A unidirectional tension-coordinated control system and method for acrylic filament feeding

Abstract

This invention relates to the field of programmable control technology, specifically to a unidirectional tension collaborative control system and method for acrylic filament. The collaborative control of tension during the molecular orientation stretching process of acrylic filament along a unidirectional path through multiple sets of rollers includes: acquiring first data at key positions along the unidirectional feeding path and the running speeds of the multiple sets of rollers; constructing a dynamic characteristic control model to analyze the first data and running speeds, generating preliminary control commands; compensating and adjusting the stretching prediction values ​​based on control deviations; adjusting the feeding process based on the preliminary control commands to obtain second data and a second running speed; constructing a distributed collaborative control model to analyze the preliminary control commands, generating a control adjustment strategy based on the global tension distribution balance and the stability of the tension gradient in key sections; and controlling the second running speed based on the control adjustment strategy using low-latency information interaction to obtain a globally optimized and coordinated speed control command.

Description

Technical Field

[0001] This invention relates to the field of program control technology, specifically to a unidirectional tension-coordinated control system and method for acrylic filament feeding. Background Technology

[0002] In continuous feeding processes, precise parameter control is crucial for ensuring processing quality, especially in unidirectional feeding applications. Unidirectional feeding processes require continuous material supply in a single direction and are widely used in stretching, winding, and other processing steps. Maintaining parameter stability is essential to prevent performance inconsistencies. However, the sensitivity of materials to external stress and their high elongation characteristics make precise control of processing parameters a technical challenge. Improper parameter settings, such as being too high or too low, can negatively impact material performance and processing results. In multi-stage or complex feeding paths, traditional single-control-point methods struggle to coordinate parameters across stages, leading to insufficient processing consistency. Furthermore, the processing demands a high system response speed, requiring real-time monitoring and dynamic adjustment of relevant parameters to cope with process changes. Existing technologies sometimes rely on a single feedback mechanism for parameter control, which is acceptable in basic applications but has limitations in achieving multi-point collaborative control and rapid response. This control method is ineffective in complex processing scenarios, especially in processes with extremely high requirements for continuity and parameter stability, leading to frequent quality problems. Existing technologies lack effective multi-point collaborative control mechanisms, which cannot meet the high precision and high stability requirements of modern processing technology.

[0003] To address this, a unidirectional tension control system and method for acrylic filament feeding is proposed. Summary of the Invention

[0004] The present invention aims to provide a unidirectional feeding tension coordinated control system and method for acrylic filaments. The coordinated control of feeding tension during the molecular orientation stretching process of acrylic filaments along a unidirectional path through multiple sets of rollers includes: acquiring first data at key positions along the unidirectional feeding path and the running speeds of the multiple sets of rollers; constructing a dynamic characteristic control model to analyze the first data and running speeds, generating preliminary control commands; compensating and adjusting the stretching prediction value based on control deviations; adjusting the feeding process based on the preliminary control commands to obtain second data and a second running speed; constructing a distributed coordinated control model to analyze the preliminary control commands, generating a control adjustment strategy based on the global tension distribution balance and the stability of the tension gradient in key sections; and controlling the second running speed based on the control adjustment strategy using low-latency information interaction to obtain a globally optimized and coordinated speed control command.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A unidirectional feeding tension coordinated control system for acrylic filaments, comprising the following aspects:

[0007] The data acquisition unit acquires the first data of key positions in the unidirectional wire feeding path and the running speed of multiple sets of rollers in real time;

[0008] The initial control command acquisition unit constructs a dynamic characteristic control model to analyze the first data and running speed, and generates preliminary control commands. The preliminary control commands include control deviation, tensile prediction value, and first running speed. The tensile prediction value is compensated and adjusted based on the control deviation. The wire feeding process is adjusted based on the preliminary control commands to obtain second data and second running speed.

[0009] The control adjustment strategy generation unit constructs a distributed collaborative control model to analyze the second data and the second running speed, and generates a control adjustment strategy based on the global tension distribution balance and the key segment tension gradient stability.

[0010] The global optimization and coordination instruction generation unit, based on low-latency information interaction between multiple sets of rollers, controls the second running speed according to the control adjustment strategy to obtain a globally optimized and coordinated speed control instruction.

[0011] Preferably, a tension sensor array is arranged along the acrylic filament feeding path. The tension sensor array is set at the entrance section, tension change section, stretching roller group and winding pre-tensioning section of the feeding path to collect real-time data of each section of acrylic filament. Based on the real-time data, the tension amplitude change rate is obtained, and the key position of the unidirectional feeding path is determined. Specifically, the tension fluctuation amplitude difference rate between two adjacent sensors is calculated. When the tension fluctuation amplitude difference rate exceeds a preset threshold, it is determined as a key position, and the tension data collected by the sensor corresponding to the key position is used as the first data.

[0012] Preferably, the dynamic feature control model includes a data preprocessing layer, a feature extraction layer, a predictive modeling layer, and a control decision layer:

[0013] The data preprocessing layer performs normalization processing on the first data and the running speed, and detects and corrects outliers.

[0014] The feature extraction layer uses wavelet transform decomposition to extract time-frequency domain features from the processed first data, and then uses principal component analysis to reduce the dimension of the data to generate a dimension-reduced feature vector.

[0015] The predictive modeling layer generates tensile prediction values ​​by analyzing the dimensionality-reduced feature vectors. The tensile prediction values ​​include the tension prediction values ​​at each key location.

[0016] The control decision layer calculates the control deviation by comparing the deviation between the actual tension and the predicted tension value, and compensates and adjusts the predicted tension value. Based on the adjusted predicted tension value, it obtains the first operating speed using a tension-velocity nonlinear mapping model and generates initial control commands.

[0017] Preferably, the tension-speed nonlinear mapping model is a dual-input single-output nonlinear regression model constructed based on historical production data. The input variables are the adjusted tension prediction value and the current roller operating speed, and the output variable is the speed adjustment amount. The operating speed is adjusted by the speed adjustment amount to obtain the first operating speed.

[0018] Preferably, the distributed cooperative control model includes a state evaluation layer, a local policy generation layer, and a global cooperative optimization layer:

[0019] The state assessment layer receives preliminary control commands, second data, and second operating speed, and calculates the global tension distribution balance and the key segment tension gradient stability.

[0020] The local strategy generation layer designs local control and adjustment strategies for different tension distribution states, including steady-state control strategies, transient control strategies, and fluctuation suppression strategies; the control and adjustment strategies include control parameters, including velocity adjustment step size coefficient, tension compensation coefficient, gradient smoothing factor, and cooperative weight coefficient.

[0021] The global collaborative optimization layer uses the analytic hierarchy process (AHP) to determine the weights of the control parameters and combines it with the particle swarm optimization algorithm for dynamic optimization. Based on the local control adjustment strategies of different rollers, it selects the optimal control parameters and generates a control adjustment strategy.

[0022] Preferably, the specific process for obtaining globally optimized and coordinated speed control commands includes:

[0023] Based on the second operating speed and control adjustment strategy, a multi-objective optimization function is constructed based on the global tension distribution balance, the key segment tension gradient stability and the minimization of speed change.

[0024] An improved differential evolution algorithm is used to iteratively solve the multi-objective optimization function. At the same time, during the iteration process, the Lagrange constraint method is used to handle the physical constraints of the system, including the upper and lower limits of the speed of each roller, the speed ratio of adjacent rollers, and the speed change rate constraint, to generate the optimal speed vector.

[0025] Based on the optimal velocity vector, a globally optimized and coordinated velocity control command is obtained.

[0026] A method for coordinated control of unidirectional feeding tension of acrylic filament includes the following:

[0027] Real-time acquisition of first-hand data at key locations along the unidirectional yarn feeding path and the running speed of multiple sets of rollers;

[0028] A dynamic feature control model is constructed to analyze the first data and the running speed, and generate preliminary control instructions. The preliminary control instructions include control deviation, tensile prediction value, and first running speed. The tensile prediction value is compensated and adjusted according to the control deviation. The first running speed is obtained by adjusting the running speed according to the tensile prediction value. The wire feeding process is adjusted according to the preliminary control instructions to obtain second data and second running speed.

[0029] A distributed collaborative control model is constructed to analyze the second data and the second operating speed. Based on the global tension distribution balance and the stability of the tension gradient in the key segment, a control adjustment strategy is generated.

[0030] Based on low-latency information interaction between multiple sets of rollers, the second operating speed is controlled according to the control adjustment strategy to obtain a globally optimized and coordinated speed control command.

[0031] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0032] 1. This invention constructs a dynamic feature control model to perform real-time preprocessing, time-frequency domain feature extraction, and principal component dimensionality reduction on tension data and roller speed at key locations along the unidirectional yarn feeding path. By generating predicted tension values, compensation adjustments are implemented at the control decision level based on the deviation between actual and predicted tension. It can capture sudden tension changes and subtle fluctuations in real time, and calculate speed adjustment amounts using a tension-speed nonlinear mapping model, achieving targeted speed control for each group of rollers. This not only makes the tension distribution of acrylic filaments more uniform during molecular orientation stretching but also significantly reduces defects such as yarn breakage, end breakage, and inconsistent fineness caused by tension fluctuations, improving the yield rate of spinning yarn and batch-to-batch consistency.

[0033] 2. This invention addresses the tension coordination problem involving complex coupling among multiple sets of rollers by proposing a distributed cooperative control model, comprising a state evaluation layer, a local strategy generation layer, and a global cooperative optimization layer. A tension sensor array is deployed at multiple points along the path entrance, tension abrupt change section, stretching roller group, and winding pre-tensioning section to collect second data and second operating speed in real time, and calculate tension distribution uniformity and gradient stability indices to evaluate the state of each key section. In the local strategy generation layer, strategies are designed based on three typical operating conditions: steady state, transient state, and fluctuation suppression. At the global level, the analytic hierarchy process (AHP) is used to determine the weights of control parameters, and then combined with particle swarm optimization (PSO) algorithm for dynamic optimization to generate the optimal control adjustment strategy. This effectively reduces the information interaction delay and control conflicts between nodes, ensuring the stability and scalability of the system under high-speed operation.

[0034] 3. In the speed control command generation stage, this invention constructs a multi-objective optimization function that balances global tension balance, tension gradient stability, and speed variation minimization based on a second operating speed and a distributed control strategy, and iteratively solves the function using a differential evolution algorithm. During the calculation process, the system's physical constraints are embedded using the Lagrangian constraint method, satisfying process quality requirements while minimizing equipment energy consumption and mechanical wear. This enables a balance between energy consumption and efficiency optimization while ensuring the quality of acrylic filament molecular orientation stretching, thereby improving the overall energy efficiency and economic benefits of the production line. Attached Figure Description

[0035] Figure 1 A schematic diagram of a unidirectional tension-coordinated control system for acrylic filament feeding is provided for this invention.

[0036] Figure 2 A schematic diagram of a method for coordinated control of unidirectional feeding tension of acrylic filament provided by the present invention;

[0037] Figure 3 A schematic diagram of the dynamic feature control model structure provided by the present invention;

[0038] Figure 4 This is a schematic diagram of the distributed collaborative control model structure provided by the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Example 1

[0041] Please see Figures 1 to 4 This invention provides a unidirectional tension-coordinated control system for acrylic filament feeding, the overall structure of which is as follows: Figure 1 As shown. Technical solution reference. Figure 2 Specifically, the coordinated control of the feeding tension of acrylic filaments during the molecular orientation stretching process through multiple sets of rollers along a unidirectional path includes:

[0042] Real-time acquisition of first-hand data at key locations along the unidirectional yarn feeding path and the running speed of multiple sets of rollers;

[0043] Furthermore, a tension sensor array is arranged along the acrylic filament feeding path. The tension sensor array is set at the entrance section, tension change section, stretching roller group and winding pre-tensioning section of the feeding path to collect real-time data of each section of acrylic filament. Based on the real-time data, the tension amplitude change rate is obtained, and the key position of the unidirectional feeding path is determined. Specifically, the tension fluctuation amplitude difference rate between two adjacent sensors is calculated. When the tension fluctuation amplitude difference rate exceeds a preset threshold, it is determined as a key position, and the tension data collected by the sensor corresponding to the key position is used as the first data.

[0044] In this embodiment, the present invention deploys an array of tension sensors along the feeding path in the inlet section, tension change section, stretching roller group and winding pre-tightening section, and combines the tension amplitude change rate analysis method to locate key positions in real time. This enables precise monitoring and rapid closed-loop adjustment of the feeding tension of acrylic filaments, effectively suppressing tension fluctuations and oscillations, significantly reducing the breakage rate, and improving fiber stretching uniformity and mechanical properties.

[0045] A dynamic feature control model is constructed to analyze the first data and the running speed, and generate preliminary control instructions. The preliminary control instructions include control deviation, tensile prediction value, and first running speed. The tensile prediction value is compensated and adjusted according to the control deviation. The first running speed is obtained by adjusting the running speed according to the tensile prediction value. The wire feeding process is adjusted according to the preliminary control instructions to obtain second data and second running speed.

[0046] Furthermore, the dynamic feature control model refers to Figure 3 It includes a data preprocessing layer, a feature extraction layer, a predictive modeling layer, and a control decision layer:

[0047] The data preprocessing layer performs normalization processing on the first data and the running speed, and detects and corrects outliers.

[0048] The feature extraction layer uses wavelet transform decomposition to extract time-frequency domain features from the processed first data, and then uses principal component analysis to reduce the dimension of the data to generate a dimension-reduced feature vector.

[0049] The predictive modeling layer generates tensile prediction values ​​by analyzing the dimensionality-reduced feature vectors. The tensile prediction values ​​include the tension prediction values ​​at each key location.

[0050] The control decision layer calculates the control deviation by comparing the deviation between the actual tension and the predicted tension value, and compensates and adjusts the predicted tension value. Based on the adjusted predicted tension value, it obtains the first operating speed using a tension-velocity nonlinear mapping model and generates initial control commands.

[0051] In this embodiment, the present invention constructs a dynamic feature control model to perform in-depth analysis and prediction of real-time tension data and roller speed information, which can generate more accurate initial control commands and adaptively compensate and adjust during operation. This enables forward-looking prediction and closed-loop optimization of the tensile tension of acrylic filaments, effectively suppressing tension fluctuations and oscillations, reducing filament breakage and rework rates, improving fiber tensile uniformity and mechanical properties, and simultaneously achieving intelligent system operation and energy consumption optimization, significantly improving production efficiency and product quality.

[0052] Furthermore, the tension-speed nonlinear mapping model is constructed based on historical production data to create a dual-input single-output nonlinear regression model. The input variables are the adjusted tension prediction value and the current roller operating speed, and the output variable is the speed adjustment amount. The operating speed is adjusted by the speed adjustment amount to obtain the first operating speed.

[0053] In this embodiment, the tension-speed nonlinear mapping model of the present invention constructs a dual-input single-output regression model based on historical production data. The compensated tension prediction value and the current roller speed are used as inputs. The roller speed is finely adjusted by the predicted speed adjustment amount, so as to achieve precise control of tension demand in a fast response. This significantly suppresses oscillation, reduces the breakage rate and rework rate, improves the uniformity of filament stretching and mechanical properties, and optimizes energy consumption and production efficiency.

[0054] A distributed collaborative control model is constructed to analyze the second data and the second operating speed. Based on the global tension distribution balance and the stability of the tension gradient in the key segment, a control adjustment strategy is generated.

[0055] Furthermore, the distributed cooperative control model refers to Figure 4 It includes a state evaluation layer, a local policy generation layer, and a global collaborative optimization layer:

[0056] The state assessment layer receives preliminary control commands, second data, and second operating speed, and calculates the global tension distribution balance and the key segment tension gradient stability.

[0057] The local strategy generation layer designs local control and adjustment strategies for different tension distribution states, including steady-state control strategies, transient control strategies, and fluctuation suppression strategies; the control and adjustment strategies include control parameters, including velocity adjustment step size coefficient, tension compensation coefficient, gradient smoothing factor, and cooperative weight coefficient.

[0058] The global collaborative optimization layer uses the analytic hierarchy process (AHP) to determine the weights of the control parameters and combines it with the particle swarm optimization algorithm for dynamic optimization. Based on the local control adjustment strategies of different rollers, it selects the optimal control parameters and generates a control adjustment strategy.

[0059] In this embodiment, the constructed distributed collaborative control model quantifies the real-time tension distribution balance and tension gradient stability through a state evaluation layer. Combined with steady-state, transient, and fluctuation suppression control schemes in the local strategy generation layer, and utilizing the analytic hierarchy process (AHP) and particle swarm optimization algorithm in the global collaborative optimization layer to dynamically determine the speed adjustment step size, tension compensation coefficient, gradient smoothing factor, and collaborative weight coefficient, it achieves the organic synergy of global tension distribution balance and local gradient stability among multi-roller groups. This not only adapts to tension disturbances under different production conditions and quickly suppresses oscillations and abrupt changes, but also enables online optimization and updating of optimal control parameters across the entire line, further reducing filament breakage and rework rates, improving filament stretching uniformity, mechanical properties, and production yield, and significantly enhancing production stability and economic benefits.

[0060] Based on low-latency information interaction between multiple sets of rollers, the second running speed is controlled according to the control adjustment strategy to obtain a globally optimized and coordinated speed control command.

[0061] Furthermore, the specific process of obtaining globally optimized and coordinated speed control commands includes:

[0062] Based on the second operating speed and control adjustment strategy, a multi-objective optimization function is constructed based on the global tension distribution balance, the key segment tension gradient stability and the minimization of speed change.

[0063] An improved differential evolution algorithm is used to iteratively solve the multi-objective optimization function. At the same time, during the iteration process, the Lagrange constraint method is used to handle the physical constraints of the system, including the upper and lower limits of the speed of each roller, the speed ratio of adjacent rollers, and the speed change rate constraint, to generate the optimal speed vector.

[0064] Based on the optimal velocity vector, a globally optimized and coordinated velocity control command is obtained.

[0065] In this embodiment, by utilizing low-latency information interaction between multiple sets of rollers and an improved differential evolution algorithm, multi-objective optimization is achieved, taking into account tension balance, gradient stability, and minimization of speed variation. In addition, Lagrangian constraints are used to constrain the upper and lower limits of roller speed, speed ratio, and rate of change in real time. This enables the generation of globally optimal coordinated speed control commands, achieving rapid suppression and uniform distribution of tension fluctuations, and significantly improving production stability and fiber quality.

[0066] This invention achieves refined tension monitoring by deploying an array of tension sensors at key locations along the acrylic filament feeding path and combining this with real-time positioning based on the rate of change of tension amplitude. A dynamic feature control model incorporating data preprocessing, wavelet decomposition, and principal component analysis is introduced to perform in-depth analysis and prediction of tension and roll speed information, generating adaptive compensation initial control commands. A tension-speed nonlinear mapping model based on historical data enables forward-looking fine-tuning of the roll speed. Furthermore, a distributed collaborative control model dynamically coordinates local strategies and global optimization, employing an improved differential evolution algorithm and Lagrange-constrained multi-objective optimization to generate globally optimal speed control commands, balancing tension distribution, gradient stability, and minimizing speed variations. This system can suppress tension fluctuations and oscillations in real-time with low latency, significantly reducing filament breakage and rework rates, improving filament stretching uniformity and mechanical properties, and enhancing production line stability, yield, and economic efficiency.

[0067] Example 2

[0068] This system is applied to the process of stretching acrylic filaments along a unidirectional path through multiple sets of rollers to achieve coordinated control of the feeding tension. Specifically, it includes: a data acquisition unit, an initial control command acquisition unit, a control adjustment strategy generation unit, and a global optimization coordination command generation unit.

[0069] The data acquisition unit is used to acquire real-time first data of key positions in the unidirectional filament feeding path and the operating speed of multiple sets of rollers. In modern acrylic fiber production lines, the molecular orientation stretching of the raw filament through multiple sets of rollers is a key process to improve the strength and elasticity of acrylic filaments. The accuracy of tension control directly affects product quality and production efficiency. In this embodiment, the data acquisition unit uses a tension sensor array arranged along the acrylic filament feeding path. The tension sensor array is set at the entrance section, tension change section, stretching roller group, and winding pre-tensioning section of the feeding path to collect real-time data of each section of acrylic filament. Based on the real-time data, the tension amplitude change rate is obtained, and the key positions of the unidirectional filament feeding path are determined. Specifically, the difference rate of tension fluctuation amplitude between two adjacent sensors is calculated. The specific calculation formula is as follows: for any two adjacent sensors... and Calculate them within the time window The tension fluctuation amplitudes within are respectively and Tension fluctuation amplitude difference rate for:

[0070] ;

[0071] in, For any two adjacent sensors and Tension fluctuation amplitude, This represents the difference in the maximum tension fluctuation amplitude.

[0072] When the difference rate of tension fluctuation exceeds a preset threshold, it is identified as a critical position, and the tension data collected by the sensor corresponding to the critical position is used as the first data.

[0073] The initial control command acquisition unit is used to construct a dynamic characteristic control model, analyze the first data and operating speed, and generate preliminary control commands. The preliminary control commands include control deviation, tensile prediction value, and first operating speed; the tensile prediction value is compensated and adjusted based on the control deviation; the wire feeding process is adjusted based on the preliminary control commands to obtain second data and a second operating speed; such as... Figure 3 As shown, the dynamic feature control model includes a data preprocessing layer, a feature extraction layer, a predictive modeling layer, and a control decision layer, forming a complete closed-loop system from data to control.

[0074] The data preprocessing layer normalizes the initial data and processing speed, detects and corrects outliers, and uses a sliding window method to smooth the original signal, with a window size of 20 sampling points. Normalization employs a min-max normalization method to scale each data point to the [0,1] interval; outlier detection uses... The criterion is to consider data points that deviate from the mean by more than three standard deviations as outliers and correct them using linear interpolation.

[0075] The feature extraction layer employs wavelet transform decomposition to extract time-frequency domain features from the processed first data, including time-domain and frequency-domain feature parameters such as mean, variance, peak value, spectral energy distribution, and rate of change, forming feature vectors. The wavelet transform uses the db4 wavelet basis function, with a decomposition level of 3, thereby obtaining signal components and their energy distributions in different frequency bands. Then, principal component analysis is used to reduce the dimensionality of the feature vectors, retaining feature vectors with a cumulative contribution rate exceeding 95%, significantly reducing the dimensionality of the feature space while preserving key information, thus generating dimensionality-reduced feature vectors.

[0076] The predictive modeling layer is based on a long short-term memory network structure. The input layer contains the dimensionality-reduced feature vectors output by the feature extraction layer. The hidden layer contains 64 LSTM units and 32 fully connected neurons. The output layer generates stretch prediction values, including the tension prediction values ​​of each key position within the next 0.5 seconds.

[0077] The control decision layer calculates the control deviation by comparing the deviation between the actual tension and the target tension, and uses an adaptive fuzzy PID controller to compensate and adjust the tension prediction value. The fuzzy control rule is designed based on the tension error and its rate of change, and the output is the PID parameter adjustment amount. After the compensation adjustment is completed, the first running speed is calculated using a tension-speed nonlinear mapping model based on the adjusted tension prediction value, thereby realizing the initial control command for the running speed of multiple sets of rollers.

[0078] The tension-velocity nonlinear mapping model is a two-input, single-output nonlinear regression model constructed based on historical production data. The input variable is the adjusted predicted tension value. and the current roller running speed The output variable is the speed adjustment amount. By adjusting the speed For the current running speed Adjustments were made to achieve the initial operating speed. .

[0079] The specific process of adjusting the wire feeding process according to the initial control command to obtain the second data and the second operating speed includes: firstly, sending the first operating speed to the frequency conversion control unit of each roller driver via fieldbus, and each frequency conversion control unit adjusting the rotational speed of the roller according to the received command; then implementing control adjustment for a predetermined duration, which is set to 2-5 seconds according to the system response characteristics; during the adjustment process, using a tension sensor array to monitor the tension changes at each key position in real time and recording them as the second data; simultaneously, acquiring the actual operating speed of each group of rollers after adjustment through an encoder and recording it as the second operating speed.

[0080] The control adjustment strategy generation unit is used to construct a distributed collaborative control model, analyze preliminary control commands, and generate control adjustment strategies based on the global tension distribution balance and the stability of the tension gradient in key segments. For example... Figure 4 As shown, the distributed collaborative control model includes a state evaluation layer, a policy generation layer, and a collaborative optimization layer.

[0081] The distributed cooperative control model includes a state evaluation layer, a local policy generation layer, and a global coordination layer.

[0082] The state assessment layer receives preliminary control commands, second data, and second operating speed, and calculates the global tension distribution balance and the key segment tension gradient stability. The tension distribution balance index is the root mean square error between the actual tension value and the tension prediction value at each key position, and the tension gradient stability index is the standard deviation of the tension change rate between adjacent key positions.

[0083] The local strategy generation layer, based on an improved model predictive control algorithm, designs a control adjustment strategy library for different tension distribution states. It includes three basic control modes: steady-state control strategy, transient control strategy, and fluctuation suppression strategy. Each mode contains multiple sets of control parameters. The control parameters include the velocity adjustment step size coefficient. Tension compensation coefficient Gradient smoothing factor and collaborative weighting coefficient Under the steady-state control strategy, the parameters are conservatively selected. Smaller values ​​prioritize maintaining system stability; under transient control strategies... A larger value improves the system response speed; under the fluctuation suppression strategy, and The value is relatively large, which is mainly used to suppress tension fluctuations.

[0084] The global collaborative optimization layer uses the analytic hierarchy process (AHP) to determine the weights of each control parameter and combines it with a particle swarm optimization algorithm for dynamic optimization. When the system is under different operating conditions, it automatically selects the optimal set of control parameters and generates a control adjustment strategy. The specific process for obtaining the control adjustment strategy is as follows: First, the deviation matrix between the current system state and the control objective is calculated. Then, based on the deviation matrix, the strategy library is queried to match the closest basic control mode. Next, the optimization algorithm is used to fine-tune the control parameters under the selected basic control mode to minimize the overall performance index. ,in , For the uniformity of tension distribution, For tension gradient stability, For speed control, , and For the corresponding weights, The position and velocity update formulas for the particle swarm optimization algorithm are as follows:

[0085] ;

[0086] ;

[0087] in, and The first The velocity and position of each particle This is the particle's historical optimal position. The optimal position globally. For inertial weights, and The learning factor for speed, and A random number within the interval [0,1]. Unit of time.

[0088] The global optimization and coordination command generation unit, based on low-latency information interaction between multiple sets of rollers, controls the second operating speed according to the control adjustment strategy to obtain a globally optimized and coordinated speed control command. The low-latency information interaction between multiple sets of rollers utilizes a distributed real-time bus system to construct a communication network between roller control units. This real-time bus system adopts the EtherCAT industrial Ethernet protocol. Each roller control unit is equipped with an edge computing module and integrates a real-time operating system. Each roller control unit can directly transmit key status information to adjacent roller control units.

[0089] The specific process of obtaining the speed control command for global optimization and coordination includes: based on the second running speed and control adjustment strategy, constructing a multi-objective optimization function based on the global tension distribution balance, the stability of the tension gradient in the key section and the minimization of speed change;

[0090] An improved differential evolution algorithm is used to solve the multi-objective optimization function. In this embodiment, the population size of the differential evolution algorithm is set to 30, the maximum number of iterations is 20, the crossover probability is 0.8, and the mutation factor is dynamically adjusted within the range of [0.4, 0.9]. In each iteration, the physical constraints of the system are handled using the Lagrange constraint method, including the upper and lower limits of the speed of each roller, the speed ratio of adjacent rollers, and the speed change rate constraint, to generate the optimal speed vector.

[0091] After the optimization solution is completed, the obtained optimal speed vector is used as the speed control command for global optimization and coordination. It is synchronously sent to each roller control unit through the real-time bus system. Each control unit accurately controls the roller speed according to the received command to achieve global coordinated control.

[0092] The experimental results demonstrate that the performance of the unidirectional yarn feeding tension coordinated control system for acrylic filaments of the present invention compared with traditional control methods are shown in Table 1.

[0093] Table 1. Comparison of the effects of the method of the present invention and the traditional method.

[0094]

[0095] As can be seen from Table 1, the method of the present invention has significant advantages over traditional control methods in key indicators such as tension fluctuation amplitude, response time, fiber breakage rate, fiber strength uniformity and batch consistency, which verifies the effectiveness and advancement of the present invention.

[0096] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A unidirectional tension-coordinated control system for acrylic filament feeding, characterized in that, The coordinated control of the feeding tension for acrylic filaments during molecular orientation stretching via multiple sets of rollers along a unidirectional path includes: The data acquisition unit acquires the first data of key positions in the unidirectional wire feeding path and the running speed of multiple sets of rollers in real time; The initial control command acquisition unit constructs a dynamic characteristic control model to analyze the first data and running speed, and generates preliminary control commands. The preliminary control commands include control deviation, tensile prediction value, and first running speed. The tensile prediction value is compensated and adjusted based on the control deviation. The wire feeding process is adjusted based on the preliminary control commands to obtain second data and second running speed. The dynamic feature control model includes a data preprocessing layer, a feature extraction layer, a prediction modeling layer, and a control decision layer. The data preprocessing layer performs normalization processing on the first data and the running speed, and detects and corrects outliers. The feature extraction layer uses wavelet transform decomposition to extract time-frequency domain features from the processed first data, and then uses principal component analysis to reduce the dimension of the data to generate a dimension-reduced feature vector. The predictive modeling layer generates tensile prediction values ​​by analyzing the dimensionality-reduced feature vectors. The tensile prediction values ​​include the tension prediction values ​​at each key location. The control decision layer calculates the control deviation by comparing the deviation between the actual tension and the predicted tension value, and compensates and adjusts the predicted tension value. Based on the adjusted predicted tension value, it obtains the first operating speed using the tension-velocity nonlinear mapping model and generates the initial control command. The control adjustment strategy generation unit constructs a distributed collaborative control model to analyze the second data and the second running speed, and generates a control adjustment strategy based on the global tension distribution balance and the key segment tension gradient stability. The global optimization and coordination instruction generation unit, based on low-latency information interaction between multiple sets of rollers, controls the second running speed according to the control adjustment strategy to obtain a globally optimized and coordinated speed control instruction.

2. The unidirectional tension-coordinated control system for acrylic filament feeding according to claim 1, characterized in that: A tension sensor array is arranged along the acrylic filament feeding path. The tension sensor array is set at the entrance section, tension change section, stretching roller group and winding pre-tensioning section of the feeding path to collect real-time data of each section of acrylic filament. Based on the real-time data, the tension amplitude change rate is obtained, and the key position of the unidirectional feeding path is determined. Specifically, the difference rate of tension fluctuation amplitude between two adjacent sensors is calculated. When the difference rate of tension fluctuation amplitude exceeds a preset threshold, it is determined as a key position. The tension data collected by the sensor corresponding to the key position is used as the first data.

3. The unidirectional tension-coordinated control system for acrylic filament feeding according to claim 1, characterized in that: The tension-speed nonlinear mapping model is a dual-input, single-output nonlinear regression model constructed based on historical production data. The input variables are the adjusted tension prediction value and the current roller operating speed, and the output variable is the speed adjustment amount. The operating speed is adjusted by the speed adjustment amount to obtain the first operating speed.

4. The unidirectional tension-coordinated control system for acrylic filament feeding according to claim 1, characterized in that: The distributed cooperative control model includes a state evaluation layer, a local policy generation layer, and a global cooperative optimization layer. The state assessment layer receives preliminary control commands, second data, and second operating speed, and calculates the global tension distribution balance and the key segment tension gradient stability. The local strategy generation layer designs local control and adjustment strategies for different tension distribution states, including steady-state control strategies, transient control strategies, and fluctuation suppression strategies; the control and adjustment strategies include control parameters, including velocity adjustment step size coefficient, tension compensation coefficient, gradient smoothing factor, and cooperative weight coefficient. The global collaborative optimization layer uses the analytic hierarchy process (AHP) to determine the weights of the control parameters and combines it with the particle swarm optimization algorithm for dynamic optimization. Based on the local control adjustment strategies of different rollers, it selects the optimal control parameters and generates a control adjustment strategy.

5. The unidirectional tension-coordinated control system for acrylic filament feeding according to claim 1, characterized in that: The specific process of obtaining globally optimized and coordinated speed control commands includes: Based on the second operating speed and control adjustment strategy, a multi-objective optimization function is constructed based on the global tension distribution balance, the key segment tension gradient stability and the minimization of speed change. An improved differential evolution algorithm is used to iteratively solve the multi-objective optimization function. At the same time, during the iteration process, the Lagrange constraint method is used to handle the physical constraints of the system, including the upper and lower limits of the speed of each roller, the speed ratio of adjacent rollers, and the speed change rate constraint, to generate the optimal speed vector. Based on the optimal velocity vector, a globally optimized and coordinated velocity control command is obtained.

6. A method for coordinated control of unidirectional feeding tension of acrylic filament, characterized in that, The unidirectional feeding tension coordinated control system for acrylic filaments according to any one of claims 1-5, comprising the coordinated control of feeding tension during the molecular orientation stretching process of acrylic filaments passing through multiple sets of rollers along a unidirectional path, includes: Real-time acquisition of first-hand data at key locations along the unidirectional yarn feeding path and the running speed of multiple sets of rollers; A dynamic feature control model is constructed to analyze the first data and the running speed, and generate preliminary control instructions. The preliminary control instructions include control deviation, tensile prediction value, and first running speed. The tensile prediction value is compensated and adjusted according to the control deviation. The first running speed is obtained by adjusting the running speed according to the tensile prediction value. The wire feeding process is adjusted according to the preliminary control instructions to obtain second data and second running speed. A distributed collaborative control model is constructed to analyze the second data and the second operating speed. Based on the global tension distribution balance and the stability of the tension gradient in the key segment, a control adjustment strategy is generated. Based on low-latency information interaction between multiple sets of rollers, the second operating speed is controlled according to the control adjustment strategy to obtain a globally optimized and coordinated speed control command.

Images