CN-121974192-A - Tension constant control system of high-capacity coiling machine

CN121974192ACN 121974192 ACN121974192 ACN 121974192ACN-121974192-A

Abstract

The application relates to the technical field of tension control of a coiling machine, in particular to a tension constant control system of a high-capacity coiling machine, which comprises a plurality of groups of detection units for collecting multidimensional state data, preprocessing the data, extracting characteristics to generate a comprehensive state vector, adjusting output tension control parameters and system response time in real time based on a self-adaptive control model, triggering a tension adjustment instruction and evaluating running stability through a nonlinear prediction model. The application can comprehensively reflect the running state of the coiling machine, eliminate external interference, ensure data consistency, fully excavate effective information of data, update model parameters in real time through dynamic learning capacity, improve the accuracy and timeliness of tension control, simultaneously plan maintenance work in advance, reduce shutdown loss, effectively ensure stable running of the high-capacity coiling machine and have higher popularization value.

Inventors

ZHANG JIANHUA
YANG YANG
YU YUE
ZHANG YICHI
QIAN ZIJIE

Assignees

江苏常铝铝业集团股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260320

Claims (10)

1. A tension constant control system of a high-capacity coiling machine is characterized by comprising the following steps of S1, collecting multi-dimensional state data in the operation process of the coiling machine through a plurality of groups of detection units, wherein the multi-dimensional state data at least comprises tension fluctuation data, load change data and speed regulation data, S2, preprocessing the multi-dimensional state data and comprises signal smoothing processing and dimension unification processing, S3, extracting characteristics of the preprocessed multi-dimensional state data to generate an integrated state vector, wherein the fluctuation amplitude, fluctuation frequency and fluctuation trend are extracted from the tension fluctuation data through a time domain analysis method, the change rate and stability indexes are respectively extracted from the load change data and the speed regulation data, the characteristics together form an integrated state vector, S4, inputting the integrated state vector into an adaptive control model obtained based on historical operation data for real-time regulation, outputting tension control parameters and system response time, the adaptive control model is provided with dynamic learning capability and can update internal parameters according to newly collected data, S5, when the response time of the system is lower than a set threshold, triggering tension regulation command is triggered, and S6, the integrated state is based on a linear operation condition prediction model under the current state of the coiling machine, and the stability is estimated.
2. The tension constant control system for a high-capacity reel as recited in claim 1, wherein in step S3, the fluctuation amplitude is a difference between a maximum value and a minimum value of the tension fluctuation data, the fluctuation frequency is a number of fluctuation periods per unit time, and the fluctuation trend is a change slope of the tension fluctuation data with time.
3. The tension constant control system for a high-capacity reel as recited in claim 1, wherein in step S3, the change rate is a change amount of the load change data and the speed adjustment data per unit time, and the stability index is a standard deviation of the load change data and the speed adjustment data.
4. The tension constant control system of a high capacity reel as claimed in claim 1, wherein in step S3, the fluctuation amplitude, fluctuation frequency, fluctuation trend, change rate and stability index are synthesized into a comprehensive state vector by a weight distribution method, and the weight distribution is dynamically adjusted according to the historical performance and actual importance of each data type.
5. The tension constant control system of a high capacity reel as recited in claim 1, wherein in step S4, the training process of the adaptive control model includes optimizing model parameters using an error feedback mechanism, the error feedback mechanism being implemented by calculating a deviation between a predicted value and an actual value.
6. The constant tension control system of a high capacity reel as recited in claim 5, wherein the training process of the adaptive control model uses a gradient optimization algorithm to iteratively update model parameters to reduce prediction errors.
7. The tension constant control system for a high-capacity coiler according to claim 1, wherein in step S4, the dynamic learning capability is realized based on an incremental learning algorithm, and a regularization term is introduced in the model parameter updating process to prevent overfitting by updating model parameters to adapt to newly acquired data.
8. The tension constant control system of a high-capacity coiler according to claim 7, wherein the parameter updating strategy in the incremental learning process adopts a piecewise linear adjustment mode, and the piecewise linear adjustment mode dynamically divides adjustment intervals according to data distribution characteristics.
9. The tension constant control system for a high-capacity reel as recited in claim 1, wherein in step S6, the training process of the nonlinear predictive model includes feature space mapping, the feature space mapping being implemented by constructing polynomial features.
10. The tension constant control system of a high capacity reel as recited in claim 9, wherein the order of the polynomial characteristic is dynamically adjusted based on data complexity.

Description

Tension constant control system of high-capacity coiling machine Technical Field The invention belongs to the technical field of control of a coiling machine, and particularly relates to a tension constant control system of a high-capacity coiling machine. Background With the continuous development of the aluminum industry, the application of the coiling machine in the aluminum strip production is more and more extensive, and the tension control is used as a key technology in the coiling process, so that the coiling quality and the production efficiency of the aluminum strip are directly affected. However, the existing coiler tension control method still has some drawbacks in coping with large-capacity, high-strength aluminum strip coiling, and in particular, there is room for improvement in terms of dynamic response speed, tension stability, and system adaptability. The existing coiling machine tension control method still has certain defects in the aspects of dynamic response speed, tension stability, system adaptability and the like when the coiling machine tension control method is used for coiling large-capacity and multi-specification aluminum strips. Therefore, the invention provides a tension constant control system of a high-capacity coiling machine, which aims to optimize a tension control algorithm, improve the dynamic response speed and adaptability of the system, simplify parameter input and calculation processes, ensure that high-efficiency and stable tension control is realized in the coiling process of aluminum strips with different specifications, and meet the requirements of the modern aluminum industry on high-quality and high-efficiency coiling equipment. Disclosure of Invention The invention solves the technical problems through the following technical proposal, and the invention comprises the following steps: step S1, acquiring multi-dimensional state data in the running process of a coiling machine through a plurality of groups of detection units, wherein the multi-dimensional state data at least comprises tension fluctuation data, load change data and speed regulation data; Step S2, preprocessing the multidimensional state data, including signal smoothing processing and dimension unification processing, so as to eliminate external interference and ensure data consistency; Step S3, extracting characteristics of the preprocessed multidimensional state data to generate a comprehensive state vector, wherein the tension fluctuation data adopts a time domain analysis method to extract fluctuation amplitude, fluctuation frequency and fluctuation trend; S4, inputting the comprehensive state vector into an adaptive control model which is trained based on historical operation data for real-time adjustment, and outputting tension control parameters and system response time; step S5, triggering a tension adjusting instruction when the response time of the system is lower than a set threshold value; and S6, based on the tension control parameters and the comprehensive state vector, evaluating the running stability of the coiling machine under the current working condition by using a nonlinear prediction model. Further, in step S3, the fluctuation amplitude, the fluctuation frequency and the fluctuation trend are obtained by time-series analysis of the tension fluctuation data, wherein the fluctuation amplitude is the difference between the maximum value and the minimum value of the data, the fluctuation frequency is the number of fluctuation periods in unit time, and the fluctuation trend is the change slope of the data with time; further, in step S3, the change rate and the stability index are obtained by statistical analysis of the load change data and the speed adjustment data, wherein the change rate is the change amount of the data in unit time, and the stability index is the standard deviation of the data; Further, in step S3, a weight distribution method is adopted to synthesize the fluctuation amplitude, the fluctuation frequency, the fluctuation trend, the change rate and the stability index into a comprehensive state vector, and the weight distribution is dynamically adjusted according to the historical performance and the actual importance of each data type; Further, in step S4, the training process of the adaptive control model includes optimizing model parameters using an error feedback mechanism, which is implemented by calculating a deviation between the predicted value and the actual value; further, the training process of the self-adaptive control model uses a gradient optimization algorithm, and model parameters are updated through iteration to reduce prediction errors; Further, in step S4, the dynamic learning capability is realized based on an incremental learning algorithm, and a regularization term is introduced in the model parameter updating process to prevent overfitting by updating the model parameters to adapt to the newly acquired data; The parameter updating strategy