Intelligent take-up tension control system

Abstract

The invention relates to the technical field of take-up tension control, and discloses an intelligent take-up tension control system. A multi-source sensing acquisition unit of the system acquires a wire distribution topological graph through a tension sensing platform, acquires a reference tension parameter of a take-up reference point, and acquires strain characteristics and movement speed characteristics of each wire section; the dynamic feature modeling unit performs multi-modal fusion on the reference tension parameters and the features to generate a multi-modal tension feature matrix; the static path analysis unit constructs a static tension conduction path based on the tension characteristic value in the matrix and the corresponding wire section; the take-up execution control unit controls a take-up track of the take-up mechanism on the adjustable path in combination with the static path conduction node value and the corresponding wire section; the dynamic compensation unit collects the dynamic tension parameter of the current wire section in the take-up process, the tension applying direction of the take-up mechanism is compensated and corrected according to the dynamic tension parameter and the static path conduction node value, and the take-up tension control stability and accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of wire take-up tension control technology, specifically to an intelligent control system for wire take-up tension. Background Technology

[0002] In industrial production fields such as wire and cable, textiles, and metal wire processing, the winding process is a crucial step in product formation, and the tension stability during winding directly affects the final quality of the wire. Currently, most winding tension control systems used in the industry employ a single sensor for tension acquisition, only acquiring tension data at a fixed point during winding. This fails to provide a comprehensive understanding of the wire's distribution throughout the entire winding path. Because the wire exhibits different topological distributions during winding, and external environmental influences, material differences, and changes in motion affect different sections of the wire, tension fluctuations can occur. Tension data from a single reference point is insufficient to reflect the overall tension distribution pattern of the wire, leaving subsequent tension control lacking a comprehensive and accurate initial basis.

[0003] Traditional control systems, when acquiring wire characteristics, often focus solely on the tension value itself, neglecting the strain and velocity variations of the wire in different sections. The strain characteristics of the wire reflect its deformation state during stress, while the velocity characteristics are directly related to the operating rhythm of the take-up mechanism. The lack of these two types of characteristics prevents the control system from establishing a correlation between tension and the actual state of the wire, making it difficult to accurately determine the root cause of tension fluctuations. In terms of tension characteristic processing, traditional methods often employ single-dimensional data processing approaches, failing to effectively integrate baseline tension parameters with multi-dimensional characteristics such as wire strain and velocity. This makes it difficult to form a feature model that comprehensively characterizes the wire tension state, thus hindering the provision of refined decision support for tension control.

[0004] In the path analysis and execution control stages, traditional systems typically perform take-up operations based on preset fixed paths, failing to consider the differences in tension characteristics across different wire sections. Due to the lack of a static tension transmission path built upon actual tension characteristics, the adjustable path of the take-up mechanism does not match the actual tension distribution of the wire. This results in the take-up trajectory being unable to adaptively adjust to the tension requirements of different wire sections, easily leading to excessive or insufficient local tension, which in turn causes problems such as wire stretching deformation, surface damage, or uneven take-up arrangement. Furthermore, the take-up process is dynamic; the wire is affected by various dynamic factors during movement, such as equipment vibration, uneven raw material composition, and changes in external resistance. Traditional control systems lack effective dynamic compensation mechanisms, cannot collect dynamic tension parameters of the current wire section in real time, and cannot promptly correct the tension application direction based on the difference between dynamic parameters and preset static parameters. This makes it difficult to maintain stable take-up tension in dynamic environments, severely impacting production efficiency and product quality. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent control system for take-up tension to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides an intelligent control system for take-up tension, the system comprising: The multi-source sensing acquisition unit acquires a wire distribution topology map through a tension sensing platform, and acquires tension at the take-up reference point based on the wire distribution topology map to obtain reference tension parameters. The multi-source sensing acquisition unit performs feature acquisition based on each wire segment in the wire distribution topology map to obtain the wire strain characteristics and motion velocity characteristics corresponding to each wire segment. The dynamic feature modeling unit performs multimodal fusion of the wire strain characteristics and motion velocity characteristics corresponding to each wire segment based on the reference tension parameters, and generates a multimodal tension feature matrix. The static path analysis unit constructs a static tension transmission path based on each tension feature value in the multimodal tension feature matrix and the corresponding wire segments. The take-up execution control unit acquires the adjustable path of the take-up mechanism and controls the take-up trajectory of the take-up mechanism on the adjustable path based on the transmission node values ​​in the static tension transmission path and the corresponding wire segments. The dynamic compensation unit collects the dynamic tension parameters of the current wire section during the take-up process, and compensates and corrects the tension application direction of the take-up mechanism based on the dynamic tension parameters of the current wire section and the transmission node values ​​in the static tension transmission path.

[0007] Preferably, the multi-source sensing acquisition unit includes: The wire sectioning subunit obtains the wire distribution topology map and performs segmentation based on the density gradient in the wire distribution topology map to obtain multiple wire sections; The strain acquisition subunit acquires the wire traction direction, selects an initial section based on the wire traction direction to acquire wire strain characteristics, selects the next wire section based on the wire traction direction to acquire wire strain characteristics, and so on, until the wire strain characteristics of each wire section are acquired. The velocity mapping subunit acquires the wire motion vector, reselects an initial segment based on the wire motion vector for motion velocity feature acquisition, selects the next wire segment based on the wire motion vector for motion velocity feature acquisition, and so on, until the motion velocity feature acquisition of each wire segment is completed.

[0008] Preferably, the dynamic feature modeling unit performs: The feature fusion depth is determined based on the aforementioned reference tension parameters; Obtain the strain characteristics and velocity characteristics of the first wire section, select a comparison section based on the position coordinates of the first wire section and the feature fusion depth, perform multimodal fusion, and generate the tension characteristic value corresponding to the first wire section. After sequentially acquiring the strain characteristics and velocity characteristics of other wire sections, the multimodal fusion operation is repeatedly performed to generate the tension characteristic values ​​of each wire section, and matrix mapping is performed according to the wire distribution topology map to generate the multimodal tension characteristic matrix.

[0009] Preferably, the static path analysis unit includes: The transmission node generates a sub-unit, and extracts each tension feature value and the corresponding wire segment coordinates from the multimodal tension feature matrix; The path topology construction sub-unit establishes a transmission node connection chain based on the spatial relationship between the coordinates of wire segments, and defines the transmission path weight value according to the gradient change of the tension characteristic value. The static path optimization subunit uses a dynamic topology optimization algorithm to iteratively calculate the transmission path weight values ​​and generate the path with the minimum transmission loss as the static tension transmission path.

[0010] Preferably, the take-up execution control unit performs the following: Obtain the values ​​of the transmission nodes in the static tension transmission path and their corresponding wire segments; Extract all adjustable path segments of the take-up mechanism and the adjacent wire sections corresponding to each path segment; The starting point for take-up is determined, and the transmission node values ​​of all adjacent wire sections corresponding to the adjustable path segment are weighted and accumulated as the priority passage coefficient of the adjustable path segment. The take-up trajectory of the take-up mechanism starting from the starting point is generated by the gradient optimization algorithm.

[0011] Preferably, the dynamic compensation unit includes: The deviation response subunit obtains the real-time deviation between the dynamic tension parameters of the current wire section and the corresponding transmission node values ​​in the static tension transmission path; The compensation decision subunit determines the compensation depth based on the tension fluctuation characteristics of the current wire section when the real-time deviation exceeds the tolerance threshold, and obtains the transmission node value of the adjacent wire section based on the compensation depth. The direction correction subunit generates a tension application direction compensation vector by combining the transmission node values ​​of adjacent wire sections, and adjusts the tension application direction of the take-up mechanism.

[0012] Preferably, the orientation correction subunit performs: The tilt adjustment of the take-up roller is calculated based on the compensation vector of the tension application direction; The roller attitude control signal is generated based on the tilt angle adjustment, and the tension application direction of the take-up mechanism is updated synchronously.

[0013] Preferably, the system further includes: The real-time feedback unit continuously collects the vibration spectrum of the wire during the winding process and converts the vibration spectrum of the wire into real-time spectral feature values. When the deviation between the real-time spectral feature value and the transmission node value of the static tension transmission path continues to increase, the path update unit re-triggers the dynamic feature modeling unit to update the multimodal tension feature matrix and drives the static path analysis unit to reconstruct the static tension transmission path.

[0014] Preferably, the multi-source sensing acquisition unit acquires the strain characteristics of the wire through a distributed optical fiber sensor and the motion speed characteristics through an encoder.

[0015] Preferably, the dynamic compensation unit and the real-time feedback unit form a closed-loop control, and the real-time spectrum characteristic value output by the real-time feedback unit serves as the compensation and correction trigger condition for the dynamic compensation unit.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This intelligent tension control system for take-up wires achieves comprehensive acquisition of wire tension data through a multi-source sensor acquisition unit. It can not only accurately acquire the reference tension parameters at the take-up reference point based on the wire distribution topology map, but also acquire the corresponding wire strain characteristics and motion velocity characteristics for each wire segment. This multi-dimensional and comprehensive feature acquisition method breaks through the limitations of traditional single sensors that can only acquire local data. It allows the control system to fully grasp the distribution and actual operating characteristics of the wire throughout the entire take-up path, providing more comprehensive and realistic basic data for subsequent tension control, making the initial basis for tension control more scientific and accurate.

[0017] The dynamic feature modeling unit employs multimodal fusion technology to integrate baseline tension parameters with the strain and velocity characteristics of each wire segment, generating a multimodal tension feature matrix. This fusion method fully explores the inherent correlations between different types of features, transforming previously scattered single-dimensional data into a multi-dimensional feature set that comprehensively characterizes the wire tension state, resulting in a more refined and comprehensive description of tension features. Through this feature matrix, the control system can clearly grasp the variation patterns and influencing factors of tension in different wire segments, providing accurate decision-making basis for subsequent path analysis and execution control, effectively avoiding control deviations caused by incomplete feature information.

[0018] The static path analysis unit constructs a static tension transmission path based on a multimodal tension feature matrix, fully considering the differences in tension characteristic values ​​corresponding to each section of the wire. This path construction method does not rely on a preset fixed path, but rather on the actual tension distribution characteristics, ensuring that the constructed static tension transmission path accurately reflects the tension requirements of different sections of the wire. Through this static path, the take-up mechanism can clearly define the tension transmission patterns of different sections before performing the take-up operation, providing clear path guidance for subsequent trajectory control and enabling the take-up mechanism to better conform to the tension distribution characteristics of the wire.

[0019] Based on the adjustable path of the take-up mechanism, the take-up execution control unit combines the transmission node values ​​in the static tension transmission path with the corresponding wire segment information to precisely control the take-up trajectory. This control method enables dynamic matching between the take-up trajectory and the wire tension distribution. The take-up mechanism no longer operates along a fixed path but adjusts its trajectory according to the tension requirements of different wire segments. This effectively avoids situations where the local tension is too high or too low, reducing the occurrence of wire stretching deformation, surface damage, and other problems, thus improving the stability of the take-up process and the forming quality of the wire.

[0020] The dynamic compensation unit collects dynamic tension parameters of the current wire section in real time during the take-up process and compares them with the values ​​of the transmission nodes in the static tension transmission path. This allows for compensation and correction of the tension application direction of the take-up mechanism. This unit enables the control system to respond to dynamic changes during take-up in real time. Tension changes caused by factors such as equipment vibration, uneven raw material composition, or fluctuations in external resistance can be promptly captured and adjusted. Through continuous dynamic compensation and correction, the tension applied by the take-up mechanism always remains consistent with the actual tension requirement of the current wire section, ensuring tension stability throughout the entire take-up process. Even in complex and variable production environments, this guarantees consistent wire quality while reducing production interruptions caused by tension instability and improving overall production efficiency. Attached Figure Description

[0021] Figure 1 This is a timing diagram of the intelligent winding tension control system described in this invention; Figure 2 A flowchart illustrating the working principle of the dynamic feature modeling unit; Figure 3 A flowchart illustrating the working principle of the static path analysis unit; Figure 4 This is a flowchart illustrating the working principle of the direction correction subunit. Detailed Implementation

[0022] 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.

[0023] Please see Figure 1 The present invention provides an intelligent control system for take-up tension, the system comprising: A multi-source sensing acquisition unit acquires a wire distribution topology map and performs tension acquisition based on this map to obtain baseline tension parameters. Simultaneously, it acquires wire strain and velocity characteristics for each wire segment in the topology map. A dynamic feature modeling unit performs multimodal fusion of the baseline tension parameters with the wire strain and velocity characteristics to generate a multimodal tension feature matrix. A static path analysis unit constructs a static tension transmission path based on the tension feature values ​​in this matrix and the corresponding wire segments. The take-up execution control unit controls the take-up mechanism's trajectory on the adjustable path based on the transmission node values ​​and the corresponding wire segments. A dynamic compensation unit acquires dynamic tension parameters during take-up, compares them with the static transmission node values, and compensates for and corrects the tension application direction to ensure accurate tension control.

[0024] Example 1: See Figure 2 The multi-source sensing acquisition unit first acquires a topology map of the wire distribution, reflecting the geometric distribution and density variations of the wire in space. The wire partitioning subunit analyzes the density gradient distribution in the topology map, identifying high-density clusters and low-density transition areas. For example, near the guide wheel assembly of the winding system, the wire forms a localized high-density area due to multiple windings, while the straight-line transmission section exhibits a uniform density. The partitioning subunit sets dividing points along the direction of density gradient change, dividing the wire path into multiple continuous segments. The length of each segment is dynamically adjusted according to the density change rate; a more refined partitioning strategy is used in areas of abrupt density changes, while the segment range is appropriately expanded in areas of gradual density change. The partitioning results form a logically continuous sequence of wire segments, each with independent spatial coordinates.

[0025] The strain acquisition subunit receives the traction direction parameters, defined by the rotation axis of the take-up roller. The acquisition process begins in the initial section where the traction direction starts. For this section, distributed fiber optic sensors are deployed along the axial direction to measure the micro-strain changes on the wire surface. The sensors capture the strain waveform at a fixed sampling frequency, and after filtering, three core features are extracted: the strain peak value reflects the maximum deformation intensity, the strain fluctuation frequency characterizes the dynamic oscillation characteristics, and the strain gradient distribution describes the attenuation trend of the deformation along the axial direction. After completing the acquisition of the initial section, the system automatically locates the next adjacent section along the traction direction and repeats the above strain feature acquisition process until all sections are covered.

[0026] The velocity mapping subunit synchronously initiates motion velocity acquisition. The wire motion vector is calculated from the encoder's measurement of the guide wheel's rotational speed, and its direction is tangent to the wire's instantaneous motion trajectory. The initial segment selection strategy for velocity acquisition differs from strain acquisition: the system selects the wire segment where the vector originates as the initial acquisition point based on the motion vector direction. Within this segment, the encoder measures the wire surface linear velocity via a contact roller, and calculates the average motion velocity based on the segment length. Simultaneously, a laser velocimeter non-contactly acquires the instantaneous velocity at multiple points within the segment, generating a velocity distribution curve and extracting two key features: the velocity standard deviation reflects motion stability, and the velocity slope change identifies acceleration abrupt changes. After completing the initial segment acquisition, the system continues acquisition along the motion vector direction to the next segment, forming a velocity feature set isomorphic to the strain features.

[0027] The dynamic feature modeling unit receives the reference tension parameter, which is directly measured by the tension sensor at the take-up reference point. The feature fusion depth is dynamically set according to the reference tension value: when the reference tension is higher than a threshold, the fusion depth expands to three adjacent segments; when it is lower than the threshold, only the current segment is fused. For the first wire segment (usually located at the take-up end), the modeling unit calls its strain feature and velocity feature datasets. The gradient distribution data in the strain features and the slope change data in the velocity features are first aligned in the time domain to eliminate differences in acquisition timing. The aligned data is input into the multimodal fusion module, where a tension correlation factor is generated through weighted cross-calculation. The fusion process introduces position coordinates as spatial weights, assigning higher weights to coordinates closer to the reference point. Finally, the tension feature value of the first segment is output, which contains a composite feature of static tension components and dynamic oscillation components.

[0028] The system processes subsequent segments sequentially by partition. For the second segment, in addition to calling the strain and velocity characteristics of this segment, it also needs to extract the tension characteristic values ​​of the preceding segment (the first segment) as a reference benchmark, according to the feature fusion depth requirements. The fusion algorithm performs differential calculations between the feature data of this segment and the reference benchmark to eliminate accumulated errors on the transmission path. The differential results are normalized to generate independent tension characteristic values ​​for the current segment. This process is executed cyclically, with the fusion calculation of each new segment referencing the output results of the preceding segment, forming a chain-like feature transfer structure.

[0029] After all segments are processed, the modeling unit maps the tension feature values ​​of each segment to a two-dimensional matrix based on the spatial coordinate relationships of the wire distribution topology. The matrix row vectors correspond to the partition sequence order, and the column vectors store the time-series data of the feature values. During the spatial mapping process, the feature values ​​of adjacent segments are smoothly transitioned through interpolation algorithms to ensure the continuity of the matrix in the spatial dimension. The resulting multimodal tension feature matrix contains holographic features of the tension state of a specific segment at a specific time, providing a data foundation for static path construction.

[0030] In the application of metal wire production lines, the wire distribution topology diagram shows that the wire forms a complex spatial path through 7 guide rollers. The partitioning subunit identifies 8 high-density areas formed by the contact points of the guide rollers and divides them into 12 continuous segments. Strain acquisition starts from the first segment near the take-up roller, where a fiber optic sensor detects a strain gradient of 0.8 mm / m. Velocity acquisition starts from the 12th segment at the unwinding end, where an encoder measures a reference velocity of 2.5 m / s. When the reference tension sensor measures 150 N, the modeling unit sets the fusion depth to 3 segments. When processing the third segment, in addition to the strain fluctuation frequency characteristic of 0.6 Hz in this segment, the velocity standard deviation data of 1.2% in the second segment is also fused, finally generating the (3,t) elements of the tension feature value matrix. The completed matrix contains a 12×500 tension state dataset (500 being the number of sampling points), completely depicting the tension distribution evolution throughout the entire wire path.

[0031] Example 2: See Figure 3 The static path analysis unit receives a multimodal tension feature matrix from the dynamic feature modeling unit. This matrix contains the tension feature values ​​of each wire segment and their corresponding spatial coordinate information. The conduction node generation subunit first parses the matrix data structure, extracts the tension feature value sequence of each wire segment, and associates it with its coordinate identifier. For example, in a cable winding system, the wire path is divided into 15 segments, and the coordinates of each segment are represented by a three-dimensional spatial position. The tension change characteristics of that segment during the winding process are also recorded.

[0032] The path topology construction sub-unit establishes a transmission node connection chain based on the spatial relationship between the coordinates of wire segments. The coordinates of adjacent segments are used to calculate the spatial distance and relative angle through vector operations, forming the physical connection relationship of the transmission path. Each transmission node contains reference pointers to the preceding and succeeding segments, forming a doubly linked list structure. The weight value of the transmission path is defined based on the gradient change of the tension characteristic value, specifically manifested as the tension difference rate between adjacent segments. For example, when the tension characteristic value difference between segment 5 and segment 6 exceeds a set threshold, the system automatically increases the weight value of the transmission path between these two segments, reflecting a high energy loss during the transition of that segment.

[0033] The static path optimization subunit employs a dynamic topology optimization algorithm to iteratively calculate the weight values ​​of the conduction paths. During algorithm initialization, the weight values ​​of all conduction paths are normalized to eliminate dimensional differences. In the first round of iteration, the algorithm detects conduction paths with abrupt weight changes, identifying potential tension conduction bottlenecks. In the cable winding example, the system finds that the conduction path weights in sections 8 and 9 are significantly higher than in other sections, indicating an abnormal tension gradient change in this area. The optimization algorithm initiates a local search for this path, recalculating the conduction path weights by adjusting the distribution of tension characteristic values ​​between the preceding and following sections. After multiple iterations, the algorithm outputs the globally optimal combination of conduction paths, where the weight values ​​of each path tend to be balanced, ultimately generating the static tension conduction path.

[0034] The take-up control unit acquires the values ​​of the transmission nodes in the static tension transmission path and their corresponding wire segment information. In the take-up machine control system, the adjustable path segment is defined as the movement trajectory range of the take-up roller, with each range corresponding to a range of wire segments. The system first establishes a mapping table between the adjustable path segments and the wire segments; for example, path segment A covers segments 1 to 3, and path segment B covers segments 4 to 6. The take-up starting point is usually set in the first segment at the unwinding end, and the system traverses the adjustable path segments from this point.

[0035] For each adjustable path segment, the system retrieves all adjacent wire segments it covers and extracts the node values ​​of these segments in the static conduction path. The node values ​​include two dimensions: tension magnitude and conduction direction, which are combined into a priority passage coefficient using weighting factors. The weighting coefficients are set considering the characteristics of the wire material; for example, the weighting coefficient for copper core cables emphasizes tension uniformity, while that for fiber optic composite cables emphasizes conduction direction stability. In the cable winding case, the priority passage coefficient calculation for path segment C (corresponding to segments 7 to 9) integrates the values ​​of segment 7 (12N), segment 8 (15N), and segment 9 (14N), as well as the rate of change of conduction angle among them.

[0036] The gradient optimization algorithm generates the take-up trajectory based on the priority pass coefficient. The algorithm constructs a three-dimensional search space, with the coordinate axes representing take-up speed, roller pressure, and path offset, respectively. The initial search point is set to the operating parameters of the current take-up mechanism. The algorithm calculates the gradient change of the objective function (priority pass coefficient) along the three-dimensional direction. Each iteration adjusts a parameter in one dimension; for example, first optimizing the take-up speed to increase the priority pass coefficient by 5%, then fine-tuning the roller pressure to obtain an additional 2% increase in the coefficient. After approximately 20 iterations, the algorithm converges to the optimal parameter combination and outputs specific control commands for the take-up mechanism.

[0037] In practical applications of optical fiber production lines, the static path analysis unit processes a multimodal tension feature matrix comprising 24 segments. The transmission node generation subunit identifies a height abrupt change in the coordinates of segment 11, causing an abnormally high transmission path weight between this segment and its adjacent segments. The path topology construction subunit inserts three virtual transmission nodes in this region and smooths the distribution of tension feature values ​​through an interpolation algorithm, bringing the weight values ​​back to the normal range. The take-up execution control unit divides the stroke of the take-up roller into 8 adjustable path segments, where the priority passage coefficient calculation for path segment E (corresponding to segments 15 to 18) incorporates a bending radius compensation factor. The final generated take-up trajectory reduces tension fluctuations in the optical fiber as it passes through the guide roller assembly, resulting in a uniform interlayer structure in the wound optical fiber spool.

[0038] Because superconducting materials are extremely sensitive to tension fluctuations, the static path analysis unit has significantly enhanced the accuracy of the conduction path weight calculation. A temperature compensation factor is incorporated into the weight calculation to reflect changes in the tension conduction characteristics of the superconducting material at low temperatures. The adjustable path segment division of the take-up execution control unit is refined to every 5 cm interval, ensuring precise control of the tension distribution. A critical current density constraint is added to the objective function of the gradient optimization algorithm to ensure that the superconducting performance is not compromised during take-up. The implemented take-up system can maintain the tension stability of the superconducting tape, meeting the requirements of precision winding.

[0039] In composite cable winding production lines, the system faces challenges due to the stacking of multiple layers of insulation materials. The static path analysis unit sets differentiated conduction path weight calculation rules for different material layers. The conductor layer uses current density-based weight correction, while the insulation layer considers the influence of the dielectric loss factor. The winding execution control unit has developed a multi-layer collaborative control strategy. For the same adjustable path segment, independent priority passage coefficients are calculated for different material layers, and then a coupled algorithm is used to generate comprehensive control commands. This approach effectively solves the interlayer stress balance problem during composite cable winding, ensuring that the wound cable maintains the integrity of each functional layer's structure.

[0040] Example 3: See Figure 4The dynamic compensation unit monitors the deviation between the dynamic tension parameters and the static transmission node values ​​of the current wire section in real time through the deviation response subunit. On the optical fiber drawing production line, the position of the current wire section is fed back in real time by the encoder, and the system collects the dynamic tension data of that section every 50 milliseconds. The deviation response subunit compares the collected dynamic tension value with the corresponding transmission node value in the static tension transmission path to calculate the real-time deviation. This deviation includes not only the difference in tension amplitude but also phase shift information, reflecting the degree of deviation between the dynamic process and the static model. For example, during the winding process of carbon fiber composite materials, when the real-time monitored dynamic tension value of section 7 is 245N, while the static transmission node value is 230N, the system determines that there is a positive deviation of 15N.

[0041] The compensation decision subunit sets a tolerance threshold of 5% of the static transmission node value. When the real-time deviation exceeds this threshold, the system initiates the compensation mechanism. The compensation depth is determined based on the tension fluctuation characteristics of the current wire section, including fluctuation frequency, amplitude change rate, and historical deviation trend. In the continuous copper wire annealing production line, the system detected that the tension fluctuation frequency of the 12th section reached 8Hz, and the amplitude change rate exceeded 3% per second, indicating that deep compensation was required. The compensation decision subunit automatically expands the search range of adjacent wire sections according to the compensation depth parameter. For shallow compensation (deviation between 5% and 10%), only the transmission node value of one adjacent section is used; for deep compensation (deviation exceeding 10%), the set of transmission node values ​​of three adjacent sections is used.

[0042] After receiving the compensation decision result, the direction correction subunit generates a tension application direction compensation vector. The calculation of this vector comprehensively considers the deviation characteristics of the current segment and the distribution of conduction node values ​​in adjacent segments. In the aluminum alloy wire rolling system, when a tension deviation occurs in segment 5, the system calls upon the conduction node values ​​of segments 4, 5, and 6, and constructs a tension distribution field through a spatial interpolation algorithm. The direction of the compensation vector is determined by the deviation gradient, and its magnitude is proportional to the compensation depth. The mathematical representation of the compensation vector is:

[0043] in: Represents the compensation vector. for The weighting coefficients of neighboring segments, Indicates the first The propagation node vectors of neighboring segments, This represents the dynamic tension vector of the current segment. The weighting coefficients are based on... The distance between segments is dynamically adjusted; the closer the segment is to the current segment, the greater its weight.

[0044] After the compensation vector is generated, the system converts it into an adjustment amount for the tilt angle of the take-up roller. This conversion process takes into account the mechanical characteristics of the roller, including the maximum adjustable angle, response speed, and motion inertia. For example, in an ultra-thin copper foil take-up system, if the compensation vector indicates a 3.5° directional correction, the system decomposes this into a roller axial tilt angle of 2.1° and a circumferential deflection of 1.4°. The calculation of the tilt angle adjustment incorporates a mechanical transmission ratio coefficient to ensure the motion accuracy of the actuator.

[0045] The roller attitude control signal is generated using an incremental encoding method, outputting a 16-bit precision control word per control cycle (10 milliseconds), containing the tilt angle change and rate of change parameters. In the multilayer polymer film production line, the system updates the roller attitude control signal in real time, synchronously adjusting the motion parameters of the three degrees of freedom. The control signal is transmitted to the servo driver via the CAN bus, driving the motor to perform precise angle adjustments. After each adjustment, the system automatically verifies the match between the actual tilt angle and the target value; if the error exceeds 0.1°, a fine-tuning mechanism is triggered.

[0046] The dynamic compensation process has a typical application in the production of special steel wire ropes. When the system detects a sudden change in the dynamic tension of section 9, the deviation response subunit completes the deviation calculation within 2 milliseconds. The compensation decision subunit analyzes the tension fluctuation pattern of this section over the last 10 cycles and determines to adopt a medium compensation depth. The direction correction subunit calls the transmission node values ​​of sections 8, 9, and 10 to generate a compensation vector pointing to section 8. The roller tilt adjustment mechanism completes a 1.8° angle correction within 15 milliseconds, restoring the tension distribution to a balanced state. The entire compensation process, from deviation detection to completion, takes no more than 30 milliseconds, meeting the real-time requirements of high-speed production.

[0047] In a composite wire co-extrusion production line, the system needs to handle the coordinated compensation of multiple materials. When tension asynchrony occurs between the carbon fiber core and the polymer sheath, the dynamic compensation unit initiates a layered compensation strategy. A compensation algorithm based on the elastic modulus is used for the carbon fiber layer, while a viscoelastic response model is employed for the polymer layer. After spatial synthesis of the compensation vectors for the two materials, they are converted into six-degree-of-freedom adjustment commands for the composite roller. This layered processing method effectively solves the tension matching problem of heterogeneous materials during co-extrusion.

[0048] The take-up system for high-temperature superconducting tapes places special demands on dynamic compensation. Due to the changes in the mechanical properties of superconducting materials at low temperatures, the compensation decision subunit integrates a temperature-tension coupling model. When the liquid nitrogen cooling system causes local temperature fluctuations, the model automatically adjusts the reference benchmark for the conduction node values. The compensation vector generated by the direction correction subunit includes a temperature compensation component, ensuring precise control of tension distribution even at a low temperature of 77K. Low-temperature adaptable materials are used for roller attitude control to avoid the impact of cold brittleness on adjustment accuracy.

[0049] For the winding application of large-diameter submarine cables, a long-distance conduction compensation algorithm was developed using a dynamic compensation unit. When the cable length exceeds 500 meters, the system divides the conduction path into multiple compensation intervals. Each interval operates its compensation mechanism independently, while maintaining coordination between intervals through boundary conditions. The calculation of the compensation vector incorporates the drag coefficient of the submarine cable in water, reflecting the influence of the fluid environment on tension transmission. The roller adjustment mechanism employs a hydraulic servo system to provide sufficient torque to handle the inertial load of the large-diameter cable.

[0050] In the production of micro-electronic wires, the dynamic compensation unit handles tension control with μm-level precision. The system employs nanometer-resolution fiber optic sensors to monitor wire strain, achieving compensation decision accuracy on the order of 0.1N. The compensation vector generated by the orientation correction subunit enables micro-arc-level roller adjustment via piezoelectric ceramic actuators. The entire control system operates in a clean environment, preventing dust particles from interfering with the ultra-fine wires. This precision compensation mechanism ensures the consistency of high-end electronic wire products.

[0051] The production of multi-material composite braided yarns requires handling more complex tension compensation scenarios. A dynamic compensation unit establishes a material property database, storing the mechanical parameters of each component material. When a tension deviation is detected, the system automatically matches the material combination of the current braiding pattern and invokes the corresponding compensation algorithm. The compensation vector generated by the direction correction subunit is implemented through a multi-axis linkage mechanism, precisely controlling the tension balance of each braiding spindle. This adaptive compensation strategy significantly improves the structural uniformity of the composite braided yarn.

[0052] Example 4: The real-time feedback unit continuously collects the vibration spectrum of the wire during the winding process through a distributed vibration sensor network. On the copper wire finishing production line, sensors capture the vibration waveform of the wire surface at a sampling frequency of 10kHz, generating a set of spectrum data for each acquisition cycle (100 milliseconds). The system uses Fast Fourier Transform to convert the time-domain vibration signal into frequency-domain features, extracting three key indicators: the main resonant frequency, harmonic energy distribution, and bandwidth. These indicators are combined to form real-time spectrum feature values, used to characterize the current vibration state of the wire. For example, during the winding process of 0.5mm diameter copper wire, typical spectrum feature values ​​show that the main resonant peak is located at 850Hz, the second harmonic energy accounts for 18%, and the bandwidth is controlled within ±75Hz.

[0053] The path update unit monitors the deviation trend between real-time spectral characteristic values ​​and the values ​​of transmission nodes in the static tension transmission path. The system establishes a dynamic evaluation window, continuously analyzing the characteristic changes over the most recent 20 sampling periods. Evaluation parameters include the absolute value of the deviation, the slope of change, and the number of duration periods. In the fiber coloring production line, when a main resonant frequency offset exceeding 50Hz for five consecutive periods is detected, and the slope of change is greater than 10Hz / period, the system determines that the transmission path needs to be updated. The trigger condition setting adopts a hierarchical strategy, dynamically adjusting the sensitivity according to the production stage: a lenient threshold is used in the acceleration stage (deviation lasting 10 periods), and a strict threshold is used in the steady-state stage (deviation lasting 5 periods).

[0054] The changes in key parameters during a path update process. The data comes from the aluminum foil composite film production line and records the entire process parameters of the system reconstructing the conduction path after detecting a continuous deviation. See Table 1.

[0055] Table 1: Key Parameter Recording Table for Path Update Process Timestamp Main frequency deviation (Hz) Harmonic deviation (%) Trigger Level Reconstruction time (ms) Number of nodes in the new path 10:15:23.120 +62 +9.2 Level 2 48 14 10:15:23.240 +58 +11.5 Level 1 52 15 10:15:23.360 +67 +13.8 Level 1 45 16 10:15:23.480 +71 +15.3 Level 1 50 17

[0056] When the update conditions are met, the system re-triggers the dynamic feature modeling unit. This unit receives the latest vibration spectrum data and the current wire distribution topology map, and initiates the update process of the multimodal tension feature matrix. In the ultra-thin stainless steel strip production line, the update process first adjusts the feature fusion depth, expanding the original 3 segments to 5 segments to enhance the ability to capture vibration propagation paths. During matrix reconstruction, a vibration attenuation factor is introduced, adjusting the tension feature weights of each segment according to the spectral energy distribution. For example, when strong vibration is detected in segment 7, the feature weights of its adjacent segments (segments 6 and 8) are increased by 30% accordingly.

[0057] The static path analysis unit receives the updated multimodal tension feature matrix and initiates the path reconstruction algorithm. The reconstruction process employs an incremental optimization strategy, retaining unaffected nodes in the original conduction path and recalculating only the topology of the deviation region. In the case of multi-layer insulated cable production, the system detected a persistent spectral deviation in sections 9-12. The reconstruction algorithm first freezes the path structure of sections 1-8, then establishes a local optimization model for the problem section. The optimization process considers material creep characteristics, setting different conduction loss coefficients for the polymer insulation layer. The final generated new conduction path adds three auxiliary nodes to the problem section, improving the uniformity of tension distribution.

[0058] The system exhibits unique adaptability in special fiber weaving applications. When periodic vibrations caused by commutation of the weaving machine are detected, the real-time feedback unit automatically switches to a high-frequency sampling mode (20kHz). Spectral analysis focuses on the commutation characteristic frequency band (1.5-3kHz), accurately capturing transient vibration waveforms. The path update unit employs a predictive algorithm to pre-generate alternative conduction paths 200 milliseconds before each commutation. This proactive approach allows the system to maintain tension stability during mechanical commutation, avoiding the post-compensation delays of traditional methods.

[0059] In the annealing and take-up system of high-temperature alloy wire, changes in ambient temperature cause drift in material stiffness parameters. The real-time feedback unit integrates temperature sensor data to perform real-time temperature compensation on spectral characteristic values. When the furnace temperature rises to 850℃, the system automatically adjusts the parameters of the resonance vibration analysis algorithm to compensate for the thermal decay effect of the material's elastic modulus. The path update unit introduces a temperature-tension coupling model to ensure that the newly generated conduction path meets the requirements of high-temperature operating conditions. The calculation of conduction node values ​​includes a thermal expansion compensation term, reflecting the influence of the temperature gradient on the tension distribution.

[0060] Long-span cable winding systems face the challenge of long-distance vibration transmission. The real-time feedback unit employs a segmented spectrum analysis method, dividing the 50-meter cable into eight monitoring sections. Each section undergoes independent vibration analysis, and the system integrates data from all sections to establish a global vibration mode. The path update unit develops a distributed reconstruction algorithm, optimizing the transmission paths of each section in parallel and then coordinating their connections through boundary conditions. This method effectively solves the interference problem caused by vibration wave reflection in long-distance cables, significantly reducing the amplitude of winding tension fluctuations.

[0061] The production of micro-electronic wires places higher demands on vibration control. This system employs a laser Doppler vibrometer to replace contact sensors, achieving nanometer-level vibration resolution. Real-time spectrum analysis tracks low-frequency micro-vibrations below 200Hz, subtle disturbances often ignored in conventional systems. The path update unit features a multi-level caching architecture, ensuring fine-tuning of the conduction path within 1 millisecond. The updated conduction path includes sub-millimeter-level precision node positioning, meeting the process standards for high-end electronic wires.

[0062] In a composite wire co-extrusion production line, the system needs to distinguish the vibration characteristics of different materials. The real-time feedback unit runs a material identification algorithm to determine the material composition of the current monitoring point based on the vibration waveform characteristics. The path update unit establishes a multi-material conduction model, configuring independent path optimization parameters for each material. When heterogeneous vibrations are detected between the carbon fiber layer and the polymer layer, the system generates a layered conduction path update scheme to optimize the tension distribution of the two materials separately. This refined processing significantly improves the interfacial bonding quality of the composite wire.

[0063] The system excels in winding non-circular cross-section wires. For flat and irregularly shaped wires, the vibration sensor array employs a three-dimensional arrangement to capture anisotropic vibration characteristics. Real-time spectral feature values ​​include the cross-sectional direction vibration ratio parameter, reflecting the impact of wire attitude changes. The path update unit introduces a cross-sectional shape factor to consider additional tension fluctuations caused by geometric characteristics during path reconstruction. The adjustment commands for the take-up rollers include a rotational component, actively compensating for periodic disturbances caused by asymmetrical cross-sections.

[0064] Example 5: The multi-source sensing acquisition unit uses a distributed fiber optic sensor network to acquire the strain characteristics of the wire. In the copper alloy wire production line, fiber optic sensors are arranged on the wire surface in a helical winding manner, and the spacing between sensing nodes is dynamically adjusted according to the wire diameter. For a 1.2mm diameter wire, the node density is set to 3 measuring points per centimeter. The sensors emit modulated light signals of a specific wavelength, and the micro-strain distribution on the wire surface is calculated by analyzing the phase shift of the backscattered light. The strain feature extraction process includes three dimensions: axial strain reflects the tensile state, circumferential strain monitors the torsional effect, and the composite strain gradient identifies the bending stress concentration area. In the case of nickel-based alloy wire production, the system successfully captured a local strain anomaly of 0.05% at the guide wheel contact point.

[0065] Motion velocity characteristics are acquired using a high-precision photoelectric encoder, which is mounted on the guide roller bearing housing and employs a non-contact measurement principle. In the polyimide film production line, the encoder captures marked points on the guide roller surface at a rate of 5000 frames per second, and the linear velocity of the wire is calculated using image processing algorithms. The velocity characteristics include three parameters: instantaneous velocity, rate of change of acceleration, and curvature of the motion trajectory. The system features a specially designed velocity anomaly detection mechanism; when the velocity jump between adjacent sampling points exceeds 5%, a data verification process is automatically triggered. For highly elastic silicone rubber wires, an additional velocity smoothing algorithm is added to eliminate measurement noise caused by material deformation.

[0066] The dynamic compensation unit and the real-time feedback unit constitute a closed-loop control system, with the spectral characteristic value output by the real-time feedback unit serving as the core input parameter for dynamic compensation. In the carbon fiber prepreg tape winding system, the closed-loop control operates at a fixed cycle: the vibration spectrum is acquired every 50 milliseconds and converted into a set of characteristic values ​​containing the dominant frequency, harmonic energy, and bandwidth; this set of characteristic values ​​is compared with a preset threshold, and if it exceeds the threshold, the compensation process is activated. The trigger condition setting adopts an adaptive mechanism, automatically adjusting the sensitivity according to the production line speed. When the winding speed increases to 30 meters per minute, the system tightens the harmonic energy deviation threshold from 15% to 8%.

[0067] The closed-loop control data flow forms a bidirectional coupling. After the dynamic compensation unit performs tension direction correction, the real-time feedback unit immediately monitors the correction effect. In the titanium alloy wire cold drawing production line, the system records the trend of spectral characteristic changes after each compensation. If the characteristic value still does not return to the normal range within three compensation cycles, the compensation decision level is automatically increased. At the same time, the roller posture adjustment data during the compensation process is fed back to the real-time feedback unit to optimize the spectral analysis algorithm parameters. This bidirectional data exchange enables the system to have continuous learning capabilities, gradually adapting to the dynamic characteristics of different materials.

[0068] The production demonstration system for special alloy wires showcases its performance under extreme conditions. When handling shape memory alloys, the distributed fiber optic sensors employ high-temperature probes capable of withstanding ambient temperatures up to 450°C. The encoder is equipped with an anti-oxidation coating to prevent measurement accuracy degradation under high-temperature environments. The real-time feedback unit sets a special monitoring frequency band for the phase transition temperature point to capture the characteristic frequency abrupt changes caused by the material's phase transition. The dynamic compensation unit integrates a phase transition prediction model, activating the compensation mechanism in advance when a characteristic frequency precursor is detected. The closed-loop control cycle is shortened to 20 milliseconds, ensuring tension stability at the phase transition critical point.

[0069] The superconducting tape take-up system demonstrates precise control capabilities. The distributed fiber optic sensors, employing a dedicated superconducting model, maintain a strain measurement accuracy of 0.001% even in a liquid nitrogen cryogenic environment. The encoder system incorporates an anti-condensation device to prevent mechanical jamming caused by low temperatures. A real-time feedback unit analyzes the 15-25kHz high-frequency vibration spectrum, a band sensitive to microscopic defects in the superconducting layer. The dynamic compensation unit utilizes a nanometer-scale compensation algorithm, achieving a roller attitude adjustment resolution of 0.001 degrees. The closed-loop system operates continuously at 77K, with a compensation trigger response time controlled within 5 milliseconds, meeting the stringent requirements of superconducting materials regarding tension fluctuations.

[0070] In the co-extrusion production line for composite wires, the system achieves simultaneous monitoring of multiple materials. Distributed fiber optic sensors, employing a multi-core structure, independently collect strain data from the core wire and sheath layer. The encoder system is equipped with dual reading heads to measure the movement velocity of different materials. The real-time feedback unit runs a material identification algorithm, automatically distinguishing the material layer where the monitoring point is located based on vibration spectrum characteristics. The dynamic compensation unit establishes a layered compensation model, generating independent compensation vectors for the core wire and sheath layer. The closed-loop control adopts a parallel processing architecture, simultaneously maintaining tension balance between the two materials. In the production of carbon fiber-PTFE composite lines, this system successfully solves the tension mismatch problem caused by the difference in shrinkage rates between the core wire and sheath layer.

[0071] The application of large-section submarine cable winding demonstrates the system's scalability. Distributed fiber optic sensors employ kilometer-level distributed measurement technology, with a single fiber covering a cable length of 500 meters. The encoder system is installed at multiple guide wheel nodes, calculating the overall movement speed through data fusion. The real-time feedback unit performs segmented spectrum analysis, dividing the long cable into eight logical segments for independent monitoring. The dynamic compensation unit develops a collaborative compensation strategy, ensuring consistent compensation actions across segments based on boundary conditions. The closed-loop control adopts a master-slave architecture, with the central controller coordinating 32 local compensation units to ensure uniform tension distribution across the kilometer-long cable.

[0072] Microelectronic bonding wire production demands ultra-high precision control. Distributed fiber optic sensors employ ultraviolet writing technology to fabricate micro-Bragg grating arrays on 25μm gold wires. The encoder system, equipped with a 100nm resolution optical scale, precisely captures the motion characteristics of the fine wires. The real-time feedback unit focuses on low-frequency micro-vibration analysis from 0-200Hz, using wavelet transform to extract sub-micron vibration features. The dynamic compensation unit, integrated with a precision motion platform, achieves nanometer-level roller adjustment via piezoelectric ceramic actuators. The closed-loop system operates in a clean environment, completing a control cycle every millisecond to maintain sub-Newtonian stability of the bonding wire tension.

[0073] The system employs a dedicated sensing solution for rectangular copper busbars: distributed fiber optic sensors are symmetrically arranged along the four edges, and the encoder uses a multi-axis measurement mode to capture complex motion trajectories. A real-time feedback unit analyzes the torsional vibration characteristics of the cross-section and establishes a three-dimensional vibration model. A dynamic compensation unit generates a spatial compensation vector, and a six-degree-of-freedom adjustment mechanism corrects the wire posture. In the case of flat electromagnetic wire production, this system effectively suppressed warping deformation of the special 5:1 aspect ratio cross-section during the winding process.

[0074] The production of multi-material composite braided yarn requires handling dynamically changing material combinations. Distributed fiber optic sensors employ rapid reconfiguration technology to automatically switch monitoring modes based on the braiding pattern. An encoder system synchronously tracks the motion trajectories of 16 braiding spindles. A real-time feedback unit uses a mode recognition algorithm to determine the material composition of the current monitoring point in real time. A dynamic compensation unit accesses a material database to match optimal compensation parameters for each material combination. The closed-loop system processes 32 channels of sensor data per millisecond, generating a coordinated tension balance scheme for the 16 spindles to ensure the structural consistency of the composite braided yarn.

[0075] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0076] 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 take-up tension intelligent control system, characterized by, The method comprises the following steps: A multi-source sensing acquisition unit obtains a wire distribution topology graph through a tension sensing platform, collects tension of a wire reference point based on the wire distribution topology graph, and obtains a reference tension parameter; The multi-source sensing acquisition unit collects characteristics of each wire section in the wire distribution topology graph to obtain wire strain characteristics and motion speed characteristics corresponding to each wire section respectively; A dynamic characteristic modeling unit performs multi-modal fusion on the wire strain characteristics and motion speed characteristics corresponding to each wire section respectively based on the reference tension parameter to generate a multi-modal tension feature matrix; A static path analysis unit constructs a static tension conduction path based on each tension feature value in the multi-modal tension feature matrix and the wire section corresponding to the tension feature value respectively; A wire collection execution control unit obtains an adjustable path of a wire collection mechanism, controls a wire collection trajectory of the wire collection mechanism on the adjustable path based on a conduction node value in the static tension conduction path and the wire section corresponding to the conduction node value respectively, and compensates and corrects a tension application direction of the wire collection mechanism according to a dynamic tension parameter of a current wire section and the conduction node value in the static tension conduction path during a wire collection process. The multi-source sensing acquisition unit comprises:

2. The intelligent take-up tension control system of claim 1, wherein, A wire partition subunit obtains the wire distribution topology graph, performs section segmentation based on a density gradient in the wire distribution topology graph, and obtains a plurality of wire sections; A strain acquisition subunit obtains a wire collection traction direction, selects an initial section for wire strain characteristic acquisition according to the wire collection traction direction, selects a next wire section for wire strain characteristic acquisition based on the wire collection traction direction, and performs wire strain characteristic acquisition on each wire section until the wire strain characteristic acquisition is completed; A speed mapping subunit obtains a wire motion vector, reselects an initial section for motion speed characteristic acquisition according to the wire motion vector, selects a next wire section for motion speed characteristic acquisition based on the wire motion vector, and performs motion speed characteristic acquisition on each wire section until the motion speed characteristic acquisition is completed. The dynamic characteristic modeling unit performs the following steps:

3. The intelligent take-up tension control system of claim 1, wherein, Determines a characteristic fusion depth according to the reference tension parameter; Obtains wire strain characteristics and motion speed characteristics corresponding to a first wire section, selects a comparison section for multi-modal fusion according to a position coordinate of the first wire section and the characteristic fusion depth, and generates a tension feature value corresponding to the first wire section; After sequentially obtaining wire strain characteristics and motion speed characteristics of other wire sections, the multi-modal fusion operation is repeatedly performed to generate tension feature values of each wire section, and a matrix mapping is performed according to the wire distribution topology graph to generate the multi-modal tension feature matrix. The static path analysis unit comprises:

4. The intelligent take-up tension control system of claim 1, wherein, A conduction node generation subunit extracts each tension feature value in the multi-modal tension feature matrix and the wire section coordinate corresponding to the tension feature value respectively; A path topology construction subunit establishes a conduction node connection chain based on a spatial relationship between wire section coordinates, and defines a conduction path weight value according to a gradient change of the tension feature value; ​ The static path optimization subunit adopts a dynamic topology optimization algorithm to iteratively calculate the conduction path weight value, and generates a minimum conduction loss path as the static tension conduction path.

5. The intelligent take-up tension control system of claim 1, wherein, The take-up execution control unit executes: Obtaining the conduction node values in the static tension conduction path and the corresponding wire sections; Extracting all adjustable path segments of the take-up mechanism and the corresponding adjacent wire sections of each path segment; Determining the take-up starting point, weighting and accumulating the conduction node values of all adjacent wire sections corresponding to the adjustable path segment as the priority passing coefficient of the adjustable path segment, and generating the take-up trajectory of the take-up mechanism starting from the take-up starting point using a gradient optimization algorithm.

6. The intelligent take-up tension control system of claim 1, wherein, The dynamic compensation unit includes: The deviation response subunit obtains the real-time deviation of the dynamic tension parameter of the current wire section and the corresponding conduction node value in the static tension conduction path; The compensation decision subunit determines the compensation depth according to the tension fluctuation characteristics of the current wire section when the real-time deviation exceeds the tolerance threshold, and obtains the conduction node value of the adjacent wire section based on the compensation depth; The direction correction subunit generates a tension application direction compensation vector combining the conduction node value of the adjacent wire section, and adjusts the tension application direction of the take-up mechanism.

7. The intelligent take-up tension control system of claim 6, wherein, The direction correction subunit executes: According to the tension application direction compensation vector, calculate the inclination adjustment amount of the take-up roller; Based on the inclination adjustment amount, generate a roller posture control signal to synchronously update the tension application direction of the take-up mechanism.

8. The intelligent take-up tension control system of claim 1, wherein, Further comprising: The real-time feedback unit continuously collects the wire vibration frequency spectrum during the take-up process, and converts the wire vibration frequency spectrum into real-time spectrum characteristic values; When the real-time spectrum characteristic values and the conduction node values of the static tension conduction path deviate continuously, the path updating unit re-triggers the dynamic characteristic modeling unit to update the multi-modal tension characteristic matrix, and drives the static path analysis unit to reconstruct the static tension conduction path.

9. The intelligent take-up tension control system of claim 1, wherein, The multi-source sensing acquisition unit acquires wire strain characteristics through distributed optical fiber sensors and motion speed characteristics through encoders.

10. The intelligent take-up tension control system of claim 1, wherein, The dynamic compensation unit and the real-time feedback unit form a closed loop control, and the real-time spectrum characteristic values output by the real-time feedback unit are used as the compensation correction trigger condition of the dynamic compensation unit.

Images