CN-121981391-A - Weaving production management system based on internet of things

CN121981391ACN 121981391 ACN121981391 ACN 121981391ACN-121981391-A

Abstract

The invention relates to the technical field of intelligent manufacturing in textile industry, and particularly discloses a weaving production management system based on the Internet of things, which comprises a data acquisition module, a production quality monitoring module, a root cause analysis module and a process optimization module, wherein full-dimension real-time data of weaving production is synchronously acquired, a process reference database is arranged, a double-layer intelligent monitoring model is deployed and operated to monitor weaving production quality in real time by combining the full-dimension real-time data, whether process early warning or quality defects occur is judged, if the quality defects occur, a defect root cause analysis flow is triggered, if the quality defects occur, a data snapshot is extracted, the defect root cause is analyzed by combining a process early warning based on a causal discovery algorithm, process data in a historical period is generated, the process data is subjected to data treatment based on the data snapshot and the corresponding defect root cause, a process optimization data set is generated, the process reference database is optimized and updated by adopting a multi-objective evolutionary algorithm, and parameters are dynamically adjusted by combining the process data.

Inventors

FAN JIANLING
ZHOU XIAOPING
CHEN QIWEI
Zheng Guancheng

Assignees

浙江齐越新材料有限公司

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (10)

1. The weaving production management system based on the Internet of things is characterized by comprising the following modules: The data acquisition module synchronously acquires all-dimensional real-time data of weaving production, including raw material data, process data, equipment data and product data; The production quality monitoring module is used for setting a process reference database, deploying and operating a double-layer intelligent monitoring model to combine full-dimension real-time data to monitor the weaving production quality in real time, judging whether to trigger process early warning or quality defect occurrence, and triggering a defect root cause analysis flow if the quality defect occurs; the root cause analysis module extracts a data snapshot if triggered, and early-warning analysis of defect root causes is performed based on a causal discovery algorithm combining process; And the process optimization module is used for carrying out data management on the process data in the historical period based on the data snapshot and the corresponding defect root cause, generating a process optimization data set, carrying out optimization updating on a process reference database by adopting a multi-objective evolutionary algorithm, and carrying out parameter dynamic adjustment by combining the process data.
2. The Internet of things-based weaving production management system of claim 1, wherein the double-layer intelligent monitoring model comprises a process state monitoring layer and a product quality monitoring layer, and monitoring tasks of the two layers are executed in parallel by taking a full-dimension real-time data stream as a unified input source.
3. The Internet of things-based weaving production management system as set forth in claim 2, wherein the judgment mode of whether to trigger the process early warning is: Setting a numerical value allowed interval for each raw material data and equipment data, setting an allowed fluctuation range for each process data, calculating the deviation value of each process data and a corresponding process data reference in a process reference database in real time by a process state monitoring layer, and judging to trigger process early warning if any raw material data and equipment data exceed the set numerical value allowed interval or the deviation value of any process data continuously exceed the corresponding allowed fluctuation range for a preset duration.
4. The Internet of things-based weaving production management system as set forth in claim 2, wherein the judging mode of whether the quality defect occurs is as follows: The product quality monitoring layer identifies defect events according to production batches, performs accumulation statistics of real-time accumulated numbers of defect events with the same defect type and defect degree grade, judges that any two defect events identified in product data adjacent in time sequence are identical defect events if the actual position coincidence rate on a fabric product exceeds a preset coincidence standard, compares the real-time accumulated numbers with the acceptable numbers of the corresponding defect types and defect degree grades in the preset product quality standard, and judges that quality defects occur if any one real-time accumulated number exceeds the corresponding acceptable number.
5. The Internet of things-based weaving production management system of claim 4, wherein the defect event is identified by: and the visual detection analysis sub-module is embedded in the product quality monitoring layer, a pre-trained deep convolutional neural network model is adopted to process the surface image of the fabric product contained in the product data, an identification result data packet containing the detected defect type, the coordinate position of the defect in the surface image of the fabric product and the confidence coefficient of an identification result is output, and if the confidence coefficient exceeds a preset confidence standard, the defect event is judged to be identified.
6. The Internet of things-based weaving production management system of claim 5, wherein the defect level determination method is as follows: The defect type is classified into direct grading or indirect grading according to the basis, the defect degree grade is slightly, generally or seriously judged for the direct grading defect type, and the defect degree grade is judged by a lightweight classifier according to the actual physical size of the defects, the distribution density in the fabric width direction and the tolerance regulation in the product quality standard for the indirect grading defect type.
7. The Internet of things-based weaving production management system of claim 1, wherein the defect root cause analysis mode is as follows: If quality defects occur, a quality defect event is generated, a data snapshot taking a quality defect event time stamp as a center is extracted, raw material data, process data and equipment data which cause the quality defect event are identified as root cause parameters, all the root cause parameters in the data snapshot are processed by adopting a PC causal discovery algorithm based on constraint, a directed acyclic graph representing potential causal relations among the root cause parameters is constructed, causal effect values are calculated, the directed acyclic graph is analyzed, and parameters which trigger process early warning and have the highest causal effect values on quality defect event result nodes in the data snapshot period are determined to be main defect root causes.
8. The Internet of things-based weaving production management system of claim 1, wherein the process optimization data set is generated in the following manner: And in a preset historical period, if the main defect root corresponding to the quality defect event is irrelevant to the process data, eliminating the process data in the period covered by the associated data snapshot, and marking all the process data which are reserved after the rule screening as a process optimization data set.
9. The Internet of things-based weaving production management system as set forth in claim 1, wherein the process reference database is optimally updated by: deploying and running a multi-objective evolutionary algorithm taking an NSGA-II algorithm as a core; Defining a group of process data benchmarks for the same raw material type in a process benchmark database as an individual coded by real numbers, wherein each process data benchmark is used as a gene value of the individual, carrying out simulated production monitoring deduction by using the individual based on a process optimization data set, designing a double-target fitness function, obtaining a pareto optimal solution through iterative evolution as an optimal individual, and updating the gene value of the optimal individual, namely the optimized process data benchmark, to a record of the corresponding raw material type in the process benchmark database.
10. The Internet of things-based weaving production management system as set forth in claim 9, wherein the design of the dual-objective fitness function is specifically: The method comprises the steps of firstly, calculating the proportion of the number of simulated production batches with quality defects to the number of total simulated production batches through simulated playback, and secondly, calculating the sum of production stability score and rotation speed potential coefficient to obtain the maximum production efficiency index, wherein the production stability score is calculated based on the time length proportion of operation of process data in a fluctuation-allowed interval, and the rotation speed potential coefficient is obtained through normalization processing of the rotation speed of a weaving machine.

Description

Weaving production management system based on internet of things Technical Field The invention relates to the technical field of intelligent manufacturing in textile industry, in particular to a weaving production management system based on the Internet of things. Background In the current background of intelligent transformation of the textile industry, fine management of weaving production has become the core for improving the competitiveness of the industry. However, when the conventional production management mode and the prior art scheme are used for coping with market demands of high quality, multiple varieties and fast delivery, the following technical problems to be solved are exposed: The data acquisition dimension is single and isolated, and global analysis is difficult to support. The prior art focuses on the acquisition and monitoring of data with single dimensions such as equipment rotating speed, yield and the like, and forms a data island. The real-time change of the raw material line diameter, the temperature and humidity and dust concentration of the production environment, microscopic defect images on the surface of the fabric and other key information often depend on manual spot check or independent system recording, and synchronous acquisition and unified association are lacking. The quality monitoring is lagged and depends on manpower, and the defect root causes are difficult to trace. At present, although part of process parameters can be recorded in a production execution system (MES) which is commonly adopted, the monitoring of the MES is mostly stopped at an overrun alarm level, and the MES belongs to post-record. The judgment of the quality defects of the fabrics is seriously dependent on the eye force and experience of a skilled quality inspector, and has the problems of low efficiency, different standards and high omission rate. More importantly, even if defects are found, due to the lack of synchronous associated snapshots of multi-source data, engineering technicians can only carry out assumption and investigation by experience when tracing the defect causes, and cannot quickly and objectively locate the causes such as raw material fluctuation, process parameter drift, equipment abnormality and the like, so that problems repeatedly occur, and the quality cost is high. Thirdly, the process reference is statically stiff, and dynamic optimization cannot be achieved. The process parameter criteria in the prior art are typically based on historical experience or fixed recipe settings and once determined are constant over a long period of time, however, dynamic shifts often occur in actual production. Aiming at the problems, the invention provides a weaving production management system based on the Internet of things. Disclosure of Invention The invention aims to provide a weaving production management system based on the Internet of things, so as to solve the background problem. The aim of the invention can be achieved by the following technical scheme that the weaving production management system based on the Internet of things comprises: The data acquisition module synchronously acquires all-dimensional real-time data of weaving production, including raw material data, process data, equipment data and product data; The production quality monitoring module is used for setting a process reference database, deploying and operating a double-layer intelligent monitoring model to combine full-dimension real-time data to monitor the weaving production quality in real time, judging whether to trigger process early warning or quality defect occurrence, and triggering a defect root cause analysis flow if the quality defect occurs; the root cause analysis module extracts a data snapshot if triggered, and early-warning analysis of defect root causes is performed based on a causal discovery algorithm combining process; The process optimization module is used for carrying out data management on the process data in the historical period based on the data snapshot and the corresponding defect root cause to generate a process optimization data set, carrying out optimization updating on a process reference database by adopting a multi-objective evolutionary algorithm, and carrying out parameter dynamic adjustment by combining the process data; further, the double-layer intelligent monitoring model comprises a process state monitoring layer and a product quality monitoring layer, and monitoring tasks of the two layers are executed in parallel by taking a full-dimension real-time data stream as a unified input source; further, the judgment mode of whether to trigger the process early warning is as follows: Setting a numerical value allowed interval for each raw material data and equipment data, setting an allowed fluctuation range for each process data, calculating the deviation value of each process data and a corresponding process data reference in a process reference database in real time by a pro