Automatic control system of three-dimensional braiding machine

Abstract

The invention relates to the technical field of composite material manufacturing equipment, in particular to an automatic control system of a three-dimensional braiding machine, which is characterized in that a client host, a braiding machine upper computer and a PLC (Programmable Logic Controller) form a control layer; the double-shaft driving module and the traction robot form an execution layer; and the IOT module and the safety monitoring component form a sensing layer. The core of the method is that a parameter self-learning optimization module is arranged, a quality prediction model is established and process parameters are iteratively optimized, so that intelligent control of the knitting process from experience dependence to data driving is realized, and the quality consistency and the production efficiency of the braided fabric are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of composite material manufacturing equipment, and particularly to an automatic control system of a three-dimensional braiding machine. BACKGROUND

[0002] Three-dimensional braiding technology is an important process method for manufacturing high-performance composite material preforms, and is widely used in the fields of aerospace, national defense and military industry, etc.

[0003] The electromagnetic type step three-dimensional rotary braiding machine disclosed in Chinese patent CN117127311A controls the movement of the yarn carrier through an electromagnetic auxiliary system, which simplifies the mechanical structure, but still has the following defects: 1. Process parameter setting relies on human experience, lacks data-driven optimization mechanism, and leads to large fluctuations in product quality; 2. The motion control precision is insufficient, and it is difficult to realize high-precision reproduction of complex three-dimensional trajectories; 3. Lack of real-time quality monitoring and process self-learning ability, unable to adaptively optimize the production process; 4. The safety protection function is imperfect, and the equipment operation reliability is low.

[0004] Therefore, there is an urgent need for an intelligent control system that can automatically optimize process parameters, improve braiding precision and stability. SUMMARY

[0005] To this end, the present application provides an automatic control system of a three-dimensional braiding machine to overcome the technical problems of unstable braiding quality, insufficient processing precision, process optimization relying on human experience, and poor equipment safety and reliability due to the lack of an integrated intelligent control system in the prior art.

[0006] To achieve the above-mentioned purpose, the present application provides an automatic control system of a three-dimensional braiding machine, comprising: a client host for providing a CAD model file; a braiding machine upper computer in communication connection with the client host for analyzing the CAD model file and generating braiding control parameters, the braiding machine upper computer comprising a process database module; a touch screen for human-computer interaction; a PLC in communication connection with the braiding machine upper computer and the touch screen for receiving the braiding control parameters and issuing control instructions to a dual-shaft drive module and a traction robot based on the braiding control parameters, driving the dual-shaft drive module and the traction robot to work cooperatively to perform braiding actions; the dual-shaft drive module for responding to the execution control instructions to drive the yarn carrier to move to form a braiding structure; The traction robot is used to respond to the execution control command and pull the woven fabric to move along a predetermined trajectory; The IoT module is used to collect system production data in real time. Security monitoring components are used to detect operational anomalies and trigger protective actions; The parameter self-learning optimization module is used to establish a quality prediction model based on the historical weaving parameters and corresponding woven fabric quality data stored in the process database module, and to iteratively optimize the weaving process parameters based on the quality prediction model in order to improve the quality of the woven fabric.

[0007] Furthermore, the parameter self-learning optimization module continuously collects historical weaving parameters and corresponding woven fabric quality scores through the IoT module and stores them in the process database module. It compares the real-time quality data of the current production cycle with the preset quality target threshold, identifies process parameter combinations that fail to meet performance standards, and establishes a mapping model between weaving parameters and quality scores based on the historical weaving parameters using a regression analysis algorithm. The mapping model is then used to generate optimized new parameters for the process parameter combinations that fail to meet performance standards, and the optimized new parameters are applied to the next production cycle until the quality score reaches or exceeds the quality target threshold.

[0008] Furthermore, the traction robot includes: The robot control module is used to receive trajectory instructions from the host computer of the knitting machine via Ethernet; Multiple rotary axis servo drive units, controlled by the robot control module, are used to drive the multi-joint movement of the robot body; The guide rail servo drive unit, controlled by the robot control module, is used to control the precise axial positioning of the robot on the guide rail.

[0009] Furthermore, it also includes: a vibration motor, which is controlled by the PLC, for vibrating the carbon fiber bundles during the weaving process; A temperature and humidity control module, which communicates with the PLC, is used to monitor and adjust the temperature and humidity of the knitting area; The PLC is configured to dynamically adjust the vibration frequency of the vibration motor based on feedback from the temperature and humidity control module.

[0010] Furthermore, the safety monitoring component includes a safety light curtain and a wire breakage limit switch; Based on the detection of personnel or objects entering its protected area by the safety light curtain, or the detection of broken wire by the broken wire limit switch, a multi-level alarm protocol is triggered. The multi-level protocol comprises local warning by a signal tower and sending a fault report to a remote monitoring end through the IOT module.

[0011] Further, the dual-shaft drive module is connected with the PLC through an EtherCAT bus, and comprises: a control module configured to receive instructions from the PLC; a drive module controlled by the control module and configured to accurately and synchronously control two servo motors.

[0012] Further, the plurality of shaft servo drive units specifically comprises six sets of shaft servo drivers and corresponding six shaft servo motors, so as to realize the coordinated movement of the traction robot in six degrees of freedom.

[0013] Further, the traction robot further comprises: a signal tower connected with the PLC and configured to intuitively display the running state of the system through different colors of LED lights and their flashing modes.

[0014] Further, the PLC is configured to support two control modes: an automatic mode, in which the PLC automatically controls the entire system operation according to parameters received from the upper computer of the braiding machine; and a manual mode, in which the PLC responds to single-point operation instructions input through the touch screen.

[0015] Further, the regression analysis algorithm comprises a linear regression or a nonlinear regression algorithm.

[0016] Compared with the prior art, the present application has the following advantages: By introducing a parameter self-learning optimization module composed of linear regression and neural network, an accurate quantitative model between process parameters and product quality is established, thereby realizing a fundamental change from experience dependence to data driving of the braiding process, and significantly improving the quality consistency and good product rate of the product.

[0017] Further, by adopting a dual-shaft drive module based on FPGA and EtherCAT bus and a traction robot with six degrees of freedom full-closed loop servo drive, a motion control network with high precision and high synchronism is constructed, thereby realizing high-precision reproduction of complex three-dimensional braiding trajectories, and meeting the precise manufacturing requirements of high-performance special-shaped components in the fields of aerospace and the like.

[0018] Further, by integrating a temperature and humidity control module and a vibration motor, and realizing a coordinated control strategy based on real-time environmental feedback by the PLC, accurate and stable control of the carbon fiber tow processing environment is realized, and the compactness and resin infiltration uniformity inside the braided fabric are effectively guaranteed.

[0019] Further, by configuring a multi-level safety monitoring assembly composed of a safety curtain, a broken wire limit switch and a hard-wired safety circuit, and establishing an alarm protocol including local warning and remote reporting, a fault protection system conforming to international safety standards is constructed, ensuring long-period stable operation of the equipment and safety of the operating personnel.

[0020] Further, by fusing an IOT module, a process database and a PLC control system supporting automatic / manual dual mode, a digital production platform integrating real-time monitoring, data tracing, process optimization and flexible operation is constructed, providing a complete technical foundation for constructing an intelligent factory. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The connection structure block diagram of the automatic control system of the three-dimensional braiding machine of the embodiment of the application is shown in the figure. Figure 2 The step flow chart of the closed-loop work of the parameter self-learning optimization module of the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0022] In order to make the purpose and advantages of the application more clear and explicit, the application is further described below in combination with embodiments; it should be understood that the specific embodiments described herein are only used to explain the application, and do not limit the application.

[0023] The preferred embodiments of the application are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the application, and are not intended to limit the protection scope of the application.

[0024] It should be noted that, in the description of the application, the terms "upper", "lower", "left", "right", "inner", "outer" and the like indicate the direction or positional relationship terms based on the direction or positional relationship shown in the drawings, which are only for the convenience of description, and do not indicate or imply that the device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the application.

[0025] In addition, it should also be noted that, in the description of the application, unless otherwise explicitly specified and limited, the terms "mounting", "connection", "connection" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integral connection; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, or the internal communication of two elements. Those skilled in the art can understand the specific meaning of the above terms in the application according to the specific circumstances.

[0026] The system operation process of the application mainly includes: system initialization and feeding, automatic weaving execution, real-time data acquisition and process self-learning optimization, safety monitoring and abnormal handling, normal stop and discharging and the like.

[0027] Referring to Figure 1 as shown, Figure 1 The application is a three-dimensional weaving machine automatic control system.

[0028] A three-dimensional weaving machine automatic control system, comprising: A customer host computer for providing a CAD model file; A weaving machine host computer in communication connection with the customer host computer for analyzing the CAD model file and generating weaving control parameters, the weaving machine host computer comprising a process database module; A touch screen for human-computer interaction; A PLC in communication connection with the weaving machine host computer and the touch screen for receiving the weaving control parameters and issuing control instructions to a double-shaft driving module and a traction robot based on the weaving control parameters, so as to drive the double-shaft driving module and the traction robot to work cooperatively to execute weaving actions; The double-shaft driving module for driving the yarn carrier to move to form a weaving structure in response to the execution control instructions; The traction robot for moving the woven fabric according to a predetermined trajectory in response to the execution control instructions; An IOT module for collecting system production data in real time; A safety monitoring component for detecting operation abnormalities and triggering protection actions; A parameter self-learning optimization module for establishing a quality prediction model through a regression analysis algorithm based on historical weaving parameters and corresponding woven fabric quality data stored in the process database module, and iteratively optimizing weaving process parameters based on the quality prediction model to improve the quality of the woven fabric.

[0029] Referring to Figure 2 as shown, Figure 2 The application is a three-dimensional weaving machine automatic control system.

[0030] Specifically, the closed-loop working process of the parameter self-learning optimization module comprises: The parameter self-learning optimization module is the core of the system to realize intelligent and self-adaptive process control, and its operation follows a clear closed loop of "data acquisition-modeling analysis-optimization decision-application feedback", and the specific process is as follows: Step 1: Multi-source data aggregation Through the IOT module, real-time collection of process running parameters (such as weaving speed, fiber tension) from PLC, environmental parameters (temperature, humidity) from temperature and humidity control module, and state parameters (vibration frequency) from vibration motor, as well as motion state data from dual-axis drive module and traction robot. At the same time, the system obtains the corresponding batch of woven quality data (such as porosity, interlayer shear strength, surface roughness) through external detection equipment (such as industrial vision system, universal testing machine), and calculates the comprehensive quality score. All data are synchronously stored in the process database module to form a structured historical data set.

[0031] Step two: quality prediction model construction and update The parameter self-learning optimization module periodically (such as every N production batches) calls the historical data set in the process database. Using regression analysis algorithm, with multi-dimensional process parameter vector as input (X) and comprehensive quality score (Y) as output, the quality prediction model is established or updated. This model can use linear regression method as claimed in claim 10 to obtain the linear influence weight of each parameter on quality; or use nonlinear regression algorithm (such as feedforward neural network) to capture the complex interaction and nonlinear relationship between parameters. The model is trained through historical data and its prediction accuracy is evaluated by cross-validation method.

[0032] Step three: process parameter optimization and verification Before or during each production cycle, the module compares the current real-time collected process parameters with the preset quality target threshold. If the predicted or measured quality does not meet the standard, the module starts the optimization program: based on the quality prediction model, using optimization algorithms (such as gradient descent method, genetic algorithm) to search within the preset safe operation space of process parameters, and solve the new parameter combination that can maximize the predicted quality score (Y). To ensure the safety and effectiveness of the optimized parameters, the new parameter combination first enters the simulation verification or small batch trial production stage. After the trial production sample is detected to be qualified, the optimization parameter scheme is marked as available.

[0033] Step four: application of optimized parameters and closed-loop iteration The verified and effective optimized parameter combination is issued by the parameter self-learning optimization module through the weaving machine host computer to the PLC, updating the control set value for the next formal production cycle. The PLC controls the dual-axis drive module, traction robot, vibration motor and other actuators accordingly. The system enters a new production cycle and performs data collection again in step one. This cycle continues, forming a continuous "perception-decision-execution-learning" closed loop, so that the process parameters can be continuously self-adapted and optimized with the accumulation of production data, and finally drive the product quality to be stable and close to the target interval.

[0034] Through the above closed-loop process, the application specifically realizes the transition from relying on fixed experience parameters to dynamically optimizing parameters based on data models, which is the core technology embodiment of the fundamental transition of the weaving process from experience dependence to data-driven in the invention content.

[0035] Specifically, before weaving starts, the customer host sends the CAD model file to the weaving machine host computer, the host computer parses the model and generates weaving control parameters and transmits them to the PLC. The operator confirms the parameters through the touch screen and starts the automatic mode.

[0036] Specifically, the parameter self-learning optimization module continuously collects historical weaving parameters and corresponding quality scores of the woven fabric through the IOT module, and stores them in the process database module. The real-time quality data of the current production cycle is compared with the preset quality target threshold, and the process parameter combination whose performance does not meet the standard is identified. Based on the historical weaving parameters, a mapping model between the weaving parameters and the quality score is established using a regression analysis algorithm. The mapping model generates new optimized parameters for the process parameter combination whose performance does not meet the standard. The new optimized parameters are applied to the next production cycle until the quality score reaches or exceeds the quality target threshold.

[0037] In the embodiment of the application, the quality data refers to a set of physical parameters used to quantitatively evaluate the quality of the woven fabric, which includes at least one or more of the following: Structural integrity data: In this embodiment, it is the porosity and weaving angle error obtained through an industrial vision detection system; Mechanical property prediction data: In this embodiment, it is obtained through a combination of online monitoring and offline sampling testing. The offline testing measures the interlaminar shear strength through a universal testing machine, and correlates the modeling with the online monitoring process parameters; Appearance defect data: In this embodiment, it is the surface roughness obtained through a surface quality detector; The quality score is a comprehensive quantitative index calculated based on one or more quality data according to a preset weight algorithm.

[0038] The parameter self-learning optimization module continuously collects historical weaving parameters and corresponding quality scores of the woven fabric through the IOT module, and stores them in the process database module. The historical weaving parameters are a five-dimensional vector containing weaving speed (unit: mm / s), fiber tension (unit: N), environmental temperature (unit: ℃), environmental humidity (unit: %RH), and vibration frequency (unit: Hz). The quality score Y of the woven fabric is a comprehensive index between 0 and 100, which is calculated by weighting the strength score (50%), uniformity score (30%), and defect score (20%); The parameter self-learning optimization module compares the real-time quality data of the current production cycle with the preset quality target threshold. The quality target threshold is set to strength ≥ 50 MPa, uniformity ≥ 95%, and comprehensive quality score Y ≥ 90 points. Through the control chart method in statistical process control (SPC), the parameter self-learning optimization module identifies the process parameter combination whose quality data exceeds the lower control limit, and determines that the performance is not up to standard; Based on the historical weaving parameters, the parameter self-learning optimization module uses a regression analysis algorithm to establish a mapping model between the weaving parameters and the quality score. The linear regression model is as follows:

[0039] Where X1 to X5 represent the above five process parameters, β is the coefficient estimated by the least squares method, and ε is the error term. The nonlinear regression algorithm uses a feedforward neural network with one hidden layer, which contains 5 input nodes, the number of hidden layer neurons can be adjusted between 8 to 16 according to the amount of data and the complexity of the model, in this embodiment, the ReLU activation function is used, and 1 output node, in this embodiment, the linear activation function is used; Using the established mapping model, the parameter self-learning optimization module generates new parameters optimized for the process parameter combination whose performance is not up to standard; The optimization process uses the gradient descent method to maximize the quality score Y as the target, and iteratively searches within the safe operating range of the process parameters; The new parameters generated by optimization first enter the simulation verification or small batch trial production stage, and after the trial production quality score is verified, the parameter self-learning optimization module will apply the optimized new parameters to the next production cycle; This process continues to iterate until the quality score reaches or exceeds the quality target threshold.

[0040] It can be understood that the least squares method minimizes the sum of squared prediction errors by solving the normal equation to determine the optimal coefficients of the linear model. The neural network adjusts its connection weights based on historical data through the backpropagation algorithm to learn the complex nonlinear relationship between parameters and quality. This data-based modeling method enables the system to adaptively find the optimal process window.

[0041] Specifically, the parameter self-learning optimization module continuously collects system production data through the IOT module, and associates the quality data of the corresponding woven fabric obtained through external detection equipment.

[0042] Specifically, the traction robot includes: The robot control module is used to receive trajectory instructions from the weaving machine host computer through Ethernet; A plurality of rotation axis servo driving units controlled by the robot control module to drive the multi-joint movement of the robot body; A guide rail servo driving unit controlled by the robot control module to control the axial precise positioning of the robot on the guide rail.

[0043] In the embodiment, the traction robot comprises a robot control module based on an ARM Cortex-A series processor, six sets of rotation axis servo driving units and a set of guide rail servo driving unit; The robot control module receives trajectory instructions from the upper computer of the braiding machine through Ethernet, and in this embodiment, an industrial Ethernet based on TCP / IP protocol is adopted. The six sets of rotation axis servo driving units adopt servo drivers supporting EtherCAT communication protocol, and in this embodiment, Panasonic MINAS A6B-E series and the matching 1kW servo motor are adopted. The guide rail servo driving unit controls the axial positioning of the robot on the ball linear guide rail with a precision level of P5, and the positioning accuracy is controlled in full closed loop by the grating ruler installed on the guide rail.

[0044] It can be understood that the robot control module is the core computing unit of the traction robot, responsible for running kinematics solving and trajectory planning algorithms. The adopted servo driver supports high-resolution encoder feedback, ensuring the accuracy and synchronization of each joint movement.

[0045] Specifically, it also includes: a vibration motor controlled by the PLC to vibrate the carbon fiber tows during the braiding process; A temperature and humidity control module in communication with the PLC to monitor and adjust the temperature and humidity of the braiding area; The PLC is configured to dynamically adjust the vibration frequency of the vibration motor according to the feedback of the temperature and humidity control module.

[0046] In the embodiment, the system is configured with a vibration motor and a temperature and humidity control module. The vibration motor is controlled by the PLC through an analog output module, and the input control signal is 0-10V DC voltage, corresponding to a vibration frequency output range of 0-200Hz; The temperature and humidity control module includes a digital temperature and humidity sensor of model SHT35 and a temperature and humidity regulator based on the PID control algorithm; The PLC reads the sensor data of the temperature and humidity control module in real time, and has a pre-stored temperature and humidity-vibration frequency relationship mapping table determined through process experiments. The PLC queries the mapping table according to the real-time collected temperature and humidity data, and dynamically adjusts the vibration frequency of the vibration motor.

[0047] It can be understood that the cooperative control is based on a preset temperature and humidity-vibration frequency relationship table in the PLC, which is determined through preliminary process experiments, and aims to activate the most effective fiber dispersion mode for different environmental conditions.

[0048] Specifically, the safety monitoring assembly includes a safety curtain and a broken filament limit switch; Based on the result that the safety curtain detects that a person or object enters its protection area, or the broken filament limit switch detects a broken filament, a multi-level alarm protocol is triggered; The multi-level protocol includes local warning through a signal light tower, and sending a fault report to a remote monitoring end through an IOT Internet of Things module.

[0049] In the embodiment of the application, the safety monitoring assembly is composed of a safety curtain of model SICK deTec4 Core and a broken filament limit switch of Omron D4B-N series. The protection area height of the safety curtain is 2 meters, and the detection resolution is 14 mm. The trigger force of the broken filament limit switch is set to 5N.

[0050] When any safety device of the safety curtain or the broken filament limit switch is triggered, the system executes a multi-level alarm protocol: The first level is that the signal light tower outputs red flashing light (with a flashing frequency of 2 Hz); The second level is that a fault report conforming to the ISO 13374 standard is sent to the remote monitoring end through the IOT Internet of Things module; The third level is that within 100 milliseconds after triggering, the main power supply of all servo drives is cut off through the safety relay module of the PLC, and in the embodiment of the application, the PLC model adopts Siemens S7-1500 series.

[0051] It can be understood that the safety system follows the redundancy and fault safety principle, and even if a single element fails, the system can enter a safety state through a hard-wired safety circuit to ensure the safety of personnel and equipment; The LED lights of different colors at least include red, yellow and green, and other color groups can also be included according to needs.

[0052] Specifically, the dual-shaft drive module is connected with the PLC through an EtherCAT bus, and includes: A control module for receiving instructions from the PLC; A drive module controlled by the control module for accurately synchronously controlling two servo motors.

[0053] In the embodiment of the application, the dual-shaft drive module is connected with the PLC through an EtherCAT field bus, and in the embodiment, the IEC 61158 standard is followed. The dual-shaft driving module comprises a main control module, in this embodiment, an Xilinx Artix-7 series FPGA chip, and a driving module, in this embodiment, a model EL7221-9014; The main control module of the dual-shaft driving module performs high-precision interpolation operation to generate synchronous position instructions of two shafts. The driving module of the dual-shaft driving module controls two servo motors respectively, adopts an electronic gear synchronization mode, and the synchronization error is less than 1 encoder pulse.

[0054] It can be understood that the parallel processing capability of the FPGA chip in the dual-shaft driving module enables it to realize a complex multi-shaft synchronization algorithm and meet the real-time requirement of an EtherCAT communication cycle of 250 μs, which is a key to realize high-precision weaving.

[0055] Specifically, the plurality of shaft servo driving units specifically include six sets of shaft servo drivers and corresponding six shaft servo motors, to realize collaborative movement of the traction robot in six degrees of freedom.

[0056] In the embodiment of the application, the six sets of shaft servo driving units adopt three-phase permanent magnet synchronous servo motors with a rated power of 750 W and a rated speed of 3000 rpm. The driver of each set of shaft servo driving unit supports the STO (safe torque off) function and is equipped with a 23-bit multi-turn absolute value encoder. The motor of the shaft servo driving unit is connected to the joint through a harmonic reducer with a reduction ratio of 1:100, in this embodiment, the harmonic reducer is a model of HARMONIC DRIVE CSF-14-100-2A.

[0057] It can be understood that the high-resolution encoder provides accurate rotor position feedback and is the basis for realizing high-precision position closed-loop control. The harmonic reducer realizes zero-backlash transmission while ensuring high-torque output.

[0058] Specifically, it further comprises: The signal tower is connected with the PLC, to intuitively display the running state of the system through different colors of LED lights and their flashing modes.

[0059] In the embodiment of the application, the signal tower is connected with the PLC through a digital output module, and the signal tower comprises a red, yellow and green LED light group, and the display logic thereof is controlled by the PLC: The constant green light indicates normal operation of the system; The yellow light flashes at a frequency of 1 Hz, indicating that the device is in standby or debugging state; The red light flashes at a frequency of 2 Hz, indicating that a general fault is detected and the system is paused; The constant red light indicates an emergency stop state or a serious fault.

[0060] It can be understood that the display logic of the signal tower is directly driven by the state word in the PLC program, and different colors and flashing modes constitute an intuitive and internationally conventional on-site state indication system.

[0061] Specifically, the PLC is configured to support two control modes: Automatic mode, in which the PLC automatically controls the entire system operation according to parameters received from the knitting machine host computer; Manual mode, in which the PLC responds to single-point operation instructions input through the touch screen.

[0062] The PLC supports both automatic and manual control modes. In automatic mode, the PLC executes the process program downloaded from the knitting machine host computer, which complies with the PLCopen motion control specification. In manual mode, the operator inputs a jog command through the touch screen, and the PLC limits the speed of each axis to below 10% of the rated speed through an internal program. Mode switching is achieved through a mechanical key switch with three positions (automatic, zero, and manual). In this embodiment, the mechanical key switch is model SICK ATS240.

[0063] It can be understood that the mechanical key switch provides the highest level of mode switching authority control, preventing misoperation. The speed limiting function in the PLC is the core logic to ensure equipment and personnel safety in manual mode.

[0064] Specifically, the regression analysis algorithm includes a linear regression or a nonlinear regression algorithm.

[0065] The regression analysis algorithm used by the parameter self-learning optimization module includes both linear regression and nonlinear regression implementations. The linear regression model uses the least squares method to estimate the coefficient vector β by solving the matrix equation (XᵀX)β=XᵀY. The nonlinear regression model is a three-layer feedforward neural network with 5 nodes in the input layer, 8 neurons in the hidden layer using ReLU activation function, and 1 node in the output layer using linear activation function. The network uses the Adam optimizer and trains with mean squared error (MSE) as the loss function.

[0066] It can be understood that the least squares method provides a globally optimal linear fitting solution. The neural network, through its nonlinear activation function, can capture more complex interactions between process parameters and product quality, making it particularly suitable for handling high-dimensional and nonlinear process optimization problems. Embodiment

[0067] I. System composition and initialization After the system is started, each module performs self-checking in turn. The traction robot receives the feeding instruction, and its six sets of shaft servo drive units and guide rail servo drive units act in concert to move the robot to the feeding station. The operator installs the mandrel to the end effector of the traction robot. Then, the robot moves to the preparation station, and the operator fixes the carbon fiber tows to the head end of the mandrel.

[0068] II. Automatic weaving process The operator switches the system to automatic mode (see claim 9) through the touch screen and presses the start button. After receiving the start instruction, the PLC performs the following coordinated control: 1. Dual shaft drive module starts: The PLC sends instructions to the drive module of the dual shaft drive module through the EtherCAT bus, and the drive module controls two dial servo motors. Each motor drives the dial to move through the chain wheel, so that the yarn carrier runs on the dial according to the preset path to form a weaving structure. The drive module collects the torque feedback of each servo motor in real time, and takes one of them as the main shaft and the others as the slave shaft to realize electronic gear synchronous control, ensuring that the synchronization error of multiple motors is less than 1 encoder pulse.

[0069] 2. Vibration motor starts: The PLC controls the vibration frequency converter through the analog output module to start the vibration motor installed on both sides of the vibration ring, and vibrates the carbon fiber tows at a preset frequency to reduce the friction between the tows and prevent fuzz and net damage.

[0070] 3. Traction robot starts: The PLC sends trajectory instructions to the robot control module of the traction robot through Ethernet. After the control module analyzes the instructions, it drives six shaft servo motors through six sets of shaft servo drives to realize the coordinated movement of the robot in six degrees of freedom; at the same time, it controls the guide rail servo motor through the guide rail servo drive to make the robot move accurately along the preset trajectory on the linear guide rail and pull the woven fabric.

[0071] 4. Environmental coordinated control: The temperature and humidity control module monitors the environment data of the weaving area in real time and feeds back to the PLC. The PLC dynamically adjusts the vibration frequency of the vibration motor according to the built-in temperature and humidity-vibration frequency mapping table to optimize the fiber dispersion effect.

[0072] III. Data acquisition and process optimization The IOT module collects real-time production data (such as weaving speed, fiber tension, environmental temperature and humidity, vibration frequency, etc.) in the PLC through the Modbus TCP protocol and uploads it to the process database module. The parameter self-learning optimization module establishes a quality prediction model based on historical weaving parameters and corresponding quality scores using linear regression or neural network algorithms, continuously optimizes the process parameter combination until the quality score reaches the preset threshold.

[0073] IV. Normal stop and abnormal handling When the weaving progress reaches the set value, the PLC sends a stop command to each execution module, the system is paused, and the signal tower displays a constant yellow light, prompting the operator to unload.

[0074] If a broken filament occurs, the limit switch triggers a signal to the PLC, which immediately executes a multi-level alarm protocol: the signal tower flashes red (with a flashing frequency of 2 Hz), the IOT module sends an alarm message to the remote monitoring terminal, and the PLC cuts off the power supply to the servo driver.

[0075] If the safety curtain detects the intrusion of personnel or objects, the same alarm and shutdown process is triggered to ensure the safety of personnel and equipment.

[0076] V. Control mode support The system supports both automatic and manual modes. In manual mode, the operator can perform single-axis point operation through the touch screen, and the PLC limits the speed of each axis to below 10% of the rated speed to ensure safe operation.

[0077] Thus, the technical solutions of the present application have been described in conjunction with the preferred embodiments shown in the accompanying drawings, but it is readily understood by those skilled in the art that the scope of protection of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to related technical features without departing from the principles of the present application, and the technical solutions after such changes or replacements will fall within the scope of protection of the present application.

[0078] The above description is only the preferred embodiments of the present application and is not intended to limit the present application; for those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. An automatic control system for a three-dimensional braiding machine, characterized in that, The application relates to a weaving system, comprising: a client host configured to provide a CAD model file; a weaving machine host connected to the client host and configured to parse the CAD model file and generate weaving control parameters, the weaving machine host comprising a process database module; a touch screen configured to interact with a user; a PLC connected to the weaving machine host and the touch screen and configured to receive the weaving control parameters and send control instructions to a dual-axis drive module and a traction robot based on the weaving control parameters, so that the dual-axis drive module and the traction robot work cooperatively to perform a weaving action; the dual-axis drive module is configured to drive a carrier to move to form a weaving structure in response to the execution control instructions; the traction robot is configured to drive the woven fabric to move along a predetermined trajectory in response to the execution control instructions; an IOT module configured to collect system production data in real time; a safety monitoring component configured to detect operation abnormalities and trigger a protection action; a parameter self-learning optimization module configured to establish a quality prediction model through a regression analysis algorithm based on historical weaving parameters and corresponding weaving quality data stored in the process database module, and iteratively optimize weaving process parameters based on the quality prediction model to improve weaving quality.

2. The automatic control system for a three-dimensional braiding machine according to claim 1, wherein The parameter self-learning optimization module continuously collects historical weaving parameters and corresponding weaving quality scores through the IOT module, stores the parameters and scores in the process database module, compares real-time quality data of a current production cycle with a preset quality target threshold, identifies a process parameter combination that does not meet performance standards, and establishes a mapping model between weaving parameters and quality scores by using a regression analysis algorithm based on the historical weaving parameters, generates new optimized parameters for the process parameter combination that does not meet performance standards by using the mapping model, applies the new optimized parameters to a next production cycle, and continues until the quality score reaches or exceeds the quality target threshold.

3. The automatic control system for a three-dimensional braiding machine according to claim 1, wherein The traction robot comprises: a robot control module configured to receive trajectory instructions from the weaving machine host through Ethernet; a plurality of shaft servo drive units controlled by the robot control module and configured to drive multi-joint movement of a robot body; a guide rail servo drive unit controlled by the robot control module and configured to control axial precise positioning of the robot on a guide rail.

4. The automatic control system for a 3-D braiding machine according to claim 1, wherein Further comprising: a vibration motor controlled by the PLC and configured to vibrate carbon fiber tows during weaving; a temperature and humidity control module connected to the PLC and configured to monitor and adjust temperature and humidity of a weaving area; wherein the PLC is configured to dynamically adjust a vibration frequency of the vibration motor according to feedback of the temperature and humidity control module.

5. The automatic control system for three-dimensional braiding machines according to claim 1, characterized in that, The safety monitoring component comprises a safety light curtain and a broken filament limiting switch; a multi-level alarm protocol is triggered based on detection of personnel or objects entering a protection area of the safety light curtain or detection of a broken filament by the broken filament limiting switch; the multi-level protocol comprises local warning through a signal light tower and sending of a fault report to a remote monitoring end through the IOT module.

6. The automatic control system for three-dimensional braiding machines according to claim 1, characterized in that, The double-shaft drive module is connected with the PLC through an EtherCAT bus, and comprises: a control module configured to receive instructions from the PLC; a drive module controlled by the control module and configured to precisely and synchronously control two servo motors.

7. The automatic control system for a 3-D braiding machine according to claim 3, wherein The plurality of shaft servo drive units specifically comprises six sets of shaft servo drivers and corresponding six shaft servo motors, so as to realize the collaborative movement of the traction robot in six degrees of freedom.

8. The automatic control system for three-dimensional braiding machines according to claim 1, characterized in that, Further comprising: a signal tower connected with the PLC and configured to intuitively display the running state of the system through different colors of LED lights and their flashing modes.

9. The automatic control system for three-dimensional braiding machines according to claim 1, characterized in that, The PLC is configured to support two control modes: an automatic mode, in which the PLC automatically controls the entire system operation according to parameters received from the weaving machine host computer; a manual mode, in which the PLC responds to single-point operation instructions input through the touch screen.

10. The automatic control system for three-dimensional braiding machines according to claim 2, characterized in that, The regression analysis algorithm comprises a linear regression or a nonlinear regression algorithm.

Images