An online upgrade method for an intelligent control system of noil distance of a combing machine

Abstract

The present invention relates to the field of online upgrade and update technology, specifically to an online upgrade method for a combing machine noil spacing intelligent control system, which realizes efficient online upgrade of the system through three steps: module operation analysis, video recognition and algorithm adaptation evaluation, and algorithm version query and upgrade. First, based on the historical monitoring parameter analysis function module operation status, accurately identify the offset module and download the corresponding update file, and complete the dynamic update of the offset module according to the real-time status; secondly, by real-time acquisition of raw material video and combing video, use image analysis to extract current raw material parameters and combing quality parameters, compare with the applicable scope of the intelligent control algorithm, evaluate the algorithm adaptability and generate an upgrade reminder; finally, query the remote server to obtain a list of high-version intelligent control algorithms, and perform algorithm upgrades according to working conditions. This method improves the accuracy and stability of the system's online upgrades, ensuring efficient operation and quality optimization of the combing process.

Description

Technical Field

[0001] The invention relates to the technical field of online upgrading and updating, and in particular to an online upgrading method of an intelligent control system for cotton drop spacing of a combing machine. Background Art

[0002] Combing machine is an important mechanical equipment in textile production, mainly used to remove impurities (such as lint, grass clippings and cotton neps, etc.) from fibers and improve the uniformity of fibers. The combing machine noil gauge refers to the distance between the front edge of the lower nipper lip and the surface of the detaching roller when the nipper swings to the frontmost position during the operation of the combing machine. The combing machine noil gauge is an important process parameter. In order to improve the automation and intelligence level of combing machine noil gauge setting, modern combing machines have gradually introduced a noil gauge intelligent control system. The system uses intelligent technology to realize real-time detection and dynamic control of the combing machine noil gauge, which can effectively improve the stability, quality and production efficiency of the combing process.

[0003] However, the existing combing machine cotton drop gauge intelligent control system generally faces significant challenges during the online upgrade process, especially in the case of system module failure, raw material fluctuations or combing quality degradation. Due to the lack of an accurate upgrade judgment mechanism, it is easy to trigger a blind upgrade problem. The existing upgrade method often ignores the comprehensive evaluation of the operating status of the intelligent control system, and cannot effectively distinguish between real performance deviations and normal parameter fluctuations, resulting in system operation interruption, resource waste, and even equipment operation chaos. Especially in a dynamically changing environment, the upgrade trigger conditions of the existing technology are too simple and lack the ability to comprehensively analyze multi-dimensional parameters and working conditions. Therefore, how to ensure the stability of the combing machine cotton drop gauge intelligent control system while achieving accurate judgment and adaptive upgrades of complex working conditions and dynamic changes has become an important issue that needs to be solved to improve production efficiency and equipment intelligence.

[0004] Therefore, an online upgrade method for the intelligent control system of comber noil spacing was proposed. Summary of the invention

[0005] The purpose of the present invention is to provide an online upgrade method for a combing machine cotton drop gauge intelligent control system. The method adopts a three-stage sequential design of module operation analysis and update, video recognition and algorithm adaptation evaluation, and algorithm version query and upgrade, which can ensure the accuracy of the system upgrade process. First, the module operation analysis and update step is used as the basis. The operation status of each functional module is analyzed through historical monitoring parameters, and the offset module that needs to be adjusted is accurately identified, so that the upgrade process is limited to the actual required range, avoiding the waste of resources and operation risks that may be caused by blind upgrades, and providing a basis for subsequent evaluation and upgrade steps. The video recognition and algorithm adaptation evaluation step is based on the real-time collected raw material video and combing video, combined with image analysis to extract the current production parameters, and compares them with the applicable parameter range of the first intelligent control algorithm to further confirm whether the current module is suitable for continued operation or needs to be upgraded, ensuring that the upgrade decision is based on the accurate evaluation of dynamic production conditions, and enhancing the flexibility and scientificity of algorithm adaptation. Finally, in the third stage, after ensuring that the module and working condition analysis has been completed in the first two stages, the remote server is queried to clarify the list of high-version intelligent control algorithms and implement upgrades, avoiding blind upgrades due to module failures and raw material fluctuations, and avoiding unnecessary upgrade interference or performance degradation. This sequential design meets the actual needs of production. First, the problem module is located, the offset module is updated and repaired, and then the real-time working conditions are evaluated. Finally, after confirming that there are no module anomalies and parameter offsets in the system, the optimization and upgrade of the higher version algorithm is performed. Through the comprehensive analysis and repair of the first two stages, conditions are created for the smooth upgrade of the higher version algorithm. Through this step-by-step approach, the security, accuracy and dynamic adaptability of the entire upgrade process are ensured, effectively avoiding the confusion and risks caused by blind upgrades.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An online upgrade method for a combing machine noil gauge intelligent control system, comprising:

[0008] S101: Acquire a first historical monitoring parameter of a designated comber noil gauge intelligent control system within a designated time period, wherein the comber noil gauge intelligent control system has a built-in first intelligent control algorithm and multiple functional modules;

[0009] S102: Analyze the operating status of each of the functional modules according to the first historical monitoring parameters, and determine the offset module;

[0010] S103: Downloading an update file corresponding to the offset module from a remote server;

[0011] S104: Loading the update file to update the offset module according to the real-time status of the offset module;

[0012] S201: collecting raw material video and combing video in real time, and obtaining current parameters by image analysis, including raw material parameters and combing quality parameters;

[0013] S202: Obtaining an applicable parameter range of the first intelligent control algorithm, including a raw material parameter range and a combing quality parameter range;

[0014] S203: determining whether the raw material parameter range is not within the expected raw material parameter range or the combing quality parameter range is not within the expected combing quality parameter range, and obtaining a determination result;

[0015] If the judgment result is yes, directly execute S301;

[0016] S204: If the judgment result is no, generating upgrade reminder information according to the current parameters;

[0017] S301: Query the first updated version corresponding to the first intelligent control algorithm on the remote server to obtain a list of higher-version intelligent control algorithms;

[0018] S302: Upgrading the first intelligent control algorithm according to the high-version intelligent control algorithm list.

[0019] Furthermore, the first historical monitoring parameter is specifically:

[0020] The first historical monitoring parameters include a mechanical parameter sequence, a raw material parameter sequence, a system control parameter sequence, an environmental parameter sequence, a combed fiber quality parameter sequence and a system fault sequence; the mechanical parameter sequence includes a timestamp, a nipper movement amplitude, a separation roller rotation speed, a cotton drop spacing dynamic control parameter and vibration data of each of the functional modules, and the vibration data includes vibration amplitude and frequency; the raw material parameter sequence includes a timestamp, an image frame sequence, an effective length, a uniformity index, a short fiber content, a foreign matter content, an average size of foreign matter particles and an image grayscale uniformity index; the system control parameter sequence includes a timestamp, a draft ratio, a tension adjustment value and a nipper pressure; the environmental parameter sequence includes a timestamp, a temperature, a humidity and a noise level; the combed fiber quality parameter sequence includes a timestamp, an image frame sequence, a number of cotton knots, a residual foreign matter rate and a combed fiber separation degree; the system fault sequence includes a timestamp, a fault type, an impact module, a subsequent processing method, a fault recovery time and a fault severity level.

[0021] Furthermore, according to the first historical monitoring parameters, the operating status of each of the functional modules is analyzed to determine the offset module, including: constructing multidimensional time series data by associating mechanical parameter sequence, raw material parameter sequence, system control parameter sequence, environmental parameter sequence and combed fiber quality parameter sequence through timestamps; inputting the preprocessed multidimensional time series data into a trained offset detection model to obtain an offset detection result, and determining the offset module according to the offset detection result.

[0022] Furthermore, the training process of the offset detection model includes:

[0023] The second historical monitoring parameters of the comber noil spacing intelligent control system within the historical time period are obtained; the second historical monitoring parameters include a historical mechanical parameter sequence, a historical raw material parameter sequence, a historical system control parameter sequence, a historical environmental parameter sequence, a historical combed fiber quality parameter sequence and a historical system fault sequence; according to the historical system fault sequence, the data segments in the second historical monitoring parameters are annotated to obtain the annotated results; the second historical monitoring parameters are preprocessed to obtain historical multidimensional time series data; the preprocessing includes data alignment, missing value processing, outlier removal and data standardization; the basic features, cross features and high-order features of the historical multidimensional time series data are extracted; the basic features, cross features and high-order features are screened by using a feature importance scoring method, redundant features are removed, and an offset detection feature matrix is ​​generated; the offset detection feature matrix is ​​divided into a training set and a test set in chronological order; the offset detection feature matrix in the training set and the corresponding annotated results are input into an LSTM model for training to obtain an initial offset detection model; the hyperparameters of the initial offset detection model are optimized using the test set to obtain the offset detection model.

[0024] Further, according to the real-time status of the offset module, loading the update file to update the offset module includes:

[0025] Perform MD5 verification and compatibility verification on the update file, the compatibility verification includes checking the matching of the functional interface between the update file and the corresponding offset module, data structure consistency and adaptability to the first intelligent control algorithm; if the MD5 verification and the compatibility verification are passed, enable the dual-module hot switching mechanism, load the copy instance of the offset module corresponding to the update file into an independent operating environment, and isolate the independent operating environment from the existing module operating environment to prevent real-time operation interruption; collect the synchronization data of the current offset module in real time, including the input data, operating status and interaction information between the offset module and the comber cotton drop spacing intelligent control system; inject the synchronization data into the update module, collect and verify the output data; analyze the The real-time status of the offset module includes real-time collection and analysis of dynamic control parameters of cotton drop spacing, separation roller speed, clamp plate swing characteristics, fiber parameters and combing quality parameters to determine whether the update conditions are met; if the update conditions are met, identify low-priority tasks and assign the low-priority tasks to the update module for operation; during the operation, the fiber parameter deviation and cotton drop spacing parameter deviation of the update module are monitored in real time. If the parameter deviation and the cotton drop spacing parameter deviation are both within the corresponding deviation range, the task load distribution is gradually adjusted according to the preset task allocation weights until the update module completely takes over all tasks and replaces the current offset module. After the task transfer is completed, the system resources of the current offset module are released to complete the update process of the offset module.

[0026] Furthermore, real-time acquisition of the raw material video includes: utilizing a first video monitoring device to acquire the video of the raw material in real time at a first acquisition point to obtain the raw material video; the first acquisition point is located on the raw material conveying path between the cotton feeder outlet and the comber feeding inlet, and the first video monitoring device is installed above the conveying channel within a specified range of the cotton feeder outlet, and fixedly captures the state of the raw material after uniform laying at a top-down angle.

[0027] Furthermore, real-time collection of the combing video includes: using a second video monitoring device to collect real-time video of the raw material after the combing process at a second collection point to obtain the combing video; the second collection point is located on the conveying channel between the output end of the combing machine and the entrance of the subsequent process, and the second video monitoring device is installed above the conveying channel to shoot the fiber state after the combing process at a fixed angle.

[0028] Furthermore, the raw material video and the combing video are collected in real time, and the current parameters are obtained by image analysis:

[0029] Performing image frame processing on the raw material video and the combed video to obtain a current raw material image and a current combed image;

[0030] Preprocessing the current raw material image and the current combed image to obtain a current raw material fiber image and a current combed fiber image; the preprocessing includes denoising, histogram equalization, background segmentation and illumination normalization;

[0031] Performing image analysis on the current fiber image of the raw material to obtain the raw material parameters, including effective length, uniformity index, short fiber content, impurity content, average size of impurity particles and image gray uniformity index;

[0032] The combed current fiber image is subjected to image analysis to obtain the combing quality parameters, including the number of neps, the residual impurity rate and the combed fiber separation degree.

[0033] Furthermore, generating the upgrade reminder information according to the current parameters includes: if any parameter of the raw material parameters exceeds the corresponding raw material parameter range, identifying a first out-of-range parameter; matching an adaptive intelligent control algorithm set according to the first out-of-range parameter to obtain a second intelligent control algorithm; if any parameter of the combed fiber quality parameters exceeds the corresponding combed fiber quality parameter range, identifying a second out-of-range parameter; matching an adaptive intelligent control algorithm set according to the second out-of-range parameter to obtain a third intelligent control algorithm; performing an intersection operation on the second intelligent control algorithm and the third intelligent control algorithm to generate an upgrade algorithm list; generating the upgrade reminder information according to the upgrade algorithm list.

[0034] Furthermore, the online upgrade method also includes: identifying the functional module whose operating status has failed according to the first historical monitoring parameter to obtain the failed module; generating failure warning information for the failed module, and publishing the failure warning information through the system interface.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. The present invention proposes an online upgrade method for a combing machine drop gauge intelligent control system based on three stages. The module operation analysis and update step of the first stage obtains historical monitoring parameters and combines a comprehensive analysis of the module operation status to accurately identify the offset module, and only performs update operations on the modules that need to be updated, thereby avoiding unnecessary resource consumption and risks caused by blind upgrades, and ensuring the pertinence and efficiency of the upgrade process. The video recognition and algorithm adaptation evaluation step of the second stage collects raw materials and combing videos in real time, extracts key parameters using image analysis, and compares them with the applicable parameter range of the first intelligent control algorithm to accurately evaluate the adaptability of the current algorithm, generate upgrade reminder information, and provide a basis for algorithm updates under dynamic conditions. At the same time, combined with algorithm adaptation analysis, the control accuracy is improved. The algorithm version query and upgrade step of the third stage queries the algorithm version of the remote server, obtains a list of high-version intelligent control algorithms, and completes the upgrade operation, ensuring that the system can obtain the latest algorithm optimization functions in a timely manner. The three-stage design fully combines the synergistic effects of module operation status evaluation, parameter identification and intelligent control algorithm upgrade. It can not only dynamically adapt to complex production conditions, but also ensure the safety, efficiency and precision of the upgrade process, providing strong technical support for the stability and production efficiency of the combing machine cotton drop spacing intelligent control system.

[0037] 2. The present invention proposes a method for realizing non-interrupted online upgrade of the combing machine's cotton drop gauge intelligent control module through a dual-module hot switching mechanism and real-time verification of synchronous data. Through MD5 verification and compatibility verification, the integrity and adaptability of the update file are ensured to avoid system failures due to file errors or incompatibility; synchronous data is collected in real time and injected into the update module to verify its functional consistency and operational stability to prevent quality problems caused by blind upgrades; combined with dynamic analysis of key control parameters (such as cotton drop gauge parameters, fiber parameters, etc.) and task priority allocation mechanism, the task takeover process is ensured to be smooth and risk-free by gradually transferring the task load; at the same time, by real-time monitoring of the operating deviation of the update module, it is ensured that it takes over the task after reaching a stable state, thereby reducing performance fluctuations or abnormal risks during the upgrade process, which helps to improve production efficiency and the level of equipment intelligence.

[0038] 3. The present invention realizes the intelligent control algorithm evaluation and accurate generation of upgrade suggestions by introducing the out-of-range parameter identification and adaptation algorithm matching mechanism in the monitoring of raw material parameters and combed fiber quality parameters. Specifically, by identifying the first out-of-range parameter in the raw material parameters and the second out-of-range parameter in the combed fiber quality parameters, the adaptive intelligent control algorithm sets are matched respectively, and the algorithm set is subjected to intersection operation to generate an upgrade algorithm list, thereby avoiding unreasonable algorithm recommendations caused by single parameter anomalies, and ensuring that the adaptation algorithm is highly consistent with the actual working conditions. This design effectively solves the efficiency and quality problems caused by insufficient response to complex working conditions or blind algorithm upgrades in traditional control systems, and provides a reliable basis for subsequent control optimization and intelligent upgrades by generating accurate upgrade reminder information, thereby significantly improving the control system's adaptability to dynamic working conditions and production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flowchart of the main steps of an online upgrade method of an intelligent control system for cotton noil spacing of a combing machine provided by an embodiment of the present invention;

[0040] Figure 2 A flowchart of the detailed steps of an online upgrade method of an intelligent control system for cotton drop spacing of a combing machine provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The gauge of comber noil directly affects the combing effect of fibers and the quality of yarn. If the gauge is too large, high-quality fibers may be wasted. If the gauge is too small, insufficient noil may occur and impurities may remain, thus reducing the quality of yarn. During the combing process, the gauge needs to be adjusted at any time due to the influence of the raw fiber state (length, fineness, etc.), mechanical wear and environmental factors (humidity, temperature). Therefore, the introduction of an intelligent control system for the noil gauge of combers is an inevitable choice for factories in the process of digital transformation.

[0043] See also Figure 1 to Figure 2 The present invention provides an online upgrade method for a combing machine noil gauge intelligent control system, and the technical solution is demonstrated in detail through the following two embodiments:

[0044] Embodiment 1:

[0045] Company A is an enterprise focusing on textile production. On one of its textile production lines, it is equipped with an intelligent control system for the cotton drop distance of the combing machine. The production line runs continuously during the whole working day to efficiently complete the fiber processing and combing process.

[0046] The comber noil gauge intelligent control system has built-in intelligent control algorithms and multiple functional modules. Through modular design, the system divides different functional requirements into independent logical units (functional modules), and takes the intelligent control algorithm as the core to collaboratively complete the dynamic control of the combing process. Functional modules are logical units that implement specific functions in the system, including nipper control module, detaching roller control module, noil gauge adjustment module, vibration monitoring module and environmental parameter monitoring module, etc. They are responsible for collecting and processing specific types of data and performing corresponding control tasks. The intelligent control algorithm is used to analyze the collected operating data (such as noil gauge parameters and equipment vibration data, etc.) in real time, and dynamically adjust each functional module according to the process requirements to achieve the optimization and stabilization of the comber operation, including adjusting the nipper movement, detaching roller speed and noil gauge setting. The intelligent control algorithm is the core decision-making logic of the system, responsible for parsing real-time data and generating optimization instructions, while the functional module is responsible for specific execution operations and data collection. Through the collaborative work of the two, efficient control of the combing process is achieved.

[0047] The built-in intelligent control algorithm and each functional module of the comber noil spacing intelligent control system are controlled by the corresponding software, and the software is all upgradeable. It can analyze and update the operating status of each module separately, so as to achieve accurate upgrade of each module. The system adopts a modular architecture, so that each functional module can independently load the update file for dynamic upgrade under the premise of meeting compatibility and real-time performance, and support the version optimization of the overall intelligent control algorithm.

[0048] The comber noil gauge intelligent control system applies an online upgrading method of the comber noil gauge intelligent control system.

[0049] See also Figure 1 S1 in the module operation analysis and update step includes the following:

[0050] See also Figure 2 S101 in which a first historical monitoring parameter of a designated comber noil gauge intelligent control system within a designated time period is obtained, wherein the comber noil gauge intelligent control system has a first intelligent control algorithm and a plurality of functional modules built therein;

[0051] In this embodiment, the specified time period can be selected according to actual requirements; the specified time period refers to a continuous time interval from a set starting point in the past to the current time point. For example, you can choose to start from the time span of the last hour, day, or week, covering the latest monitoring data of the current system in real time. This time period contains all relevant data of the system under the current production conditions, so it is real-time and dynamic, and is not limited to static historical data analysis, ensuring that the analysis results can reflect the current status and make more accurate decisions based on recent operating trends.

[0052] Furthermore, the first historical monitoring parameter is specifically:

[0053] The first historical monitoring parameter is time series data containing timestamps, including mechanical parameter sequence, raw material parameter sequence, system control parameter sequence, environmental parameter sequence, combed fiber quality parameter sequence and system fault sequence; the mechanical parameter sequence includes timestamp, nipper movement amplitude, separation roller speed, cotton drop spacing dynamic control parameters and vibration data of each functional module, and the vibration data includes vibration amplitude and frequency; the raw material parameter sequence includes timestamp, image frame sequence, effective length, uniformity index, short fiber content, impurity content, average size of impurity particles and image grayscale uniformity index; the system control parameter sequence includes timestamp, draft ratio, tension adjustment value and nipper pressure; the environmental parameter sequence includes timestamp, temperature, humidity and noise level; the combed fiber quality parameter sequence includes timestamp, image frame sequence, number of cotton knots, impurity residual rate and combed fiber separation degree; the system fault sequence includes timestamp, fault type, affecting module, subsequent processing method, fault recovery time and fault severity level.

[0054] The present invention comprehensively integrates and correlates mechanical parameters, raw material parameters, system control parameters, environmental parameters, combed fiber quality parameters and system fault information by constructing multidimensional time series data containing timestamps, and realizes accurate monitoring and comprehensive evaluation of the dynamic regulation process of comber cotton drop spacing. Mechanical parameters (such as nipper plate movement amplitude, separation roller speed, etc.) directly reflect the operating status of the equipment, raw material parameters (such as effective length, impurity content, etc.) and combed fiber quality parameters (such as cotton knot number, separation degree, etc.) provide the key connection between processing raw materials and combing process quality, and system control parameters and environmental parameters further compensate for the influence of external conditions and operation control on overall regulation. Through the integration of multidimensional parameters, not only can potential anomalies in equipment operation be effectively identified, but also fault tracing and recovery guidance can be provided through fault information. Compared with the monitoring method of a single parameter, the present invention can capture the complex interactive relationship in the production process, improve the accuracy of regulation and the stability of system operation, and provide a strong guarantee for achieving efficient combing processing and refined management.

[0055] According to the first historical monitoring parameter, the functional module whose running status has failed is identified to obtain the failed module (the abnormal functional module that cannot be restored by automatic update); failure warning information is generated for the failed module, and the failure warning information is released through the system interface or remote notification function.

[0056] By analyzing the first historical monitoring parameter, the present invention can quickly and accurately identify functional modules with failed operating status and generate failure warning information. This method significantly improves the system's fault perception ability and avoids chain failures caused by module failures that are not discovered in time. In particular, for abnormal modules that cannot be restored through automatic updates, accurate fault location and clear processing instructions are provided, thereby shortening downtime, reducing the impact of faults on production, and further improving the stability and reliability of system operation.

[0057] See also Figure 2 In S102, after the failed module is repaired, the operating state of each functional module is analyzed according to the first historical monitoring parameter to determine the offset module;

[0058] Furthermore, according to the first historical monitoring parameter, the operating status of each functional module is analyzed, and the offset module is determined, including: constructing multidimensional time series data by associating mechanical parameter sequence, raw material parameter sequence, system control parameter sequence, environmental parameter sequence and combed fiber quality parameter sequence through timestamp; inputting the preprocessed multidimensional time series data into the trained offset detection model to obtain the offset detection result, and determining the offset module according to the offset detection result.

[0059] The present invention can realize a comprehensive analysis of the operating status of each functional module by constructing multidimensional time series data. The introduction of multidimensional time series data enables the dynamic interaction relationship between different parameters to be revealed, providing data support for the accurate identification of the offset module. Through the trained offset detection model, the state deviation degree of the module is automatically analyzed, and potential abnormal modules or performance degradation modules can be discovered in time, improving the system's fault prediction ability and operational stability, laying the foundation for subsequent upgrades, avoiding blind upgrades, and making system upgrades more targeted and effective, thereby ensuring the balance between system optimization and operation.

[0060] Furthermore, the training process of the offset detection model includes:

[0061] The second historical monitoring parameters of the combing machine noil spacing intelligent control system within the historical time period are obtained; the second historical monitoring parameters include historical mechanical parameter sequence, historical raw material parameter sequence, historical system control parameter sequence, historical environmental parameter sequence, historical combed fiber quality parameter sequence and historical system fault sequence; the historical system fault sequence includes timestamp, fault type, impact module, subsequent processing method, fault recovery time and fault severity level.

[0062] According to the historical system fault sequence, the data segments in the second historical monitoring parameter are annotated to obtain the annotation results; the second historical monitoring parameter is preprocessed to obtain historical multidimensional time series data; the preprocessing includes data alignment, missing value processing, outlier removal and data standardization; the historical multidimensional time series data is feature engineered to extract basic features, cross features and high-order features; the basic features, cross features and high-order features are screened using the feature importance scoring method, redundant features are removed, and the offset detection feature matrix is ​​generated; the offset detection feature matrix is ​​divided into a training set and a test set in chronological order; the offset detection feature matrix in the training set and the corresponding annotation results are input into the LSTM model for training to obtain the initial offset detection model; the hyperparameters of the initial offset detection model are optimized using the test set to obtain the offset detection model. The offset detection model realizes high-precision identification of module operation state deviation. This method effectively utilizes historical system fault data, models the dynamic change characteristics of time series data through a deep learning model, and can accurately capture the potential abnormal trends of module performance. This method significantly improves the automation level of fault identification and the accuracy of prediction, provides reliable decision-making support for module operation status analysis, optimization adjustment and subsequent system upgrades, and further ensures the stability and efficient operation of the combing machine system.

[0063] Specifically, some basic features obtained after extracting basic features are shown in Table 1; some cross features constructed are shown in Table 2; and some high-order features are shown in Table 3.

[0064] Table 1 Basic feature example table

[0065]

[0066] Table 2 Cross-feature example table

[0067]

[0068] Table 3 High-level feature example table

[0069]

[0070] See also Figure 2 In S103, downloading the update file corresponding to the offset module from the remote server;

[0071] Specifically, according to the software update requirements corresponding to the offset module matching, an offset module update list is generated; according to the offset module update list, the corresponding update file is queried and downloaded on the remote server; the update file is a software update package or a patch file.

[0072] See also Figure 2In S104, according to the real-time status of the offset module, the update file is loaded to update the offset module;

[0073] Further, according to the real-time status of the offset module, loading the update file to update the offset module includes:

[0074] Perform MD5 verification and compatibility verification on the update file. The compatibility verification includes checking the matching of the functional interface between the update file and the corresponding offset module, data structure consistency, compatibility with other functional modules and compatibility with the first intelligent control algorithm. If the MD5 verification and compatibility verification are passed, enable the dual-module hot-swap mechanism, load the copy instance of the offset module corresponding to the update file into an independent operating environment, and isolate the independent operating environment from the existing module operating environment to prevent real-time operation interruption. At the same time, collect the synchronization data of the current offset module in real time, including the input data, operating status and interaction information between the offset module and the comber cotton drop spacing intelligent control system of the current offset module. Inject the synchronization data into the update module, collect and verify the output data. The update module is used to load the update. file; analyze the real-time status of the offset module, including real-time collection and analysis of dynamic control parameters of cotton drop spacing, separation roller speed, clamp plate swing characteristics, fiber parameters and combing quality parameters, to determine whether the update conditions are met; if the update conditions are met, identify low-priority tasks and assign low-priority tasks to the update module for operation; during the operation, monitor the fiber parameter deviation and cotton drop spacing parameter deviation of the update module in real time. If the parameter deviation and the cotton drop spacing parameter deviation are both within the corresponding deviation range, gradually adjust the task load distribution according to the preset task allocation weights until the update module completely takes over all tasks and replaces the current offset module. After the task transfer is completed, release the system resources of the current offset module to complete the update process of the offset module.

[0075] Specifically, the nipper plate swing characteristics include the maximum nipper plate swing amplitude and the nipper plate swing frequency; the fiber parameters include the effective fiber length, short fiber content and uniformity index; the quality parameters after combing include the fiber separation degree after combing, the number of cotton knots and the residual impurity rate.

[0076] The present invention introduces MD5 verification to ensure the integrity and security of the update file and prevent system anomalies caused by file damage or tampering; combined with multi-level compatibility verification (functional interface matching, data structure consistency, module and algorithm adaptability), it effectively avoids upgrade failures caused by file mismatch. By real-time acquisition and analysis of the operating status of the offset module, including dynamic control parameters of cotton drop spacing, separation roller speed, clamp plate swing characteristics, fiber parameters and combing quality parameters, the present invention realizes accurate evaluation of module operation status, avoids blind updates under high load or abnormal conditions, and ensures system stability. By giving priority to assigning low-priority tasks to the update module for trial operation and dynamically adjusting the task load distribution, a smooth transition of the update process is ensured to reduce production interference. Through a comprehensive verification and dynamic adjustment mechanism, high security, high reliability and high efficiency of the online upgrade process are achieved. At the same time, through the close combination of precise control and status evaluation, the stability of combing machine production and product quality are further improved.

[0077] See also Figure 1 S2 in the video recognition and algorithm adaptation evaluation step includes the following:

[0078] See also Figure 2 S201 in which raw material video and combing video are collected in real time, and current parameters including raw material parameters and combing quality parameters are obtained by image analysis;

[0079] Furthermore, real-time collection of raw material video includes: using a first video monitoring device to collect real-time video of the raw material at a first collection point to obtain the raw material video; the first collection point is located on the raw material conveying path between the cotton feeder outlet and the comber feeding inlet, and the first video monitoring device is installed above the conveying channel within a specified range of the cotton feeder outlet, and fixedly shoots the state of the raw material after uniform laying at a top-down angle.

[0080] Preferably, the first collection point is located at the outlet of the cotton feeder, connected to the discharge channel of the cotton feeder, specifically arranged at the end of the discharge channel of the cotton feeder, and located above the conveying channel between the cotton feeder and the feeding inlet of the comber, close to the upstream position of the feeding inlet of the comber. The collection point ensures that the state of the raw material after uniform layering of the cotton feeder can be collected. At the same time, since the raw material has not yet entered the feeding inlet of the comber main machine, the collection point can fully reflect the characteristics of the raw material before entering the comber, and provide accurate basic data for subsequent regulation. The present invention realizes accurate monitoring of the state of the raw material before the feeding inlet of the comber. By using the first video monitoring device to take a fixed shot at a top-down angle, the morphological characteristics of the raw material after uniform layering can be clearly captured, providing a high-quality data basis for the calculation and analysis of subsequent raw material parameters. Not only does it ensure the comprehensiveness of the monitoring coverage, but it also effectively avoids the parameter deviation caused by uneven layering or dynamic conveying of the raw material, thereby improving the accuracy of raw material analysis and the stability of the combing process. In addition, the introduction of real-time monitoring provides technical support for the early identification and rapid response of abnormal raw material status, thereby reducing interference with downstream process flows and improving production efficiency and product quality.

[0081] Optionally, other collection points that can collect the state of the raw materials after completing the previous process can also be selected, as long as the collection point can fully reflect the characteristics of the raw materials before entering the combing process; the first video monitoring equipment is usually installed above the conveying channel of the collection point, and fixedly shoots the state of the raw materials after uniform laying at a top-down angle to ensure that accurate and complete raw material video data is collected.

[0082] Furthermore, real-time collection of combing video includes: using a second video monitoring device to collect a video of the raw material after the combing process in real time at a second collection point to obtain a combing video; the second collection point is located on the conveying channel between the output end of the combing machine and the entrance of the subsequent process, and the second video monitoring device is installed above the conveying channel to shoot the fiber state after the combing process at a fixed viewing angle. The present invention realizes real-time collection and precise monitoring of the fiber state after the combing process. The setting of a fixed viewing angle ensures the stability and continuity of the fiber image, avoids the degradation of image quality due to angle offset or motion blur, and thus lays the foundation for the accurate extraction of combing quality parameters (such as the number of cotton knots, fiber separation degree and impurity residual rate, etc.). This real-time monitoring design can timely reflect the actual effect of the combing process, provide highly timely data support, help to discover and correct abnormal conditions in the combing process, and further improve the control accuracy of the combing process.

[0083] Optionally, the second collection point can be flexibly set at any position on the conveying path after the combing process is completed and before entering the next process, as long as the fiber state after the combing process (without external interference) can be fully photographed.

[0084] On the production line, high frame rate cameras (such as 120fps) combined with short exposure time (such as 1 / 1000 second) can clearly capture the subtle features of fibers in high-speed motion, effectively avoiding motion blur and loss of details. The collected video is accelerated by the back-end GPU to complete frame segmentation, feature extraction and parameter calculation in real time, ensuring accurate acquisition of the dynamic characteristics of the fiber in a high-speed production environment, providing efficient data support for real-time monitoring and intelligent control. This combination not only meets the speed requirements of the industrial site, but also takes into account the accuracy of fiber quality analysis.

[0085] Furthermore, the raw material video and combing video are collected in real time, and the current parameters are obtained by image analysis:

[0086] Performing image frame processing on the raw material video and the combed video to obtain a current raw material image and a current combed image;

[0087] Preprocessing the current raw material image and the current combed image to obtain the current raw material fiber image and the current combed fiber image; the preprocessing includes denoising, histogram equalization, background segmentation and illumination normalization;

[0088] Perform image analysis on the current fiber image of the raw material to obtain raw material parameters, including effective length, uniformity index, short fiber content, impurity content, average size of impurity particles and image gray uniformity index;

[0089] The current fiber image after combing is analyzed to obtain combing quality parameters, including the number of neps, residual impurity rate and combed fiber separation degree.

[0090] By real-time acquisition of raw material video and combing video and combining image analysis, the present invention can efficiently and accurately obtain current raw material parameters and combing quality parameters. Through image frame processing and preprocessing technology (such as denoising, histogram equalization, background segmentation and illumination normalization), the image quality and accuracy of feature extraction are significantly improved, thereby ensuring the reliability of parameter calculation. The real-time acquisition of raw material parameters (such as effective length, uniformity index, etc.) and combing quality parameters (such as cotton knot number, fiber separation, etc.) not only improves the monitoring ability of combing process quality, but also provides key data support for optimizing the production process and quickly adjusting process parameters. It effectively overcomes the subjective errors and response delays existing in traditional manual detection, realizes real-time control of raw material quality and combing effect, greatly improves the intelligence level of the production line and the consistency of product quality, and reduces the scrap rate and cost investment.

[0091] Optionally, the fiber distribution characteristics of each frame can be extracted by performing image segmentation processing on the raw material video and combing video collected in real time; for the acquisition of raw material parameters, the fiber length distribution is calculated by detecting the fiber contour and morphological analysis of the raw material image, and the effective length and uniformity index are calculated by combining the mean and standard deviation; the uniformity of fiber distribution is identified by pixel grayscale changes, and the impurity particles in the fiber are further detected by combining the impurity recognition algorithm, and the average size and density of the impurity particles are calculated to obtain the impurity content; for short fiber content, the proportion of fibers shorter than the specified threshold is obtained by fiber length histogram analysis. At the same time, for the acquisition of combing quality parameters, the separation characteristics of the fibers in the combing video and the image characteristics of the nep detection area are analyzed, and the number of neps and separation index are counted by combining image segmentation and deep learning classification models; the impurity residual rate is calculated by detecting the proportion of non-fiber components in the fiber image.

[0092] The raw material parameter calculation formula includes:

[0093] The fiber length distribution calculation formula is as follows:

[0094] ;

[0095] in, For the The length of the fiber, is the first detected The outline of the fiber strip; Function to calculate fiber contour length.

[0096] Effective length The calculation formula is as follows:

[0097] ;

[0098] in, is the number of fibers detected, is the fiber length;

[0099] Uniformity Index The calculation formula is as follows:

[0100] ;

[0101] in, is the arithmetic mean of fiber length.

[0102] Image gray uniformity index The calculation formula is as follows:

[0103] ;

[0104] in, is the gray value of the raw image pixel, and are the standard deviation and mean of the gray value respectively;

[0105] Impurity rate The calculation formula is as follows:

[0106] ;

[0107] in, is the total area of ​​impurities detected, is the total area of ​​the image.

[0108] Average size of impurity particles The calculation formula is as follows:

[0109] ;

[0110] in, For the The area of ​​impurity particles, is the total number of foreign particles detected.

[0111] Short fiber content The calculation formula is as follows:

[0112] ;

[0113] in, is the length threshold of short fibers, It means counting the total number of target elements in a piece of data;

[0114] The formulas for calculating combing quality parameters include:

[0115] Number of Neps The calculation formula is as follows:

[0116] ;

[0117] in, is the cotton nep area detected by image segmentation and classification;

[0118] Fiber separation The calculation formula is as follows:

[0119] ;

[0120] in, represents the detected fiber overlap area, is the total fiber area.

[0121] Residual impurities rate The calculation formula is as follows:

[0122] ;

[0123] in, It is the area of ​​impurities remaining after combing; is the total area of ​​the combed fiber.

[0124] See also Figure 2 In S202, obtaining an applicable parameter range of a first intelligent control algorithm, including a raw material parameter range and a combing quality parameter range;

[0125] See also Figure 2 In S203, it is determined whether the raw material parameter range is not within the expected raw material parameter range or the combing quality parameter range is not within the expected combing quality parameter range.

[0126] In this embodiment, the expected raw material parameter range and the expected combing quality parameter range refer to the parameter threshold range set according to the current production standard. The expected raw material parameter range is used to measure whether the quality characteristics of the raw material entering the combing machine are within an acceptable range to ensure that the raw material is suitable for subsequent combing process processing. The expected combing quality parameter range is used to measure whether the quality of the combed fiber meets the requirements to ensure that the output fiber meets the established production standards.

[0127] If the raw material parameter range is not within the expected raw material parameter range or the combing quality parameter range is not within the expected combing quality parameter range, or both occur at the same time, directly execute S301;

[0128] See also Figure 2 In S204, if the raw material parameter range is within the expected raw material parameter range and the combing quality parameter range is within the expected combing quality parameter range, an upgrade reminder message is generated according to the current parameters;

[0129] Furthermore, generating upgrade reminder information based on current parameters includes: if any parameter in the raw material parameters exceeds the corresponding raw material parameter range, identifying a first out-of-range parameter; matching an adaptive intelligent control algorithm set based on the first out-of-range parameter to obtain a second intelligent control algorithm; if any parameter in the combed fiber quality parameters exceeds the corresponding combed fiber quality parameter range, identifying a second out-of-range parameter; matching an adaptive intelligent control algorithm set based on the second out-of-range parameter to obtain a third intelligent control algorithm; performing an intersection operation on the second intelligent control algorithm and the third intelligent control algorithm to generate an upgrade algorithm list; generating upgrade reminder information based on the upgrade algorithm list.

[0130] The process of generating upgrade reminder information based on current parameters is mainly used to provide upgrade suggestions rather than directly trigger automatic upgrade operations. It can avoid unnecessary automatic upgrades caused by accidental or short-term fluctuations (such as instantaneous deviations of raw material parameters or combing quality parameters), which may lead to system misadjustments or production instability. Through this upgrade algorithm list, the system generates upgrade reminder information and reminds staff to manually confirm through interface prompts or notifications. Staff can decide whether to perform the upgrade operation based on the current production status, system operation trends, and a comprehensive assessment of raw material fluctuations. This reminder mechanism effectively avoids the risk of automatic upgrades due to raw material fluctuations or occasional anomalies, while retaining strict control over system stability and production continuity, improving the intelligence level of the system and the flexibility of on-site operations.

[0131] See also Figure 1 The steps for querying and upgrading the algorithm version in S3 include the following:

[0132] See also Figure 2 In S301, a first updated version corresponding to a first intelligent control algorithm on a remote server is queried to obtain a list of higher-version intelligent control algorithms; the first updated version is a higher-version that can be updated currently.

[0133] See also Figure 2 In S302, the first intelligent control algorithm is upgraded according to the list of higher-version intelligent control algorithms.

[0134] The first intelligent control algorithm for upgrading adopts the same dual-module hot switching mechanism and real-time verification method as the offset module update to ensure the safety of the upgrade and the continuity of system operation. During the algorithm upgrade process, the new version of the intelligent control algorithm is loaded into an operating environment isolated from the existing algorithm in the form of an independent copy to ensure that the existing system logic remains in normal operation during the upgrade process and avoid production interruptions caused by the upgrade. At the same time, the system's operating input data, control parameters and module status information are collected in real time and injected into the new algorithm copy to verify whether its output results under the existing operating conditions are consistent with the current algorithm, thereby ensuring the stability and compatibility of the new algorithm. After meeting the verification conditions, some control tasks are gradually transferred from the current algorithm to the new algorithm for trial operation, and the reliability of the new algorithm operation is ensured by real-time monitoring of the deviation range of key performance indicators (such as cotton drop spacing parameters and fiber separation). The new algorithm that performs well during the trial operation will gradually take over more tasks and eventually completely replace the old version of the algorithm. In addition, the upgrade process includes comprehensive verification of the functional interface, data structure and adaptability with each module to ensure that the new algorithm can be seamlessly integrated into the existing system and work together. Therefore, whether it is the update of the offset module or the upgrade of the intelligent control algorithm, it is based on a consistent upgrade logic framework, thereby achieving accurate and efficient upgrades while ensuring the stability of system operation.

[0135] In order to verify the effectiveness of the online upgrade method of the comber noil gauge intelligent control system of the present invention, the following comparative experiment was conducted. The comber noil gauge intelligent control system of the experimental group adopted the online upgrade method proposed by the present invention; the control group 1 adopted the traditional upgrade method of manual inspection and manual operation, and the comber noil gauge intelligent control system itself was consistent with the experimental group (hereinafter referred to as the traditional manual upgrade method); the control group 2 was the traditional automatic upgrade method adopted by the system before the application of the method of the present invention, specifically a comparative experimental group that only supports the automatic upgrade function but does not include the multi-level control method of the present invention (hereinafter referred to as the automatic upgrade method). The main difference of this experimental group is that the accurate judgment of the offset detection model and the task classification hot switching mechanism are not used in the automatic upgrade process, and the video recognition and intelligent algorithm adaptation functions are not combined.

[0136] The control group and the experimental group were run on three production lines, with each production line running for 8 hours a day and monitored continuously for one week.

[0137] Table 4 Comparative experimental results data table

[0138]

[0139] Referring to Table 4, the experimental results show that the method of the present invention shortens the average downtime to 5.2 minutes / time through the optimized online upgrade mechanism, which is significantly lower than the 8.7 minutes / time of the automatic upgrade method and the 15.4 minutes / time of the traditional manual upgrade method, and minimizes the abnormal downtime of the production line. In addition, its upgrade failure rate is only 0.5%, which is much lower than 3.2% of the control group 2 and 7.8% of the control group 1. Thanks to the dual-module hot switching and strict compatibility verification mechanism, the reliability of the system upgrade is effectively guaranteed. In terms of production efficiency, the unit time output of the method of the present invention reaches 150.5kg / h, which is 5.8% and 13.4% higher than the 142.3kg / h of the automatic upgrade method and the 132.7kg / h of the traditional manual upgrade method, respectively.

[0140] More importantly, the method of the present invention brings additional effects. By introducing real-time monitoring and analysis of raw material images and combing images in the upgrade method, the changes in fiber parameters and combing quality parameters are accurately captured, and dynamic regulation of the production process and early identification of abnormal parameters are achieved, thereby significantly reducing the scrap rate to only 1.2%. In comparison, the scrap rates of the automatic upgrade method and the traditional manual upgrade method are 2.4% and 4.1%, respectively. This significant reduction in scrap rate is due to the precise evaluation of the key parameter range and the efficient switching of the real-time adaptation algorithm during the upgrade process of the method of the present invention, which avoids the generation of unqualified products due to untimely regulation or parameter drift, and provides higher quality assurance and economic benefits for the production process.

[0141] The present invention adopts a three-stage design, aiming to comprehensively cover the abnormal conditions and normal working conditions of the combing machine production environment to ensure efficient operation, precise control and intelligent upgrading of the system. The core goal of the module operation analysis and update step (S1) in the first stage and the video recognition and algorithm adaptation evaluation step (S2) in the second stage is to identify and handle abnormal conditions. Among them, S1 analyzes the historical monitoring parameters of the system function modules to quickly locate the modules that may be offset, and dynamically updates them in combination with the real-time status to ensure the stable operation of the system under abnormal conditions. S2 identifies abnormal conditions in the current raw materials and production process, such as parameters exceeding the expected range, through real-time collection and analysis of raw material videos and combing videos, providing real-time data support and intelligent decision-making basis for the precise processing of abnormal conditions.

[0142] The algorithm version query and upgrade step (S3) in the third phase targets the normal working state of the system and ensures that the system always runs under the optimal control logic by dynamically matching the latest intelligent control algorithm version. Compared with the exception handling in the first two phases, the third phase focuses on performance optimization and function expansion under stable operating conditions, thereby achieving continuous improvement of the system.

[0143] By clearly distinguishing the processing logic of abnormal and normal working states, a comprehensive upgrade framework has been built, which can not only quickly update and upgrade to restore production under abnormal conditions, but also continuously improve system performance under normal conditions. This phased and hierarchical design ensures the stability and continuity of the system, while significantly improving production efficiency and product quality, providing an efficient, intelligent and precise control solution for the combing machine production process.

[0144] Embodiment 2:

[0145] On the textile production line of Factory B, an intelligent control system for the cotton drop gauge of combing machines is applied. The system is equipped with high-precision sensors and intelligent control modules, which can monitor the raw material parameters and post-combing quality parameters of fibers in real time. In order to improve the intelligence level of the system, Factory B introduced an online upgrade method for the intelligent control system for the cotton drop gauge of combing machines. The method is divided into three steps from S1 to S3, covering the entire process of module analysis and update, algorithm evaluation and adaptation, version query and upgrade.

[0146] S1: Module operation analysis and update steps, including the following:

[0147] Acquire the first historical monitoring parameters of the designated comber noil gauge intelligent control system within a designated time period, the comber noil gauge intelligent control system having a built-in first intelligent control algorithm and multiple functional modules;

[0148] According to the first historical monitoring parameter, a functional module whose running state fails is identified to obtain a failed module; failure warning information is generated for the failed module, and the failure warning information is released through a system interface or a remote notification function.

[0149] Analyze the operating status of each functional module according to the first historical monitoring parameter and determine the offset module;

[0150] Download the update file corresponding to the offset module from the remote server;

[0151] According to the real-time status of the offset module, load the update file to update the offset module;

[0152] S2: Video recognition and algorithm adaptation evaluation steps, including the following:

[0153] Collect raw material video and combing video in real time, and use image analysis to obtain current parameters, including raw material parameters and combing quality parameters;

[0154] Obtaining an applicable parameter range of the first intelligent control algorithm, including a raw material parameter range and a combing quality parameter range; if the raw material parameter range is not within the expected raw material parameter range or the combing quality parameter range is not within the expected combing quality parameter range, directly executing S301;

[0155] Generate upgrade reminder information based on current parameters;

[0156] S3: Algorithm version query and upgrade steps, including the following:

[0157] Query the first updated version corresponding to the first intelligent control algorithm on the remote server to obtain a list of higher-version intelligent control algorithms; and upgrade the first intelligent control algorithm according to the list of higher-version intelligent control algorithms.

[0158] Furthermore, the first historical monitoring parameter is specifically:

[0159] The first historical monitoring parameter is time series data containing time stamps, including a mechanical parameter sequence, a raw material parameter sequence, a system control parameter sequence, an environmental parameter sequence, a combed fiber quality parameter sequence, and a system fault sequence;

[0160] The mechanical parameter sequence includes timestamp, nipper movement amplitude, detaching roller speed, cotton drop spacing dynamic control parameters and vibration data of each functional module. The vibration data includes vibration amplitude and frequency.

[0161] The raw material parameter sequence includes timestamp, image frame sequence, effective length, uniformity index, short fiber content, impurity content, average size of impurity particles and image gray uniformity index;

[0162] The system control parameter sequence includes time stamp, draft ratio, tension adjustment value and clamp pressure;

[0163] Environmental parameter sequences include timestamp, temperature, humidity, and noise level;

[0164] The combed fiber quality parameter sequence includes timestamp, image frame sequence, nep number, impurity residue rate and combed fiber separation degree;

[0165] The system fault sequence includes timestamp, fault type, affected module, subsequent processing method, fault recovery time and fault severity level.

[0166] Further, according to the first historical monitoring parameter, the operating state of each functional module is analyzed, and the offset module is determined to include:

[0167] Multidimensional time series data are constructed by associating mechanical parameter sequence, raw material parameter sequence, system control parameter sequence, environmental parameter sequence and combed fiber quality parameter sequence by timestamp. The preprocessed multidimensional time series data are input into the trained offset detection model to obtain the offset detection result, and the offset module is determined according to the offset detection result.

[0168] Furthermore, the training process of the offset detection model includes:

[0169] Obtain the second historical monitoring parameters of the comber noil spacing intelligent control system within the historical time period;

[0170] The second historical monitoring parameters include a historical mechanical parameter sequence, a historical raw material parameter sequence, a historical system control parameter sequence, a historical environmental parameter sequence, a historical combed fiber quality parameter sequence, and a historical system failure sequence;

[0171] According to the historical system fault sequence, the data segment in the second historical monitoring parameter is labeled to obtain a labeling result;

[0172] Preprocessing the second historical monitoring parameter to obtain historical multi-dimensional time series data; the preprocessing includes data alignment, missing value processing, outlier removal and data standardization;

[0173] Perform feature engineering on historical multi-dimensional time series data to extract basic features, cross-features, and high-order features;

[0174] The feature importance scoring method is used to screen basic features, cross features and high-order features, eliminate redundant features, and generate an offset detection feature matrix;

[0175] Divide the offset detection feature matrix into a training set and a test set in chronological order;

[0176] The offset detection feature matrix and the corresponding annotation results in the training set are input into the LSTM model for training to obtain the initial offset detection model;

[0177] The test set is used to optimize the hyperparameters of the initial drift detection model to obtain the drift detection model.

[0178] Further, according to the real-time status of the offset module, loading the update file to update the offset module includes:

[0179] Perform MD5 verification on the update file; if the MD5 verification passes, load the update file into the cache area; perform compatibility verification on the update file in the cache area, including checking the matching of the functional interface between the update file and the corresponding offset module, data structure consistency and compatibility with the first intelligent control algorithm; if the verification passes, enable the dual-module hot switching mechanism, load the copy instance of the offset module corresponding to the update file into an independent operating environment, and isolate the independent operating environment from the existing module operating environment to prevent real-time operation interruption; at the same time, collect the synchronization data of the current offset module in real time, including the input data, operating status and interaction information between the offset module and the comber cotton drop spacing intelligent control system of the current offset module; inject the synchronization data into the update module, collect and verify the output data; the update module is the copy instance of the offset module after the update file is loaded; analyze the real-time status of the offset module, including real-time collection and analysis of the cotton drop spacing dynamics The state control parameters, separation roller speed, nipper plate swing characteristics, fiber parameters and combing quality parameters are used to determine whether the update conditions are met; the update condition is that the offset module is in a non-high load state, and the current task supports phased transfer, and the verification output data of the update module is consistent with the output data of the current offset module; if the update conditions are met, low-priority tasks are identified and assigned to the update module for operation; during the operation, the fiber parameter deviation and cotton drop spacing parameter deviation of the update module are monitored in real time. If the parameter deviation and the cotton drop spacing parameter deviation are both within the corresponding deviation range, the task load distribution is gradually adjusted according to the preset task allocation weights until the update module completely takes over all tasks and replaces the current offset module. After the task transfer is completed, the system resources of the current offset module are released to complete the update process of the offset module; otherwise, the update module is stopped and the task is rolled back to the current offset module immediately.

[0180] Furthermore, real-time collection of raw material video includes:

[0181] Using a first video monitoring device to collect a video of the raw material at a first collection point in real time to obtain a video of the raw material;

[0182] The first collection point is located on the raw material conveying path between the cotton feeder outlet and the comber feeding inlet. The first video monitoring device is installed above the conveying channel within the specified range of the cotton feeder outlet to take a fixed bird's-eye view of the state of the raw material after uniform laying.

[0183] Furthermore, real-time collection of combing video includes:

[0184] Using a second video monitoring device to collect a video of the raw material after the combing process at a second collection point in real time to obtain a combing video;

[0185] The second collection point is located on the conveying channel between the output end of the combing machine and the entrance of the subsequent process. The second video monitoring equipment is installed above the conveying channel to shoot the fiber state after the combing process at a fixed angle.

[0186] Furthermore, the raw material video and combing video are collected in real time, and the current parameters are obtained by image analysis:

[0187] Performing image frame processing on the raw material video and the combed video to obtain a current raw material image and a current combed image;

[0188] Preprocessing the current raw material image and the current combed image to obtain the current raw material fiber image and the current combed fiber image; the preprocessing includes denoising, histogram equalization, background segmentation and illumination normalization;

[0189] Perform image analysis on the current fiber image of the raw material to obtain raw material parameters, including effective length, uniformity index, short fiber content, impurity content, average size of impurity particles and image gray uniformity index;

[0190] The current fiber image after combing is analyzed to obtain combing quality parameters, including the number of neps, residual impurity rate and combed fiber separation degree.

[0191] Furthermore, generating upgrade reminder information according to current parameters includes:

[0192] If any parameter among the raw material parameters exceeds the corresponding raw material parameter range, identify the first out-of-range parameter; match the adaptive intelligent control algorithm set according to the first out-of-range parameter to obtain the second intelligent control algorithm; if any parameter among the combed fiber quality parameters exceeds the corresponding combed fiber quality parameter range, identify the second out-of-range parameter; match the adaptive intelligent control algorithm set according to the second out-of-range parameter to obtain the third intelligent control algorithm; perform intersection operation on the second intelligent control algorithm and the third intelligent control algorithm to generate an upgraded algorithm list; generate upgrade reminder information according to the upgraded algorithm list.

[0193] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An online upgrade method for a combing machine noil gauge intelligent control system, characterized in that: include: S101: Acquire a first historical monitoring parameter of a designated comber noil gauge intelligent control system within a designated time period, wherein the comber noil gauge intelligent control system has a built-in first intelligent control algorithm and multiple functional modules; S102: Analyze the operating status of each of the functional modules according to the first historical monitoring parameters, and determine the offset module; S103: Downloading an update file corresponding to the offset module from a remote server; S104: Loading the update file to update the offset module according to the real-time status of the offset module; S201: collecting raw material video and combing video in real time, and obtaining current parameters by image analysis, including raw material parameters and combing quality parameters; S202: Obtaining an applicable parameter range of the first intelligent control algorithm, including a raw material parameter range and a combing quality parameter range; S203: determining whether the raw material parameter range is not within the expected raw material parameter range or the combing quality parameter range is not within the expected combing quality parameter range, and obtaining a determination result; If the judgment result is yes, directly execute S301; S204: If the judgment result is no, generating upgrade reminder information according to the current parameters; S301: Query the first updated version corresponding to the first intelligent control algorithm on the remote server to obtain a list of higher-version intelligent control algorithms; S302: Upgrading the first intelligent control algorithm according to the high-version intelligent control algorithm list.

2. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: The first historical monitoring parameter is specifically: The first historical monitoring parameters include a mechanical parameter sequence, a raw material parameter sequence, a system control parameter sequence, an environmental parameter sequence, a combed fiber quality parameter sequence and a system fault sequence; the mechanical parameter sequence includes a timestamp, a nipper motion amplitude, a separation roller rotation speed, a cotton drop spacing dynamic control parameter and vibration data of each of the functional modules, and the vibration data includes vibration amplitude and frequency; the raw material parameter sequence includes a timestamp, an image frame sequence, an effective length, a uniformity index, a short fiber content, a foreign matter content, an average size of foreign matter particles and an image grayscale uniformity index; the system control parameter sequence includes a timestamp, a draft ratio, a tension adjustment value and a nipper pressure; the environmental parameter sequence includes a timestamp, a temperature, a humidity and a noise level; the combed fiber quality parameter sequence includes a timestamp, an image frame sequence, a number of cotton neps, a foreign matter residual rate and a combed fiber separation degree; The system fault sequence includes timestamp, fault type, affected module, subsequent processing method, fault recovery time and fault severity level.

3. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: According to the first historical monitoring parameters, the operating status of each of the functional modules is analyzed to determine the offset module, including: constructing multidimensional time series data by associating mechanical parameter sequence, raw material parameter sequence, system control parameter sequence, environmental parameter sequence and combed fiber quality parameter sequence through timestamps; inputting the preprocessed multidimensional time series data into a trained offset detection model to obtain an offset detection result, and determining the offset module according to the offset detection result.

4. The online upgrade method of the comber noil gauge intelligent control system according to claim 3 is characterized in that: The training process of the offset detection model includes: The second historical monitoring parameters of the comber noil spacing intelligent control system within the historical time period are obtained; the second historical monitoring parameters include a historical mechanical parameter sequence, a historical raw material parameter sequence, a historical system control parameter sequence, a historical environmental parameter sequence, a historical combed fiber quality parameter sequence and a historical system fault sequence; according to the historical system fault sequence, the data segments in the second historical monitoring parameters are annotated to obtain the annotated results; the second historical monitoring parameters are preprocessed to obtain historical multidimensional time series data; the preprocessing includes data alignment, missing value processing, outlier removal and data standardization; the basic features, cross features and high-order features of the historical multidimensional time series data are extracted; the basic features, cross features and high-order features are screened by using a feature importance scoring method, redundant features are removed, and an offset detection feature matrix is ​​generated; the offset detection feature matrix is ​​divided into a training set and a test set in chronological order; the offset detection feature matrix in the training set and the corresponding annotated results are input into an LSTM model for training to obtain an initial offset detection model; the hyperparameters of the initial offset detection model are optimized using the test set to obtain the offset detection model.

5. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: According to the real-time state of the offset module, loading the update file to update the offset module comprises: Perform MD5 verification and compatibility verification on the update file, the compatibility verification includes checking the matching of the functional interface between the update file and the corresponding offset module, data structure consistency and adaptability to the first intelligent control algorithm; if the MD5 verification and the compatibility verification are passed, enable the dual-module hot switching mechanism, load the copy instance of the offset module corresponding to the update file into an independent operating environment, and isolate the independent operating environment from the existing module operating environment to prevent real-time operation interruption; collect the synchronization data of the current offset module in real time, including the input data, operating status and interaction information between the offset module and the comber cotton drop spacing intelligent control system; inject the synchronization data into the update module, collect and verify the output data; analyze the The real-time status of the offset module includes real-time collection and analysis of dynamic control parameters of cotton drop spacing, separation roller speed, clamp plate swing characteristics, fiber parameters and combing quality parameters to determine whether the update conditions are met; if the update conditions are met, identify low-priority tasks and assign the low-priority tasks to the update module for operation; during the operation, the fiber parameter deviation and cotton drop spacing parameter deviation of the update module are monitored in real time. If the parameter deviation and the cotton drop spacing parameter deviation are both within the corresponding deviation range, the task load distribution is gradually adjusted according to the preset task allocation weights until the update module completely takes over all tasks and replaces the current offset module. After the task transfer is completed, the system resources of the current offset module are released to complete the update process of the offset module.

6. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: Real-time acquisition of the raw material video includes: using a first video monitoring device to acquire the video of the raw material in real time at a first acquisition point to obtain the raw material video; the first acquisition point is located on the raw material conveying path between the cotton feeder outlet and the comber feeding inlet, and the first video monitoring device is installed above the conveying channel within a specified range of the cotton feeder outlet, and fixedly captures the state of the raw material after uniform laying at a top-down angle.

7. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: Real-time collection of the combing video includes: using a second video monitoring device to collect real-time video of the raw material after the combing process at a second collection point to obtain the combing video; the second collection point is located on the conveying channel between the output end of the combing machine and the entrance of the subsequent process, and the second video monitoring device is installed above the conveying channel to shoot the fiber state after the combing process at a fixed angle.

8. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: The raw material video and the combing video are collected in real time, and the current parameters are obtained by image analysis: Performing image frame processing on the raw material video and the combed video to obtain a current raw material image and a current combed image; Preprocessing the current raw material image and the current combed image to obtain a current raw material fiber image and a current combed fiber image; the preprocessing includes denoising, histogram equalization, background segmentation and illumination normalization; Performing image analysis on the current fiber image of the raw material to obtain the raw material parameters, including effective length, uniformity index, short fiber content, impurity content, average size of impurity particles and image gray uniformity index; The combed current fiber image is subjected to image analysis to obtain the combing quality parameters, including the number of neps, the residual impurity rate and the combed fiber separation degree.

9. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: Generating the upgrade reminder information according to the current parameters includes: if any parameter of the raw material parameters exceeds the corresponding raw material parameter range, identifying a first out-of-range parameter; matching an adaptive intelligent control algorithm set according to the first out-of-range parameter to obtain a second intelligent control algorithm; if any parameter of the combed fiber quality parameters exceeds the corresponding combed fiber quality parameter range, identifying a second out-of-range parameter; matching an adaptive intelligent control algorithm set according to the second out-of-range parameter to obtain a third intelligent control algorithm; performing an intersection operation on the second intelligent control algorithm and the third intelligent control algorithm to generate an upgrade algorithm list; generating the upgrade reminder information according to the upgrade algorithm list.

10. The online upgrade method of the comber noil gauge intelligent control system according to claim 1, characterized in that: The online upgrade method further includes: identifying a functional module whose operating status fails according to the first historical monitoring parameter to obtain a failed module; generating failure warning information for the failed module, and publishing the failure warning information through a system interface.

Images