Intelligent yarn low tension control system

Abstract

The application discloses a kind of intelligent yarn low tension control systems, it is related to textile control system technical field, the tension estimation module is with the torque and speed signal of servo driver output with physical tension sensor input, and after being handled to the model predictive controller module and adaptive PID module with filter, provide tension data signal;The visual detection unit and telescopic reed driving mechanism are coupled through control signal and timing synchronization interface, and visual detection unit is according to the segmented position information corresponding to the width section output to edge real-time controller with telescopic reed position.The application is fused by the perception architecture of multi-channel physical sensing and virtual tension estimation, it can still provide stable, redundant tension estimation and output confidence information when there is insufficient or local interference, overcome the problem that traditional single-point sensing is susceptible to local droplet and dirt interference, high cost and installation space limited.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of textile control systems, in particular to an intelligent yarn low-tension control system. BACKGROUND

[0002] In high-density, multi-head cloud dyeing and spray dyeing textile production lines, yarns are extremely susceptible to tension fluctuations, parallel crowding and shutdown adhesion when passing through dyeing tanks, cloud dyeing wheels and drying sections due to speed difference, liquid surface fluctuations, mist disturbance and changes in temperature and humidity during shutdown. Traditional tension control methods rely on single-point contact sensing or open-loop speed-based simple PID regulation, lacking the ability to perceive large-scale parallel yarn paths, visual position discrimination and process linkage response, resulting in uneven dyeing, limited production capacity and high manual intervention rates.

[0003] Patent CN103305998B discloses a yarn tension control system for a ball warp dividing machine, which achieves balanced tension control of the entire machine and good consistency in high and low speed tension control.

[0004] The above-mentioned patent has the advantages of simple structure, low cost, good consistency in high and low speed tension control, and feedback of the feedback power of the tension control motor to the winding frequency converter for effective energy saving, but lacks the ability to perceive large-scale parallel yarn paths, visual position discrimination and process linkage response.

[0005] Therefore, the present application proposes an intelligent yarn low-tension control system that can provide stable and redundant tension estimation and output confidence information even in the presence of insufficient measurement points or local interference. SUMMARY

[0006] The present application aims to provide an intelligent yarn low-tension control system to solve the technical problems of traditional single-point sensing being susceptible to local liquid droplets and dirt, high cost of additional points and limited installation space.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solution: an intelligent yarn low-tension control system, the control system comprising: a multi-channel tension measurement unit, a visual detection unit, an actuator unit, an edge real-time controller, an upper PLC, a prescription management and human-computer interaction unit;

[0008] The edge real-time controller is respectively provided with a tension estimation module, a model prediction controller module and an adaptive PID module, the tension estimation module takes the output of a physical tension sensor and the torque and speed signals of a servo driver as input, and provides a tension data signal to the model prediction controller module and the adaptive PID module after filter processing; the visual detection unit and the telescopic reed driving mechanism are coupled through a control signal and a timing synchronization interface, and the visual detection unit outputs segmented position information corresponding to the telescopic reed position to the edge real-time controller according to the width.

[0009] Preferably, the multi-channel tension measurement unit is provided with physical contact type tension sensors and at least one non-contact optical tension measurement device at several positions along the yarn conveying path.

[0010] The multi-channel tension measurement unit comprises: a plurality of contact type micro-cantilever strain tension sensors arranged in an array along the key nodes of the yarn arrangement; and at least one non-contact laser vibration or photoelectric measurement unit installed according to the width partition and physically connected to the tension estimation module of the edge real-time controller.

[0011] Preferably, the visual detection unit is arranged as a camera array along the width direction and is provided with a backlight and an anti-fog device.

[0012] The visual detection unit is composed of a plurality of high-speed industrial cameras, the frame rate, field of view and installation position of the industrial cameras are distributed according to the width segment and the yarn density, the camera array, the backlight unit, the electric heating anti-fog cover and the image preprocessing unit are in the same protective shell, and the segmented yarn parallel and spacing mapping data are transmitted to the edge real-time controller through the image bus.

[0013] Preferably, the actuator unit is provided with a plurality of groups according to control groups, each group includes at least one servo yarn feeding driver, at least one brake and clamping device, and a telescopic reed driving mechanism corresponding to the camera array and a set of active pressure rollers.

[0014] The actuator unit is divided into a control unit for 16-32 yarns per group, and each control unit includes: one servo driver, one yarn feeding shaft with torque measurement function, one electromagnetic or pneumatic brake, and a micro-step or servo side shift driver for telescopic reed side shift.

[0015] The servo driver is connected to the edge real-time controller through the EtherCAT bus point-to-point, and the rotational speed and torque sampling signals are exposed.

[0016] Preferably, the sley driving mechanism comprises a plurality of individually position-adjustable sley tooth modules, each sley tooth module is equipped with a micro driver, a position encoder and a mechanical stopper, the sley tooth modules are arranged in several independently controllable sley sections according to the width of the fabric, and the control bus of each sley section is connected to the sley position control interface of the edge real-time controller.

[0017] Preferably, the tension estimation module comprises a virtual tension calculation submodule based on the torque and speed of the servo motor and a Kalman filter submodule for fusing the output of the physical tension sensor and the virtual tension;

[0018] The virtual tension calculation submodule and the physical sensor are connected to the edge real-time controller through a sampling interface, respectively, and the state dimension, observation matrix and noise covariance of the Kalman filter submodule are stored in the edge real-time controller as configurable parameters.

[0019] Preferably, the edge real-time controller comprises a model predictive controller module, a feedforward compensation module and an adaptive PID module.

[0020] The model predictive controller module stores the control model parameters in the form of a discrete-time state-space model, a prediction step N and an input-output constraint matrix in the edge real-time controller, and the model predictive controller module and the adaptive PID module share the tension estimation value, the visual distribution mapping and the actuator state information through an internal data bus; at the same time, the edge real-time controller is configured with a control sampling clock source of at least 1 kHz and provides a real-time synchronization signal externally.

[0021] Preferably, the control system further comprises an event-driven switching unit, the event-driven switching unit is connected to the liquid inlet pump controller, the spraying controller and the dryer controller of the process unit through an industrial Ethernet, the event-driven switching unit manages a plurality of sets of control parameters with the prescription identifier as the index, and when receiving the state change signal of the process unit, the event-driven switching unit indexes and sends the corresponding control parameter set to the edge real-time controller.

[0022] Preferably, the upper PLC, prescription management and human-computer interaction unit comprises a prescription database, a running history record database and a parameter version management module, the prescription database is associated with the control parameter set, the visual threshold set and the sley position configuration of the prescription identifier, and the parameter version management module timestamps the change record of the prescription and replays the historical parameters to the edge real-time controller through the network interface when needed.

[0023] Preferably, the control system further comprises an anti-sticking relaxation unit composed of several light vibration actuators, air pulse generators and local temperature and humidity sensors arranged in the out-drying zone and the yarn separating zone, the light vibration actuators and the air pulse generators being connected with the edge real-time controller and the event-driven switching unit through a dedicated control bus, and the measurement signals of the local temperature and humidity sensors being recorded in the operation history record database at the same time.

[0024] Compared with the prior art, the present application has the following advantages:

[0025] 1. The present application realizes stable and redundant tension estimation and outputs confidence information even in the presence of insufficient measurement points or local interference through the fusion perception architecture of multi-channel physical sensing and virtual tension estimation, overcomes the problems of traditional single-point sensing being easily disturbed by local droplets and dirt, high cost of supplementing points and limited installation space, improves measurement coverage and system robustness, improves tension estimation accuracy and reliability, reduces misjudgment and downtime;

[0026] 2. The present application realizes independent speed, torque and position control and horizontal displacement adjustment to redistribute yarn spacing by group through the active actuator structure of grouping and channel and the segment-by-segment positioning mechanism of the expansion reed, solves the problem of local congestion, empty space and parallel misalignment in large-width parallel yarns that cannot be adjusted locally, avoids global errors caused by single overall adjustment, and improves parallel uniformity and consistency of tension per group;

[0027] 3. The present application realizes simultaneous consideration of high-speed real-time response and medium-term disturbance predictive compensation through the hybrid control of MPC and adaptive PID + event-driven prescription switching, overcomes the problem that single PID cannot consider system delay and coupling and responds slowly to sudden disturbances, avoids adjustment lag caused by manual intervention during process switching, and improves system robustness and disturbance suppression capability;

[0028] 4. The present application realizes real-time detection and correction of parallelism, congestion, automatic execution of physical anti-sticking action and recording of events when stopped or temperature and humidity are abnormal through the visual-mechanical coupling and anti-sticking event triggering subsystem, eliminates geometric parallelism errors that cannot be detected by tension control alone and yarn sticking problems after shutdown, improves automation of protective measures during shutdown, reduces dyeing defects, and improves production continuity and product quality. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 The present application is a yarn low-tension control cycle flowchart. DETAILED DESCRIPTION

[0030] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0031] Embodiment 1

[0032] Please refer to Figure 1 An intelligent yarn low-tension control system, applicable scenarios: existing cloud dyeing and spray dyeing combined machine, width ≈ 750 mm, about 400 yarn heads, and low-cost modification is expected to achieve low-tension control of each group;

[0033] Further, the edge real-time controller: a real-time embedded industrial controller, supporting EtherCAT, bus cycle ≤ 1 ms (for real-time data sampling and PID closed loop); multi-channel physical tension sensor: micro-cantilever strain pressure sensor, measurement range 0.1-100 cN, resolution ≤ 0.1 cN, output analog ± 10V or digital SPI / I2C, at least 1 per group (arranged according to every 32 roots as a group); servo driver: EtherCAT slave station, encoder resolution ≥ 5000 PPR, supporting speed and torque feedback; brake unit: electromagnetic brake or pneumatic gripper, response time ≤ 50 ms; camera and lighting: 1 low-speed detection camera for overall parallel inspection; HMI / PLC: upper industrial computer for prescription management and historical record;

[0034] Each control unit = 32 yarns, each group containing: 1 servo drive, 1 physical tension sensor point, and a number of winding and guide wheels;

[0035] The real-time controller, servo, and tension sensor are connected through EtherCAT, real-time Ethernet; sampling period: 1 kHz for tension sensor and servo data; PID closed loop execution frequency: 1 kHz; upper PLC synchronizes prescription or alarm with edge controller every 500 ms;

[0036] Virtual tension estimation: mapping (torque-tension) based on servo motor current, torque constant, and speed is established as virtual tension; Kalman filter: first-order state Kalman filter is realized in the edge controller, fusing physical sensor (y1) and virtual tension estimation (y2), and outputting estimated tension x; sampling time T s = 1 ms, initial process noise Q and measurement noise R can be set as Q = 1e-4, R physical = 1e-2, R virtual = 5e-2 (only initial value, field tuning);

[0037] Controller implementation: inner loop speed closed loop (provided by servo drive) + outer loop tension PID; PID structure and adaptive rules: when the yarn feeding speed < 50 m / min, PID gain set A is applicable (Kp: 0.8-1.5, Ki: 0.5-1.0; Kd: 0-0.05); when the yarn feeding speed ≥ 50 m / min, gain set B is switched (Kp: 0.4-0.9, Ki: 0.2-0.5; Kd: 0-0.02); feedforward compensation: based on the acceleration feedforward uff= a · (dw / dt), a is identified through experiments (a is initially set as 0.02);

[0038] Debugging and calibration steps:

[0039] Install tension sensors at the inlet and outlet of each group and ground shield; read the servo current and torque under the conditions of no load and known tension (through weights or standard force sensing), and fit the torque-tension linear model (least squares); estimate the noise matrix Q, R with three types of data: static, light disturbance, and large disturbance; single group closed loop step test (small amplitude speed step), adjust Kp, Ki according to the response; load the prescription and run at low speed for 30 min, check the uniformity of the finished product and alarm;

[0040] Event: if the tension estimation error > 20% and continuously for 100 ms, the edge controller triggers speed reduction and sends an alarm, switches to the relaxation mode (reduces the tension target by 30%) for this group, and notifies the upper PLC;

[0041] Data recording: 1 kHz data is summarized (average, peak) for each group and stored to the upper PLC or database, and the prescription ID and running file are saved each time of production.

[0042] Example 2

[0043] Please refer to Figure 1 An intelligent yarn low-tension control system, applicable scenarios: new machines or modified machines, width ≥ 1200 mm, yarn head 800-1600, requiring high-performance disturbance suppression and segment-by-segment distributed control;

[0044] Further, the edge controller: a high-performance real-time industrial computing unit, supporting dual-core division: 1 core for 1 kHz inner loop, 1 core for 100 Hz MPC solving and vision data processing, supporting EtherCAT; camera array: deploy an industrial camera per 200 mm width, specification ≥ 2MP, frame rate 200fps, with backlight and anti-fog cover; telescopic reed: reed segment width 25-50 mm, independently adjustable by micro-step; high-density servo drive: 16 yarns per group, servo drive with torque measurement; laser and photoelectric non-contact tension unit: several for detecting high dynamic disturbance (integrated into tension estimation);

[0045] Inner loop 1kHz: Adaptive PID + feedforward, for real-time control of servo drive and fast disturbance rejection;

[0046] Outer loop 100Hz: MPC module runs at 100Hz, predicts future N-step tension response based on discrete state-space model and optimizes control action, MPC output sent as tension reference correction to inner loop;

[0047] Vision cycle: camera 200fps, but vision processing aggregates every 50ms (20Hz) to output lap distribution matrix and side-by-side, crowded events as MPC disturbance inputs;

[0048] Vision-tension coupling:

[0049] Image preprocessing: denoising - binarization - vertical projection to find yarn centerline - CNN-assisted identification of side-by-side errors (crowded and empty);

[0050] Mapping rule: vision output lap width coordinate is mapped to control group index (camera pixel x - lap width mm - group number), each group gets a crowdedness coefficient ci ∈ [0, 1] as MPC exogenous disturbance di (corresponding constraint or weight adjustment);

[0051] Calibration and identification (model establishment):

[0052] Experimental identification is used: for several representative operating points, real-time step, pulse control (speed step change, brake pulse), record group tension response, use least squares or subspace identification to get state-space model (A, B, C, D); Discretize the identification model to T smpc = 10ms step and use it as the internal model of MPC;

[0053] Operation, fault strategy:

[0054] Vision detects "side-by-side crowdedness" and lasts > 100ms - MPC reduces the reference tension of the corresponding group by ΔT = -10% for a short time and instructs the expansion reed to move slightly (10-50μm step) to expand the yarn surface, while recording the event;

[0055] When MPC fails to solve (real-time quadratic programming solver does not converge within 9ms) - fallback strategy: use the last MPC output or turn on conservative PID gain, and record the alarm in the log.

[0056] Example 3

[0057] Please refer to Figure 1 An intelligent yarn low-tension control system, suitable for scenarios: research and development or high-end demonstration line, independent tension measurement and accurate control for each yarn (one by one, single head drive);

[0058] Further, each yarn head is equipped with: micro servo or stepper drive (with torque sensor), micro guide roller and micro pressure roller mechanism; encoder resolution ≥20000CPR; independent tension sensor for each yarn: ultra-thin contact force sensor (installed between guide rollers, measurement range 0.05-50cN); high-density camera or line array camera: for detecting the displacement of single yarn and yarn surface morphology; edge controller: multi-channel different IO, strong real-time processing capability (can process 400 independent channels);

[0059] Each yarn realizes inner loop speed closed loop and outer loop tension closed loop (two-level independent control); control law: each uses cascade control (speed inner loop 2kHz, tension outer loop 500Hz), outer loop uses state feedback + sliding mode compensation to suppress nonlinear friction; tension filtering: second-order filter, band-pass filtering to suppress high-frequency vibration interference;

[0060] The micro pressure roller spacing and material are optimized to minimize friction; the pressure roller uses ceramic coating to reduce wear; the tension sensor of each yarn is installed in a detachable slide groove, which is convenient for maintenance and can be quickly replaced;

[0061] Each yarn is calibrated statically: a known microgram-level force is applied to calibrate the sensor linearity; dynamic characteristic determination: frequency scanning (1-200Hz) is performed to observe the system frequency response and set the filter bandwidth.

[0062] Embodiment 4

[0063] Please refer to Figure 1 An intelligent yarn low-tension control system, suitable for scenarios where: the device structure cannot directly install contact sensors or the cost and downtime are limited, and only existing motor signals and a small amount of physical sensors can be used for modification;

[0064] Further, only one physical tension sensor is installed in each group for periodic calibration and calibration, and the rest rely on virtual tension estimation of servo current, torque and speed; virtual tension model: T virtual =k t ·l motor +k w ·w+b (k t , k w , b are uncalibrated coefficients);

[0065] Under the conditions of empty yarn and standard tension samples, N=10 working points are measured: collect (l motor , w, T physical ); fit k t , k w , b using least squares method, calculate R 2 , if R 2< 0.98 then increase sample points or check mechanical hysteresis; online correction triggered automatically every 6 hours: short production speed reduction and reading of physical sensors with known small braking force (e.g. 5% of rated torque) and update of parameters;

[0066] Kalman filter configuration (virtual + diluted physical): due to the sparsity of physical measurement points, the measurement noise R of the Kalman filter is diluted virtual The process noise Q should be increased (e.g. R virtual = 1e-1) to allow model updates;

[0067] Risk: virtual estimation sensitive to machine friction, temperature rise; mitigation: online temperature monitoring and re-calibration triggered when temperature change > 5°C; switch to safe relaxation mode and alarm immediately when a severe disturbance occurs (visual detection of yarn face anomalies).

[0068] Example 5

[0069] See Figure 1 , an intelligent yarn low tension control system, applicable scenarios: engineering production and operation and maintenance after deployment of any embodiment;

[0070] Further, mechanical installation acceptance: mechanical base surface error of expansion reed, press roller, camera cover and sensor ≤ 0.5mm; electrical acceptance: shielding ground, wiring terminal without looseness, EtherCAT bus terminal resistance correct, ground loop impedance < 1Ω; software preparation: upload prescription database, set initial parameter set (PID, MPC initial model, Kalman matrix), and enable HMI to read logs;

[0071] System power on, all actuators idle at low speed (10% rated speed) to check reverse, limit and emergency stop loop; tension sensor measures zero point and records; camera does white balance and intrinsic parameter calibration; calibrate virtual model according to Example 4; do identification experiment (step and pulse) according to Example 2; save template file version; single group closed loop step test (speed step 10%-30%), record overshoot and steady state error and adjust Kp, Ki; run MPC simulation for 8 hours in simulation environment (using identified model), observe QP solving success rate and behavior; then run online at 50% speed for 4 hours; do 10 parallel error trigger experiments for typical yarn types, record expansion reed side shift and vision threshold adjustment value, determine congestion coefficient trigger threshold;

[0072] Check tension sensor baseline drift ≤ ± 1 cN per shift (8 hours); if the deviation is out of limit, perform online correction procedure; clean camera cover and backlight every week, check light source attenuation and correct white balance; perform model re-identification (short step and pulse) once a month, and submit new model to version management module;

[0073] Fault, retreat logic;

[0074] Controller software error or QP solving timeout: edge controller goes to "degraded operation" - close MPC, only enable conservative PID, trigger alarm and record diagnostic log;

[0075] Tension sensor short circuit, failure: automatically switch to non-virtual tension estimation (if there is no physical sensor in the group) and reduce the maximum allowed speed to 60%;

[0076] Camera occlusion or backlight failure: visual data is marked as failed and triggers the expansion reel to return to the safe mid position, the system runs with conservative parameters and notifies the operator;

[0077] Shutdown anti-sticking process (event-driven) - implemented by an event-driven switching unit:

[0078] 1. Receive a shutdown or drying temperature loss signal;

[0079] 2. Issue a "anti-sticking relaxation" parameter set: tension target reduced by 30%, expansion reel widened by 10% position, start the drying zone light vibration (frequency 5-10 Hz, amplitude micron level) and intermittent air pulse (pulse period 2s, duration 10min);

[0080] 3. Record and replay the event process;

[0081] Run data and event records use hierarchical storage: real-time buffer, periodic upload, long-term archive; file format JSON; remote diagnostic interface: HTTPS secure channel; support remote log reading and model download.

[0082] Working principle: several physical contact tension sensors and several non-contact optical and laser vibration measurement units are arranged at key nodes of the yarn conveying path, and speed and torque signals are read from the servo driver. These measurement quantities are first processed by a pre-processing unit in the edge real-time controller, then fused by Kalman filter and state estimator with physical measurement and virtual tension estimation based on motor signal, and reliable tension estimation value and confidence at current time for each control group are output for upper layer control;

[0083] The edge real-time controller implements a double-rate control structure internally: the high-speed inner loop runs adaptive PID and feedforward compensation to ensure servo response and speed closed-loop stability; the low-speed outer loop runs model predictive control, which uses a discrete event state space model, a prediction step N, and a constraint matrix to predict future tension changes and optimize control inputs. The vision module inputs yarn parallelism, crowding, and width distribution information as exogenous disturbances or constraints to the MPC, thereby achieving predictive control and trajectory correction. The system is equipped with an event-driven switching unit that switches control parameter sets according to the prescription index when receiving process unit state change signals;

[0084] The controller decomposes the control amount calculated by the MPC and the adaptive PID into servo speed, torque instruction, brake torque instruction, and extension and contraction reed, side shift, and press roller position instructions, and sends them to the corresponding actuator units through industrial Ethernet with a determined timing; at the same time, the anti-sticking relaxation subsystem is executed in linkage with the controller according to event triggering, to ensure that physical measures are taken to reduce the risk of sticking when the machine is stopped or the temperature and humidity are abnormal, and the entire execution layer has a fault retreat strategy to ensure a continuous closed-loop action sequence.

[0085] It is apparent for a person skilled in the art that the present application is not limited to the details of the above-described exemplary embodiments, but can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all respects as illustrative and not restrictive, the scope of the present application being defined by the appended claims rather than the above description, and it is intended to embrace all changes and modifications that fall within the meaning and scope of the equivalent elements of the claims. Any reference signs in the claims should not be considered as limiting the claims involved.

Claims

1. An intelligent yarn low tension control system characterized by: The control system comprises a multi-channel tension measurement unit, a visual detection unit, an actuator unit, an edge real-time controller, an upper PLC, a prescription management and human-computer interaction unit; The edge real-time controller is respectively provided with a tension estimation module, a model prediction controller module and an adaptive PID module, the tension estimation module takes the output of a physical tension sensor and the torque and speed signals of a servo driver as input, and provides a tension data signal to the model prediction controller module and the adaptive PID module after filter processing, the visual detection unit and the telescopic reed driving mechanism are coupled through a control signal and a timing synchronization interface, the visual detection unit outputs segmented position information corresponding to the telescopic reed position to the edge real-time controller according to the width section; The multi-channel tension measurement unit is provided with physical contact type tension sensors and at least one non-contact optical tension measurement device at several positions along the yarn conveying path; The multi-channel tension measurement unit comprises a plurality of contact type micro-cantilever strain type tension sensors arranged in an array form along the key nodes of the yarn arrangement, and at least one non-contact laser vibration or photoelectric measurement unit installed according to the width partition and physically connected to the tension estimation module of the edge real-time controller; The edge real-time controller comprises a model prediction controller module, a feedforward compensation module and an adaptive PID module; The model prediction controller module stores control model parameters in the form of a discrete time state space model, a prediction step N and an input-output constraint matrix in the edge real-time controller, and the model prediction controller module and the adaptive PID module share tension estimation values, visual distribution mapping and actuator state information through an internal data bus; at the same time, the edge real-time controller is configured with a control sampling clock source of at least 1 kHz and provides a real-time synchronization signal externally.

2. The intelligent yarn low tension control system according to claim 1, wherein: The visual detection unit is arranged as a camera array along the width direction and is configured with a backlight and an anti-fog device; The visual detection unit is composed of a plurality of high-speed industrial cameras, the frame rate, field of view and installation position of the industrial cameras are distributed according to the width section and the yarn density, the camera array, the backlight unit, the electric heating anti-fog cover and the image preprocessing unit are in the same protective shell, and the segmented yarn parallel and spacing mapping data are transmitted to the edge real-time controller through an image bus.

3. The intelligent yarn low tension control system of claim 1, wherein: The actuator unit is provided with a plurality of groups according to the control group, each group comprises at least one servo yarn feeding driver, at least one brake and clamping device, and a telescopic reed driving mechanism corresponding to the camera array and a set of active pressure rollers; The actuator unit is divided into a control unit according to 16-32 yarns per group, and each control unit comprises a servo driver, a yarn feeding shaft with torque measurement function, an electromagnetic or pneumatic brake and a micro-step or servo side shift driver for telescopic reed side shift; The servo driver is connected to the edge real-time controller through an EtherCAT bus point-to-point connection, and exposes the rotation speed and torque sampling signals.

4. The intelligent yarn low tension control system of claim 1, wherein: The stretch reed driving mechanism comprises a plurality of reed tooth modules with adjustable positions, each reed tooth module is equipped with a micro driver, a position encoder and a mechanical stopper, the reed tooth modules are arranged in several reed sections according to the width, and the control bus of each reed section is connected to the reed position control interface of the edge real-time controller.

5. The intelligent yarn low tension control system of claim 1, wherein: The tension estimation module comprises a virtual tension calculation submodule based on the torque and speed of the servo motor and a Kalman filter submodule for fusing the output of the physical tension sensor and the virtual tension; The virtual tension calculation submodule and the physical sensor are connected to the edge real-time controller through a sampling interface, and the state quantity dimension, the observation matrix and the noise covariance of the Kalman filter submodule are stored as configurable parameters in the edge real-time controller.

6. The intelligent yarn low tension control system of claim 1, wherein: The control system further comprises an event-driven switching unit connected to the liquid inlet pump controller, the spraying controller and the dryer controller of the process unit through an industrial Ethernet, the event-driven switching unit manages several sets of control parameter sets with the prescription identifier as the index, and when receiving the state change signal of the process unit, the corresponding control parameter set is sent to the edge real-time controller according to the index.

7. The intelligent yarn low tension control system of claim 1, wherein: The upper PLC, prescription management and human-computer interaction unit comprises a prescription database, a running history record database and a parameter version management module, the prescription database is associated with the control parameter set, the visual threshold set and the stretch reed position configuration, and the parameter version management module records the timestamp of the prescription change and plays back the historical parameters to the edge real-time controller through the network interface when needed.

8. The intelligent yarn low tension control system of claim 1, wherein: The control system further comprises an anti-sticking relaxation unit composed of a plurality of light vibration actuators, air pulse generators and local temperature and humidity sensors arranged in the drying area and the yarn separating area, the light vibration actuators and the air pulse generators are connected to the edge real-time controller and the event-driven switching unit through a special control bus, and the measurement signals of the local temperature and humidity sensors are recorded in the running history record database.

Images