A sewing thread tension control system and method based on piezoelectric film self-sensing

Abstract

The application discloses a sewing thread tension control system and method based on piezoelectric film self-sensing, and the control system comprises a piezoelectric film self-sensing module, a working condition self-adaptive decision module and a hierarchical collaborative control module. The application improves the synchronism and accuracy of signal acquisition, reduces signal delay and installation error, and improves the overall performance and reliability of the system. The application solves the problem that the traditional discrete adjustment mode is difficult to cope with continuous change complex scenes, greatly improves the adaptation ability and adjustment precision of the system to complex working conditions, reduces the adjustment lag time, improves the response speed and adjustment precision of the system in the extreme high-speed sewing or sudden change of sewing material thickness scene, improves the intelligent level and working condition adaptation ability of the system, expands the adaptability of the system to various special fabrics, and guarantees the stability and reliability of the system under various working conditions.

Description

Technical Field

[0001] This invention relates to the field of sewing equipment, and more specifically to a sewing thread tension control system based on piezoelectric film self-sensing. Background Technology

[0002] A sewing machine is a machine that uses one or more sewing threads to create one or more stitches on fabric, allowing one or more layers of fabric to interweave or sew together. During the sewing process, thread tension is a key parameter affecting the tightness and evenness of the stitches and the appearance quality of the finished product. With the continuous improvement of automation and intelligence in industrial sewing equipment, higher demands are being placed on the precision and flexibility of thread tension control during the sewing process.

[0003] Traditional sewing machines, employing independent tension sensors and actuators, suffer from several problems. First, the traditional design leads to signal delays and installation errors, making it difficult to handle continuously changing and complex scenarios. Furthermore, it is prone to adjustment lag in extreme high-speed sewing or situations involving sudden changes in fabric thickness. Second, existing electronic thread tensioners use electromagnets as the driving source. The electromagnet's stroke is fixed, and the stroke that moves the thread tensioning plate is also fixed. This results in uncontrollable and non-linear thread tension applied to the sewing thread by the electronic thread tensioner, failing to adequately meet diverse sewing needs.

[0004] To address these issues, the industry has been exploring new technological solutions. For example, some research has attempted to use electromagnets as actuators for thread tension adjustment, regulating the clamping force by changing the electromagnet current. However, this approach still has significant drawbacks: electromagnets generate significant heat during prolonged operation, and achieving high-precision thread tension adjustment is difficult, failing to meet the precise thread tension control requirements of high-end sewing processes. Furthermore, some research has attempted to integrate piezoelectric materials as actuators and sensors into a single unit to achieve a multifunctional modular design. However, this approach still faces the problem of electric field interference when multiple piezoelectric materials work together.

[0005] Therefore, there is an urgent need for a new type of sewing thread tension control system that can achieve high-precision control, reduce tension fluctuations under complex working conditions, improve system robustness, reduce adjustment lag, and expand the range of working conditions adaptable.

[0006] Several invention patents have been developed to address issues related to the accuracy of sewing thread tension control, adaptability to working conditions, and adjustment lag.

[0007] For example, CN106637727A discloses an electronically controlled wire tension surface wire end stabilization mechanism, which includes a voltage regulating electrical assembly, a return spring, a braking wire clamp, and a tightening ring assembly. The return spring, braking wire clamp, and tightening ring assembly pass sequentially through the connecting rod of the voltage regulating electrical assembly, and are controlled by a certain control method. However, this patent still has insufficient stability issues in the structural design of the voltage regulating electrical assembly and the braking wire clamp.

[0008] CN112048843A discloses an intelligent thread tension adjustment device and its adjustment method, which includes a thread tension component, a fabric thickness detection component, and a control component. During sewing, it autonomously senses the fabric thickness and automatically and intelligently adjusts the thread tension based on the fabric thickness. However, this patent still has shortcomings in the precise control of thread tension adjustment, making it difficult to cope with the continuously changing requirements under complex working conditions.

[0009] The existing technology has the following drawbacks: 1. Traditional sewing equipment uses independent tension sensors and actuators, which leads to signal delays and installation errors. It is difficult to cope with continuously changing complex scenarios, and it is prone to adjustment lag in extreme high-speed sewing or sudden changes in fabric thickness, affecting sewing quality and efficiency.

[0010] 2. Existing electronic thread tensioners use electromagnets as the driving source. When the electromagnet is activated, its stroke is fixed, and the stroke that pushes the thread tensioning plate is also fixed. This makes the thread tension applied to the sewing thread by the electronic thread tensioner uncontrollable and non-linear, and cannot well meet different sewing needs.

[0011] 3. Traditional adjustment methods suffer from severe electromagnet overheating during prolonged operation and struggle to achieve high-precision thread tension adjustment, failing to meet the requirements for precise thread tension control in high-end sewing processes.

[0012] 4. In existing technologies, multiple piezoelectric materials still face the problem of electric field interference when working together, which affects the overall performance and reliability of the system.

[0013] 5. Existing sewing thread tension control systems suffer from insufficient stability in the structural design of the voltage regulating electrical components and the braking thread clamp, affecting the long-term performance and reliability of the system. Summary of the Invention

[0014] Existing technologies for sewing thread tension control systems suffer from problems such as signal delay, large installation errors, severe adjustment lag, and poor adaptability to various operating conditions. Therefore, to address these issues, this invention provides a sewing thread tension control system based on piezoelectric thin film self-sensing.

[0015] A sewing thread tension control system based on piezoelectric thin film self-sensing, characterized in that it includes: Piezoelectric film self-sensing module: The PVDF piezoelectric film is located on the inner surface of the movable clamping piece, aligned with the suture channel formed by the fixed clamping piece and the movable clamping piece, and uses the positive piezoelectric effect to detect the line tension in real time; Working condition adaptive decision-making module: Based on a deep learning model, it realizes automatic identification of fabric and thread types, and provides initial parameters and working condition weights for the hierarchical collaborative control module; The hierarchical collaborative control module: the upper layer coarsely adjusts the response to slow-changing operating conditions, the lower layer finely adjusts the response to suppress rapid fluctuations, and at the same time, it links with the operating condition adaptive decision-making module to realize dynamic parameter optimization, ensuring the optimal adjustment effect under all operating conditions.

[0016] Preferably, the working condition adaptive decision module has a built-in CNN-LSTM hybrid neural network model, which automatically identifies 8 common fabrics and 5 wire specifications by analyzing the characteristics of piezoelectric signals, and dynamically adjusts PID parameters and mode switching thresholds to achieve intelligent adaptive control.

[0017] Preferably, the hierarchical collaborative control module includes a coarse adjustment submodule and a fine adjustment submodule; The coarse adjustment submodule: The stepper motor is connected to the movable tension plate through a transmission mechanism to respond to changes in the thickness of the sewing material, establish a basic tension adjustment benchmark, and ensure the stability of the initial tension under different thicknesses of sewing material; The fine-tuning module: The electromagnet is connected to the movable tension clamp through a transmission mechanism to suppress instantaneous tension fluctuations during high-speed sewing and improve tension stability under dynamic working conditions.

[0018] Preferably, it also includes a constant current-constant voltage hybrid drive module, which is connected to the fine-tuning sub-module; Start-up phase: constant pressure drive, rise time <5ms, rapid response to tension fluctuations or pre-adjustment commands; Steady-state stage: constant current control, current accuracy ±2%, suppressing the influence of voltage fluctuations on magnetic force, and ensuring the stability of tension adjustment.

[0019] Preferably, it also includes a fuzzy PID module: The fuzzy PID controller calls the fuzzy rule base, based on the tension deviation value e and the rate of change of deviation. Dynamically adjust PID parameter correction amount The coarse adjustment module optimizes the integral coefficient Ki to suppress steady-state error; the fine adjustment module optimizes the differential coefficient Kd. The initial range of PID parameters is superimposed with the fuzzy PID correction amount to ensure that the corrected parameters are still within the effective operating range.

[0020] Preferably, it also includes a working condition prediction module, which includes a sewing trajectory prediction submodule, a coarse adjustment pre-positioning submodule, and a fine adjustment pre-energy storage submodule. Sewing trajectory prediction submodule: By collecting speed and stitch position signals from the sewing machine spindle encoder and combining them with fabric recognition results, it predicts upcoming changes in working conditions and sends pre-adjustment commands to the coarse / fine adjustment submodule 0.1-0.2ms in advance. Coarse adjustment pre-positioning submodule: For scenarios with sudden changes in fabric thickness, the coarse adjustment module detects the seam position in advance using a visual sensor and pre-adjusts the reference displacement in advance to avoid sudden changes in tension caused by untimely adjustment when the seam passes through. Fine-tuning the pre-energy storage submodule: When a high-frequency tension fluctuation scenario is anticipated, the fine-tuning submodule switches to a pre-activated state in advance to maintain low-power standby and shorten the startup response time.

[0021] Preferably, the spindle encoder is installed at the end of the sewing machine spindle and is coaxially connected to the spindle to collect speed and stitch position signals; The vision sensor is installed on the frame beam 5-8cm in front of the presser foot of the sewing machine. The lens is vertically facing the surface of the fabric, and the center of the lens is aligned with the center line of the fabric feed. It is used to detect the position of the fabric seam in advance and provide a trigger signal for the pre-positioning of the coarse adjustment module.

[0022] Preferably, it also includes a multimodal collaborative weighting submodule: dynamically calculating the collaborative weighting factor by real-time detection of the frequency and amplitude of the tension signal. f is the frequency of tension change, and A is the amplitude of tension deviation. , The normalized frequency of tension variation. The normalized deviation magnitude, where α and β are weighting coefficients and ; Default α=0.6, β=0.4, adaptive adjustment based on operating conditions; When ω approaches 1, the coarse adjustment submodule dominates the adjustment, and the fine adjustment submodule assists in the correction; When ω approaches 0, the fine adjustment submodule dominates the high-value compensation, and the coarse adjustment submodule maintains the baseline.

[0023] A control method for a sewing thread tension control system based on piezoelectric thin film self-sensing, characterized by the following steps: Before sewing starts or when the fabric thickness changes, the stepper motor moves first, adjusting the reference displacement of the thread clamping plate according to the fabric thickness to establish the initial tension for the corresponding working condition. During high-speed sewing, the piezoelectric film detects the thread tension signal in real time. If a momentary fluctuation occurs, the electromagnet responds immediately and compensates for the tension deviation through high-frequency adjustment. The working condition adaptive decision module dynamically adjusts the switching threshold and initial PID parameters of the coarse adjustment submodule and the fine adjustment submodule based on the fabric / thread type identified by the CNN-LSTM model, to ensure basic collaborative effect under different working conditions. The stepper motor in the layered collaborative control module is responsible for coarse adjustment of the reference displacement, responding to changes in the fabric thickness; the electromagnet is responsible for fine adjustment of dynamic compensation, suppressing tension fluctuations during high-speed sewing.

[0024] Compared with the prior art, the present invention provides a sewing thread tension control system based on piezoelectric thin film self-sensing, which has the following beneficial effects: 1. By integrating a PVDF piezoelectric film with specific parameters onto the surface of the tension plate, the traditional independent tension sensor is replaced, realizing the integration of "detection-execution" functions. This effectively reduces the complexity of the system structure, simplifies the system installation process, improves the synchronization and accuracy of signal acquisition, significantly reduces signal delay and installation errors, and improves the overall performance and reliability of the system. 2. The original architecture is upgraded by adopting fuzzy PID control to form a collaborative control system of "original architecture foundation + fuzzy PID optimization", which realizes continuous dynamic optimization of parameters, effectively solves the problem that traditional discrete adjustment mode is difficult to cope with continuously changing complex scenarios, and greatly improves the system's adaptability to complex working conditions and adjustment accuracy. 3. By introducing a "condition pre-judgment mechanism" and a multimodal collaborative weighting mechanism, combined with data acquisition from the sewing machine spindle encoder and vision sensor, accurate prediction of the seam position of the sewing material and real-time analysis of the condition characteristics are achieved, effectively reducing the adjustment lag time and significantly improving the system's response speed and adjustment accuracy in extreme high-speed sewing or sudden changes in sewing material thickness. 4. The working condition recognition technology based on the CNN-LSTM hybrid model, combined with the dynamic parameter optimization of fuzzy PID, forms a complete adaptive system of "working condition recognition - initial matching - real-time correction", which greatly improves the intelligence level and working condition adaptability of the system and expands the system's adaptability to various special fabrics (such as elastic fabrics and carbon fiber threads). 5. By using a reasonably designed fuzzy rule base and parameter connection rules, the effective superposition of fuzzy PID and original PID parameters is ensured, avoiding adjustment failure caused by parameter overflow, and guaranteeing the stability and reliability of the system under various operating conditions. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the installation of the spindle encoder and vision sensor.

[0026] Figure 2 This is a flowchart of a sewing thread tension control method based on piezoelectric thin film self-sensing according to the present invention.

[0027] Figure 3 This is a schematic diagram of a sewing thread tension control method based on piezoelectric thin film self-sensing according to the present invention. Detailed Implementation

[0028] 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.

[0029] Example 1 This invention employs integrated piezoelectric thin film sensing and actuation technology. By integrating a PVDF piezoelectric thin film with specific parameters onto the surface of the wire clamp, it replaces the traditional independent tension sensor, achieving integrated "detection-actuation" functions. The PVDF piezoelectric thin film, with sensing as its core function, captures the wire tension signal in real time through the positive piezoelectric effect. Simultaneously, it forms an integrated layout with the wire clamp, stepper motor, electromagnet, and other actuators, simplifying the system structure and eliminating installation errors and signal delays between independent sensors and actuators, thus achieving high-precision control.

[0030] Piezoelectric film type: PVDF piezoelectric film Film thickness: 50μm Curie temperature: 135°C (suitable for normal operating temperature range of sewing equipment) Sensing performance: Real-time detection of thread tension is achieved by utilizing the positive piezoelectric effect, with a sensitivity of 8.2mV / cN and a response time of <2ms, ensuring real-time capture of tension signals in high-speed sewing scenarios.

[0031] As a sensing component integrating sensing and actuation, the mounting structure of PVDF piezoelectric film directly determines the tension detection accuracy, as detailed below: Installation location: Install on the inner surface of the movable clamping piece (the side in contact with the sewing thread). The installation area should be precisely aligned with the "sewing thread channel" formed by the fixed clamping piece and the movable clamping piece to ensure that the sewing thread fits tightly against the film when passing through, so as to achieve distortion-free transmission of tension signals. A 3mm safety distance should be reserved between the edge of the film and the edge of the clamping piece to avoid wear and detachment of the film caused by installation or clamping actions.

[0032] The wire clamp assembly is existing technology and will not be described in detail here.

[0033] Fixing method: Epoxy resin adhesive is used for bonding. The thickness of the adhesive layer is strictly controlled to ≤5μm. After bonding, visual inspection is required to confirm that there are no bubbles or wrinkles, ensuring a rigid connection between the film and the clamping piece without relative displacement. The inner surface of the movable clamping piece needs to be precisely processed, with a flatness of ≤0.01mm / 20mm and a roughness Ra≤0.8μm, to improve the bonding firmness and eliminate the influence of surface gaps on signal transmission.

[0034] Signal lead-out structure: The two-pole leads of the thin film (0.8mm in diameter) are led out along the cable avoidance groove (1mm wide and 2mm deep) opened on the side of the movable clamping plate, and connected to the main control module through a waterproof connector. The lead fixing point is 5mm away from the edge of the film to avoid poor contact caused by pulling, and to ensure that the signal transmission delay is <2ms.

[0035] Structural adaptation requirements: After integration, the surface of the film should be flush with the inner surface of the movable clamping piece, with an error of ≤0.02mm; the total thickness of the movable clamping piece should be ≤5.1mm (including 50μm of film + 5μm of adhesive layer), without affecting the clamping piece spacing adjustment range (0-5mm).

[0036] The integrated piezoelectric film self-sensing solution integrates a 50μm thick PVDF piezoelectric film (Curie temperature 135°C) on the surface of the thread tensioning piece, utilizing the positive piezoelectric effect to detect thread tension in real time. When the sewing thread applies pressure to the thread tensioning piece, the PVDF film generates an electric charge signal (sensitivity 8.2mV / cN) with a response time of <2ms, achieving integrated execution and detection and eliminating the need for a separate sensor.

[0037] Adaptive Decision-Making Module: Adaptive Parameter Adjustment The system has a built-in adaptive decision-making module for working conditions, which uses a deep learning model to automatically identify the type of fabric and thread, thereby providing initial parameters and working condition weights for the hierarchical collaborative control module and improving the system's adaptability to complex working conditions.

[0038] Recognition Model: Convolutional Neural Network (CNN-LSTM Hybrid Model) – Combining the feature extraction capabilities of CNN with the temporal dependency modeling advantages of LSTM to improve recognition accuracy. Specifically, CNN is responsible for extracting spatial features from piezoelectric signals (such as signal amplitude changes, high-frequency fluctuation details, etc.) and filtering redundant noise; LSTM is responsible for capturing the temporal dependencies of the signal, mining the correlation patterns of tension changes at different sewing stages, and adapting to continuous temporal tension signal analysis scenarios. This hybrid model architecture enables the system to accurately identify the signal feature differences corresponding to different fabrics and threads, providing a reliable basis for subsequent parameter optimization. Model performance: Inference time <15ms (ensuring real-time parameter adjustment, not lagging behind changes in sewing conditions); Identification scope: 8 common fabrics (cotton, polyester, nylon, etc.) and 5 yarn specifications; Parameter output capability: Output fabric / thread type identifier as "initial working condition weight", and simultaneously output initial PID parameters (proportional coefficient Kp range 0.8-2.5, integral coefficient Ki 0.05-0.3, derivative coefficient Kd 0.1-0.8) and module switching threshold benchmark value, providing basic data support for fuzzy PID optimization and hierarchical collaboration.

[0039] The adaptive decision-making module has a built-in CNN-LSTM hybrid neural network model (inference time <15ms). By analyzing the characteristics of piezoelectric signals, it automatically identifies 8 common fabrics (cotton, polyester, nylon, etc.) and 5 types of wire specifications, and dynamically adjusts the PID parameters (Kp range 0.8-2.5) and mode switching threshold to achieve intelligent adaptive control.

[0040] Hierarchical collaborative control module: coarse and fine dual-modal mechanism To adapt to the tension adjustment needs under different sewing conditions, the system adopts a hierarchical collaborative control architecture. Through the coordinated operation of stepper motors and electromagnets, it achieves coarse adjustment of the reference displacement and fine adjustment of dynamic tension, forming a "coarse and fine" adjustment system that effectively balances the adjustment range and accuracy. The core logic of this architecture is "first establish a reference, then dynamically correct," breaking down the tension adjustment task into layers according to the characteristics of the working conditions. The upper layer coarse adjustment responds to slowly changing working conditions, while the lower layer fine adjustment suppresses rapidly changing fluctuations. At the same time, it links with an intelligent decision-making module to achieve dynamic parameter optimization, ensuring optimal adjustment results under all working conditions.

[0041] Coarse adjustment module: Stepper motor reference displacement adjustment In response to changes in fabric thickness, a basic tension adjustment benchmark is established to ensure stable initial tension under different fabric thicknesses.

[0042] 42 stepper motors approach the wire clamping plate assembly and form a transmission linkage with the wire clamping plate.

[0043] Motor model: 42 stepper motor (can be replaced with a closed-loop stepper motor) Step angle: 0.9° (to ensure fine-grained displacement adjustment) Adjustment range: 0-5mm (covering the range of fabric thickness variations in mainstream sewing scenarios) Positioning accuracy: ±10μm (original configuration); upgraded ±5μm (adds position feedback function to reduce coarse adjustment reference deviation). In conjunction with the core function of the stepper motor in this solution—"reference displacement adjustment" (adapting to changes in fabric thickness and establishing initial tension)—the connection between the stepper motor and the thread clamp must meet the requirements of precise displacement transmission, smooth and seamless adjustment, and a compact structure that fits within the sewing machine's space. The specific connection scheme is as follows (including mechanical structure, transmission method, and key parameters): The stepper motor converts angular displacement into linear displacement of the clamping plate through a precision transmission mechanism, thereby achieving fine adjustment of the clamping gap (0-5mm adjustment range). The core logic is as follows: the stepper motor receives a control signal and rotates → the transmission mechanism converts the rotational motion into linear motion → pushes the clamping plate seat to move along the guide structure → changes the gap between the clamping plate and the fixed clamping plate → adjusts the initial tension.

[0044] Screw and nut drive + guide shaft structure 1. Mechanical structure composition Motor and lead screw connection: A flexible coupling (outer diameter 8mm, length 12mm) is used to compensate for installation coaxiality deviation (≤0.1mm), avoid transmission jamming, and ensure smooth rotation of the stepper motor; The clamping plate seat and nut seat are rigidly fixed by M3 screws, and the bottom is equipped with a guide bushing (made of brass and with self-lubrication). The clearance between the guide bushing and the guide shaft is ≤0.02mm to ensure the accuracy of linear motion. Installation of movable wire clamping plates: Fix them to the front end of the wire clamping plate seat with countersunk screws, ensuring that the parallelism deviation between the movable clamping plates and the fixed clamping plates is ≤0.03mm, and ensuring that the wire clamping gap is uniform (to avoid uneven load during tension adjustment). Overall fixation: The fixing bracket is connected to the sewing machine frame / beam by expansion screws or clips, and the bottom is reserved with adjustment holes to finely adjust the level of each transmission mechanism (to adapt to different equipment installation standards).

[0045] With a ball screw lead of 1mm and a stepper motor step angle of 0.9°, a minimum displacement increment of ≈2.78μm can be achieved (meeting the positioning accuracy requirement of ±5μm). The overall transmission mechanism has a width of ≤30mm and a length of ≤50mm, making it suitable for the limited installation space inside a sewing machine. The dual guide shaft design prevents the clamping plate seat from deflecting, and the flexible coupling absorbs installation errors, reducing wear during long-term operation. The bracket adopts a modular design, which can be adapted to the frame structure of different models of industrial sewing machines without the need for major equipment modifications.

[0046] As the core actuator of the coarse adjustment module, the stepper motor's installation structure must ensure both displacement adjustment accuracy and transmission stability, as detailed below: Installation location: Place it close to the thread clamping plate assembly, with the installation center 40mm from the reference surface of the thread clamping plate assembly. It is rigidly fixed to the sewing machine frame by an L-shaped bracket to avoid the vibration of the motor affecting the detection and adjustment accuracy. The installation direction must ensure that the motor output shaft is parallel to the direction of the thread clamping plate spacing adjustment to reduce transmission error.

[0047] Connection structure: The motor output shaft (shaft diameter 5mm) is connected to the miniature ball screw through a precision flexible coupling. The coupling is 15mm long and is used to compensate for the coaxiality deviation between the motor and the screw (maximum compensation amount 0.1mm). The ball screw has a lead of 1mm, a total length of 60mm, and an outer diameter of 8mm. Its nut and sliding seat are integrated. The sliding seat and the movable clamping plate are rigidly connected by M3 hexagon socket bolts (bolt spacing 12mm), forming a complete transmission chain of "motor-coupling-screw-sliding seat-clamping plate".

[0048] Guiding and limiting structure: The sliding seat is equipped with a 50mm long linear guide rail with a 20mm spacing between the rails and a parallelism tolerance of ≤0.02mm / 100mm with the lead screw, which limits the sliding seat to move only in the tension adjustment direction; limit switches are installed at both ends of the linear guide rail, 5mm away from the extreme position of the sliding seat, and the switch trigger stroke is 0.5mm to prevent the motor from running over the range and damaging the components.

[0049] Precision assurance structure: Equipped with a position feedback sensor to form a closed-loop control with the motor. The sensor signal is linked with the main control module through a high-speed interface to ensure positioning accuracy of ±5μm; the coaxiality tolerance of the motor mounting hole is φ0.02mm, and the bolt preload torque is 5±0.5N·m to prevent loosening during high-speed adjustment.

[0050] Fine-tuning module: Electromagnetic dynamic tension compensation It suppresses instantaneous tension fluctuations during high-speed sewing and improves tension stability under dynamic working conditions.

[0051] Electromagnet rated power: 12W Magnetic adjustment range: 0.5-8N (adapts to the tension requirements of different fabrics and threads) Adjustable frequency: 0-200Hz (can quickly respond to high-frequency tension fluctuations during high-speed sewing) Suitable sewing speed: up to 5000 stitches / minute Response time: <8ms The electromagnet converts electromagnetic attraction into minute linear displacement of the clamping plate through an elastic transmission component, enabling dynamic fine-tuning of the clamping gap (adjustment stroke 0.1-1mm). The core logic is: the electromagnet receives a dynamic adjustment signal → generates controllable electromagnetic attraction → pushes the movable clamping plate to a minute displacement through the transmission component → fine-tunes the clamping pressure and tension → suppresses instantaneous tension fluctuations; at the same time, the elastic transmission structure is compatible with the reference displacement adjustment driven by a stepper motor and does not interfere with the normal operation of the coarse adjustment mechanism.

[0052] Elastic push rod + limiting guide structure (1) Mechanical structure composition Electromagnet and push rod connection: One end of the elastic push rod is fixed to the electromagnet armature by a thread, and the other end is in contact with the clamping plate connecting seat by a spherical contact, which reduces the lateral force caused by installation deviation and ensures that the thrust is transmitted along the clamping line direction; the push rod is nested with a compression spring, which keeps the push rod and the connecting seat slightly in contact in the natural state, avoiding adjustment lag caused by gap.

[0053] Compatible with stepper motor structure: The electromagnet mounting base and the stepper motor mounting bracket are integrated into the design, and the movable clamping plate is nested in the clamping plate seat driven by the stepper motor, forming a dual adaptation structure of "coarse adjustment bearing + fine adjustment drive"; when the stepper motor drives the clamping plate seat to move as a whole, the elastic push rod can move synchronously with the clamping plate, and the spring extension and contraction compensate for the displacement difference without interfering with the coarse adjustment action.

[0054] Guiding and limiting protection: The guide sleeve of the push rod is rigidly connected to the electromagnet fixing seat to ensure that the coaxiality of the push rod movement trajectory and the clamping direction is ≤0.02mm; a limiting boss is set at the clamping plate connecting seat to limit the maximum fine adjustment stroke of the movable clamping plate (≤1mm) to avoid damage to the wire due to excessive adjustment.

[0055] Rapid response: The elastic push rod and small gap guide design, combined with the electromagnet's response speed of <5ms, can quickly transmit adjustment force and accurately suppress high-frequency tension fluctuations; Good synergy: Modular integrated design and flexible transmission structure enable interference-free adaptation with the coarse adjustment mechanism of the stepper motor, and smooth connection between coarse and fine adjustment actions; Uniform stress distribution: Spherical contact + coaxial guide design avoids uneven stress on the wire clamping pieces, ensuring uniform wire clamping gap and stable tension adjustment; Compact structure: The overall component width is ≤25mm. After integration with the stepper motor transmission mechanism, it does not occupy too much additional equipment space and is suitable for the internal layout of sewing machines.

[0056] The above connection scheme enables precise and rapid transmission between the electromagnet and the clamping plate assembly, while ensuring coordinated adaptation with the coarse adjustment mechanism of the stepper motor, providing reliable mechanical structural support for the layered coordinated control of "coarse adjustment + fine adjustment".

[0057] As the core actuator of the fine-tuning module, the electromagnet's installation structure must balance magnetic force transmission efficiency, anti-interference, and space compatibility. Installation position: Located within 15mm directly below or to the side of the thread clamping plate assembly, fixed to the pre-set mounting holes (8mm diameter) on the sewing machine frame using an L-shaped metal bracket (Q235 steel, 3mm thick); the coaxiality between the center of the electromagnet core and the center of the thread clamping plate's sewing channel is ≤0.5mm, and the direction of magnetic force is consistent with the direction of thread tension, ensuring that the magnetic force directly acts on the tension adjustment area; the minimum distance between the core and the movable thread clamping plate is controlled at 3-5mm, compatible with the 0-5mm adjustment range of the thread clamping plate, avoiding mechanical interference.

[0058] Fixing and positioning structure: The bracket has an elongated hole (8mm in diameter, with an adjustment margin of ±2mm) to facilitate fine-tuning of the position during installation; it is calibrated with a 4mm diameter positioning pin with a tolerance of H7 to ensure installation accuracy; the bracket is fixed with M5 self-tapping screws (with external toothed washers) with a pre-tightening torque of 3±0.5N·m to improve vibration resistance.

[0059] Protection and anti-interference structure: The electromagnet is wrapped with a 0.2mm thick copper foil shielding layer and grounded at one end (grounding resistance ≤1Ω) to reduce the interference of the magnetic field on precision components such as PVDF piezoelectric film and spindle encoder; it is equipped with an IP54 grade ABS protective shell to cover the coil and terminal block. The cable outlet hole of the shell is 5mm in diameter and equipped with a waterproof sealing ring to prevent wire chips and oil stains from affecting heat dissipation and insulation performance.

[0060] Wiring structure: The coil leads (4 wires, 0.25mm diameter enameled copper wire, insulation class 180) are arranged according to color coding (red - positive, black - negative, yellow - current sampling +, blue - current sampling -), and are led out through the cable outlet of the protective shell and connected to the nearest terminal block (≤30cm from the electromagnet). The lead length is ≤30cm and the wire diameter is ≥0.5mm² to reduce line loss.

[0061] The tension control mode using stepper motors and electromagnets is existing technology and will not be described in detail here.

[0062] Based on the working characteristics of electromagnets, a constant current-constant voltage hybrid drive strategy is adopted to balance response speed and steady-state control accuracy, ensuring stable and reliable magnetic force output and providing a guarantee for high-frequency precise adjustment of the fine-tuning module.

[0063] Drive logic and parameters Start-up phase: Constant voltage drive (supply voltage 24V) – enables rapid start-up of the electromagnet with a rise time of <5ms (after optimization), and quick response to tension fluctuations or pre-adjustment commands; Steady-state stage: Constant current control—current accuracy ±2%, effectively suppressing the influence of voltage fluctuations on magnetic force and ensuring the stability of tension adjustment; Switching mechanism: The drive mode is adaptively switched through the main control module, with a switching transition time of <0.5ms, no impact interference, and avoids tension fluctuations during the switching process.

[0064] During the start-up phase, the electromagnet uses constant voltage drive (24V) to achieve a fast response. During the steady-state phase, it switches to constant current control (current accuracy ±2%) to ensure magnetic stability, thus optimizing dynamic response and steady-state accuracy.

[0065] Workflow: Before sewing starts or when the fabric thickness changes, the stepper motor moves first, adjusting the reference displacement of the thread clamping plate according to the fabric thickness to establish the initial tension for the corresponding working condition. During high-speed sewing, the piezoelectric film detects the thread tension signal in real time. If there is a momentary fluctuation (such as due to stitch switching or changes in fabric texture), the electromagnet responds immediately and compensates for the tension deviation through high-frequency adjustment. The intelligent decision-making module dynamically adjusts the switching threshold and initial PID parameters (Kp0.8-2.5) of the two-level modules based on the fabric / thread type identified by the CNN-LSTM model, ensuring basic collaborative performance under different working conditions.

[0066] The stepper motor in the hierarchical collaborative control architecture is responsible for coarse adjustment of the reference displacement (adjustment range 0-5mm, positioning accuracy ±10μm), responding to changes in fabric thickness; the electromagnet (rated power 12W, magnetic force range 0.5-8N) is responsible for fine adjustment of dynamic compensation (adjustment frequency 0-200Hz), suppressing tension fluctuations during high-speed sewing (up to 5000 stitches / minute). Together, they form a hierarchical collaborative mechanism combining coarse and fine adjustments, resolving the issue of functional redundancy.

[0067] Example 2 The hierarchical collaborative architecture in Example 1 has achieved basic "coarse-fine combined" tension adjustment through a dual-modal mechanism of "stepper motor coarse adjustment + electromagnet fine adjustment." However, considering the needs of complex industrial sewing conditions (such as multi-material mixed fabrics, extreme high-speed sewing, and sudden changes in fabric thickness), improvements are made in three dimensions: deepening adjustment accuracy, enhancing adaptability to working conditions, and improving collaborative efficiency. This enhances system robustness, reduces adjustment lag, and expands the range of working conditions covered. This example features deep collaborative linkage in the following aspects: Fuzzy PID-enabled hierarchical adjustment parameters The parameter adjustment mode of the hierarchical collaborative architecture in Example 1 is a discrete mode of "condition identification - parameter matching". Although it can adapt to the conditions of conventional fabrics and yarns, it is difficult to cope with continuously changing complex scenarios (such as gradual changes in fabric texture and small fluctuations in yarn thickness). Fuzzy PID control is introduced to upgrade the original architecture, forming a collaborative control system of "original architecture foundation + fuzzy PID optimization", which realizes continuous dynamic optimization of parameters and enhances the collaborative accuracy of coarse and fine adjustment modules.

[0068] Implementation details: Constructing a fuzzy rule base: Based on historical sewing data, establish a rule base with "tension deviation value (e), deviation change rate" as the basis for the rule base. "Input" refers to the PID parameter correction amount. "The output is a fuzzy rule library that covers different combinations of fabrics, threads, and sewing speeds." Independent optimization of hierarchical parameters: For the characteristics of the coarse adjustment module (stepper motor) and the fine adjustment module (electromagnet), a dedicated fuzzy PID controller is designed respectively. The coarse adjustment module focuses on "steady-state accuracy" and optimizes the integral coefficient Ki to suppress steady-state error. The fine adjustment module focuses on "dynamic response" and optimizes the derivative coefficient Kd to reduce overshoot. Linked with the intelligent decision-making module: The fabric / thread type identified by the CNN-LSTM model is used as the "initial weight of working condition" for the fuzzy PID, which quickly narrows the parameter adjustment range. Then, the parameters are dynamically corrected by the deviation and rate of change of the real-time tension signal, so as to achieve the dual guarantee of "first coarse matching (original architecture PID parameters) and then fine optimization (fuzzy PID correction)". Parameter connection rules: Clearly define the superposition logic of the initial range of the original PID parameters and the fuzzy PID correction, and set... , , To ensure that the corrected parameters remain within the effective operating range (e.g., Kp remains at 0.8-2.5, Ki at 0.05-0.3, and Kd at 0.1-0.8), avoid parameter overflow that could lead to adjustment failure.

[0069] This solution upgrades parameter adjustment from "discrete adaptation" to "continuous optimization," reducing the tension fluctuation amplitude under complex working conditions. It is expected to improve the tension control accuracy from ±3% to within ±2%.

[0070] Add a pre-judgment mechanism to reduce the lag in dynamic adjustment. The collaborative process in Example 1 is a passive response mode of "tension fluctuation occurrence - piezoelectric film detection - adjustment module response". In extreme high-speed sewing (such as above 5000 stitches / minute) or sudden changes in fabric thickness, adjustment lag is prone to occur, resulting in instantaneous tension exceeding the tolerance. This example adds a "condition pre-judgment mechanism" to upgrade from "passive response" to "active prediction", initiating adjustment actions in advance.

[0071] like Figure 1 As shown, this embodiment includes a sewing machine spindle encoder 2: installed at the end 1 of the sewing machine spindle, fixed by a flange, and coaxially connected to the spindle to ensure accurate acquisition of speed and stitch position signals, with installation deviation controlled within 0.02mm; used to acquire speed and stitch position signals, providing data support for sewing trajectory prediction; Vision sensor 3: Installed on the frame beam 5-8cm in front of the presser foot of the sewing machine, with the lens vertically facing the fabric surface and the center of the lens aligned with the center line of the fabric feed, ensuring that features such as fabric seams and thickness changes can be clearly captured; used to detect the position of fabric seams in advance and provide a trigger signal for the pre-positioning of the coarse adjustment module; the area in front of the presser foot is the feeding direction, and the fabric enters the presser foot position after passing through the vision sensor.

[0072] The aforementioned sensors are linked to the system's main control module via a high-speed interface, with a data transmission delay of <0.05ms, ensuring the real-time performance of the predicted signals.

[0073] Installation and positioning accuracy: The main shaft encoder must be coaxially connected to the sewing machine main shaft, with a flange positioning and fixing deviation of ≤0.02mm and no transmission backlash, to avoid distortion of speed and stitch position signals; the vision sensor lens is vertically oriented towards the fabric surface, with its center aligned with the fabric conveying center line, and the installation height adapted to the needle plate size to ensure complete coverage of the conveying path and clear focus.

[0074] Environmental protection adaptation: Equipped with a dedicated protective housing, the spindle encoder adopts a sealed flange, and the vision sensor is equipped with a dustproof transparent cover to prevent the intrusion of lint, fibers, and oil mist; the protective housing is reserved with heat dissipation holes to avoid the sensor from overheating during long-term operation.

[0075] Electromagnetic interference resistant design: The installation location is far away from strong interference sources such as electromagnets and stepper motors (spacing ≥10cm); shielded cables are used to transmit signals, and the sensor housing is grounded to ensure stable data transmission delay <0.05ms.

[0076] Wiring and interference avoidance: Wear-resistant and bend-resistant industrial-grade wires are used and fixedly arranged along the frame, with allowance for movement of moving parts; the vision sensor avoids the sewing material conveying channel, presser foot trajectory and needle and thread movement range, and the spindle encoder does not affect the spindle bearing rotation and the original transmission structure.

[0077] Specific implementation details are as follows: Figure 2 , 3 As shown: Sewing trajectory prediction: By collecting speed and stitch position signals from the sewing machine spindle encoder and combining them with fabric recognition results, the system predicts upcoming changes in working conditions (such as stitch switching and fabric seam position) and sends a "pre-adjustment command" to the coarse / fine adjustment module 0.1-0.2ms in advance. Coarse adjustment module pre-positioning: For scenarios with sudden changes in fabric thickness, the seam position is detected in advance by a visual sensor, and the coarse adjustment module (stepper motor) completes the pre-adjustment of the reference displacement in advance to avoid sudden changes in tension due to untimely adjustment when the seam passes through. Fine-tuning module pre-energy storage: When a high-frequency tension fluctuation scenario is anticipated (such as high-speed sewing of synthetic fiber fabrics), the fine-tuning module (electromagnet) switches to the "pre-activated state" in advance to maintain low-power standby and shorten the startup response time (from the current <8ms to <5ms).

[0078] This embodiment reduces the adjustment lag time by more than 30%, effectively suppresses instantaneous tension deviation under extreme working conditions, and expands the upper limit of the sewing speed that the system can adapt to (covering high-speed scenarios of more than 6,000 stitches / minute).

[0079] Construct a multimodal collaborative weighting mechanism to improve adaptability to complex working conditions. In the architecture of Example 1, the division of labor between the coarse adjustment and fine adjustment modules is fixed (coarse adjustment is responsible for slow changes, and fine adjustment is responsible for fast changes). However, under mixed working conditions (such as alternating thick and thin fabrics + high-speed sewing), the fixed division of labor can easily lead to redundancy or insufficiency in adjustment. A "multimodal collaborative weighting mechanism" is constructed to dynamically allocate the adjustment weights of the two-level modules, so as to achieve adaptive collaboration under working conditions.

[0080] Define the collaborative weighting factor: Introduce two core indicators, "tension change frequency (f)" and "tension deviation amplitude (A)," to construct the collaborative weighting factor. The calculation logic is clearly defined as follows: (in The normalized frequency of tension variation. The normalized deviation magnitude, where α and β are weighting coefficients and (Default α=0.6, β=0.4, can be adaptively adjusted based on operating conditions) — When ω approaches 1, the coarse adjustment module dominates the adjustment; when ω approaches 0, the fine adjustment module dominates the adjustment. Dynamic weight allocation logic: By detecting the frequency and amplitude of the tension signal in real time, the ω value is dynamically calculated—such as in the case of thick material joints. ω=0.8, the coarse adjustment module mainly adjusts the baseline, and the fine adjustment module assists in correction; such as in high-speed scenarios with thin materials. ω=0.2, the fine-tuning module dominates high-frequency compensation, while the coarse-tuning module maintains the baseline stability; Weight self-learning optimization: Based on reinforcement learning algorithms, the weight allocation model is trained through historical operating data, enabling the system to gradually master the optimal weight ratio under different operating conditions and improve the stability of long-term operation.

[0081] This embodiment breaks the fixed division of labor boundaries and realizes the "flexible collaboration" of the two-level adjustment modules, which significantly improves the tension stability under mixed working conditions and makes the system adaptable to the sewing needs of more special fabrics (such as elastic fabrics and carbon fiber threads).

[0082] The system performance indicators using the technical solution of Example 2 are as follows: System power consumption: <18W (low power design, suitable for energy-saving requirements of industrial equipment). Tension control accuracy: within ±2% (meets the requirements of high-precision sewing processes for tension stability); Suitable sewing speed: 0-6000 stitches / minute (covering both regular and extremely high-speed sewing scenarios); Adjusting lag time: more than 30% shorter than existing solutions; Working conditions adaptability: 8 types of common fabrics, 5 types of yarn specifications, compatible with special working conditions such as mixed fabrics, elastic fabrics, and carbon fiber yarns; Cost control: The cost increase compared to the traditional solution is no more than 25%, and the overall cost after hardware upgrade is still controlled within 35% of the traditional solution (balancing performance improvement and commercial feasibility).

[0083] Innovation of this invention: Integrated design: The "sensing-execution integrated layout" is achieved through PVDF piezoelectric film. With the high-precision sensing function of PVDF as the core, it is integrated with the clamping plate and the actuator to replace the traditional independent sensor. This simplifies the structure, improves signal synchronization, and reduces system error. Compared with the traditional separate design, the signal delay is reduced by more than 80%. Dual-modal collaborative optimization mechanism: Based on the "stepper motor coarse adjustment + electromagnet fine adjustment", three core functions are added: continuous optimization of fuzzy PID parameters, pre-judgment of working conditions, and dynamic weight allocation. The key logics such as parameter connection and weight calculation are clarified, and the adjustment range, dynamic accuracy and adaptability to complex working conditions are taken into account to achieve "flexible collaboration" in all scenarios. Intelligent adaptive capability: Based on the CNN-LSTM hybrid model for work condition recognition, combined with the dynamic parameter optimization of fuzzy PID, a full-process adaptive system of "work condition recognition - initial matching - real-time correction" is formed, which improves the system's versatility and robustness. In summary, this solution, through the design of "self-sensing integration + hierarchical collaborative closed-loop control + hardware and software collaborative optimization", improves the accuracy of line tension control and expands the range of working conditions adaptable, while taking into account the characteristics of low power consumption and low cost, and is suitable for the upgrading and transformation of various industrial sewing equipment.

[0084] Example 3 Based on Example 1, the CNN-LSTM hybrid model in the adaptive decision-making module for specific working conditions adopts the following configuration: the CNN uses 5 convolutional layers with a kernel size of 5x5 and a stride of 1, employing the ReLU activation function; the LSTM uses 2 LSTM layers with 128 hidden units, employing the sigmoid activation function; the total number of model parameters is 1.2 million. The fuzzy rule base contains 20 rules, each describing working conditions for different combinations of fabrics and threads. In the parameter connection rules, the initial range of Kp is 0.8-2.5, the initial range of Ki is 0.05-0.3, and the initial range of Kd is 0.1-0.8.

[0085] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A sewing thread tension control system based on piezoelectric thin film self-sensing, characterized in that: include: Piezoelectric film self-sensing module: The PVDF piezoelectric film is located on the inner surface of the movable clamping piece, aligned with the suture channel formed by the fixed clamping piece and the movable clamping piece, and uses the positive piezoelectric effect to detect the line tension in real time; Working condition adaptive decision-making module: Based on a deep learning model, it realizes automatic identification of fabric and thread types, and provides initial parameters and working condition weights for the hierarchical collaborative control module; The hierarchical collaborative control module: the upper layer coarsely adjusts the response to slow-changing operating conditions, the lower layer finely adjusts the response to suppress rapid fluctuations, and at the same time, it links with the operating condition adaptive decision-making module to realize dynamic parameter optimization, ensuring the optimal adjustment effect under all operating conditions.

2. The sewing thread tension control system based on piezoelectric thin film self-sensing according to claim 1, characterized in that: The adaptive decision-making module incorporates a CNN-LSTM hybrid neural network model. By analyzing the characteristics of piezoelectric signals, it automatically identifies eight common fabrics and five wire specifications, dynamically adjusts PID parameters and mode switching thresholds, and achieves intelligent adaptive control.

3. The sewing thread tension control system based on piezoelectric thin film self-sensing according to claim 2, characterized in that: The hierarchical collaborative control module includes a coarse adjustment submodule and a fine adjustment submodule; The coarse adjustment submodule: The stepper motor is connected to the movable clamping plate through the transmission mechanism to respond to changes in the thickness of the sewing material, establish a basic tension adjustment benchmark, and ensure the stability of the initial tension under different thicknesses of sewing material; The fine adjustment module: The electromagnet is connected to the movable tension clamp through a transmission mechanism to suppress instantaneous tension fluctuations during high-speed sewing and improve tension stability under dynamic working conditions.

4. The sewing thread tension control system based on piezoelectric thin film self-sensing according to claim 3, characterized in that: It also includes a constant current-constant voltage hybrid drive module, which is connected to the fine-tuning sub-module; Start-up phase: constant pressure drive, rise time <5ms, rapid response to tension fluctuations or pre-adjustment commands; Steady-state stage: constant current control, current accuracy ±2%, suppressing the influence of voltage fluctuations on magnetic force, and ensuring the stability of tension adjustment.

5. The sewing thread tension control system based on piezoelectric thin film self-sensing according to claim 3, characterized in that: It also includes a fuzzy PID module: Fuzzy PID controller: used to call the fuzzy rule base, based on the tension deviation value e and the rate of change of deviation. Dynamically adjust PID parameter correction amount The coarse-tuning submodule optimizes the integral coefficient Ki to suppress steady-state error. The fine-tuning submodule optimizes the differential coefficient Kd; the initial range of PID parameters is superimposed with the fuzzy PID correction amount to ensure that the corrected parameters are still within the effective working range.

6. The sewing thread tension control system based on piezoelectric thin film self-sensing according to claim 3, characterized in that: It also includes a working condition prediction module, which comprises a sewing trajectory prediction submodule, a coarse adjustment pre-positioning submodule, and a fine adjustment pre-energy storage submodule. Sewing trajectory prediction submodule: By collecting speed and stitch position signals from the sewing machine spindle encoder and combining them with fabric recognition results, it predicts upcoming changes in working conditions and sends pre-adjustment commands to the coarse / fine adjustment submodule 0.1-0.2ms in advance. Coarse adjustment pre-positioning submodule: For scenarios with sudden changes in fabric thickness, the coarse adjustment module detects the seam position in advance using a visual sensor and pre-adjusts the reference displacement in advance to avoid sudden changes in tension caused by untimely adjustment when the seam passes through. Fine-tuning the pre-energy storage submodule: When a high-frequency tension fluctuation scenario is anticipated, the fine-tuning submodule switches to a pre-activated state in advance to maintain low-power standby and shorten the startup response time.

7. The sewing thread tension control system based on piezoelectric thin film self-sensing according to claim 6, characterized in that: The main shaft encoder (2) is installed at the end (1) of the sewing machine main shaft and is coaxially connected to the main shaft. It is used to collect speed and stitch position signals. The vision sensor (3) is installed on the frame beam 5-8cm in front of the presser foot of the sewing machine. The lens is vertically facing the surface of the sewing material, and the center of the lens is aligned with the center line of the sewing material conveyor. It is used to detect the seam position of the sewing material in advance and provide a trigger signal for the pre-positioning of the coarse adjustment module.

8. The sewing thread tension control system based on piezoelectric thin film self-sensing according to claim 7, characterized in that: It also includes a multimodal collaborative weighting submodule: dynamically calculating the collaborative weighting factor by detecting the frequency and amplitude of the tension signal in real time. f is the frequency of tension change, and A is the amplitude of tension deviation. , This represents the normalized frequency of tension variation. The normalized deviation magnitude, where α and β are weighting coefficients and ; The default values ​​are α=0.6 and β=0.4, which are adaptively adjusted based on operating conditions. When ω approaches 1, the coarse adjustment submodule dominates the adjustment, while the fine adjustment submodule assists in the correction. When ω approaches 0, the fine-tuning submodule dominates high-value compensation, while the coarse-tuning submodule maintains the baseline.

9. A control method for a sewing thread tension control system based on piezoelectric thin film self-sensing as described in any one of claims 3-8, characterized in that: Includes the following steps: Before sewing starts or when the fabric thickness changes, the stepper motor moves first, adjusting the reference displacement of the thread clamping plate according to the fabric thickness to establish the initial tension for the corresponding working condition. During high-speed sewing, the piezoelectric film detects the thread tension signal in real time. If a momentary fluctuation occurs, the electromagnet responds immediately and compensates for the tension deviation through high-frequency adjustment. The working condition adaptive decision module dynamically adjusts the switching threshold and initial PID parameters of the coarse adjustment submodule and the fine adjustment submodule based on the fabric / thread type identified by the CNN-LSTM model, to ensure basic collaborative effect under different working conditions. The stepper motor in the layered collaborative control module is responsible for coarse adjustment of the reference displacement, responding to changes in the fabric thickness; the electromagnet is responsible for fine adjustment of dynamic compensation, suppressing tension fluctuations during high-speed sewing.

10. The sewing thread tension control method based on piezoelectric thin film self-sensing according to claim 9, characterized in that: an electromagnet... Start-up phase: constant pressure drive, rise time <5ms, rapid response to tension fluctuations or pre-adjustment commands; In the steady-state phase of the electromagnet: constant current control with a current accuracy of ±2% suppresses the influence of voltage fluctuations on the magnetic force and ensures the stability of tension adjustment.

Images