An electronic drafting system for fine yarn manufacturing in spinning frames

Abstract

The present invention belongs to the technical field of yarn manufacturing, and specifically refers to an electronic drafting system for a spinning frame for fine yarn manufacturing, including a data integration module, a twin model module, a real-time analysis module, an automatic control module, and an interactive interface. The present invention accurately reflects the dynamic characteristics of the physical system by establishing a digital twin model of the spinning frame drafting equipment, collects spinning frame operation data, trains the digital twin model, uses the trained digital twin model, and formulates corresponding control strategies according to the current system state and input. The actual state of the drafting equipment is monitored in real time through a virtual sensor, and compared with the predicted state of the digital twin model. According to the difference between the actual state and the predicted state, the control strategy is adjusted in real time to improve the accuracy and effectiveness of the predictive control.

Description

Technical Field

[0001] The invention belongs to the technical field of yarn manufacturing, and specifically refers to an electronic drafting system of a spinning frame for manufacturing fine yarn. Background Art

[0002] The drafting equipment of the spinning frame in the textile industry is an important link in realizing the transformation from coarse yarn to fine yarn. In the traditional spinning frame, the drafting equipment usually completes the drafting process through multiple gears meshing with each other and the difference in gear ratio. However, the current drafting equipment of the spinning frame has the following shortcomings:

[0003] (1) During the drafting process, the speed and tension cannot be accurately controlled, resulting in unsatisfactory straightness and parallelism of the fibers, affecting the quality of the yarn. The yarn is unevenly drafted in the length direction, resulting in uneven thickness and unstable strength of the yarn.

[0004] (2) Different types of fibers have different properties. The drafting equipment is not adaptable enough to some special fibers and cannot give full play to their performance. Changing the draft ratio or setting the production mode of different types of yarns requires a complex adjustment process, which consumes a lot of time and labor.

[0005] (3) During the drafting process, the lack of a monitoring system for the drafting equipment may lead to insensitivity to subtle changes in the yarn during the production process, making it impossible to make timely adjustments to optimize production;

[0006] (4) High-speed drafting equipment may generate loud noise and vibration, which may cause production defects. Summary of the invention

[0007] To address these problems, the present invention establishes a digital twin model of the spinning frame drafting equipment to accurately reflect the dynamic characteristics of the physical system of the drafting equipment, collects spinning frame operation data, trains the digital twin model, and uses the digital twin model to formulate corresponding control strategies based on the current system status and input. The actual status of the drafting equipment is monitored in real time through virtual sensors and compared with the predicted status by the digital twin model. According to the difference between the actual status and the predicted status, the control strategy is adjusted in real time to improve the accuracy and effectiveness of predictive control.

[0008] The system includes: a data integration module, a twin model module, a real-time analysis module, an automatic control module and an interactive interface;

[0009] The data integration module is connected to the twin model module, the twin model module is connected to the real-time analysis module, the real-time analysis module is connected to the automatic control module, the automatic control module feedback is connected to the real-time analysis module, and the interactive interface connects the data integration module, the twin model module, the real-time analysis module and the automatic control module.

[0010] The data integration module collects operation data in real time;

[0011] The twin model module constructs a digital twin model to reflect the actual state of the spinning frame drafting equipment;

[0012] The real-time analysis module performs performance analysis on the digital twin model to obtain analysis results;

[0013] The automatic control module realizes automatic control of the drafting process and optimizes the drafting ratio, tension and speed of the drafting equipment based on the analysis results of the real-time analysis module;

[0014] The interactive interface provides visualization of operation data and analysis results, and an operator can control the operation of the spinning frame drafting device through the interactive interface.

[0015] The data integration module collects operation data using a data integration method, and the data integration method includes the following steps:

[0016] Step D1: Collecting operation data;

[0017] Step D2: Clean, denoise and format the running data.

[0018] Further, step D1 includes the following steps:

[0019] Step D11: collecting the operating parameters of the spinning frame, including the speed and tension of the drafting device;

[0020] Step D12: collecting yarn material properties, including fiber type, strength and elasticity of the yarn;

[0021] Step D13: Collecting the operating status of the spinning frame, including vibration, noise and temperature.

[0022] The twin model module uses a digital twin model method to model the drafting equipment, and the digital twin model method includes the following steps:

[0023] Step M1: Use CREO software to build a geometric model, i.e. a three-dimensional model of the spinning frame drafting equipment;

[0024] Step M2: Establishing a physical model, including a mechanical model and a thermodynamic model of the drawing process, by using a finite element analysis method;

[0025] Step M3: establishing a control model of the drafting equipment, including a tension control model and a speed control model;

[0026] Step M4: Integrate the geometric model, physical model, and control model into the digital twin model;

[0027] Step M5: Update and adjust the digital twin model based on the operating data to ensure that the digital twin model is consistent with the drafting equipment.

[0028] Further, step M5 includes the following steps:

[0029] Step M51: Set different types of virtual sensors, and represent different types of virtual sensors as different types of nodes. Interconnected nodes are neighbor nodes. The temperature node attribute is XT, the vibration node attribute is XV, the noise node attribute is XN, the fiber type node attribute is XF, the strength node attribute is XE, and the elasticity node attribute is XS.

[0030] Step M52: using GNN to extract the operating data of each type of virtual sensor;

[0031] Step M53: Use GAT to calculate the attention weight of the node, initialize the feature vector of each node, and identify the importance of different types of nodes by aggregating the feature vectors;

[0032] Step M54: Use the virtual sensor function F(XT, XV, XN, XF, XE, XS, ST, VT)=Yt to analyze the tension and speed of the stretching device Yt at a given time t, where the tension is ST and the speed is VT.

[0033] Further, step M53 specifically includes the following steps:

[0034] Step M531: Initialize the feature vector of each node, assign an initial weight to each node, and obtain the initial attention coefficient between the node and its neighboring nodes;

[0035] Step M532: Take the feature vector of the node and the feature vector of the neighboring node as input and output a new attention coefficient. The formula used is as follows:

[0036] ;

[0037] in, and are the feature vectors of node i and neighbor node j, is a learnable parameter vector, || represents the feature vector concatenation operation, LReLU is the activation function, is the new attention coefficient of node i to neighbor node j;

[0038] Step M533: Use the softmax function to normalize the new attention coefficient to obtain the normalized attention coefficient. The formula used is as follows:

[0039] ;

[0040] in, represents the set of neighbor nodes of node i, is the normalized attention function, k is the current traversal node, exp() is the exponential function operation, is the exponential function operation of the new attention coefficient, is the new attention coefficient of node i for the currently traversed node k, It is the exponential function operation of the new attention coefficient of node i for the currently traversed node k;

[0041] Step M534: According to the normalized attention coefficient, perform weighted summation on the feature vectors of neighboring nodes to obtain a new feature vector of the node. The formula used is as follows:

[0042] ;

[0043] in, is the activation function, which performs a nonlinear transformation on the aggregated feature vector. is the new eigenvector;

[0044] Step M535: Repeat steps M532-M534, use a multi-layer GAT structure, perform aggregation operations on the feature vectors, and finally represent the node as a unified feature vector of the multi-layer virtual sensor.

[0045] The real-time analysis module uses a parameter vector updating method to perform error calibration operations, and the parameter vector updating method includes the following steps:

[0046] Step A1: Use the transfer function model to describe the relationship between the speed of the drafting device and the yarn speed, and the relationship between the tension control signal and the yarn tension. The formula used in the transfer function model is as follows:

[0047] G(s)=Y(s) / U(s);

[0048] Where s is the complex variable in the Laplace transform, G(s) is the transfer function, Y(s) is the Laplace transform of the output speed and tension control signal, and U(s) is the Laplace transform of the input speed and tension control signal;

[0049] Step A2: Analyze the poles and zeros of the transfer function model, record the stability log of the yarn speed of the drafting device, analyze the frequency response of the transfer function model, and record the response characteristics of the drafting device to tension control signals of different frequencies;

[0050] Step A3: Generate simulation context to correct the deviation between the digital twin model and the drafting equipment operation data;

[0051] Step A4: Use the online AMS calibration strategy based on the digital twin model to accelerate the optimization convergence of the AMS calibration strategy.

[0052] Further, step A3 includes the following steps:

[0053] Step A31: Use the parameterized function to set the target model. The formula used is as follows:

[0054] P = g(c|θ);

[0055] Where P represents the output of the target model, c is the feature vector, θ is the parameter vector of the target model, and g(c|θ) describes the mapping relationship from the feature vector and parameter vector to the output of the target model;

[0056] Step A32: Use the target model to generate a context loss value, set the time step, and update the parameter vector at each time step using the online gradient descent algorithm.

[0057] Further, step A32 includes the following steps:

[0058] Step A321: randomly select a node as the initial node, set the learning rate α, initialize the parameter vector, and calculate the gradient of the parameterized function at the initial node;

[0059] Step A322: moving the parameter vector in the opposite direction of the gradient to update the parameter vector;

[0060] Step A323: Minimize the context loss value of the simulated context, using the following formula:

[0061] ;

[0062] Where γ>0, γ is the time step set at time t, is the gradient obtained by the drafting device at time t, is the parameter vector at time t, is the parameter vector at time t+1;

[0063] Step A324: Set a threshold value for the change in the parameterized function, and repeat steps A321-A323 until the change in the parameterized function is less than the threshold value.

[0064] The automatic control module adopts a PLC control method to control the draft ratio, speed and tension, and the PLC control method includes the following steps:

[0065] Step P1: According to the twin model module, the yarn material characteristics are identified and the control parameters of the drafting device are obtained by the real-time analysis module to initialize the drafting device;

[0066] Step P2: Automatically adjust control parameters according to the online AMS calibration strategy used by the real-time analysis module;

[0067] Step P3: Control the servo motor of the drafting equipment through PLC to achieve precise lifting and lowering of the steel ring plate.

[0068] Further, step P1 includes the following steps:

[0069] Step P11: Analyze the yarn type by finite element analysis method, and predict the drafting process according to the yarn type;

[0070] Step P12: Analyze the yarn characteristics according to the yarn type and obtain the characteristic results;

[0071] Step P13: Analyze the characteristic results according to the transfer function model of the real-time analysis module to obtain the analysis results.

[0072] Further, step P3 includes the following steps:

[0073] Step P31: using a servo motor driver to receive analysis results;

[0074] Step P32: Install an encoder on the front roller shaft to provide speed and position feedback information;

[0075] Step P33: Use the reducer together with the servo motor driver to convert the high-speed rotation of the servo motor into the low-speed rotation of the roller, and realize the electronic drafting function through the encoder of the front roller and the servo motor driver of the rear roller;

[0076] Step P34: Set the actuator, which includes a cylinder and an electronic cam, to control the lifting and lowering of the steel collar plate and the rotation speed of the front, middle and rear rollers, and adjust the draft ratio, tension and speed.

[0077] The beneficial effects achieved by the present invention are as follows:

[0078] (1) The system simulates the drafting process through a digital twin model and analyzes the influence of control parameters on drafting efficiency and product quality through a real-time analysis module, thereby finding the optimal control parameters, improving production efficiency and product quality, simulating the drafting process under different operating data, predicting potential problems, and improving production stability;

[0079] (2) The system can simulate energy consumption under different operating conditions, find the best operating parameters, reduce yarn breakage and hairiness, improve yarn quality, reduce material waste and energy consumption, and save production costs;

[0080] (3) The digital twin model established by the system can simulate different production scenarios, identify different yarn types, adapt to different production speeds, and facilitate users to plan production, adjust production parameters, and adapt to different production needs;

[0081] (4) The digital twin model established by the system can simulate the drafting process of different textile product design schemes, quickly evaluate the pros and cons of the design schemes, and accelerate product development. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 A module diagram of an electronic drafting system for a spinning frame for manufacturing fine yarns proposed by the present invention;

[0083] Figure 2 This is an architecture diagram of an electronic drafting system for a spinning frame for fine yarn manufacturing proposed by the present invention. DETAILED DESCRIPTION

[0084] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0085] Example 1: Reference Figure 1 , this embodiment provides an electronic drafting system for a spinning frame for fine yarn manufacturing, the system comprising a data integration module, a twin model module, a real-time analysis module, an automatic control module and an interactive interface;

[0086] The data integration module is connected to the twin model module, the twin model module is connected to the real-time analysis module, the real-time analysis module is connected to the automatic control module, the automatic control module feedback is connected to the real-time analysis module, and the interactive interface connects the data integration module, the twin model module, the real-time analysis module and the automatic control module.

[0087] Embodiment 2: This embodiment is based on the above embodiment, and the data integration module uses a data integration method to collect operation data. The data integration method includes the following steps:

[0088] Step D1: Collecting operation data;

[0089] Step D2: Clean, denoise and format the running data.

[0090] Embodiment 3: This embodiment is based on the above embodiment, and step D1 includes the following steps:

[0091] Step D11: collecting the operating parameters of the spinning frame, including the drafting speed and tension, the drafting speed is 366 m / min, and the tension is 120 N;

[0092] Step D12: collecting yarn material properties, including the fiber type, strength and elasticity of the yarn. The fiber type is cotton and linen blended yarn, the strength is 12-20 cN / tex, and the elasticity is an elongation of 3.5%;

[0093] Step D13: Collect the running status of the spinning frame, including vibration, noise and temperature. The vibration is 3 mm / s, the noise is 73 decibels, and the temperature is 86 degrees Celsius.

[0094] Embodiment 4: This embodiment is based on the above embodiment. The twin model module adopts a digital twin model method to model the drafting equipment. The digital twin model method includes the following steps:

[0095] Step M1: Use CREO software to build a geometric model, i.e. a three-dimensional model of the spinning frame drafting equipment;

[0096] Step M2: Establishing a physical model, including a mechanical model and a thermodynamic model of the drawing process, by using a finite element analysis method;

[0097] Step M3: establishing a control model of the drafting equipment, including a tension control model and a speed control model;

[0098] Step M4: Integrate the geometric model, physical model, and control model into the digital twin model;

[0099] Step M5: Update and adjust the digital twin model based on the operating data to ensure that the digital twin model is consistent with the drafting equipment.

[0100] Embodiment 5: This embodiment is based on the above embodiment, and step M5 includes the following steps:

[0101] Step M51: Set different types of virtual sensors, and represent different types of virtual sensors as different types of nodes. Interconnected nodes are neighbor nodes. The temperature node attribute is XT, the vibration node attribute is XV, the noise node attribute is XN, the fiber type node attribute is XF, the strength node attribute is XE, and the elasticity node attribute is XS.

[0102] Step M52: using GNN to extract the operating data of each type of virtual sensor;

[0103] Step M53: Use GAT to calculate the attention weight of the node, initialize the feature vector of each node, and identify the importance of different types of nodes by aggregating the feature vectors;

[0104] Step M54: Use the virtual sensor function F(XT, XV, XN, XF, XE, XS, ST, VT)=Yt to analyze the tension and speed of the stretching device Yt at a given time t. The tension is ST, the speed is VT, the tension is 122N, and the speed is 366m / min.

[0105] Embodiment 6: This embodiment is based on the above embodiment, and step M53 specifically includes the following steps:

[0106] Step M531: Initialize the feature vector of each node, assign an initial weight to each node, and use a uniform distribution method to assign the same initial weight to all nodes. The initial weight is set to 0.5, and the initial attention coefficient between the node and its neighboring nodes is 0.3;

[0107] Step M532: Take the feature vector of the node [7,4,4] and the feature vector of the neighbor node [7,6,2] as input and output a new attention coefficient of 0.35. The formula used is as follows:

[0108] ;

[0109] in, and are the feature vectors of node i and neighbor node j, is a learnable parameter vector, || represents the feature vector concatenation operation, LReLU is the activation function, is the new attention coefficient;

[0110] Step M533: Use the softmax function to normalize the new attention coefficient to obtain a normalized attention coefficient of 0.33. The formula used is as follows:

[0111] ;

[0112] in, represents the set of neighbor nodes of node i, is the normalized attention function, k is the current traversal node, exp() is the exponential function operation, is the exponential function operation of the new attention coefficient, is the new attention coefficient of node i for the currently traversed node k, It is the exponential function operation of the new attention coefficient of node i for the currently traversed node k;

[0113] Step M534: According to the normalized attention coefficient, the feature vectors of the neighboring nodes are weighted summed to obtain a new feature vector of the node [7, 8, 4]. The formula used is as follows:

[0114] ;

[0115] in, is the activation function, which performs a nonlinear transformation on the aggregated feature vector. is the new eigenvector;

[0116] Step M535: Repeat steps M532-M534, use a three-layer GAT structure, and finally the node is represented as a unified feature vector of multi-layer virtual sensors [7, 3, 4].

[0117] Embodiment 7: This embodiment is based on the above embodiment. The real-time analysis module adopts a parameter vector updating method to perform error calibration operation. The parameter vector updating method includes the following steps:

[0118] Step A1: Use the transfer function model to describe the relationship between the roller speed of the drafting device and the yarn speed, and the relationship between the tension control signal and the yarn tension. The formula used in the transfer function model is as follows:

[0119] G(s)=Y(s) / U(s);

[0120] Where s is the complex variable in the Laplace transform, G(s) is the transfer function, Y(s) is the Laplace transform of the output speed and tension control signal, and U(s) is the Laplace transform of the input speed and tension control signal;

[0121] Step A2: Analyze the poles and zeros of the transfer function model and record the yarn speed stability log of the drafting equipment. The stability log includes:

[0122] (1) Specific time: 2024-07-10-11:00:00;

[0123] (2) Speed: 121m / min;

[0124] (3) Yarn fiber type: Batch001;

[0125] (4) Machine number: Machine005;

[0126] (5) Temperature: 25 degrees Celsius;

[0127] (6) Operator: Operator002;

[0128] (7) Record the speed fluctuation amplitude and frequency: the speed fluctuation is within ±5 m / min, and fluctuates twice per minute;

[0129] (8) Description of abnormal situation: Due to equipment failure, the speed suddenly dropped to 80m / min at 10:30;

[0130] Based on the stability log, the frequency response of the transfer function model is analyzed to record the response characteristics of the drafting equipment to tension control signals of different frequencies;

[0131] Step A3: Generate simulation context to correct the deviation between the digital twin model and the drafting equipment operation data;

[0132] Step A4: Use the online AMS calibration strategy based on the digital twin model and use the operating data to accelerate the optimization convergence of the AMS calibration strategy.

[0133] Embodiment 8: This embodiment is based on the above embodiment, and step A3 includes the following steps:

[0134] Step A31: Use the parameterized function to set the target model. The formula used is as follows:

[0135] P = g(c|θ);

[0136] Where P represents the output of the target model, c is the feature vector, θ is the parameter vector of the target model, and g(c|θ) describes the mapping relationship from the feature vector and parameter vector to the output of the target model;

[0137] Step A32: Use the target model to generate a context loss value, set the time step, and update the parameter vector at each time step using the online gradient descent algorithm.

[0138] Embodiment 9: This embodiment is based on the above embodiment, and step A32 includes the following steps:

[0139] Step A321: Randomly select a node as the initial node, set the learning rate α=0.001, initialize the parameter vector variance to 2 / n_in, where n_in is the input dimension, and calculate the gradient of the parameterized function at the initial node, with a gradient of 19;

[0140] Step A322: moving the parameter vector in the opposite direction of the gradient to update the parameter vector;

[0141] Step A323: Minimize the context loss value of the simulated context, using the following formula:

[0142] ;

[0143] Where γ>0, γ is the time step set at time t, is the gradient obtained by the drafting device at time t, is the parameter vector at time t, is the parameter vector at time t+1;

[0144] Step A324: Set the threshold of the parameterized function change to 0.005, and repeat steps A321-A323 until the parameterized function change is less than the threshold.

[0145] Embodiment 10: This embodiment is based on the above embodiment. The automatic control module adopts a PLC control method to control the draft ratio, speed and tension. The PLC control method includes the following steps:

[0146] Step P1: According to the twin model module, the yarn material characteristics are identified as cotton and linen blended yarn, and the control parameters of the drafting device are obtained by the real-time analysis module to initialize the drafting device;

[0147] Step P2: Automatically adjust control parameters according to the online AMS calibration strategy of the real-time analysis module;

[0148] Step P3: Control the servo motor through PLC to control the precise lifting and lowering of the steel collar plate.

[0149] Embodiment 11: This embodiment is based on the above embodiment, and step P1 includes the following steps:

[0150] Step P11: Analyze the yarn type by finite element analysis method, and predict the drafting process according to the yarn type. During the drafting process in the rear zone, the cross-sectional shape of the roving changes from a circle to a flat ribbon, and then returns to a quasi-circular shape. The factors affecting the drafting of the spun yarn are the friction force boundary, drafting force, control force and speed change point distribution;

[0151] Step P12: Analyze the yarn characteristics according to the yarn type and obtain the characteristic results. The roving weight is 45 g / m, and the twist is 4.01 and 4.48 twists / m;

[0152] Step P13: Analyze the characteristic results according to the real-time analysis module to obtain the analysis results. The analysis results show that the additional friction generated by the twisting of the roving effectively controls the fiber movement, and the fiber speed change points are concentrated, which is conducive to the stability of the drafting process.

[0153] Embodiment 12: This embodiment is based on the above embodiment, and step P3 includes the following steps:

[0154] Step P31: using a servo motor driver to receive the analysis results and control the movements of various components of the spinning frame drafting equipment: the drafting multiple of the spinning rear zone is 1.098, the center distance of the spinning rear zone is 70 mm, and the spinning distance block is 2.0 mm;

[0155] Step P32: Install an encoder on the front roller shaft to provide speed and position feedback information;

[0156] Step P33: Use the reducer together with the servo motor driver to convert the high-speed rotation of the servo motor into the low-speed rotation of the roller, and realize the electronic drafting function through the encoder of the front roller and the servo motor driver of the rear roller;

[0157] Step P34: Set the actuator, which includes a cylinder and an electronic cam, to control the lifting and lowering of the steel collar plate and the rotation speed of the front, middle and rear rollers, and adjust the draft ratio to 1.079, the tension to 123N and the speed to 375m / min.

[0158] The present invention and its implementation methods are described above. Such description is not restrictive. If a person skilled in the art is inspired by it and creatively designs structures and implementation methods similar to the technical solution without departing from the purpose of the present invention, they should all fall within the protection scope of the present invention.

Claims

1. An electronic drafting system for a spinning frame for fine yarn manufacturing, characterized in that: The system includes data integration module, twin model module, real-time analysis module, automatic control module and interactive interface; The data integration module is connected to the twin model module, the twin model module is connected to the real-time analysis module, the real-time analysis module is connected to the automatic control module, the automatic control module feedback is connected to the real-time analysis module, and the interactive interface is connected to the data integration module, the twin model module, the real-time analysis module and the automatic control module; The data integration module collects operation data using a data integration method, and the data integration method includes the following steps: Step D1: Collecting operation data; Step D2: cleaning, denoising and formatting the running data; Step D1 includes the following steps: Step D11: collecting spinning frame operating parameters, including speed and tension; Step D12: collecting yarn material properties, including fiber type, strength and elasticity of the yarn; Step D13: collecting the operating status of the spinning frame, including vibration, noise and temperature; The twin model module uses a digital twin model method to model the drafting equipment, and the digital twin model method includes the following steps: Step M1: Establishing the geometric model of the drafting equipment; Step M2: Establishing a physical model of the drafting equipment; Step M3: Establishing a control model for the drafting equipment; Step M4: Integrate the geometric model, physical model, and control model into the digital twin model; Step M5: Update and adjust the digital twin model based on the operating data; Step M5 includes the following steps: Step M51: setting different types of virtual sensors, and representing different types of virtual sensors as different types of nodes; Step M52: using GNN to extract operation data of each type of virtual sensor; Step M53: Use GAT to calculate the attention weight of the node and identify the importance of different types of nodes; Step M54: Analyze the tension and speed of the drafting device at a given moment.

2. The electronic drafting system for a spinning frame for fine yarn manufacturing according to claim 1 is characterized in that: The real-time analysis module uses a parameter vector updating method to perform error calibration operations, and the parameter vector updating method includes the following steps: Step A1: using a transfer function model to describe the relationship between the speed of the drafting device and the yarn speed, and the relationship between the tension control signal and the tension; Step A2: Analyze the poles and zeros of the transfer function model, record the speed stability log of the drafting equipment, analyze the frequency response of the transfer function model, and record the response characteristics of the drafting equipment to tension control signals of different frequencies; Step A3: Transmit the operating data through the data integration module to correct the deviation between the digital twin model and the operating data of the drafting equipment; Step A4: Use the online AMS calibration strategy based on the digital twin model to accelerate the optimization convergence of the AMS mapping using the operational data; Step A3 includes the following steps: Step A31: Use the parameterized function to set the target model; Step A32: Generate a simulation context using the target model, set the time step, and update the parameter vector using the online gradient descent algorithm at each time step; Step A32 includes the following steps: Step A321: randomly select a node as the initial starting point, initialize the parameter vector, and calculate the gradient of the parameterized function at the current node; Step A322: moving the parameter vector in the opposite direction of the gradient to update the parameter vector; Step A323: Minimize the context loss value of the simulated context; Step A324: Set a threshold value for the change in the parameterized function, and repeat steps A321-A323 until the change in the parameterized function is less than the threshold value.

3. The electronic drafting system for a spinning frame for fine yarn manufacturing according to claim 2 is characterized in that: The automatic control module adopts a PLC control method to control the draft ratio, speed and tension, and the PLC control method includes the following steps: Step P1: According to the twin model module, the yarn material characteristics are identified, and according to the real-time analysis module, the simulation context is obtained and the control parameters of the drafting device are initialized; Step P2: Automatically adjust control parameters according to the online AMS calibration strategy of the real-time analysis module; Step P3: Control the servo motor through PLC to achieve precise lifting and lowering of the steel ring plate and optimize the draft ratio, tension and speed of the drafting equipment.

Images