High-elastic polyester yarn production method for intelligent manufacturing
By integrating intelligent manufacturing models and multi-source data, the problem of insufficient adaptability of process parameters in the production of high-elastic polyester yarn was solved, achieving stability in product quality and improving production efficiency, while reducing energy consumption and scrap rate.
Patent Information
- Application Number
- CN202511759632.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-27
- Publication Date
- 2026-02-27
AI Technical Summary
In the current production of high-elastic polyester yarn, process parameters are difficult to control adaptively, resulting in large fluctuations in product elasticity and insufficient quality consistency. In particular, when producing multiple varieties and small batches of flexible products, traditional methods rely on manual experience, leading to low production efficiency and unstable quality.
By constructing an intelligent manufacturing model and combining online detection and self-learning iteration, an intelligent control system for the production of high-elastic polyester yarn is established. Online viscosity sensors are used to detect melt viscosity, tension sensors and laser scanners are used to measure fiber diameter, and multi-source data is integrated to make dynamic optimization decisions and achieve adaptive adjustment of process parameters.
It achieves precise control over the production process of high-elastic polyester yarn, improves product quality consistency and production efficiency, reduces scrap rate, and enhances equipment utilization and process robustness.
Smart Images
Figure CN121578764A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of chemical fiber manufacturing technology, specifically to a method for producing high-elastic polyester yarn for intelligent manufacturing. Background Technology
[0002] The current industrial production of high-elastic polyester yarn mainly relies on mature spinning and texturing processes. Its core processes include polyester melt conveying, precise metering with metering pumps, forming nascent fibers using spinnerets, and side-blowing cooling and solidification. The fibers then enter the stretching and deformation stages, undergoing key steps such as heating, false twisting, and setting on a texturing machine to achieve good elasticity and bulk. Existing technologies can stably produce polyester elastic yarn that meets basic market demands. The level of automation in production equipment continues to improve, with process control systems commonly used to monitor and adjust core process parameters such as temperature, speed, and tension. There is also ongoing research within the industry on optimizing single or multiple variables such as raw material characteristics, spinning speed, and heating box temperature, aiming to improve fiber strength, elongation, and dyeing uniformity. Some companies are beginning to experiment with data collection, hoping to use historical production data analysis to assist in quality control.
[0003] The existing technological system suffers from a fundamental limitation: a severe lack of intelligence across the entire production process. This makes it difficult to achieve precise, dynamic, and adaptive closed-loop control between process parameters and final product quality. Existing control models largely rely on pre-set, fixed process formulas, failing to make autonomous decisions and dynamically optimize based on real-time fluctuations in raw material characteristics, workshop environment, and equipment operating status. This directly results in unstable control precision for the core indicator of high elasticity. Key indicators such as fiber elastic recovery rate and shrinkage elongation vary significantly between the same batch and even different rolls, making it difficult to guarantee product quality uniformity. Particularly when facing the flexible production demands of multiple varieties and small batches, traditional methods rely on manual experience for production line adjustments, leading to low changeover efficiency and a high dependence on operator skill levels, further exacerbating product quality fluctuations. This deficiency has become a major bottleneck restricting the advancement of high-elastic polyester filament performance and the realization of truly intelligent production models. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a method for producing high-elastic polyester yarn for intelligent manufacturing. The technical problem this invention aims to solve is: how to address the issues of difficult adaptive control of process parameters, large fluctuations in product elasticity indicators, and insufficient quality consistency in the production of high-elastic polyester yarn by establishing a manufacturing process based on intelligent modeling, online detection, dynamic optimization, and self-learning iteration.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for producing high-elastic polyester yarn for intelligent manufacturing, comprising: S1. Offline training and processing of historical high-elastic polyester filament production data to construct a basic intelligent manufacturing model; S2. The received production order information is parsed and processed to obtain the target elastic recovery rate specification. The polyester raw materials fed into the production line are detected online to generate melt viscosity index. The online detection uses an online viscosity sensor. S3. Input the target elastic recovery rate specification and the melt viscosity index as a joint vector into the basic intelligent manufacturing model. The basic intelligent manufacturing model performs comprehensive calculation on the joint vector to generate initial process parameters. S4. Apply the initial process parameters on the production line and produce high-elastic polyester yarn according to the initial process parameters. The production includes a yarn winding process. Construct a synchronous acquisition cycle according to the length of the yarn in the yarn winding process. Perform fusion acquisition and processing on the multi-dimensional operation status data of the synchronous acquisition cycle to form a multi-source synchronous dataset. The multi-dimensional operation status data includes the tension value of the stretching zone and the fiber diameter measurement value. S5. Input the multi-source synchronous dataset into the basic intelligent manufacturing model. The basic intelligent manufacturing model performs dynamic optimization decision-making on the multi-source synchronous dataset and outputs adjustment instructions. The adjustment instructions are then sent to the production line. The dynamic optimization decision-making adopts conflict coordination rules. S6. After the yarn is wound to form a complete packaged product, the measured elastic recovery rate of the complete packaged product is obtained. The measured elastic recovery rate, the multi-source synchronous dataset, and the adjustment instructions are integrated and processed to form a self-learning sample package. The self-learning sample package is fed back to the basic intelligent manufacturing model to form parameter iterative updates.
[0006] Preferably, the sources of the historical high-elastic polyester filament production data include different raw material batches, environmental conditions, and product specifications.
[0007] Preferably, the parsing process includes extracting a target elastic recovery rate range from the production order information, taking the median of the target elastic recovery rate range as the target elastic recovery rate specification, wherein the target elastic recovery rate range is 85%-95%, and the online viscosity sensor is installed on the polyester melt conveying pipeline of the production line.
[0008] Preferably, the steps of the comprehensive calculation process are as follows: S31. Perform feature fusion processing on the target elastic recovery rate specification and the melt viscosity index to form the joint vector; S32. Input the joint vector into the fully connected network layer of the basic intelligent manufacturing model to perform nonlinear transformation and generate a high-dimensional feature vector; S33. The high-dimensional feature vector is subjected to regression processing at the output layer of the basic intelligent manufacturing model to output the initial process parameters.
[0009] Preferably, the comprehensive calculation process employs a random forest regression algorithm, which includes performing a comprehensive calculation on the joint vector. The comprehensive calculation uses multiple decision trees, and the average output of the multiple decision trees is used as the initial process parameters.
[0010] Preferably, the construction includes a synchronous acquisition cycle when the length of the silk thread reaches 500-1000 meters, and the fusion acquisition processing includes data alignment of the multi-dimensional operational status data to form the multi-source synchronous dataset.
[0011] Preferably, the tension value in the drafting zone is obtained using a tension sensor installed between the drafting rollers of the production line, and the fiber diameter measurement value is obtained using a laser scanning diameter measuring instrument.
[0012] Preferably, the conflict coordination rule includes prioritizing the uniform adjustment direction when the stabilization adjustment direction of the tension value in the drafting zone conflicts with the homogenization adjustment direction of the fiber diameter measurement value, and simultaneously relaxing the constraint range on the stabilization adjustment direction.
[0013] Preferably, the adjustment instructions include a main draw ratio fine-tuning instruction, a first hot box temperature compensation instruction, and a second hot box temperature compensation instruction. The adjustment range of the main draw ratio fine-tuning instruction is ±0.5% to ±2% of the current set value, the adjustment range of the first hot box temperature compensation instruction is ±1℃ to ±5℃ of the current set value, and the adjustment range of the second hot box temperature compensation instruction is ±1℃ to ±3℃ of the current set value.
[0014] Preferably, the measured elastic recovery rate in the self-learning sample package ranges from 85% to 98%, and the parameter iterative update includes optimizing the parameters of the basic intelligent manufacturing model with the measured elastic recovery rate as the target value, and applying the optimized basic intelligent manufacturing model to a production task with the same target elastic recovery rate specification and melt viscosity index.
[0015] This invention provides a method for producing high-elastic polyester yarn for intelligent manufacturing. It has the following beneficial effects: This method for producing high-elastic polyester yarn for intelligent manufacturing achieves intelligent generation and dynamic optimization of process parameters throughout the entire production process by constructing a basic intelligent manufacturing model, integrating a joint vector of the target elastic recovery rate specification and melt viscosity index, and introducing multi-source synchronous datasets. Through joint analysis and model inference of production orders, raw material viscosity, and operational status data, this invention automatically outputs initial process parameters and continuously adjusts the main draw ratio and hot box temperature, enabling the production process to more accurately meet the target elastic recovery rate requirements.
[0016] By employing a random forest regression algorithm, a multi-dimensional operational status fusion acquisition mechanism, and dynamic optimization decision-making based on conflict coordination rules, the formulation of process parameters becomes more robust and adaptable. By integrating measured elastic recovery rate, multi-source synchronous data, and adjustment instructions into a self-learning sample package and using iterative model updates, the model's intelligent decision-making capability is enhanced. This enables the production line to achieve more stable tension control and fiber diameter uniformity under different raw material batches, environmental conditions, and operating conditions, thereby improving the consistency and quality stability of high-elastic polyester filament products. Attached Figure Description
[0017] Figure 1 It is a process flow diagram for realizing an invention; Figure 2 It is a flowchart for model training and application in order to realize an invention; Figure 3 It is an online optimization and self-learning flowchart for realizing an invention; Figure 4 It is a flowchart of the internal integrated operation of a model for realizing an invention; Figure 5 It is a dynamic optimization decision-making flowchart for realizing an invention. Detailed Implementation
[0018] 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.
[0019] Example 1 like Figure 1-3 As shown, this embodiment of the invention provides a method for producing high-elastic polyester yarn for intelligent manufacturing, including: S1. Constructing a basic intelligent manufacturing model by offline training processing of historical high-elastic polyester yarn production data. The sources of historical high-elastic polyester yarn production data include different raw material batches, environmental conditions, and product specifications.
[0020] S2. The received production order information is parsed and processed to obtain the target elastic recovery rate specification. The polyester raw materials fed into the production line are monitored online to generate melt viscosity indicators. Online monitoring uses an online viscosity sensor. The parsing and processing includes extracting the target elastic recovery rate range from the production order information, and taking the median value of the target elastic recovery rate range as the target elastic recovery rate specification. The target elastic recovery rate range is 85%-95%. The online viscosity sensor is installed on the polyester melt conveying pipeline of the production line.
[0021] S3. Input the target elastic recovery rate specification and melt viscosity index as a joint vector into the basic intelligent manufacturing model. The basic intelligent manufacturing model performs comprehensive calculations on the joint vector to generate initial process parameters. The steps of the comprehensive calculations are as follows: S31. The target elastic recovery rate specification and melt viscosity index are fused together to form a joint vector.
[0022] S32. Input the joint vector into the fully connected network layer of the basic intelligent manufacturing model and perform nonlinear transformation to generate high-dimensional feature vectors.
[0023] S33. The high-dimensional feature vectors are subjected to regression processing in the output layer of the basic intelligent manufacturing model to output the initial process parameters.
[0024] The comprehensive processing adopts the random forest regression algorithm, which includes comprehensive operation on the joint vector. The comprehensive operation uses multiple decision trees, and the average of the output of multiple decision trees is used as the initial process parameters.
[0025] S4. Apply initial process parameters on the production line to produce high-elastic polyester yarn. Production includes a yarn winding process. A synchronous acquisition cycle is constructed based on the yarn length during the winding process. Multi-dimensional operational data from this synchronous acquisition cycle is fused and processed to form a multi-source synchronous dataset. This multi-dimensional operational data includes the tension value in the drafting zone and the fiber diameter measurement. A synchronous acquisition cycle is defined as when the yarn length reaches 500-1000 meters. The fusion and processing involves aligning the multi-dimensional operational data to form the multi-source synchronous dataset. The tension value in the drafting zone is acquired using a tension sensor installed between the drafting rollers on the production line. The fiber diameter measurement is obtained using a laser scanning diameter measuring instrument.
[0026] S5. Input the multi-source synchronous dataset into the basic intelligent manufacturing model. The basic intelligent manufacturing model performs dynamic optimization decisions on the multi-source synchronous dataset and outputs adjustment instructions. These adjustment instructions are then sent to the production line. The dynamic optimization decision adopts conflict coordination rules. The conflict coordination rules include prioritizing the uniform adjustment direction when the stabilization adjustment direction of the tension value in the drafting zone conflicts with the homogenization adjustment direction of the fiber diameter measurement value, while simultaneously relaxing the constraints on the stabilization adjustment direction. The adjustment instructions include a main draft ratio fine-tuning instruction, a first hot box temperature compensation instruction, and a second hot box temperature compensation instruction. The adjustment range of the main draft ratio fine-tuning instruction is ±0.5% to ±2% of the current set value; the adjustment range of the first hot box temperature compensation instruction is ±1℃ to ±5℃ of the current set value; and the adjustment range of the second hot box temperature compensation instruction is ±1℃ to ±3℃ of the current set value.
[0027] S6. After the yarn is wound into a complete packaged product, the measured elastic recovery rate of the complete packaged product is obtained. The measured elastic recovery rate, multi-source synchronous dataset, and adjustment instructions are integrated and processed to form a self-learning sample package. The self-learning sample package is fed back to the basic intelligent manufacturing model to form parameter iterative updates. The measured elastic recovery rate in the self-learning sample package ranges from 85% to 98%. The parameter iterative update includes optimizing the parameters of the basic intelligent manufacturing model with the measured elastic recovery rate as the target value. The optimized basic intelligent manufacturing model is then applied to production tasks with the same target elastic recovery rate specification and melt viscosity index.
[0028] By constructing a manufacturing model based on historical data, the system can automatically identify the differences in melt rheological properties between different batches of raw materials and generate a basis for process adjustment, thereby enhancing the adaptability of the process to raw material viscosity fluctuations and maintaining product performance stability.
[0029] By inputting order performance requirements and raw material status parameters into the model, the process settings have a clear performance orientation, making the elastic recovery rate of the final product more close to the target range, thus achieving controllability and stability of key quality indicators.
[0030] The synchronous acquisition of multi-dimensional operational status data enables the system to simultaneously grasp the stretching conditions and fiber geometric characteristics, and form multi-source correlations in the model, thereby improving the ability to identify subtle deviations in the forming process and reducing quality fluctuations caused by abnormal conditions.
[0031] By introducing conflict coordination rules, the system can automatically determine the priority and adjust the constraint range when multiple quality objectives conflict, thereby avoiding global performance degradation caused by single-parameter forced optimization and improving the system's stable control capability under complex conditions.
[0032] The setting of multiple independent adjustment commands enables the system to make smaller, hierarchical, and continuous adjustments to key process parameters, reducing operating disturbances caused by step actions and improving the precision of process control.
[0033] Multi-source synchronous datasets are continuously input into the model during the shaping stage, enabling the system to output adjustment actions in a timely manner based on the current operating trend, achieving rapid response to sudden disturbances and improving the continuous stability of the production process. The construction of self-learning sample packages allows the system to automatically accumulate actual operating characteristics after each production cycle, gradually improving the model's matching accuracy for different working conditions, forming an intelligent manufacturing system with continuous optimization capabilities.
[0034] By precisely controlling the draw ratio and heat treatment temperature, overheating and overdrawing can be reduced, thus lowering energy consumption. Simultaneously, by reducing waste yarn and abnormal package quantities, overall manufacturing costs can be lowered.
[0035] Intelligent control mechanisms can improve fiber diameter stability, tension fluctuation range, and forming consistency, providing a more stable raw material base for subsequent textile processing and improving the overall processing adaptability of finished products.
[0036] The system automatically generates corresponding process parameters based on the performance requirements of different orders, enabling the production line to have higher process reconfigurability and specification compatibility, reducing changeover time and improving equipment utilization.
[0037] Example 2 like Figure 4 As shown, this embodiment is a high-elastic polyester yarn production method based on intelligent manufacturing. By combining an intelligent manufacturing model with a real-time data acquisition system, key process parameters in the high-elastic polyester yarn production process are optimized to ensure stable product quality and improve production efficiency. The specific implementation method is as follows: 1. Production line preparation and synchronous data acquisition cycle setting This embodiment is based on a high-elastic polyester yarn production line in a textile factory, and adopts the following equipment and processes: Production equipment: Spinning machine: An automated device used to process polyester raw materials into high-elastic polyester yarn.
[0038] The drafting system includes multiple drafting rollers, with tension sensors installed between them to monitor the tension in the drafting zone in real time.
[0039] Laser diameter measuring instrument: installed at the production line exit to measure fiber diameter in real time.
[0040] Automated control system: integrates tension and temperature regulation modules to automatically adjust various parameters during the production process.
[0041] The production line is set to collect data every 750 meters. When the yarn reaches 750 meters, a data collection is triggered, and key data in the production process is collected synchronously.
[0042] Synchronous acquisition cycle setting: A complete synchronous data acquisition cycle begins whenever the length of the wound thread reaches 750 meters. The synchronous data acquisition cycle includes: Time range: Each acquisition cycle lasts approximately 80 seconds, and the acquired data includes tension values in the drafting zone and fiber diameter measurements.
[0043] During the synchronous acquisition cycle, the production line records the tension value of the stretching zone and the fiber diameter in real time.
[0044] 2. Data Acquisition and Processing Tension value in the stretching zone: Tension sensors monitor the tension in the drawing zone in real time. The following is the data collected during a specific production cycle: Table 1: Data collection table of tension values in the stretching zone.
[0045]
[0046] The data above shows that the tension value gradually increased to 12.3 cN, and then stabilized between 11.6 and 12.3 cN. The increase in tension may be due to changes in raw material temperature or slight fluctuations in equipment condition.
[0047] Fiber diameter: The laser scanning diameter measuring instrument measures the diameter of the filament in real time. The following are the diameter data during the synchronous acquisition cycle: Table 2: Fiber diameter data collection table.
[0048]
[0049] The data shows that the fiber diameter remained relatively stable within the range of 14.8-15.3 μm, but experienced a slight decrease at 50 seconds before returning to normal. This indicates that there may be slight production fluctuations during the synchronous acquisition cycle.
[0050] 3. Data Alignment and Synchronization: The Formation of Datasets During the synchronous acquisition period, the tension values in the drafting zone and the fiber diameter values are aligned according to time points to ensure synchronous analysis of multi-source data. The following is the aligned dataset, in seconds: Table 3: Aligned Synchronized Dataset Table.
[0051]
[0052] The above data will be fed into the intelligent manufacturing model as input to support subsequent process adjustments.
[0053] 4. Intelligent Manufacturing Model and Dynamic Optimization After data collection and the formation of a multi-source synchronous dataset, the system feeds the data into the intelligent manufacturing model. The intelligent manufacturing model, based on machine learning and deep learning algorithms, performs dynamic optimization and decision-making. The system uses historical and real-time data to calculate adjustment instructions to optimize production parameters.
[0054] Production optimization and dynamic adjustment: When the model detects that the fiber diameter deviates from the standard range, the system will adjust based on the following decisions: Tension adjustment: When the fiber diameter deviates, the system will increase the tension value in the stretching zone by about 0.5 cN to increase fiber stretching and control the diameter.
[0055] Temperature compensation: Based on the model's optimization suggestions, the system will adjust the hot box temperature by ±2℃ to improve the uniformity of heat treatment.
[0056] 5. Adjust command issuance and execution The optimized adjustment instructions are transmitted to the production line's automated control system. The system dynamically changes parameters such as equipment tension, draw ratio, and temperature based on these instructions. Tension adjustment: Increase the tension in the drawing zone to 12.8 cN; Temperature adjustment: Adjust the hot box temperature to +2℃ to improve temperature distribution; Fine-tuning of the drawing ratio: Fine-tune the drawing ratio by ±1%.
[0057] The above adjustments will be implemented automatically during the next production cycle to ensure that the production line is always in optimal operating condition.
[0058] 6. Results Validation and Performance Optimization After the above adjustments, the fiber diameter stability during the production process was improved, with the final product diameter stabilizing within the range of 15.0±0.2μm and the tension in the drawing zone stabilizing at 11.5±0.3cN. This improved stability helps increase production efficiency while reducing the scrap rate.
[0059] Quality verification: During the production process, the sample yarn was sent to the quality inspection laboratory for final inspection. The inspection results showed that the quality of the yarn met the standards and the customer's customization requirements.
[0060] By implementing the above steps, precise control and optimization of the high-elastic polyester filament production process were achieved. The intelligent manufacturing model, combined with multi-source synchronous datasets, dynamically adjusted production parameters, controlling the stability of fiber diameter and tension, and reducing fluctuations during production. Ultimately, the quality stability of the filament products was improved, with the diameter stabilized within the range of 15.0±0.2μm and the tension in the drawing zone stabilized at 11.5±0.3cN. This reduced the scrap rate and improved production efficiency, demonstrating the effectiveness of the aforementioned intelligent manufacturing method in improving product quality and production efficiency.
[0061] Example 3 like Figure 5 As shown, this embodiment is a high-elastic polyester yarn production method based on intelligent manufacturing. Through intelligent manufacturing optimization methods, process parameters are dynamically adjusted during the high-elastic polyester yarn production process to improve product quality and production efficiency. The specific implementation method is as follows: 1. Production data collection and input In the production process of high-elastic polyester yarn, the order information received by the production line requires a target elastic recovery rate of 90%. Based on this target, the system obtains the necessary production data through the following methods: Target elasticity recovery rate specification: 90%, the target elasticity recovery rate is specified through production orders, with a range of 85%-95%.
[0062] Melt viscosity index: 850 Pa·s. Melt viscosity is detected in real time by an online viscosity sensor, which indicates the melt flowability of polyester raw materials.
[0063] At this point, all data is input into the basic intelligent manufacturing model to begin optimizing the production process.
[0064] 2. Synchronous data acquisition and data fusion During the production process, as the yarn is wound, the system sets a synchronous data acquisition cycle based on the yarn length. The system performs a synchronous data acquisition every time the yarn reaches 700 meters in length, and the key parameters acquired include: Tension value in the drafting zone: The tension value in the drafting zone is measured in real time by a tension sensor installed between the drafting rollers. It is 120N and directly affects the fiber stretching and the final fiber quality.
[0065] Fiber diameter measurement: The fiber diameter was measured in real time as 0.20 mm using a laser scanning diameter measuring instrument. Fiber diameter is a crucial parameter to ensure that the fiber meets specifications.
[0066] After synchronous data acquisition, the system inputs the tension value of the drawing zone and the fiber diameter measurement value, along with the target elastic recovery rate specification and melt viscosity index, into the basic intelligent manufacturing model.
[0067] 3. Dynamic optimization decision-making of intelligent manufacturing models After receiving input data, the intelligent manufacturing model uses the random forest regression algorithm for dynamic optimization decisions. The random forest regression algorithm processes the multidimensional input data and outputs adjustment instructions, including the following: Main draw ratio fine-tuning command: The current setting is 10.0, and the adjustment range is ±1.5%, meaning the set value range after fine-tuning is 9.85 to 10.15. Main draw ratio fine-tuning is used to ensure the stability of the stretching process in the draw zone.
[0068] First heating chamber temperature compensation command: The current setting is 180℃, with an adjustment range of ±3℃, meaning the adjusted temperature range is 177℃ to 183℃. This is used to optimize temperature fluctuations during the heat treatment process and ensure consistent fiber melting state.
[0069] Second heating chamber temperature compensation command: The current setting is 190℃, and the adjustment range is ±2℃, meaning the adjusted temperature range is 188℃ to 192℃. This is used to ensure that the temperature of the second heating chamber meets the product production requirements and to avoid temperature fluctuations affecting fiber quality.
[0070] 4. Application of conflict resolution rules During the adjustment process, the system detected a conflict between the stabilization adjustment direction of the tension value in the drafting zone and the homogenization adjustment direction of the fiber diameter measurement value. According to the conflict reconciliation rules: When the tension adjustment direction conflicts with the diameter adjustment direction, the system prioritizes the fiber diameter uniformity adjustment direction. In this case, the system prioritizes adjusting the main draw ratio and the oven temperature to optimize the uniformity of the fiber diameter.
[0071] The system relaxes the constraints on the tension value in the drafting zone to stabilize the tension value, allowing for a moderate increase in the range of fine-tuning of the draft ratio, thus ensuring the consistency of fiber diameter and product quality.
[0072] 5. Adjustment and output of production results After dynamic optimization and adjustment, the system continues to monitor and output the following optimized production parameters: The adjusted main draw ratio is 9.95, which ensures good fiber stretching within the predetermined range.
[0073] The adjusted fiber diameter is 0.198mm, which is more precise than the original control and optimizes the uniformity of the fiber, ensuring that the final fiber quality meets the requirements.
[0074] The adjusted first heating box temperature is 180℃. Fine-tuning is performed within the set range to reduce the impact of temperature fluctuations on production.
[0075] The adjusted temperature of the second hot box is 189℃. After adjustment, the temperature is maintained within a suitable range to ensure the stability of the melt state.
[0076] The final product has an elastic recovery rate of 91%, which meets the target specification requirement of 90% ± 5% through actual testing.
[0077] 6. Self-learning and feedback in the production process After the yarn is wound into a complete packaged product, the system obtains a measured elastic recovery rate of 91% for the complete packaged product. It then integrates the measured elastic recovery rate, the synchronized dataset, and adjustment instructions to form a self-learning sample package. This self-learning sample package will be fed back into the basic intelligent manufacturing model for iterative parameter updates.
[0078] After model optimization, it learns from the new sample package, using the measured elastic recovery rate as the target value to optimize model parameters. The optimized model will be applied in the next batch of production to ensure smooth production when the target elastic recovery rate and melt viscosity are the same.
[0079] In summary, by applying a basic intelligent manufacturing model and a multi-source synchronous dataset, this embodiment achieves dynamic optimization of process parameters in the production of high-elastic polyester yarn. Adjustments to key parameters such as draw ratio and hot box temperature improve the uniformity of fiber diameter and elastic recovery rate, ensuring the final product meets target specifications. Conflict coordination rules play an optimizing role in the relationship between tension in the draw zone and fiber diameter adjustment, guaranteeing the stability of the production process and a high degree of consistency in product quality. This method provides a feasible solution for intelligent process optimization in the production of high-elastic polyester yarn.
[0080] 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 method for producing high-elastic polyester yarn for intelligent manufacturing, characterized in that, include: S1. Offline training and processing of historical high-elastic polyester filament production data to construct a basic intelligent manufacturing model; S2. The received production order information is parsed and processed to obtain the target elastic recovery rate specification. The polyester raw materials fed into the production line are detected online to generate melt viscosity index. The online detection uses an online viscosity sensor. S3. Input the target elastic recovery rate specification and the melt viscosity index as a joint vector into the basic intelligent manufacturing model. The basic intelligent manufacturing model performs comprehensive calculation on the joint vector to generate initial process parameters. S4. Apply the initial process parameters on the production line and produce high-elastic polyester yarn according to the initial process parameters. The production includes a yarn winding process. Construct a synchronous acquisition cycle according to the length of the yarn in the yarn winding process. Perform fusion acquisition and processing on the multi-dimensional operation status data of the synchronous acquisition cycle to form a multi-source synchronous dataset. The multi-dimensional operation status data includes the tension value of the stretching zone and the fiber diameter measurement value. S5. Input the multi-source synchronous dataset into the basic intelligent manufacturing model. The basic intelligent manufacturing model performs dynamic optimization decision-making on the multi-source synchronous dataset and outputs adjustment instructions. The adjustment instructions are then sent to the production line. The dynamic optimization decision-making adopts conflict coordination rules. S6. After the yarn is wound to form a complete packaged product, the measured elastic recovery rate of the complete packaged product is obtained. The measured elastic recovery rate, the multi-source synchronous dataset, and the adjustment instructions are integrated and processed to form a self-learning sample package. The self-learning sample package is fed back to the basic intelligent manufacturing model to form parameter iterative updates.
2. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The sources of the historical high-elastic polyester yarn production data include different raw material batches, environmental conditions, and product specifications.
3. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The analysis process includes extracting the target elastic recovery rate range from the production order information, taking the median of the target elastic recovery rate range as the target elastic recovery rate specification, wherein the target elastic recovery rate range is 85%-95%, and the online viscosity sensor is installed on the polyester melt conveying pipeline of the production line.
4. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The steps of the comprehensive calculation and processing are as follows: S31. Perform feature fusion processing on the target elastic recovery rate specification and the melt viscosity index to form the joint vector; S32. Input the joint vector into the fully connected network layer of the basic intelligent manufacturing model to perform nonlinear transformation and generate a high-dimensional feature vector; S33. The high-dimensional feature vector is subjected to regression processing at the output layer of the basic intelligent manufacturing model to output the initial process parameters.
5. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The comprehensive calculation process employs a random forest regression algorithm, which includes performing a comprehensive calculation on the joint vector. This comprehensive calculation uses multiple decision trees, and the average output of the multiple decision trees is used as the initial process parameters.
6. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The construction includes a synchronous acquisition cycle when the length of the silk thread reaches 500-1000 meters, and the fusion acquisition processing includes data alignment of the multi-dimensional operational status data to form the multi-source synchronous dataset.
7. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The tension value in the drafting zone is obtained using a tension sensor installed between the drafting rollers of the production line, and the fiber diameter is obtained using a laser scanning diameter measuring instrument.
8. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The conflict coordination rule includes prioritizing the uniform adjustment direction when the stabilization adjustment direction of the tension value in the drafting zone conflicts with the homogenization adjustment direction of the fiber diameter measurement value, while simultaneously relaxing the constraint range on the stabilization adjustment direction.
9. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The adjustment commands include a main draw ratio fine-tuning command, a first hot box temperature compensation command, and a second hot box temperature compensation command. The adjustment range of the main draw ratio fine-tuning command is ±0.5% to ±2% of the current set value, the adjustment range of the first hot box temperature compensation command is ±1℃ to ±5℃ of the current set value, and the adjustment range of the second hot box temperature compensation command is ±1℃ to ±3℃ of the current set value.
10. The method for producing high-elastic polyester yarn for intelligent manufacturing according to claim 1, characterized in that: The measured elastic recovery rate in the self-learning sample package ranges from 85% to 98%. The parameter iterative update includes optimizing the parameters of the basic intelligent manufacturing model using the measured elastic recovery rate as the target value, and applying the optimized basic intelligent manufacturing model to production tasks with the same target elastic recovery rate specification and melt viscosity index.