Transverse movement cooperative control system and method for super high speed warp knitting machine

Abstract

The present application relates to the technical field of textile machinery control, especially to a transverse movement cooperative control system and method for a super-speed warp knitting machine. The system comprises a transverse movement cooperative processing unit embedded in a main movement controller of the warp knitting machine, and a multi-axis cooperative prediction compensation algorithm running on the transverse movement cooperative processing unit. The transverse movement cooperative processing unit is a special hardware architecture. Through the software and hardware cooperative design of the special hardware architecture and the deep customized algorithm, the present application breaks the technical bottleneck of the traditional general architecture, and realizes the improvement of the super-speed transverse movement control performance. A super-high performance control hardware platform based on heterogeneous computing and hardware acceleration is constructed, and the real-time performance and synchronization problem of the general hardware are solved.

Description

Technical Field

[0001] This invention relates to the field of textile machinery control technology, and in particular to a lateral movement coordinated control system and method for ultra-high speed warp knitting machines. Background Technology

[0002] With the textile industry's ever-increasing demand for high-efficiency and highly flexible production, warp knitting machines are rapidly developing towards ultra-high speed, high precision, and intelligence. The electronic transverse system, as a core component of modern high-speed warp knitting machines, has completely replaced the mechanical pattern cam, and its performance directly determines the machine's maximum speed, fabric quality, and the flexibility of pattern changes.

[0003] Currently, most mainstream electronic traverse control systems adopt a modular architecture of "general-purpose motion controller or PLC + standard servo drive". While this architecture achieves basic CNC functions, it has gradually revealed the following inherent bottlenecks when dealing with ultra-high-speed knitting conditions where the spindle speed exceeds 2700 rpm: General-purpose motion controllers need to handle multiple tasks simultaneously, including spindle synchronization, warp feeding, traction and winding, and lateral movement, resulting in the distribution of computational resources and bus bandwidth. This is difficult to guarantee on a shared resource platform, easily leading to communication jitter and latency. Limitations in the control algorithm make it difficult to overcome high-speed dynamics problems. The decoupled hardware and software design restricts in-depth performance optimization.

[0004] Therefore, the industry urgently needs a solution that breaks away from traditional architecture. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art described in the background art, the present invention provides a lateral movement cooperative control system and method for ultra-high speed warp knitting machines.

[0006] The technical solution adopted by the present invention to solve its technical problem is: a transverse movement cooperative control system for an ultra-high speed warp knitting machine, including a transverse movement cooperative processing unit embedded in the main motion controller of the warp knitting machine, and a multi-axis cooperative prediction and compensation algorithm running on the transverse movement cooperative processing unit; the transverse movement cooperative processing unit is a dedicated hardware architecture, which provides computing power and deterministic execution environment for the multi-axis cooperative prediction and compensation algorithm; The transverse traverse collaborative processing unit adopts a heterogeneous computing architecture, including a dual-core ARM Cortex-A53 application processor, a dual-core Cortex-R5F real-time processor, and a large-scale FPGA hardware acceleration circuit. It also integrates an EtherCAT-based gigabit industrial Ethernet bus master station. The application processor is responsible for upper-level process analysis, human-machine interaction, and data management. The real-time processor is dedicated to executing the transverse traverse core control algorithm. The FPGA is used to realize nanometer-level precision hardware position comparison and instruction triggering, with a control cycle real-time performance of ≤100 microseconds and synchronization jitter <±1 microsecond. The EtherCAT gigabit industrial Ethernet bus master station optimizes the communication protocol for linear servo drives to achieve nanosecond-level synchronization of transverse traverse servo units with up to 32 axes.

[0007] According to another embodiment of the present invention, the FPGA further includes an ultra-high precision position comparator, a high-speed instruction output engine, a high-speed data acquisition engine, and a dedicated adder module integrated within it. The ultra-high precision position comparator is physically connected to the spindle encoder signal to achieve nanosecond-level pulse matching and action triggering; the high-speed command output engine is used to send synchronization command packets to the linear servo driver; the high-speed data acquisition engine is used to receive feedback data from each axis driver and onboard sensor; and the dedicated adder module is used for real-time synthesis of current commands.

[0008] According to another embodiment of the present invention, the multi-axis collaborative prediction compensation algorithm further includes a speed-adaptive dual-mode control strategy, which intelligently and smoothly switches between a standard optimized control mode and an ultra-high-speed collaborative enhancement mode based on the real-time spindle speed of the warp knitting machine; wherein, the standard optimized control mode corresponds to a spindle speed ≤2700rpm, and the ultra-high-speed collaborative enhancement mode corresponds to a spindle speed >2700rpm.

[0009] According to another embodiment of the present invention, the ultra-high-speed collaborative enhancement mode is further comprising, on the basis of the standard optimized control mode, a dynamic dynamics feedforward compensation module and a multi-axis coupling active damping module superimposed; the dynamic dynamics feedforward compensation module is used to calculate and compensate for the comb inertial force, Coriolis force and frictional torque, and the multi-axis coupling active damping module is used to suppress the structural resonance of the multi-comb motion under ultra-high speed.

[0010] A method for coordinated control of lateral movement in an ultra-high-speed warp knitting machine, wherein the hardware system workflow includes the following steps: S1. Data Loading and Parsing: After the process document is issued by the BY Touch OS operating system, the general-purpose CPU of the main motion controller receives and initially parses it, extracts the original chain block data of the transverse movement, and loads it into the tightly coupled memory TCM of the transverse movement co-processing unit through the AXI internal high-speed bus. S2. High-speed trajectory planning and pre-calculation: The real-time processor calls the built-in dynamic model to perform online secondary planning of the lateral trajectory, calculates the velocity, acceleration, and jerk profile of the cam curve, and optimizes it smoothly according to the principle of minimizing vibration. The optimized trajectory data and feedforward compensation are stored in the FPGA's high-speed buffer register. S3. Data Distribution and Synchronization Trigger: The FPGA distributes the planning data to each hardware functional module. The ultra-high precision position comparator receives the spindle encoder pulse and completes nanosecond-level matching, automatically triggering the execution action. S4. High-speed command execution and feedback: The high-speed command output engine extracts the target position and feedforward force / current value from the high-speed buffer register, and sends synchronous command packets to the linear servo driver through the EtherCAT slave controller at a cycle of ≤100µs. At the same time, the high-speed data acquisition engine receives feedback data and sends it back to the real-time processor. S5. Health Monitoring and Adaptive Adjustment: The real-time processor runs a monitoring algorithm to analyze feedback data. If performance degradation is detected, the feedforward model parameters are fine-tuned or the compensation amount is dynamically updated via the FPGA. Running logs and performance data are also recorded. According to another embodiment of the present invention, the execution flow of the software algorithm further includes a standard optimized control mode that integrates feedforward compensation and filtering technology of a fixed model on the basis of traditional PID control to achieve accurate and stable control of lateral motion.

[0011] According to another embodiment of the present invention, the execution steps of the dynamic dynamics feedforward compensation module of the ultra-high-speed cooperative enhancement mode in the execution flow of the software algorithm further include: A1. Real-time status acquisition: Within a 100µs control cycle k, acquire the real-time angular velocity ω(k) of the spindle, the real-time acceleration a_i (k) of the i-th comb, the real-time velocity v_i (k) of the i-th axis comb, and the real-time axis current I_i (k) of the i-th axis servo driver from the FPGA high-speed buffer. A2. Dynamic feedforward force calculation: Calculate the total feedforward compensation force F_ff_i (k) of the i-th comb, where F_ff_i (k) is the vector sum of inertial force compensation F_inertia_i (k), Coriolis force compensation F_coriolis_i (k), and dynamic friction force compensation F_friction_i (k), i.e., F_ff_i (k) = F_inertia_i (k) + F_coriolis_i (k) + F_friction_i (k); Where F_inertia_i (k) = m_eff_i * a_i (k) + J_i * d (a_i (k)) / dt, m_eff_i is the equivalent mass of the i-th axis referred to the motor side, J_i is the equivalent moment of inertia of the i-th axis transmission system, and d(a_i (k)) / dt is the differential of the acceleration; F_coriolis_i (k) = 2 * m_eff_i * Ω(k) × v_i (k), where Ω(k) is the coupled rotation vector determined by the machine structure; F_friction_i (k) = [σ0 * z_i (k) + σ1 * dz_i (k) / dt]+ σ2 * v_i (k), where z_i (k) is the internal friction state variable, and σ0, σ1, and σ2 are the static friction, dynamic friction, and viscous friction coefficients, respectively. dz_i (k) / dt is the first derivative of z_i (k), which is the speed factor describing the frictional motion.

[0012] A3. Command Synthesis and Output: F_ff_i (k) is converted into an additional current command ΔI_ff_i (k) through the servo system force constant K_t, ΔI_ff_i (k) = F_ff_i (k) / K_t. The ΔI_ff_i (k) and the current command output by the position loop are synthesized in the FPGA dedicated adder module and sent to the driver through EtherCAT.

[0013] According to another embodiment of the present invention, the execution steps of the multi-axis coupled active damping module in the ultra-high-speed cooperative enhancement mode in the execution flow of the software algorithm further include: B1. Vibration State Sensing and Extraction: Receive the rack high-response acceleration sensor signal accel_j (k) transmitted by FPGA, and extract the vibration component vib_j (k) of a specific frequency band through a parallel digital bandpass filter. The center frequency of the digital bandpass filter matches the first and second natural frequencies of the mechanical structure. B2. Coupled Path Analysis and Compensation Command Generation: A multi-degree-of-freedom coupled transfer matrix H obtained from experimental modal analysis is preset. When the vibration energy E_vib_j (k) at measurement point j exceeds the safety threshold, the main source axis m is located according to H and the real-time motion state of each axis. An anti-phase damping command ΔS_damp_m (k) is generated for the source axis m. ΔS_damp_m (k) = -K_damp * P ( a_m (k) )_filtered, where K_damp is the adaptive damping gain and P ( a_m (k))_filtered is the filtered signal of the acceleration of axis m. B3. Multi-axis coordinated damping execution: The coordinated compensation command of ΔS_damp_m (k) and adjacent associated axes is injected as the position loop fine adjustment amount to control the comb to generate electronic damping motion to counteract the vibration force.

[0014] According to another embodiment of the present invention, when the software algorithm switches from the standard optimized control mode to the ultra-high-speed collaborative enhancement mode, the following collaborative configuration is executed synchronously at the hardware level: activating the ultra-high-speed signal processing channel in the FPGA to improve the sampling rate and data throughput; switching the system control cycle from 250µs to 100µs; and dynamically adjusting the task scheduling priority of the real-time processor core to allocate deterministic computation time slices for the enhancement algorithm.

[0015] The beneficial effects of this invention are that, through the collaborative design of dedicated hardware architecture and deeply customized algorithms, it breaks through the technical bottlenecks of traditional general-purpose architectures, achieving a significant improvement in ultra-high-speed lateral movement control performance. It constructs an ultra-high-performance control hardware platform based on heterogeneous computing and hardware acceleration, solving the real-time and synchronization problems of general-purpose hardware.

[0016] An advanced, controllable control algorithm for ultra-high-speed operating conditions has been developed, which effectively solves the problems of high-speed dynamics and multi-axis coupled vibration. The system has undergone rigorous production practice verification. Warp knitting machines equipped with this system can achieve ultra-high-speed knitting at 3700rpm for a long time, and the lateral positioning accuracy has been maintained within ±0.01mm for a long time, resulting in excellent fabric forming quality. Detailed Implementation

[0017] This invention provides a lateral movement collaborative control system and method for ultra-high-speed warp knitting machines. The core of this invention lies in constructing a dedicated hardware architecture for the lateral movement collaborative processing unit and running a multi-axis collaborative prediction and compensation algorithm. Through deep integration and collaborative optimization of software and hardware, nanosecond-level synchronization and microsecond-level response of the lateral movement control of ultra-high-speed warp knitting machines are achieved, solving the technical problems of real-time performance, dynamic disturbance, and multi-axis coupled vibration in traditional systems, and realizing ultra-high-speed knitting at 3700 rpm.

[0018] The core hardware of this invention system is our company's custom-designed transverse collaborative processing unit, which is embedded in the motion controller of the warp knitting machine. The processing unit adopts a heterogeneous architecture of "ARM Cortex-A53 application processor + Cortex-R5F real-time processor + FPGA hardware acceleration chip," and also integrates a gigabit industrial Ethernet bus master module based on EtherCAT. The FPGA internally integrates modules such as an ultra-high-precision position comparator, a high-speed instruction output engine, a high-speed data acquisition engine, and a dedicated adder module. Its workflow follows the logic of "planning - synchronous execution - monitoring and optimization," specifically including the following steps: Step 1: Data Loading and Parsing After the engineers set the program and parameters, the operators send the process file to the warp knitting machine's main motion controller via the BY Touch OS operating system. The main controller's CPU first receives the process file and performs preliminary parsing, extracting the raw chain block data related to the lateral movement control. Subsequently, the raw chain block data is transmitted through the AXI's internal high-speed bus and loaded into the dedicated tightly coupled memory (TCM) of the lateral movement coprocessing unit. The tightly coupled memory ensures that the real-time processor can access the data with minimal latency, laying the foundation for subsequent high-speed processing.

[0019] Step 2: High-speed trajectory planning and pre-calculation The dual-core Cortex-R5F real-time processor of the transverse traverse coprocessing unit will start immediately. Instead of directly manipulating the original chain block data, it will call the built-in warp-knitted transverse traverse multibody dynamics model to first perform secondary planning of the transverse traverse trajectory. During this process, the real-time processor will calculate the velocity, acceleration, and jerk profile of each electronic cam curve in real time, and strictly optimize the trajectory according to the principle of "vibration minimization" to avoid unnecessary vibration effects caused by sudden acceleration changes at ultra-high speed. All optimized trajectory data and corresponding feedforward compensation amounts will be pre-calculated and stored in the FPGA. When needed, the high-speed buffer register in the FPGA can be directly accessed to realize the pre-storage of instructions.

[0020] Step 3: Data Distribution and Synchronization Trigger The FPGA distributes the block data in the high-speed buffer register to various internal hardware functional modules such as the ultra-high precision position comparator and the high-speed command output engine. The ultra-high precision position comparator is physically connected directly to the spindle encoder signal. During machine operation, each pulse emitted by the spindle encoder is sent to the main controller. The comparator completes the matching of the pulse with the planned data within nanoseconds and automatically triggers the execution action. The entire process can be completed without the intervention of the real-time processor software program, eliminating the jitter caused by software task scheduling delay and ensuring a high degree of synchronization of action triggering.

[0021] Step 4: High-speed instruction execution and feedback Once the action is triggered, the high-speed command output engine within the FPGA immediately retrieves data such as the target position and feedforward force / current value from the high-speed buffer register. Through the dedicated EtherCAT slave controller integrated in the FPGA, it sends the synchronization command packet to each servo driver at a fixed ultra-short cycle of ≤100µs, realizing high-speed issuance of lateral movement control commands. At the same time, the high-speed data acquisition engine within the FPGA receives feedback data from each axis servo driver and mechanical acceleration and position sensors in real time, and packages the feedback data back to the real-time processor for subsequent monitoring and adjustment.

[0022] Step 5: Health Monitoring and Adaptive Adjustment While the real-time processor executes the core control algorithm, a health monitoring algorithm runs continuously in the background to analyze the feedback data sent back by the high-speed data acquisition engine in real time, identifying characteristic performance degradation signals such as changes in vibration spectrum and positioning accuracy deviations. Once performance degradation is detected, the real-time processor immediately fine-tunes the feedforward model parameters or dynamically updates the compensation amount through the FPGA to achieve adaptive optimization of the control algorithm. At the same time, the system records all operation logs and performance data throughout the process, providing data support for subsequent equipment maintenance and algorithm iteration.

[0023] The core of the control software in this invention is a multi-axis collaborative predictive compensation algorithm. This algorithm adopts an innovative speed-adaptive dual-mode control strategy, which intelligently and seamlessly switches between the standard optimized control mode and the ultra-high-speed collaborative enhancement mode based on the real-time rotational speed of the warp knitting machine spindle. This ensures optimal control performance across the entire speed range while maximizing the system's operating efficiency and stability. The applicable scenarios and execution logic for the two modes are as follows: (1) Mode 1: Standard optimized control mode (spindle speed ≤ 2700 rpm)

[0024] When the warp knitting machine operates at a low to medium speed of 2700 rpm or below, the system automatically activates the standard optimized control mode. This mode integrates feedforward compensation technology based on a fixed model and digital filtering technology on the basis of traditional position loop PID control. Through in-depth optimization of PID control parameters and the cancellation of conventional disturbances by fixed feedforward compensation, it can fully guarantee the positioning accuracy and stability of the multi-comb lateral movement, meeting the production needs of most high-precision fabrics. At the same time, the system consumes less hardware resources in this mode, and can maximize control stability and reliability.

[0025] The standard optimized control mode uses a positional PID controller, and the control law is: u (k) = Kp・e (k) + Ki・Ts・Σe (k) + Kd・(e (k)−e (k−1)) / Ts Where e(k) is the position error signal, e(k-1) is the position error at the previous moment, Kp, Ki, and Kd are the proportional, integral, and derivative coefficients, respectively, Ts is the control period, and u(k) is the position loop control output in the k-th control period. This formula shows that in the standard control mode, the output control quantity is calculated based on the real-time position error to achieve high-precision closed-loop position control of the comb lateral movement.

[0026] The fixed model feedforward compensation is calculated based on the nominal inertia J_nom and the nominal viscous friction coefficient B_nom, specifically as follows: F_ff_fixed = J_nom・a_ref + B_nom・v_ref Where a_ref is the reference acceleration and v_ref is the reference velocity. (2) Mode 2: Ultra-high speed synergistic enhancement mode (spindle speed > 2700 rpm)

[0027] When the system detects that the spindle speed of the warp knitting machine exceeds the threshold of 2700 rpm through the spindle encoder, the control algorithm will automatically and seamlessly switch to the ultra-high speed collaborative enhancement mode. To avoid frequent mode switching caused by spindle speed fluctuations around the 2700rpm threshold, the system is equipped with a mode switching lock mechanism: when the spindle speed first exceeds 2700rpm, the system enters continuous monitoring mode. If the speed is consistently above 2700rpm for five consecutive control cycles, it switches to the ultra-high-speed collaborative enhancement mode. After mode switching, it only reverts to the standard optimized control mode when the speed remains below 2700rpm and the same cycle lock condition is met, thus ensuring the stability of mode switching.

[0028] This mode does not completely replace the standard optimized control mode algorithm. Instead, it dynamically activates and superimposes two core enhancement modules: a dynamic dynamics feedforward compensation module and a multi-axis coupled active damping module. Through the collaborative work of the two modules, it addresses the challenges of nonlinear dynamics and multi-axis coupled vibration problems at ultra-high speeds. The specific execution steps are as follows: Enhancement Module A: Dynamics Feedforward Compensation Module The core function of this module is to accurately calculate the inertial force, Coriolis force, and frictional torque generated by the combing motion based on real-time operating conditions at ultra-high speeds, and to achieve precise cancellation through feedforward compensation, thereby avoiding vibration and positioning deviations caused by nonlinear disturbances. The specific execution steps are as follows: Step A1: Real-time Status Acquisition Within each 100µs control cycle k, the module acquires four core data items in real time from the FPGA's high-speed buffer: the real-time angular velocity ω(k) of the spindle (which can be directly converted into the spindle speed), the real-time acceleration a_i(k) of the i-th comb (given by the second derivative of the target trajectory and the state observer), the real-time velocity v_i(k) of the i-th comb, and the real-time shaft current I_i(k) of the i-th axis servo driver (which directly reflects the electromagnetic thrust of the motor).

[0029] The real-time acceleration a_i(k) of the i-th comb is calculated jointly by the second derivative of the target trajectory and the state observer. The target trajectory is the motion trajectory optimized by high-speed trajectory planning and vibration smoothing. The state observer, built based on the real-time feedback signal from the encoder, is used to obtain the deviation between the actual motion and the theoretical trajectory, and outputs the compensation acceleration in real time. The final real-time acceleration is the sum of the optimized trajectory acceleration and the observer's compensation acceleration. The state observer is a software-based, real-time oscilloscope-like program designed to observe the motion curve. It is the simulation software used during measurement. You can find corresponding oscilloscope software for observation.

[0030] Step A2: Dynamic Feedforward Force Calculation Based on the collected real-time status data, the total feedforward compensation force F_ff_i(k) is calculated for each comb. This compensation force is the vector sum of three components: inertial force compensation, Coriolis force compensation, and dynamic friction force compensation. The calculation formula is as follows: F_ff_i (k) = F_inertia_i (k) + F_coriolis_i (k) + F_friction_i (k); The internal friction state z_i(k) of the LuGre friction model satisfies the following differential equation: dz_i / dt = v_i (k) − (|v_i (k)| / g (v_i (k)))・z_i (k) Where g (v_i (k)) = μ_c + (μ_s − μ_c)・exp (−(v_i (k) / v_s)²), μ_s is the static friction coefficient, μ_c is the Coulomb friction coefficient, and v_s is the Stribeck velocity.

[0031] in: ① Inertial force compensation F_inertia_i (k): used to counteract the inertial disturbances of the comb and moving parts at ultra-high speed. The calculation formula is F_inertia_i (k) = m_eff_i * a_i (k) + J_i * d (a_i (k)) / dt; m_eff_i is the equivalent mass of the i-th axis referred to the motor side (including all moving parts such as the comb and needle bed), J_i is the equivalent rotational inertia of the i-th axis transmission system, and d (a_i (k)) / dt is the derivative of the acceleration (jerk), used to compensate for the torque change caused by the sudden acceleration.

[0032] ② Coriolis force / centripetal force compensation F_coriolis_i (k): This is not negligible under ultra-high-speed conditions with coordinated movement of multiple comb shafts. The calculation formula is F_coriolis_i (k) = 2 * m_eff_i * Ω(k) × v_i (k); Ω(k) is the coupled rotation vector determined by the mechanical structure of the warp knitting machine, which is preset in the system after experimental calibration.

[0033] ③ Dynamic friction compensation F_friction_i (k): A simplified dynamic friction observation method based on the LuGre model is adopted to accurately compensate for the dynamic friction of the transmission system. The calculation formula is F_friction_i (k) = [σ0 * z_i(k) + σ1 * dz_i (k) / dt] + σ2 * v_i (k); z_i (k) is the internal friction state variable, which is obtained by observing the differential dv_i (k) / dt of the comb's real-time speed. σ0, σ1, and σ2 are the static friction, dynamic friction, and viscous friction coefficients, respectively, which have been experimentally calibrated.

[0034] Step A3: Instruction Synthesis and Output The calculated feedforward compensation force F_ff_i (k) is converted into an additional current command ΔI_ff_i (k) for the i-th axis through the force constant K_t of the servo system. The calculation formula is ΔI_ff_i (k) = F_ff_i (k) / K_t. This additional current command ΔI_ff_i (k) and the current command output by the position loop PID control are instantaneously synthesized in the dedicated adder module of the FPGA. The synthesized total current command is sent to the corresponding servo driver through the high-speed EtherCAT bus at fixed intervals, realizing accurate feedforward compensation that occurs "before the disturbance occurs", thus canceling the influence of nonlinear disturbances at the source.

[0035] Enhancement Module B: Multi-axis Coupled Active Damping Module This module is the core component for addressing structural resonance in multi-comb motion at ultra-high speeds. By real-time monitoring of vibration status, locating the vibration source, and generating anti-phase damping commands, it creates an "electronic damper" effect, actively absorbing vibration energy and suppressing overall mechanical vibration to a low level. The specific execution steps are as follows: Step B1: Vibration State Sensing and Extraction The algorithm receives the high-response acceleration sensor signal accel_j(k) from the FPGA and installed at the key frame position of the warp knitting machine in real time; it filters the sensor signal through a set of parallel digital bandpass filters to extract the vibration component vib_j(k) of a specific frequency band. The center frequency of the digital bandpass filter is preset to be near the first and second natural frequencies of the warp knitting machine's mechanical structure (such as 80Hz and 220Hz), and only the characteristic frequency vibration that causes structural resonance is extracted.

[0036] Step B2: Coupling Path Analysis and Compensation Command Generation The system has a pre-defined simplified multi-degree-of-freedom coupling transfer matrix H, which is obtained through experimental modal analysis. This matrix H, obtained through experimental modal analysis, characterizes the transfer relationship between the excitation of each comb shaft and the structural vibration response. During construction, the hammer impact method was used to conduct on-site modal tests on the warp knitting machine frame and comb mechanism, extracting the frequency response function and mode shape vector of each measuring point, thus forming a coupling transfer matrix suitable for this machine model. The matrix parameters are pre-fixed in the system. This matrix accurately describes the vibration transfer relationship of the motion of any comb shaft (excitation source m) to any measuring point j (or another comb shaft n) on the mechanical frame. When the vibration energy E_vib_j(k) (the square of the vibration component vib_j(k)) at any measuring point j exceeds the system's pre-set safety threshold, the algorithm immediately initiates reverse calculation. i represents the i-th comb, j represents the measurement point obtained on the comb, and has no specific numerical relationship with i. m represents the number of the source axis being measured, which does not necessarily have to be the same as i. n represents the axis number affected by the vibration of axis m.

[0037] First, based on the coupling transfer matrix H and the real-time motion states of all comb axes [a_i (k), v_i (k)], the source axis m that contributes the most to the vibration of measurement point j is accurately located; Subsequently, a small-amplitude, anti-phase damping command ΔS_damp_m(k) is generated for the source axis m, calculated as ΔS_damp_m(k) = -K_damp * P(a_m(k))_filtered; where K_damp is the adaptive damping gain, which is adjusted online according to the vibration energy, specifically: K_damp = K_damp_base・(1+ γ・(E_vib_j(k)−E_th) / E_th); where K_damp_base is the base damping gain, γ is the adjustment factor, and E_th is the vibration energy threshold. This can be fine-tuned online according to the vibration amplitude. P(a_m(k))_filtered is the real-time acceleration a_m(k) of the source axis m, filtered to match the vibration frequency, ensuring that the damping command is in phase with the vibration signal.

[0038] P(am(k))_filtered represents the signal after processing the real-time acceleration signal am(k) of the source axis m using a second-order Butterworth digital bandpass filter and a zero-phase filter. The center frequency of the filter is matched with the first and second natural frequencies of the mechanical structure to extract the resonant frequency components and avoid interference with normal motion. The filter transfer function is: G (s) = (2πf0)²・s / (s² + 2πf0・ξ・s + (2πf0)²) Where s is the Laplace operator; f0 is the center frequency of the filter, which matches the first and second natural frequencies of the warp knitting machine's mechanical structure; and ξ is the damping ratio, used to adjust the filter's bandwidth and frequency selectivity. This filter can accurately extract the characteristic vibration components that induce structural resonance, providing a reliable input for multi-axis coupled active damping.

[0039] Step B3: Multi-axis coordinated damping execution The calculated anti-phase damping command ΔS_damp_m(k) for the source axis m, along with the collaborative compensation command for the adjacent associated axes n calculated from the coupling transfer matrix H, are injected into the control command as fine-tuning amounts for the position loop. Based on the coupling transfer matrix H, the collaborative compensation commands for the adjacent associated axes n can be calculated from the vibration state of the source axis m. The anti-phase damping command ΔS_damp_m(k) for the source axis m is superimposed with the collaborative compensation commands for each associated axis n, and injected into the control command as fine-tuning amounts for the position loop, achieving multi-axis collaborative damping suppression. The total fine-tuning amount = source axis m damping fine-tuning amount + Σ associated axis n collaborative fine-tuning amount. This fine-tuning amount controls the corresponding comb to generate a small-amplitude "electronic damping" motion. This motion is opposite in phase to the original motion that caused the vibration, actively absorbing vibration energy before the mechanical resonance is significantly amplified, thus effectively suppressing multi-axis coupled vibration.

[0040] When the control algorithm of this invention switches from the standard optimized control mode to the ultra-high-speed collaborative enhancement mode, it does not simply activate two enhancement algorithm modules. Simultaneously, the hardware layer undergoes collaborative configuration adjustments to ensure that the hardware platform can provide sufficient computing power, communication bandwidth, and execution efficiency for the enhancement algorithms. This achieves deep integration and collaborative empowerment of software and hardware. Specific hardware configuration adjustments include: FPGA Ultra-High-Speed ​​Signal Processing Channel Activation: The ultra-high-speed signal processing channel built into the FPGA is activated immediately. This channel has a higher sampling rate and data throughput capability, and provides a dedicated channel for high-speed data acquisition, transmission and processing of dynamic dynamics feedforward compensation module and multi-axis coupling active damping module, ensuring real-time data exchange without delay. Shortened system control cycle: The system control cycle is automatically switched from 250µs in the conventional medium speed range to 100µs in the ultra-high speed range, which greatly increases the frequency of control command issuance and meets the higher requirements of the enhanced algorithm for control bandwidth. Real-time processor task priority adjustment: Dynamically adjust the task scheduling priority of the Cortex-R5F real-time processor core to allocate the system's core computing resources to the two enhancement algorithm modules, ensuring that the enhancement algorithms are allocated absolute and deterministic computing time slices and avoiding interference from other non-core tasks.

[0041] The ultra-high-speed warp knitting machine lateral movement collaborative control system and method of the present invention have completed rigorous performance verification and industrial application in actual production: the new generation of ultra-high-speed warp knitting machines equipped with this system can achieve long-term continuous ultra-high-speed knitting at 3700rpm, which is 37% higher than the maximum speed of the traditional system; the lateral movement positioning accuracy has been maintained within ±0.01mm for a long time, and the positioning accuracy has been significantly improved; at the same time, the vibration amplitude of the guide bar lateral movement axis at the reversing point is reduced by more than 60%, which solves the vibration problem under ultra-high speed and results in excellent fabric forming quality.

[0042] Currently, the system has been applied as a standard configuration in the large-scale production and commercial sales of ultra-high-speed warp knitting machines. It has achieved 24 / 7 uninterrupted and stable operation in many downstream textile enterprises without any downtime or defective fabrics caused by control problems. This fully demonstrates the reliability, stability, and mature engineering level of the invention, and can provide core technical support for the ultra-high-speed and high-precision upgrade of the textile machinery industry.

[0043] Testing has shown that the control system of this invention fully meets the requirements of high-end warp knitting production, and its performance in all aspects is superior to existing technologies.

[0044] Additional notes: The same warp knitting machine was used in the experiment, only the transverse control system was changed to eliminate the influence of the equipment itself on the experimental results; all data are the average ± standard deviation of 3 parallel experiments, the experimental environment temperature was 25±2℃ and the humidity was 60±5%RH, which meets the standard experimental conditions of the textile industry.

[0045] The above embodiments are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A lateral movement cooperative control system for an ultra-high-speed warp knitting machine, characterized in that, It includes a transverse shift collaborative processing unit embedded in the main motion controller of the warp knitting machine, and a multi-axis collaborative prediction and compensation algorithm running on the transverse shift collaborative processing unit; the transverse shift collaborative processing unit is a dedicated hardware architecture that provides computing power and a deterministic execution environment for the multi-axis collaborative prediction and compensation algorithm; The transverse traverse collaborative processing unit adopts a heterogeneous computing architecture, including a dual-core ARM Cortex-A53 application processor, a dual-core Cortex-R5F real-time processor, and a large-scale FPGA hardware acceleration circuit. It also integrates an EtherCAT-based gigabit industrial Ethernet bus master station. The application processor is responsible for upper-level process analysis, human-machine interaction, and data management. The real-time processor is dedicated to executing the transverse traverse core control algorithm. The FPGA is used to realize nanometer-level precision hardware position comparison and instruction triggering, with a control cycle real-time performance of ≤100 microseconds and synchronization jitter <±1 microsecond. The EtherCAT gigabit industrial Ethernet bus master station optimizes the communication protocol for linear servo drives to achieve nanosecond-level synchronization of transverse traverse servo units with up to 32 axes.

2. The transverse movement cooperative control system for an ultra-high-speed warp knitting machine according to claim 1, characterized in that, The FPGA integrates an ultra-high precision position comparator, a high-speed instruction output engine, a high-speed data acquisition engine, and a dedicated adder module. The ultra-high precision position comparator is physically connected to the spindle encoder signal to achieve nanosecond-level pulse matching and action triggering; the high-speed command output engine is used to send synchronization command packets to the linear servo driver; the high-speed data acquisition engine is used to receive feedback data from each axis driver and onboard sensor; and the dedicated adder module is used for real-time synthesis of current commands.

3. The transverse movement cooperative control system for an ultra-high-speed warp knitting machine according to claim 1, characterized in that, The multi-axis collaborative prediction and compensation algorithm adopts a speed-adaptive dual-mode control strategy, which intelligently and smoothly switches between the standard optimized control mode and the ultra-high-speed collaborative enhancement mode based on the real-time spindle speed of the warp knitting machine. The standard optimized control mode corresponds to a spindle speed ≤2700rpm, while the ultra-high-speed collaborative enhancement mode corresponds to a spindle speed >2700rpm.

4. The transverse movement cooperative control system for an ultra-high-speed warp knitting machine according to claim 3, characterized in that, The ultra-high-speed collaborative enhancement mode superimposes a dynamic dynamics feedforward compensation module and a multi-axis coupling active damping module on the basis of the standard optimized control mode. The dynamic dynamics feedforward compensation module is used to calculate and compensate for the comb inertial force, Coriolis force and friction torque, and the multi-axis coupling active damping module is used to suppress the structural resonance of the multi-comb motion under ultra-high speed.

5. A method for lateral movement coordinated control of an ultra-high-speed warp knitting machine based on any one of claims 1-4, characterized in that, The hardware system workflow includes the following steps: S1. Data Loading and Parsing: After the process document is issued by the BY Touch OS operating system, the general-purpose CPU of the main motion controller receives and initially parses it, extracts the original chain block data of the transverse movement, and loads it into the tightly coupled memory TCM of the transverse movement co-processing unit through the AXI internal high-speed bus. S2. High-speed trajectory planning and pre-calculation: The real-time processor calls the built-in dynamic model to perform online secondary planning of the lateral trajectory, calculates the velocity, acceleration, and jerk profile of the cam curve, and optimizes it smoothly according to the principle of minimizing vibration. The optimized trajectory data and feedforward compensation are stored in the FPGA's high-speed buffer register. S3. Data distribution and synchronous triggering: The FPGA distributes the planning data to each hardware functional module. The ultra-high precision position comparator receives the main shaft encoder pulse and completes nanosecond-level matching, automatically triggering pedestrian actions. S4. High-speed command execution and feedback: The high-speed command output engine extracts the target position and feedforward force / current value from the high-speed buffer register, and sends synchronous command packets to the linear servo driver through the EtherCAT slave controller at a cycle of ≤100µs. At the same time, the high-speed data acquisition engine receives feedback data and sends it back to the real-time processor. S5. Health monitoring and adaptive adjustment: The real-time processor runs a monitoring algorithm to analyze feedback data. If performance degradation is detected, the feedforward model parameters are fine-tuned or the compensation amount is dynamically updated through the FPGA. The operation log and performance data are recorded.

6. The method for lateral movement coordinated control of an ultra-high-speed warp knitting machine according to claim 5, characterized in that, In the execution flow of the algorithm, the standard optimized control mode integrates feedforward compensation and filtering technology of a fixed model on the basis of traditional PID control to achieve accurate and stable control of lateral motion.

7. The method for lateral movement coordinated control of an ultra-high-speed warp knitting machine according to claim 5, characterized in that, In the execution flow of the algorithm, the dynamic feedforward compensation module of the ultra-high-speed cooperative enhancement mode executes the following steps: A1. Real-time status acquisition: Within a 100µs control cycle k, acquire the real-time angular velocity ω(k) of the spindle, the real-time acceleration a_i (k) of the i-th comb, the real-time velocity v_i (k) of the i-th axis comb, and the real-time axis current I_i (k) of the i-th axis servo driver from the FPGA high-speed buffer. A2. Dynamic feedforward force calculation: Calculate the total feedforward compensation force F_ff_i (k) of the i-th comb, where F_ff_i (k) is the vector sum of inertial force compensation F_inertia_i (k), Coriolis force compensation F_coriolis_i (k), and dynamic friction force compensation F_friction_i (k), i.e., F_ff_i (k) = F_inertia_i (k) + F_coriolis_i (k) + F_friction_i (k); Where F_inertia_i (k) = m_eff_i * a_i (k) + J_i * d (a_i (k)) / dt, m_eff_i is the equivalent mass of the i-th axis referred to the motor side, J_i is the equivalent moment of inertia of the i-th axis transmission system, and d (a_i(k)) / dt is the differential of the acceleration; F_coriolis_i (k) = 2 * m_eff_i * Ω(k) × v_i (k), where Ω(k) is the coupled rotation vector determined by the machine structure; F_friction_i (k) = [σ0 * z_i (k) + σ1 * dz_i (k) / dt] + σ2 * v_i (k), where z_i(k) is the internal friction state variable, and σ0, σ1, and σ2 are the static friction, dynamic friction, and viscous friction coefficients, respectively. A3. Command Synthesis and Output: F_ff_i (k) is converted into an additional current command ΔI_ff_i (k) through the servo system force constant K_t, ΔI_ff_i (k) = F_ff_i (k) / K_t. The ΔI_ff_i (k) and the current command output by the position loop are synthesized in the FPGA dedicated adder module and sent to the driver through EtherCAT.

8. The method for lateral movement coordinated control of an ultra-high-speed warp knitting machine according to claim 5, characterized in that, In the execution flow of the algorithm, the execution steps of the multi-axis coupled active damping module of the ultra-high-speed cooperative enhancement mode include: B1. Vibration state perception and extraction: Receive the rack high-response acceleration sensor signal accel_j(k) transmitted by FPGA, and extract the vibration component vib_j(k) of a specific frequency band through a parallel digital bandpass filter. The center frequency of the digital bandpass filter matches the first and second natural frequencies of the mechanical structure. B2. Coupled Path Analysis and Compensation Command Generation: A multi-degree-of-freedom coupled transfer matrix H obtained from experimental modal analysis is preset. When the vibration energy E_vib_j (k) at measurement point j exceeds the safety threshold, the main source axis m is located according to H and the real-time motion state of each axis. An anti-phase damping command ΔS_damp_m (k) is generated for the source axis m. ΔS_damp_m (k) = -K_damp * P ( a_m (k) )_filtered, where K_damp is the adaptive damping gain and P ( a_m (k) )_filtered is the filtered signal of the acceleration of axis m. B3. Multi-axis coordinated damping execution: The coordinated compensation command of ΔS_damp_m (k) and adjacent associated axes is injected as the position loop fine adjustment amount to control the comb to generate electronic damping motion to counteract the vibration force.

9. The method for lateral movement coordinated control of an ultra-high-speed warp knitting machine according to claim 5, characterized in that, When the algorithm switches from the standard optimized control mode to the ultra-high-speed collaborative enhancement mode, the following collaborative configurations are executed synchronously at the hardware level: the ultra-high-speed signal processing channel in the FPGA is activated to improve the sampling rate and data throughput; the system control cycle is switched from 250µs to 100µs; and the task scheduling priority of the real-time processor core is dynamically adjusted to allocate deterministic computation time slices for the enhanced algorithm.

Images