Cooperative control optimization system for warp knitting multi-axial weft laying movement process

Abstract

The invention discloses a cooperative control optimization system for a warp knitting multi-axial weft laying movement process, and relates to the technical field of textile machinery manufacturing, the cooperative control optimization system comprises a weft laying execution unit, a sensing detection unit and a control system, the weft laying execution unit comprises a warp feeding device, a weft laying device and a traction device, and the sensing detection unit is used for detecting operation parameters of each execution unit. According to the cooperative control optimization system for the warp knitting multi-axial weft insertion movement process, global optimization control over the warp knitting multi-axial weft insertion process is achieved by constructing a cooperative control framework integrating a physical mechanism and data driving, and the multi-modal sensing unit is adopted for collecting operation data in real time; by simplifying parallel processing of a mechanism model and a digital twin virtual mapping system and combining a double-check mechanism of a dynamic optimization engine, a control instruction giving consideration to global performance optimization can be generated online, the contradiction between local optimization and global collaboration in a traditional control method is overcome, and the capacity of the system for coping with complex disturbance is improved.

Description

Technical Field

[0001] This invention relates to the field of textile machinery manufacturing technology, specifically to a collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting. Background Technology

[0002] Existing technologies typically employ control strategies combining pre-programmed procedures and closed-loop feedback. For example, they configure independent controllers for units such as warp feeding, weft laying, and traction, relying on sensors to monitor and adjust individual parameters. This method can meet basic production requirements under ideal, stable operating conditions. However, in actual operation, the system continuously faces numerous complex disturbances, such as sudden changes in yarn tension due to different batches of raw materials, or characteristic drift caused by long-term wear of mechanical transmission components. Faced with these disturbances, existing control systems often exhibit limitations. Their decentralized, localized compensation mechanisms, while temporarily mitigating deviations in a particular unit, easily disrupt the overall coordination of motion. For instance, abruptly adjusting the warp feeding speed to compensate for tension fluctuations in a specific yarn creel can indirectly lead to the yarn laying angle being out of sync with other yarns, or mismatch with the main traction speed. This conflict between local optimization and global performance reduces fabric surface uniformity and can even cause defects, severely impacting the quality and production efficiency of high-end products. Therefore, the current technological bottleneck lies in the lack of an intelligent control capability that can perceive the global state online and dynamically optimize all coordinating parameters to adapt to complex and changing operating conditions. The current technical challenge lies in how to achieve online dynamic optimization of the system's global collaborative parameters to cope with complex disturbances. Summary of the Invention

[0003] The purpose of this invention is to provide a collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting, so as to solve the problems mentioned in the background art.

[0004] To solve the above technical problems, the present invention provides the following technical solution: a collaborative control and optimization system for warp knitting multi-axis weft laying motion process, comprising a weft laying execution unit, a sensing and detection unit and a control system, wherein the weft laying execution unit includes a warp feeding device, a weft laying device and a traction device, the sensing and detection unit is used to detect the operating parameters of each execution unit, and the control system is connected to the weft laying execution unit and the sensing and detection unit; The control system includes: The multimodal sensing unit is used to collect and process real-time operating data from each execution unit; An edge computing gateway, connected to the multimodal sensing unit, is used to filter and fuse the collected data to generate a global state dataset; The dynamic optimization core, connected to the edge computing gateway, includes a simplified mechanism model based on physical laws and a digital twin virtual mapping system, used to receive the global state dataset and perform parallel processing; The simplified mechanism model is used to deduce the theoretical control parameters of each execution unit based on physical laws; The digital twin virtual mapping system simulates and predicts the current global state by reproducing the characteristics of physical devices in virtual space. The dynamic optimization engine connects the simplified mechanism model and the digital twin virtual mapping system, and is used to perform collaborative verification and screening of the parameters output by the two models to generate optimization control instructions. The instruction issuing unit sends optimized control instructions to the controllers of each execution unit.

[0005] Preferably, the multimodal sensing unit includes a combination of tension sensors, position sensors, angle sensors, and velocity sensors.

[0006] Preferably, the edge computing gateway uses timestamp alignment to process multi-source data to ensure data consistency.

[0007] Preferably, the dynamic optimization engine includes a first verification module, used to compare the consistency and feasibility of the theoretical parameters output by the simplified mechanism model with the optimization parameters predicted by the digital twin virtual mapping system.

[0008] Preferably, the dynamic optimization engine further includes a second verification module. When the first verification module detects a parameter conflict, it triggers a secondary verification mechanism to feed back the conflicting parameter set and the current global state data to the digital twin virtual mapping system for closed-loop simulation.

[0009] Preferably, the secondary verification mechanism includes driving the digital twin virtual mapping system to perform short-term simulation, simulating the system state in a future time period, and outputting predicted fabric quality indicators.

[0010] Preferably, the dynamic optimization engine selects the parameter set with the best global performance as the final output based on the results of the secondary simulation.

[0011] Preferably, the digital twin virtual mapping system includes a machine learning agent model, which is trained on historical production data and is specifically designed to predict the coupling effects of various actuators and their impact on fabric uniformity under complex disturbances.

[0012] Preferably, the simplified mechanism model includes a differential equation model that considers yarn elasticity, moment of inertia, and kinematic relationships.

[0013] Preferably, the instruction issuing unit communicates with the PLC and servo driver of each execution unit through the industrial bus protocol to achieve coordinated control of the warp feeding, weft laying, and traction devices.

[0014] This invention provides a collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting. It has the following beneficial effects: This collaborative control and optimization system for the multi-axis weft laying motion process in warp knitting achieves global optimization control of the process by constructing a collaborative control architecture that integrates physical mechanisms and data-driven approaches. The system employs multi-modal sensing units to collect operational data in real time. Through parallel processing of a simplified mechanism model and a digital twin virtual mapping system, combined with a dual verification mechanism of a dynamic optimization engine, it can generate control commands online that balance global performance. This method effectively overcomes the contradiction between local optimization and global synergy in traditional control methods, enhancing the system's ability to cope with complex disturbances.

[0015] This collaborative control and optimization system for the warp-knitted multi-axial weft laying process utilizes predictive simulation and multi-objective optimization decision-making via a digital twin system. While ensuring fabric quality, the system optimizes the coordination of various actuators, reduces defects caused by localized adjustments, and simultaneously improves production efficiency and reduces energy consumption. Ultimately, it achieves a stable improvement in the quality of warp-knitted multi-axial fabrics and enhances production economics. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the module interaction of a collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting according to the present invention; Figure 2 This is a timing diagram of the control logic of a collaborative control and optimization system for the multi-axial weft laying motion process of warp knitting according to the present invention. Detailed Implementation

[0017] 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.

[0018] Please see Figure 1 and Figure 2 The present invention provides a technical solution: a collaborative control and optimization system for the multi-axial weft laying motion process of warp knitting, including a weft laying execution unit, a sensing and detection unit and a control system. The weft laying execution unit includes a warp feeding device, a weft laying device and a traction device. The sensing and detection unit is used to detect the operating parameters of each execution unit. The control system is connected to the weft laying execution unit and the sensing and detection unit. The control system includes: The multimodal sensing unit is used to collect and process real-time operating data from each execution unit; The edge computing gateway connects to the multimodal sensing unit and is used to filter and fuse the collected data to generate a global state dataset. The dynamic optimization core connects to the edge computing gateway, including a simplified mechanism model based on physical laws and a digital twin virtual mapping system, which is used to receive global state datasets and perform parallel processing. The simplified mechanism model is used to deduce the theoretical control parameters of each execution unit based on physical laws; Digital twin virtual mapping systems simulate and predict the current global state by reproducing the characteristics of physical devices in a virtual space. The dynamic optimization engine connects the simplified mechanism model and the digital twin virtual mapping system to perform collaborative verification and screening of the parameters output by the two models, and generate optimization control instructions. The instruction issuing unit sends optimized control instructions to the controllers of each execution unit.

[0019] It should be further explained that, in the specific implementation process, the collaborative control optimization system for the warp knitting multi-axial weft laying motion process is implemented as follows: After the system starts, the multimodal sensing unit continuously collects the yarn tension of the warp feeding device, the position and angle of the weft laying device, the linear velocity of the traction device, and the rotational speed signal of the spindle. This real-time data is transmitted to the edge computing gateway for data filtering and multi-source information fusion, and forms a global state dataset with time consistency through time-synchronized processing. This dataset is synchronously transmitted to the simplified mechanism model and digital twin virtual mapping system of the dynamic optimization core. Based on preset physical laws, including the elastic deformation of the yarn, the moment of inertia of the moving parts, and the kinematic relationship of the system, the simplified mechanism model calculates the theoretical control parameters that each execution unit should adopt under the current working condition. At the same time, the digital twin virtual mapping system uses its built-in machine learning agent model trained with historical data to simulate the dynamic behavior of the actual equipment in the virtual environment, predict the impact of different control parameters on the weft laying effect and the final fabric uniformity under the presence of disturbances, and outputs a set of optimization parameter suggestions.

[0020] After receiving the two sets of parameters, the dynamic optimization engine first performs a co-verification, including comparing the consistency between the two sets of parameters, determining whether the parameters recommended by the proxy model are within the physically feasible range determined by the mechanistic model, and evaluating the potential impact of these parameter adjustments on the overall system energy consumption and operational stability. If the verification passes, a preliminary optimization instruction is generated; if a non-negligible difference is found between the two sets of parameters, a secondary verification process is initiated. At this point, the engine re-inputs the current global state data and the discrepancy set of parameters into the digital twin virtual mapping system, requesting it to perform a short-term fine closed-loop simulation. This simulation simulates the system's behavior under different parameter sets over a future period and outputs key fabric quality prediction indicators. Based on these simulation results, the dynamic optimization engine selects the set of parameters that optimizes global performance as the final determined optimization control instruction. Subsequently, the instruction issuing unit transmits these instructions through the industrial network to the programmable logic controllers and servo drives corresponding to the warp feeding, weft laying, and traction devices, respectively, to achieve precise co-control of each actuator, thereby ensuring that the warp knitting weft laying process maintains high quality and high stability even when facing internal and external disturbances.

[0021] Through the above implementation methods, the system has achieved a transformation from single local control to global collaborative optimization, effectively solving the technical problem of disrupting overall synergy due to local adjustments.

[0022] The multimodal sensing unit comprises various combinations of tension sensors, position sensors, angle sensors, and speed sensors. It should be further explained that, in specific implementation, the multimodal sensing unit includes tension sensors located at each yarn creel to monitor yarn tension changes in real time; position and angle sensors installed on the weft-laying machine to continuously provide feedback on the lateral displacement and laying angle of the weft-laying device; speed sensors located at the traction rollers to detect the fabric traction speed; and a main shaft encoder connected to acquire the real-time rotational speed of the main shaft.

[0023] These sensors transmit multi-source signals to the data acquisition module via an industrial fieldbus. Different types of sensor signals are standardized by signal conditioning circuits. Analog signals undergo analog-to-digital conversion, and digital signals undergo level matching and isolation to ultimately form a unified digital signal.

[0024] During data processing, the system assigns timestamps to various sensor data and removes abnormal jump values ​​through data preprocessing algorithms to ensure the reliability and timeliness of the input data, providing an accurate multi-dimensional field data foundation for subsequent collaborative optimization control.

[0025] The edge computing gateway employs timestamp alignment to process multi-source data, ensuring data consistency. Further explanation is needed: in the specific implementation process, after receiving multi-source data uploaded by the multimodal sensing unit, the edge computing gateway initiates a timestamp alignment process. This process first parses and extracts the built-in timestamps of each sensor data packet, then establishes a unified time reference system based on the system's master clock. The gateway internally sets up a data buffer queue to temporarily store and wait for data packets that have not yet arrived synchronously due to transmission delays or different acquisition cycles. When all relevant data arrives within a preset time window, the alignment algorithm rearranges them according to the timestamps of each data packet, ensuring that tension, position, velocity, and other state information belonging to the same moment are aggregated into the same data frame. For data packets with missing or incorrect timestamps due to anomalies, the system uses an interpolation estimation algorithm to calculate reasonable values ​​based on valid data from preceding and following moments to fill in the gaps, thereby forming a globally consistent state dataset, providing an accurate data foundation for subsequent collaborative optimization.

[0026] The dynamic optimization engine includes a first verification module, used to compare the consistency and feasibility of the theoretical parameters output by the simplified mechanism model and the optimization parameters predicted by the digital twin virtual mapping system. It should be further explained that, in the specific implementation process, after receiving the theoretical control parameters output by the simplified mechanism model and the optimization parameter suggestions generated by the digital twin virtual mapping system, the first verification module in the dynamic optimization engine initiates a collaborative verification and screening process. This module first performs a structured comparison of the two sets of parameters, analyzing the numerical differences in key control variables such as warp feed speed, weft laying angle, and traction rate. The verification logic is based on a preset tolerance threshold. If the differences of all variables are within the threshold range, the two sets of parameters are determined to be consistent. If variables exceeding the threshold are found, the parameters suggested by the digital twin system are further checked to see if they are within the physically feasible domain defined by the simplified mechanism model. This feasible domain is jointly determined by equipment mechanical limits, material properties, and process constraints. Meanwhile, the module evaluation uses digital twin-based suggested parameters to assess the potential impact on the overall system energy consumption and operational stability, calculating the expected change relative to the theoretical parameters. When the parameter difference is within the tolerance range or exceeds the tolerance but is physically feasible and the stability cost is acceptable, the verification is passed. If the parameter difference is significant and may lead to stability risks, it is determined to be a conflict, triggering the subsequent processing mechanism.

[0027] The dynamic optimization engine also includes a second verification module. When the first verification module detects a parameter conflict, it triggers a secondary verification mechanism, feeding back the conflicting parameter set and the current global state data to the digital twin virtual mapping system for closed-loop simulation. It should be further explained that, in the specific implementation process, when the first verification module determines that there is a non-negligible conflict between the theoretical parameters of the simplified mechanism model and the optimization suggestion parameters of the digital twin virtual mapping system, the dynamic optimization engine automatically activates its second verification module and initiates the secondary verification mechanism. This mechanism first repackages the conflicting parameter set along with the current global state dataset and feeds it back to the digital twin virtual mapping system. After receiving this data, the digital twin system, based on its high-fidelity virtual model, uses the two conflicting sets of parameters as input conditions to initiate a refined closed-loop simulation for a future time period. The simulation process simulates the dynamic response process and mutual coupling of the execution units of warp feeding, weft laying, and traction after the system adopts these two sets of parameters respectively, and finally predicts the corresponding key indicators of fabric quality, such as weft density uniformity. The second verification module receives the two sets of simulation results returned by the digital twin system, and performs comparative analysis based on the preset global performance optimization goals, such as prioritizing more energy-efficient or more stable parameter sets while ensuring fabric quality. This allows the optimal set to be selected from the conflicting parameter options, completing the secondary verification and decision-making, and ensuring that the final control commands are optimal in terms of technical feasibility and overall performance.

[0028] The secondary verification mechanism includes driving the digital twin virtual mapping system to perform short-term simulations, simulating the system state in a future time period, and outputting predicted fabric quality indicators. It should be further explained that, in the specific implementation process, the short-term simulation process in the secondary verification mechanism is executed by the digital twin virtual mapping system. This system uses the received conflicting parameter set and the current global state as initial conditions to simulate the dynamic behavior of the system within a specific future time frame in the virtual environment. The simulation process progresses step by step, calculating how the speed adjustment of the warp feed device will affect the yarn tension, how the position change of the weft laying device will interact with other motion axes, and the matching of traction speeds, if each set of conflicting parameters is used within that time frame. The simulation calculation not only considers the independent response of each execution unit, but also focuses on calculating the coupling effect between them. Finally, the digital twin system outputs the fabric quality index predicted based on each set of parameters. This index is a key parameter for quantifying the uniformity of weft yarn laying. The dynamic optimization engine receives these prediction results and, based on the pre-set optimization objective function, such as seeking to minimize system energy consumption or achieve the most stable operation while meeting the minimum uniformity requirements, comprehensively evaluates and ranks the prediction performance corresponding to each set of parameters, thereby providing an objective decision-making basis for the final selection.

[0029] Based on the results of the secondary simulation, the dynamic optimization engine selects the parameter set with the best global performance as the final output. It should be further explained that in the specific implementation process, after receiving multiple sets of prediction results from the digital twin virtual mapping system based on short-term simulations using different parameter sets, the dynamic optimization engine initiates the final decision-making process. This process first analyzes the fabric quality indicators, system energy consumption estimates, and operational stability parameters contained in each set of prediction results. The engine has a pre-set multi-objective optimization decision-making logic. This logic does not simply compare the merits of a single indicator, but rather comprehensively weighs various performance indicators. The decision logic first checks whether the predicted fabric quality corresponding to each set of parameters meets the minimum standards specified by the production process, eliminating any parameter sets that may lead to substandard quality. For candidate parameter sets that meet the quality requirements, a weighted score is then applied based on the current system's optimization priority configuration. For example, if the current priority is energy saving, the focus is on comparing energy consumption estimates; if the priority is stability, the focus is on analyzing stability parameters. Finally, the parameter set with the highest comprehensive score is selected as the global optimal solution, generating corresponding optimized control commands, thereby ensuring that the system's control decisions not only guarantee the final product quality but also consider operational efficiency and reliability.

[0030] The digital twin virtual mapping system includes a machine learning proxy model trained on historical production data. This model is specifically designed to predict the coupling effects of various actuators and their impact on fabric evenness under complex disturbances. It's important to further clarify that, in the implementation process, the machine learning proxy model integrated into the digital twin virtual mapping system is built and trained based on a large amount of historical production data. This data covers various operating conditions under different raw material batches, equipment operating states, and external environmental conditions, including control parameters of each actuator, operating status data collected by sensor detection units, and the final fabric quality inspection results. During the model training phase, a supervised learning algorithm is used, employing historical operating status data and historical control parameters as input features, and the corresponding fabric evenness evaluation index as the training objective. An iterative optimization algorithm adjusts the model's internal parameters, enabling the model to learn the complex nonlinear mapping relationship between system state and control commands to the final quality. The trained proxy model is embedded in the digital twin system. During actual operation, it receives the current global state dataset as input and quickly infers the possible fabric evenness change trends and other system coupling effects under different control parameter combinations, providing data-driven optimization parameter suggestions for the dynamic optimization engine.

[0031] The simplified mechanism model includes a differential equation model considering yarn elasticity, moment of inertia, and kinematic relationships. It should be further explained that, in practical implementation, the simplified mechanism model is based on the physical principles of warp knitting multi-axial weft-laying equipment. Its core is a set of differential equations describing the dynamic characteristics of the system. This model treats the yarn as an elastic body with a specific elastic modulus, and the transmission mechanisms of the warp feed, weft-laying, and traction devices as rotational motion systems containing moments of inertia. It strictly defines the velocity and displacement coupling between each execution unit through kinematic relationships. The model input is the current state of each motion axis extracted from the global state dataset. By solving the differential equation set, it deduces the theoretically required control parameters for each execution unit to maintain the preset weft-laying process requirements under ideal physical laws, including the setpoint of the warp feed motor speed, the lateral movement speed of the weft-laying carriage, and the linear speed of the traction roller. This model has a fast computation speed, providing a physically sound benchmark for dynamic optimization, ensuring that the optimization process does not generate control commands that violate basic physical laws.

[0032] The command issuing unit communicates with the PLCs and servo drives of each execution unit via the industrial bus protocol to achieve coordinated control of the warp feeding, weft laying, and traction devices. Further explanation is needed: in the specific implementation process, the command issuing unit establishes a communication connection with the controllers of each execution unit via the industrial bus protocol. This unit receives the final optimized control command set from the dynamic optimization engine. This command set includes the target rotational speed of the warp feeding device, the target position and angle of the weft laying device, and the target linear velocity of the traction device, determined after multiple verifications. The command issuing unit encapsulates the above commands according to a predefined communication data frame format. The data frame contains the target device address, control command code, and specific parameter values. The encapsulated data frame is sent to the corresponding warp feeding device PLC, weft laying device servo drive, and traction device controller via the industrial bus network. Upon receiving the command, the controllers of each execution unit immediately convert it into internal control signals to drive the motor or actuator, thereby precisely realizing the coordinated motion control of warp feeding, weft laying, and traction. The entire command issuing process has a defined communication cycle, ensuring the real-time and synchronous nature of the control commands and guaranteeing the high precision and stability of the entire weft laying motion process.

[0033] It should be further explained that, in the specific implementation process, the system achieves online dynamic optimization of the global collaborative parameters of the weft laying process by integrating multi-source sensor data, constructing a virtual simulation environment, and implementing a multi-verification decision mechanism. After the system starts up, sensors deployed at each key node begin to continuously collect operational data. Tension sensors installed on the yarn frame monitor the tension changes of each yarn in real time; encoders and tilt sensors integrated on the weft laying machine detect its lateral movement position and laying angle, respectively; speed sensors on one side of the traction roller measure the fabric output speed; and the main shaft encoder provides the main shaft rotation speed signal. These sensor signals from different physical quantities and different acquisition cycles are standardized by the signal conditioning circuit, converted into a unified digital signal format, and transmitted to the data acquisition module via an industrial fieldbus.

[0034] The data acquisition module sends the received multi-source data to the edge computing gateway for processing. One of the gateway's core tasks is to solve the data synchronization problem. It assigns a precise timestamp to each data packet and aligns it with the system's master clock. For data that fails to arrive simultaneously due to network latency or slight differences in acquisition time, the gateway temporarily stores it in a buffer, waiting for a very short time window until all relevant data arrives. Then, it reassembles the data into a globally consistent state data frame based on the timestamps. For the very few data points with abnormal timestamps or that are missing, an interpolation algorithm based on historical data is used to reasonably fill in the gaps, ensuring the integrity and reliability of the dataset.

[0035] The preprocessed and synchronized global state dataset is fed in parallel into two key components of the dynamic optimization core: a simplified mechanistic model and a digital twin virtual mapping system. The simplified mechanistic model is built upon classical Newtonian mechanics and kinematic principles. It simplifies the yarn into a linear model with specific elastic properties, treats the transmission system as a rotating body with inertia, and rigorously describes the kinematic coupling relationships between actions such as warp feeding, weft laying, and traction through a series of differential equations. Using the current system state as initial conditions, this model quickly calculates a set of control parameters that should be adopted under ideal physical laws, providing a benchmark that conforms to fundamental physical laws for the entire optimization process.

[0036] Meanwhile, the digital twin virtual mapping system operates in parallel in virtual space. This system includes a machine learning proxy model trained on a large amount of historical production data. The proxy model's learning process is based on massive amounts of historical operating data, including equipment operating parameters, sensor readings, and ultimately measured fabric quality indicators. Through supervised learning algorithms, a complex nonlinear mapping relationship is established between system state and control commands and output quality. During actual operation, the model receives current global state data and quickly deduces several sets of possible optimized control parameters and their predicted execution effects.

[0037] The dynamic optimization engine receives output from the two channels mentioned above. Its built-in first verification module first compares the theoretical parameters calculated by the mechanistic model with the optimized parameters recommended by the digital twin. The comparison not only focuses on whether the numerical differences are within the preset engineering tolerance range, but more importantly, it determines whether the digital twin's recommended parameters are within the physically feasible domain defined by the mechanistic model, while also assessing the potential impact of adopting these parameters on the overall system stability and energy consumption. If the verification passes, the optimized parameters are adopted.

[0038] If the first verification detects a substantial conflict between the two sets of parameters, the second verification module is immediately activated, initiating a secondary verification mechanism. This mechanism feeds back the conflicting parameter set along with the current global state to the digital twin system, requesting it to perform a short-term, refined closed-loop simulation. The digital twin system simulates the dynamic response of the system over a future time period using different parameter sets and accurately predicts the final key fabric quality indicators, such as weft density uniformity.

[0039] Based on these simulation predictions, the dynamic optimization engine makes a final decision according to a preset multi-objective optimization strategy. The preset multi-objective optimization strategy includes pursuing the lowest energy consumption or the most stable operation while ensuring quality, and selecting a set of parameters with the best overall performance.

[0040] The final optimized control instruction set is sent from the instruction issuing unit to each execution unit via the industrial bus protocol. This unit encapsulates the instructions into data frames according to a predefined format and accurately sends them to the corresponding programmable logic controllers and servo drivers. This drives the warp feed motor, weft laying servo, and traction motor to coordinate their actions, thereby achieving high-precision, high-performance collaborative control of the weft laying motion process. This effectively addresses various internal and external disturbances and ensures the high-quality production of the final fabric.

[0041] Through the above implementation methods, the system has achieved a leap from local single-point control to global collaborative optimization, and from passive response to proactive predictive decision-making, effectively solving the core technical problem of global synergy being destroyed by local adjustments.

[0042] A further implementation of a collaborative control and optimization system for the multi-axis weft laying motion process in warp knitting is as follows: The system achieves global optimization control of the multi-axis motion in the weft laying process by constructing a collaborative control architecture that integrates physical mechanisms and data-driven approaches.

[0043] The simplified mechanism model of the system is based on the following core differential equation, which describes the dynamic characteristics of the yarn drive system: T=J×(d²θ / dt²)+B×(dθ / dt)+K×θ; where: T represents the torque output by the drive motor, J represents the equivalent moment of inertia of the transmission system, θ represents the displacement of the rotation angle, B represents the damping coefficient of the system, K represents the elastic coefficient of the yarn, t represents the time variable, d²θ / dt² represents the angular acceleration, and dθ / dt represents the angular velocity.

[0044] For the speed control of the warp feeding system, the following kinematic equation is used: v = r × ω; where: v represents the linear velocity of the yarn conveyor, r represents the radius of the warp feeding roller, and ω represents the angular velocity of the warp feeding motor.

[0045] The machine learning agent model in the digital twin virtual mapping system is trained using the following optimization objective function: minΣ[α×(y pred -y target )²+β×Δu²];where: y pred y represents the fabric quality index predicted by the model. target Δu represents the target quality index, α represents the range of change of the control quantity, and α and β represent weighting coefficients used to balance tracking accuracy and control smoothness.

[0046] The first verification module of the dynamic optimization engine uses the following criteria to check parameter consistency: |u mech -u digital |≤ε;where: u mech u represents the theoretical control parameters calculated by the mechanistic model. digital ε represents the optimization parameters recommended by the digital twin, and ε represents the tolerance threshold set based on engineering experience.

[0047] When the parameter difference exceeds the tolerance range, the second verification module initiates the following decision function for optimization: J=w1×Q+w2×E+w3×S; where: Q represents the predicted fabric quality index, E represents the estimated system energy consumption, S represents the operational stability index, and w1, w2, and w3 represent the weight coefficients of each index.

[0048] The system collects the following physical quantities through a multimodal sensing unit: Tension sensor measured values: F = {F1, F2, ..., F...} n}; Position encoder reading: P = {P} x P y P z}; Speed ​​sensor data: V = {V1, V2, ..., V} n}; Angle sensor measurement: θ = {θ1, θ2, ..., θ n}; The edge computing gateway uses the following data synchronization algorithm: For each data sampling time t k Define the synchronization data frame as: X(t) k )={F(t k ), P(t) k ), V(t) k ), θ(t) k )}; Interpolation algorithm is used to handle missing data: x(t) = x(t) k-1 )+(tt k-1 )×(x(t k )-x(t k-1 )) / (t k -t k-1 ); The instruction issuing unit communicates using the following data frame format: [address code][instruction type][parameter value][checksum]; where the checksum uses the CRC cyclic redundancy check algorithm to ensure the reliability of data transmission.

[0049] Through the synergistic effect of the aforementioned mathematical models and algorithms, the system achieves precise control over the multi-axial weft laying process in warp knitting. When yarn tension fluctuations are detected, the system first calculates the theoretical compensation amount using a mechanistic model, then predicts the impact of different compensation strategies on the global system using a digital twin system, and finally selects the optimal control strategy through an optimization algorithm to ensure that global synergy is maintained while eliminating local disturbances.

[0050] This system achieves a technological breakthrough by deeply integrating physical mechanisms and data-driven approaches, moving from passive compensation to active optimization and from local control to global collaboration. It effectively resolves the contradiction between local optimization and global performance in traditional control methods.

[0051] A collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting, further comprising a collaborative control and optimization method, including the following steps: Step S1: The operating parameters of each actuator during the weft laying process are collected in real time through the multi-modal sensing unit, including yarn tension, position and angle of the weft laying device, traction speed and spindle speed. The collected multi-source data is filtered and synchronized to form a global state dataset. Step S2: Input the processed global state dataset in parallel into the simplified mechanism model and the digital twin virtual mapping system. The simplified mechanism model derives the theoretical control parameters of each execution unit based on physical laws, and the digital twin virtual mapping system predicts the optimal control parameters under disturbance conditions through the built-in machine learning proxy model. Step S3: The dynamic optimization engine receives two sets of parameter outputs and first performs a collaborative verification and screening, including comparing parameter consistency, evaluating whether the parameters are within the physical feasible domain, and analyzing the impact on system stability and energy consumption. Step S4: If the first verification finds that there is a conflict in the parameters, the second verification mechanism is started. The conflicting parameter set and the current global state data are fed back to the digital twin virtual mapping system for refined closed-loop simulation to predict the system behavior and fabric quality indicators under different parameters. Step S5: Based on the results of the second simulation, the dynamic optimization engine selects the parameter set with the best global performance as the final optimization control command based on the multi-objective optimization decision logic; Step S6: The instruction issuing unit sends the optimized control instructions to the controllers of each execution unit through the industrial bus protocol, driving the warp feeding device, weft laying device and traction device to achieve coordinated motion control; Step S7: The system continuously monitors the operating status. When a new disturbance or status change is detected, it returns to step S1 for a new round of optimization and adjustment to form closed-loop control.

[0052] By constructing a collaborative control architecture that integrates physical mechanisms and data-driven approaches, global optimization control of the warp knitting multi-axis weft laying process is achieved. The system employs a multimodal sensing unit to collect operational data in real time. Through parallel processing of a simplified mechanistic model and a digital twin virtual mapping system, combined with a dual verification mechanism of a dynamic optimization engine, it can generate control commands online that balance optimal global performance. This method effectively overcomes the contradiction between local optimization and global synergy in traditional control methods, enhancing the system's ability to cope with complex disturbances.

[0053] Through predictive simulation and multi-objective optimization decision-making using a digital twin system, the system optimized the coordination of various actuators while ensuring fabric quality. This reduced the production of defective products caused by local adjustments, while simultaneously improving production efficiency and reducing energy consumption. Ultimately, this resulted in a stable improvement in the quality of warp-knitted multiaxial fabrics and enhanced production economics.

[0054] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0055] 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 collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting, characterized in that, It includes a weft laying execution unit, a sensing and detection unit, and a control system. The weft laying execution unit includes a warp feeding device, a weft laying device, and a traction device. The sensing and detection unit is used to detect the operating parameters of each execution unit. The control system is connected to the weft laying execution unit and the sensing and detection unit. The control system includes: The multimodal sensing unit is used to collect and process real-time operating data from each execution unit; An edge computing gateway, connected to the multimodal sensing unit, is used to filter and fuse the collected data to generate a global state dataset; The dynamic optimization core, connected to the edge computing gateway, includes a simplified mechanism model based on physical laws and a digital twin virtual mapping system, used to receive the global state dataset and perform parallel processing; The simplified mechanism model is used to deduce the theoretical control parameters of each execution unit based on physical laws; The digital twin virtual mapping system simulates and predicts the current global state by reproducing the characteristics of physical devices in virtual space. The dynamic optimization engine connects the simplified mechanism model and the digital twin virtual mapping system, and is used to perform collaborative verification and screening of the parameters output by the two models to generate optimization control instructions. The instruction issuing unit sends optimized control instructions to the controllers of each execution unit.

2. The collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting according to claim 1, characterized in that: The multimodal sensing unit includes a combination of tension sensors, position sensors, angle sensors, and velocity sensors.

3. The collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting according to claim 2, characterized in that: The edge computing gateway uses timestamp alignment to process multi-source data, ensuring data consistency.

4. The collaborative control and optimization system for the multi-axial weft laying motion process of warp knitting according to claim 3, characterized in that: The dynamic optimization engine includes a first verification module, which is used to compare the consistency and feasibility of the theoretical parameters output by the simplified mechanism model with the optimization parameters predicted by the digital twin virtual mapping system.

5. The collaborative control and optimization system for the multi-axial weft laying motion process of warp knitting according to claim 4, characterized in that: The dynamic optimization engine also includes a second verification module. When the first verification module detects a parameter conflict, it triggers a secondary verification mechanism to feed back the conflicting parameter set and the current global state data to the digital twin virtual mapping system for closed-loop simulation.

6. The collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting according to claim 5, characterized in that: The secondary verification mechanism includes driving the digital twin virtual mapping system to perform short-term simulation, simulating the system state in a future time period, and outputting predicted fabric quality indicators.

7. The collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting according to claim 6, characterized in that: The dynamic optimization engine selects the parameter set with the best global performance as the final output based on the results of the secondary simulation.

8. The collaborative control and optimization system for the multi-axial weft laying motion process of warp knitting according to claim 7, characterized in that: The digital twin virtual mapping system includes a machine learning agent model, which is trained on historical production data and is specifically designed to predict the coupling effects of various actuators and their impact on fabric uniformity under complex disturbances.

9. The collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting according to claim 8, characterized in that: The simplified mechanism model includes a differential equation model that considers yarn elasticity, moment of inertia, and kinematic relationships.

10. The collaborative control and optimization system for the multi-axial weft laying motion process in warp knitting according to claim 9, characterized in that: The instruction issuing unit communicates with the PLC and servo driver of each execution unit through the industrial bus protocol to achieve coordinated control of the warp feeding, weft laying, and traction devices.

Images