results are shown in the table below: See attached document Figure 8 The figure shows a comparison of the effects of different processing methods on target feature decoupling and extraction when facing interference from local bright and highly reflective backgrounds in the same detection area.

[0100] Figure 8 Figure a shows the original grayscale image acquired by the front-end sensor. This physical region contains a large area of ​​irregularly scattering high-brightness interference, such as highly reflective water stains, localized overexposure, or areas of strong scattering, resulting in a sharp increase in overall grayscale. Simultaneously, to the right of this high-brightness interference area, at the location of the elliptical bounding box, lies a hidden local three-dimensional anomaly, such as a tiny bulge or depression. In the original image, the drastic brightness change of the bright background significantly weakens and masks the subtle light-dark gradient features of this three-dimensional shape itself.

[0101] Figure 8 Figure b shows the feature map output using a traditional mixed photometric evaluation algorithm. Because this algorithm fails to physically separate two-dimensional reflectivity brightness anomalies from three-dimensional morphological anomalies, large areas of high-brightness interference directly dominate the feature response, leading to severe signal saturation and grayscale whitening in this region. It is clearly visible from the elliptical marked area that the originally contained minute morphological undulations are completely swallowed and lost by the high-brightness background. Subsequent judgments are highly likely to misclassify this entire region as simple surface reflection or contamination, resulting in the missed detection of true structural defects.

[0102] Figure 8Figure c shows the shape gradient extraction result output by the feature decoupling method proposed in this invention. Through independent benchmark compensation and normalized difference operation, the large-area high-brightness common-mode interference that originally dominated the image is completely suppressed, the two-dimensional brightness fluctuation is perfectly filtered out, and the background is smoothed and flattened to a uniform steady-state grayscale benchmark. The real three-dimensional shape features that were originally hidden in the bright areas, namely the three-dimensional gradient contour with the left dark and right bright areas presented in the elliptical marked areas, are accurately peeled off and clearly revealed. This set of comparisons intuitively proves that the solution of this invention can achieve pure extraction of the underlying three-dimensional shape features under extreme light intensity abrupt interference, unaffected by surface reflection or brightness anomalies.

[0103] See attached document Figure 9 The figure further compares the dynamic curves of the transmission system tension response of the traditional closed-loop control system and the linkage control system of this invention when an abnormal cold point suddenly enters the rolling mill stand. The horizontal axis represents the production line running time, and the vertical axis represents the relative tension fluctuation rate of the transmission system. When the running time reaches the physical moment when the cold point enters the stand, the traditional closed-loop control system, relying on passive error response, inevitably experiences significant tension overshoot and severe damping oscillations in the motor as the material deformation resistance increases dramatically. This mechanical impact can easily trigger microscopic tension contraction or even macroscopic strip breakage accidents. In contrast, the tension curve of the linkage control system of this invention maintains extremely high smoothness. Because the system extracts the enthalpy offset after spatial alignment in advance, the feedforward control intervenes before the cold point contacts the deformation zone, and the sudden impact is completely offset by actively adding smooth torque. There is no obvious tension drop or oscillation phenomenon during the entire process.

[0104] The experimental results show that this invention overcomes the limitations of traditional single control or single detection methods, establishing a rigorous cross-domain linkage mechanism between optical feature calibration, deduction queue shifting, and dynamic differential verification. On the one hand, in complex surface inspection, the system can effectively eliminate common-mode interference artifacts such as high brightness or water vapor, achieving accurate extraction of microscopic morphology. On the other hand, when facing harsh industrial environments with both water vapor splashing and transmission slippage, the system successfully removes sensor artifacts through polarity matching of the first-order derivative and adaptively corrects kinematic deduction errors in real time by capturing true dynamic extrema. Through spatially precise alignment and smooth feedforward torque compensation, the severe tension oscillations inevitably caused by cold, hard bands protruding into the frame are completely suppressed. This invention, addressing both physical sensing and underlying drive, eliminates the risks of missed detections, misjudgments, and band breakage from the root, significantly improving the overall operational stability and comprehensive product yield of continuous manufacturing lines.

[0105] 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, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for linkage control of a copper rod processing and manufacturing process, characterized in that, include: S10, the physical billet is divided into a virtual slice sequence according to the pulse signal of the first encoder, and a data queue mapped to the physical space is constructed; S20: Receive surface temperature signal and flow rate signal; when the time change gradient of the surface temperature signal is greater than a threshold, use historical enthalpy data for interpolation to replace the current data; calculate the enthalpy characteristic index corresponding to each virtual slice; perform differential calculation on the enthalpy characteristic index to extract feature slices and mark them as feature anchor points. S30, the data queue shift calculation is driven by the linear velocity signal of the second encoder to deduce the theoretical timestamp of the virtual slice reaching the downstream rack; S40, within the time window corresponding to the theoretical timestamp, collect the actual electromagnetic torque data of the downstream rack strain frequency converter and perform first-order differential operation; compare the differential polarity direction with the enthalpy change direction of the feature anchor point. If the direction does not match, eliminate the feature anchor point. If the direction matches, extract the actual arrival timestamp to correct the theoretical timestamp. S50, a feedforward torque command is generated based on the enthalpy characteristic index and sent to the frequency converter, and the feedforward torque command is revoked after the virtual slice leaves the deformation zone of the downstream rack.

2. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The process of dividing the physical billet into a virtual slice sequence based on the pulse signal of the first encoder is specifically as follows: Obtain the pulse increment value fed back by the first encoder within the control scan cycle, and calculate the physical displacement increment by combining the single-turn pulse resolution and the working pitch circle diameter of the crystallizing wheel; An accumulation operation is performed on the physical displacement increment. When the accumulated value reaches the slice length limit, a virtual slice with an incrementing number is generated. Subtract the slice length from the accumulated value and retain the difference as the initial base for the next round of accumulation.

3. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The method of using historical enthalpy data interpolation to replace the current data specifically involves: Actively discard the current abnormal surface temperature signal and trace back to extract the effective enthalpy data of adjacent historical nodes in the data queue; A spatial polynomial is constructed using the effective enthalpy data, and the theoretical enthalpy estimate of the current spatial location is calculated using an interpolation algorithm and stored as alternative data in the feature data area corresponding to the virtual slice.

4. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The calculation of the enthalpy characteristic index corresponding to each virtual slice is specifically as follows: The effective surface temperature signal is multiplied by a first weighting coefficient to obtain a first product; The ratio of the instantaneous volumetric flow rate corresponding to the flow signal to the linear velocity of the current billet is calculated, and the ratio is multiplied by the second weighting coefficient to obtain the second product; Subtracting the second product from the first product yields the enthalpy characteristic index of the current virtual slice.

5. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The step of performing differential calculation on the enthalpy characteristic index to extract feature slices and mark them as feature anchor points is as follows: Calculate the range of the thermal enthalpy characteristic index within a sliding data window of a set length; When the range is greater than the steady-state dead zone threshold, the enthalpy gradient values ​​of each virtual slice within the sliding data window are compared. The virtual slice with the largest gradient change is selected as the feature slice, and the feature anchor is written into the corresponding cell of the data queue.

6. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The theoretical timestamp for the virtual slice reaching the downstream rack is as follows: Calculate the remaining physical distance between the virtual slice and the corresponding rack. The remaining physical distance is the total physical path length minus the product of the queue index number of the virtual slice and the slice length limit. The remaining physical distance is divided by the transmission speed after correction by the forward slip coefficient to obtain the transmission time. The transmission time is then added to the current time base value to obtain the theoretical timestamp.

7. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The comparison of the differential polarity direction with the enthalpy change direction of the feature anchor point is specifically as follows: When the direction of the enthalpy change is negative and the polarity direction corresponding to the first-order differential operation is positive, or when the direction of the enthalpy change is positive and the polarity direction is negative, the direction is determined to be a match. When the direction of the thermal enthalpy change is negative and the polarity direction is negative, or when the direction of the thermal enthalpy change is positive and the polarity direction is positive, a direction mismatch is determined, and the operation of eliminating the feature anchor point is performed. Specifically, this includes: clearing the theoretical timestamp corresponding to the feature anchor point and blocking the feedforward torque command issuance process for the corresponding virtual slice.

8. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The step of extracting the actual arrival timestamp to correct the theoretical timestamp specifically involves: The time deviation is calculated by subtracting the theoretical timestamp from the actual arrival timestamp. The spatial displacement compensation is obtained by multiplying the time deviation by the current actual linear velocity. The pointer index offset is calculated using the spatial displacement compensation and then added to the address addressing pointer of the data queue. The time deviation is input to the proportional-integral controller, and the output result is used to sum and correct the forward slip coefficient of the previous cycle, which is then used for the timestamp extrapolation of subsequent racks.

9. The linkage control method for copper rod processing and manufacturing process according to claim 1, characterized in that, The step of generating a feedforward torque command based on the enthalpy characteristic index and sending it to the frequency converter specifically involves: Calculate the difference between the reference enthalpy index and the enthalpy characteristic index, and multiply the difference by the material stiffness conversion coefficient to calculate the theoretical feedforward torque compensation. A first-order inertial filtering algorithm is used to smooth the theoretical feedforward torque compensation amount to generate the feedforward torque command, and the feedforward torque command is superimposed on the basic torque command channel of the frequency converter.

10. A linkage control system for the copper rod processing and manufacturing process, characterized in that, The method for linkage control of the copper rod processing and manufacturing process as described in any one of claims 1-9 includes: The controller is used to perform logical operations, maintain the data queue, and issue feedforward torque commands. The upstream casting area equipment includes a crystallizing wheel driven by a motor. The first encoder is coaxially mounted on the main shaft of the crystallizing wheel. A thermometer is installed at the outlet end of the crystallizing wheel to collect the surface temperature signal. A flow meter is installed in the cooling water circuit of the crystallizing wheel to collect the flow rate signal. The transition zone equipment includes a speed measuring roll located between the crystallizing wheel and the downstream continuous rolling zone, and the second encoder is installed on the shaft end of the speed measuring roll; The downstream continuous rolling zone equipment includes multiple stands arranged in series as the downstream stands, each stand is driven by a corresponding motor, and each motor is independently controlled by a frequency converter that supports torque control mode.

Images