CN117994236A

Movatterモバイル変換

Info

Publication number: CN117994236A
Application number: CN202410190873.7A
Authority: CN
Inventors: 龚正晖; 沈绒; 孙晓雯; 吴丽清
Original assignee: Xuzhou Pulutong Textile Technology Co ltd
Current assignee: Xuzhou Pulutong Textile Technology Co ltd
Priority date: 2024-02-21
Filing date: 2024-02-21
Publication date: 2024-05-07
Anticipated expiration: 2044-02-21
Also published as: CN117994236B

Abstract

The invention relates to the technical field of spinning monitoring, in particular to an intelligent vortex spun broken yarn detection method and system based on image analysis. The method comprises the following steps: collecting spinning image data of the turbo spinning equipment to generate spinning image data; performing image optimization adjustment on the spinning image data to generate optimized spinning image data; performing yarn breakage feature analysis on the optimized spinning image data to generate yarn breakage feature image data; performing yarn breakage classification processing according to the yarn breakage characteristic image data to generate classified yarn breakage image data; modeling a mathematical model of vortex spun yarn breakage detection according to the classified yarn breakage image data to generate a vortex spun yarn breakage detection model; and (3) carrying out yarn breakage detection on the turbo spinning equipment based on the vortex spinning yarn breakage detection model to obtain yarn breakage detection data, and transmitting the yarn breakage detection data to a terminal to execute yarn breakage intelligent feedback operation. According to the invention, through intelligent monitoring of spinning, more accurate broken yarn identification and feedback are realized, and the spinning processing quality is ensured.

Description

Vortex spun broken yarn intelligent detection method and system based on image analysis

Technical Field

The invention relates to the technical field of spinning monitoring, in particular to an intelligent vortex spun broken yarn detection method and system based on image analysis.

Background

Spinning is an important process for making textiles by forming yarns by slender and closely ordering raw material fibers, providing a basic raw material for the subsequent spinning process. In modern spinning processing procedures, a vortex spinning machine is used as efficient spinning equipment, yarn breakage is a common problem in processing in the general spinning process, and broken yarns can cause quality degradation of subsequent textile products. The spinning state in the processing process is identified and analyzed in real time through the image, abnormal broken yarns are monitored, and screening is timely carried out, so that the influence on the quality of textiles is avoided. However, in the conventional vortex spun yarn breakage detection method, whether the spun yarn is broken is usually judged manually by naked eyes, so that whether the spun yarn is broken is judged to have the problem of omission and negligence, and the quality of the textile finished products processed later may be different.

Disclosure of Invention

Based on the above, the invention provides an intelligent detection method and system for vortex spun broken yarn based on image analysis, so as to solve at least one of the above technical problems.

In order to achieve the above purpose, an intelligent detection method for vortex spun broken yarn based on image analysis comprises the following steps:

Step S1: collecting spinning image data of the turbo spinning equipment to generate spinning image data; performing image optimization adjustment on the spinning image data to generate optimized spinning image data;

step S2: performing yarn breakage feature analysis according to the optimized spinning image data to generate yarn breakage feature image data;

step S3: performing yarn breakage classification processing according to the yarn breakage characteristic image data to generate classified yarn breakage image data;

Step S4: modeling a mathematical model of vortex spun yarn breakage detection according to the classified yarn breakage image data to generate a vortex spun yarn breakage detection model;

Step S5: and carrying out yarn breakage detection on the turbo spinning equipment based on the vortex spinning yarn breakage detection model to obtain yarn breakage detection data, and transmitting the yarn breakage detection data to a terminal to execute yarn breakage intelligent feedback operation.

According to the invention, the spinning image data is acquired by the turbo spinning equipment, the spinning process is monitored in real time by the high-performance camera equipment, a large amount of original spinning image data is acquired, and the advanced optical technology and the high-resolution sensor are utilized to accurately capture key information such as the movement of the spinning machine, the fiber state and the like, so that an accurate and clear spinning image is ensured to be acquired, and a reliable data basis is provided for the identification and the monitoring of the subsequent broken yarn. The spinning image data is subjected to image optimization adjustment, the quality and definition of the spinning image can be effectively improved by adjusting parameters such as brightness, contrast, color balance and the like of the spinning image, and noise, blurring or distortion possibly existing can be removed by utilizing algorithms such as self-adaptive filtering, histogram equalization and the like, so that the spinning image data is optimized, and analysis is conducted for texture, state and the like of subsequent spinning conveniently. By carrying out yarn breakage feature analysis on optimized spinning image data, feature information related to yarn breakage in a spinning image can be accurately extracted, for example, key features such as fiber breakage, yarn tension change and the like in the spinning image are detected and quantitatively analyzed, and abundant and accurate feature data are provided for classifying and processing of the type of the subsequent broken yarn. The yarn breakage classification processing is carried out according to the yarn breakage characteristic image data, so that a richer yarn breakage characteristic classification set is provided for machine learning models and yarn breakage data analysis, different yarn breakage scenes can be dealt with, and the accuracy and the efficiency of yarn breakage detection are improved. The method is characterized in that a mathematical model modeling of vortex spun yarn breakage detection is carried out by utilizing classified yarn breakage image data, a mathematical model capable of automatically distinguishing the turbine spun yarn breakage is created, different types of yarn breakage can be accurately distinguished by learning a large amount of classified yarn breakage image data, key yarn breakage characteristics are extracted through the characteristics, then model training is carried out by utilizing the characteristics, and different types of yarn breakage can be accurately distinguished and alarm can be rapidly fed back through the vortex spun yarn breakage detection model by using a support vector machine, a neural network or other deep learning algorithms. The vortex spinning broken yarn detection model is used for detecting broken yarns of the turbine spinning equipment, the broken yarn condition can be rapidly and accurately identified, fiber breakage or other abnormal states can be located, the obtained broken yarn detection data comprise key information such as time, position and type of broken yarns, the broken yarn detection data are transmitted to a terminal to execute broken yarn intelligent feedback operation, detailed broken yarn information is provided for operation staff, the operation staff can rapidly respond and take necessary correction measures, and the influence of broken yarns on textile quality and production efficiency is minimized.

Preferably, step S1 comprises the steps of:

Step S11: collecting spinning image data of the turbo spinning yarn-breaking equipment in real time by using a sensor, and generating initial spinning image data;

Step S12: gray scale processing is carried out on the initial spinning image data to generate gray scale spinning image data;

step S13: performing image enhancement processing according to the gray spinning image data to generate enhanced spinning image data;

Step S14: noise reduction processing is carried out on the enhanced spinning image data by Gaussian filtering, and spinning image data are generated;

step S15: and carrying out image optimization adjustment on the spinning image data to generate optimized spinning image data.

The invention utilizes the sensor to collect the spinning image data of the turbo spinning yarn breaking equipment in real time, can obtain the image data in the spinning process in real time, and can collect and record the spinning state in real time. And gray processing is carried out on the spinning image data, so that the complexity of the image is reduced, the data volume is reduced, and the efficiency of the subsequent spinning image processing is improved. And (3) carrying out image enhancement processing on the gray spinning image data, improving the quality, contrast, characteristics and other significance of the spinning image, strengthening the spinning characteristics in the image, and highlighting and clearing the subtle changes of spinning. Noise reduction processing is carried out on the enhanced spinning image data by utilizing Gaussian filtering, smoothing operation is carried out by adopting a Gaussian filter so as to inhibit noise and tiny interference in an image, thereby generating clearer and cleaner spinning image data, and irregular fluctuation in the spinning image is smoothed by fuzzy processing, so that the noise in the image is effectively removed, and the quality and the readability of the image are improved. The spinning image data acquired by the sensor equipment is processed in multiple steps, including gray scale processing, enhancement processing, noise reduction processing and image optimization adjustment, so that the final optimized spinning image data has the advantages of clarity, high quality and noise reduction.

Preferably, step S15 comprises the steps of:

performing pixel frequency separation on the spinning image data to generate spinning frequency image data;

Performing pixel interpolation processing according to the spinning frequency image data to generate spinning interpolation image data;

and carrying out pixel frequency reconstruction according to the spinning interpolation image data to generate optimized spinning image data.

The invention separates the pixel frequency of the spinning image data, can better analyze the image information under different frequencies to make the image information more targeted in the subsequent processing, the pixel interpolation processing is helpful for eliminating the image defects possibly caused by insufficient sampling rate or sensor distortion and other reasons, the integrity of the image is improved, and finally, the high-frequency and low-frequency information is integrated through the pixel frequency reconstruction, so as to generate more detailed and optimized spinning image data.

Preferably, step S2 comprises the steps of:

Step S21: performing region division on the optimized spinning image data by utilizing an edge detection algorithm to generate regional spinning image data;

Step S22: spinning characteristic extraction is carried out on the regional spinning image data, and spinning characteristic image data are generated;

Step S23: and carrying out yarn breakage feature extraction according to the spinning feature image data to generate yarn breakage feature image data.

According to the invention, the optimized spinning image data is subjected to region division by utilizing the edge detection algorithm, so that the edge characteristics of a spinning part in the image can be accurately captured, the image is divided into different regions, the edge contour of the spinning region is highlighted, and the yarn breakage condition can be accurately identified in subsequent analysis. The spinning characteristic extraction is carried out on the regional spinning image data, and key spinning characteristic information including important characteristics such as fiber shape, density and texture and the like, possibly including fiber fracture shape, yarn tension change and the like is extracted from the regional spinning image, so that the yarn breakage condition can be accurately judged. And carrying out yarn breakage feature extraction according to the spinning feature image data, wherein the extraction comprises the accurate identification and extraction of key features such as fiber state, yarn breakage shape, spinning area change and the like, and deep mining of fine differences in the spinning feature image.

Preferably, step S23 comprises the steps of:

Carrying out spinning texture feature analysis on the spinning feature image data to generate texture feature image data;

performing spinning texture feature abnormal comparison analysis according to the texture feature image data to generate texture abnormal data;

And carrying out broken yarn texture characteristic image extraction on the spinning characteristic image data according to the texture abnormal data to generate broken yarn characteristic image data.

The invention analyzes the spinning texture characteristics of the spinning characteristic image data, specifically comprises the steps of analyzing the shape of the fiber, including the length, the width, the curvature and the like of the fiber, so as to comprehensively understand the state of the fiber and more precisely understand all aspects of the spinning image. And carrying out spinning texture feature abnormal comparison analysis according to the texture feature image data, wherein the spinning texture feature abnormal comparison analysis comprises comparison analysis of spinning texture features of different time points or areas, and distinguishing abnormal changes of spinning area textures, such as abrupt changes of texture details or unusual pattern appearance. The yarn breakage texture feature image extraction is carried out on the yarn breakage feature image data by using the texture abnormality data, so that abnormal texture features in the yarn breakage image can be accurately positioned, and the effective positioning and feature extraction of abnormal yarn breakage are realized.

Preferably, step S3 comprises the steps of:

step S31: performing pixel matrix analysis on the yarn breakage characteristic image data to obtain yarn breakage matrix pixel data;

Step S32: performing matrix transformation on the broken yarn matrix pixel data according to preset matrix transformation parameters to generate transformed broken yarn matrix pixel data;

step S33: performing comprehensive matrix fusion according to the transformed broken yarn matrix pixel data to generate transformed broken yarn characteristic image data;

step S34: performing yarn breakage region abnormality calculation on the converted yarn breakage characteristic image data by using a yarn breakage abnormality detection algorithm to generate yarn breakage image abnormality data;

Step S35: and carrying out yarn breakage classification processing on the converted yarn breakage characteristic image data according to the yarn breakage image abnormal data to generate classified yarn breakage image data.

The invention carries out pixel matrix deep analysis on the yarn breakage characteristic image data, extracts the detailed information of the yarn breakage region, can carry out fine analysis on the numerical value, color, brightness and the like of each pixel in the yarn breakage characteristic image, and is helpful for deep understanding of the position and the characteristics of yarn breakage. The method comprises the steps of carrying out matrix transformation on broken yarn matrix pixel data according to preset matrix transformation parameters, including rotation, translation, scaling and the like, carrying out corresponding transformation on the broken yarn matrix pixel data according to the parameters, dynamically adjusting the shape, the size, the position and the like of an image, and adapting to different actual broken yarn conditions better. And (3) carrying out comprehensive matrix fusion according to the transformed broken yarn matrix pixel data, integrating information after different matrix transformation, and carrying out weighted fusion or other mathematical operations so as to enable the characteristics of different matrix transformation to be reasonably integrated, form more comprehensive and rich broken yarn characteristic images, and ensure that the finally generated transformed broken yarn characteristic image data is more accurate and representative. The yarn breakage abnormal detection algorithm is utilized to perform yarn breakage region abnormal calculation on the converted yarn breakage characteristic image data, the converted yarn breakage characteristic image is analyzed through the yarn breakage abnormal detection algorithm, abnormal conditions of the yarn breakage region are positioned and calculated, the fine analysis on the texture, the color, the shape and the like of the yarn breakage region of the image is included, the degree or the confidence of the abnormality is calculated, and the generated yarn breakage image abnormal data provides important information for subsequent decision and feedback. The yarn breakage classification processing is carried out on the converted yarn breakage characteristic image data according to the yarn breakage image abnormal data, the operation of carefully marking, dividing or marking different yarn breakage types is carried out on the abnormal region, the yarn breakage region can be divided into different categories, thereby realizing the accurate classification of the yarn breakage condition, and the generated classified yarn breakage image data provides specific and meaningful references for the subsequent intelligent yarn breakage detection.

Preferably, the yarn breakage abnormality detection algorithm in step S34 is as follows:

Where X is an abnormal value of the broken yarn image, θ is an overall gray value attenuation rate of the partial gray image, n is pixel data in broken yarn feature image data, w_i is gray value weight of the ith broken yarn pixel, X_i is broken yarn gray value data of the ith pixel, μ_i is an average broken yarn gray feature value of the ith pixel, ε is an average broken yarn gray value attenuation rate, and p is a broken yarn gray adjustment value.

Preferably, step S4 comprises the steps of:

Step S41: carrying out fine recognition of yarn breakage areas on the classified yarn breakage image data to generate yarn breakage area image data;

Step S42: dynamically detecting yarn breakage edges of the yarn breakage area image data to generate dynamic yarn breakage image data;

Step S43: and modeling a mathematical model of vortex spinning broken yarn detection based on a convolutional neural network algorithm and the dynamic broken yarn image data, and generating a vortex spinning broken yarn detection model.

According to the invention, the classified broken yarn image data is subjected to refined recognition of the broken yarn region, and the precise positioning and distinguishing capability of the broken yarn region is improved by a refined recognition method, so that more accurate and fine broken yarn region information is provided for the broken yarn image data. The edge change of the broken yarn area is monitored and captured in real time, so that dynamic broken yarn image data are generated, the broken yarn edge is dynamically tracked and detected, and the position and the shape change of the broken yarn edge are captured in real time. By adopting a convolutional neural network to combine dynamic yarn breakage image data, a powerful and accurate mathematical model is established, key features in the dynamic yarn breakage image can be automatically learned and captured, intelligent processing of the dynamic yarn breakage image data is realized, and powerful support is provided for timely identifying and processing yarn breakage anomalies in textile production.

Preferably, step S41 comprises the steps of:

Extracting yarn breakage threshold values of the yarn breakage area image data to generate yarn breakage edge threshold value data;

performing linear yarn breakage threshold analysis on the yarn breakage edge threshold data to generate linear yarn breakage threshold data;

performing yarn breakage threshold weighting according to the linear yarn breakage threshold data to generate weighted yarn breakage threshold data;

And carrying out edge dynamic matching on the classified yarn breakage image data according to the weighted yarn breakage threshold value data to generate dynamic yarn breakage image data.

According to the invention, the image data of the broken yarn area is analyzed, the corresponding broken yarn edge threshold value data is extracted, the edge characteristics of the broken yarn area are determined, the broken yarn edge threshold value data is subjected to linear broken yarn threshold value analysis, the threshold value is further refined so as to more accurately reflect the gray level change trend of the broken yarn edge, the broken yarn threshold value weighting is carried out according to the linear broken yarn threshold value data, the weighted broken yarn threshold value data are generated through a reasonable weighting mechanism, the broken yarn edges of different parts can be processed to different degrees in subsequent processing, the weighted broken yarn threshold value data are utilized to carry out edge dynamic matching on the classified broken yarn images, and the accurate and dynamic matching on the broken yarn edge is realized through multi-level broken yarn threshold value extraction and analysis in combination with dynamic matching, so that a finer and flexible data base is provided for subsequent broken yarn detection and processing.

Preferably, step S5 comprises the steps of:

step S51, updating the real-time spinning image data of the turbo spinning equipment to generate real-time spinning image data;

s52, detecting yarn breakage image data of the real-time spinning image data by using the vortex spinning yarn breakage detection model to generate real-time yarn breakage image data, and updating model parameters of the vortex spinning yarn breakage detection model in real time by using the real-time spinning image data to generate an updated vortex spinning yarn breakage detection model;

Step S53, yarn breakage detection processing is carried out according to the real-time yarn breakage image data, and yarn breakage detection data are generated;

And S54, transmitting yarn breakage detection data to a terminal to execute yarn breakage intelligent feedback operation.

The invention monitors the state of the spinning equipment in real time and updates the image data, and a production manager can acquire the working state, spinning condition and potential yarn breakage problem of the current turbo-spinning equipment at any time. The real-time spinning image data is detected by using the trained vortex spinning broken yarn detection model, so that accurate real-time broken yarn image data is generated, and meanwhile, model parameters of the vortex spinning broken yarn detection model are updated in real time by using the real-time spinning image data, so that different spinning conditions and environmental changes can be more accurately dealt with. By analyzing and processing the real-time spinning image data, the yarn breakage condition in the turbo spinning equipment can be accurately detected, and detailed and accurate yarn breakage detection data are generated. The yarn breakage detection data are transmitted to the terminal to execute yarn breakage intelligent feedback operation, and through real-time yarn breakage information feedback, a manager can quickly respond to abnormal conditions on a production line to carry out timely maintenance and adjustment, so that the production efficiency is improved to the greatest extent, and the production loss is reduced.

The present disclosure provides an image analysis-based intelligent detection system for vortex spun broken yarn, which is configured to execute the above-described image analysis-based intelligent detection method for vortex spun broken yarn, where the image analysis-based intelligent detection system for vortex spun broken yarn includes:

the spinning acquisition module is used for acquiring spinning image data of the turbo spinning equipment and generating spinning image data; performing image optimization adjustment on the spinning image data to generate optimized spinning image data;

the yarn breakage feature analysis module is used for performing yarn breakage feature analysis according to the optimized spinning image data to generate yarn breakage feature image data;

The broken yarn characteristic classification module is used for carrying out broken yarn classification processing according to the broken yarn characteristic image data to generate classified broken yarn image data;

the vortex spun broken yarn detection module is used for modeling a mathematical model of vortex spun broken yarn detection according to the classified broken yarn image data to generate a vortex spun broken yarn detection model;

and the yarn breakage intelligent monitoring module is used for detecting yarn breakage of the turbine spinning equipment based on the vortex spinning yarn breakage detection model to obtain yarn breakage detection data, and transmitting the yarn breakage detection data to a terminal to execute yarn breakage intelligent feedback operation.

The intelligent vortex spinning broken yarn detection method based on image analysis has the advantages that the image data of spinning is collected by the turbine spinning equipment, spinning texture details are optimized on the spinning image data, broken yarn conditions are better identified, analysis of broken yarn characteristics is performed, classification mathematical model modeling is performed by using the broken yarn characteristics, real-time intelligent monitoring can be performed on broken yarns in a self-adaptive mode, real-time broken yarn monitoring data are fed back, and spinning generation can be more efficient and rapid.

Drawings

FIG. 1 is a schematic flow chart of steps of an intelligent detection method for vortex spun broken yarn based on image analysis;

FIG. 2 is a detailed flowchart illustrating the implementation of step S1 in FIG. 1;

FIG. 3 is a detailed flowchart illustrating the implementation of step S2 in FIG. 1;

FIG. 4 is a flowchart illustrating the detailed implementation of step S3 in FIG. 1;

FIG. 5 is a flowchart illustrating the detailed implementation of step S4 in FIG. 1;

FIG. 6 is a flowchart illustrating the detailed implementation of step S5 in FIG. 1;

The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.

Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.

It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

In order to achieve the above objective, referring to fig. 1 to 6, the present invention provides an intelligent detection method for vortex spun broken yarn based on image analysis, comprising the following steps:

In the embodiment of the present invention, as described with reference to fig. 1, the step flow diagram of the method for detecting vortex spun broken yarn based on image analysis of the present invention is provided, and in the embodiment, the method for detecting vortex spun broken yarn based on image analysis includes the following steps:

In the embodiment of the invention, spinning image data is acquired for the turbo spinning device, a high-resolution camera (for example, with a resolution of 1920x1080 pixels) is used for shooting the turbo spinning device so as to capture a detailed spinning process, for example, a turbo spinning device is used for real-time shooting by connecting an industrial camera with a resolution of 1920x1080 pixels, and the acquired image data contains information such as yarn details, tension changes and the like in the spinning process. The spinning image data is subjected to image optimization adjustment, the definition and contrast of the image can be improved through image processing technologies such as gray level equalization, filtering and the like, so that spinning details can be better displayed, for example, an adaptive histogram equalization algorithm can be adopted, the fact that the yarn details in the image are more prominent is ensured, the image optimization adjustment is beneficial to the follow-up broken yarn intelligent detection method to accurately identify possible problems in the spinning process, and the stability and reliability of detection are improved.

In the embodiment of the invention, the image processing and computer vision technology are used, the optimized spinning image data are analyzed, the key characteristics of broken yarns are extracted, the characteristic image data for broken yarn detection are generated, for example, an edge detection algorithm such as Canny edge detection is adopted, the edge characteristics of yarns in the image are highlighted, the threshold value of the Canny edge detection is set to be 50 and 150 so as to ensure that fine yarn edges are detected, morphological operations such as expansion and corrosion are applied, noise is removed and broken yarn edges are connected, a more complete characteristic image is formed, further, texture analysis is carried out by utilizing the texture characteristics of the yarns, and further, the broken yarn characteristic image data containing the key characteristics of broken yarn problems are obtained, so that a foundation is laid for subsequent intelligent broken yarn detection.

in the embodiment of the invention, the broken yarn characteristic image is processed by utilizing pixel matrix analysis, matrix transformation and an abnormality detection algorithm, and finally classified broken yarn image data is generated, for example, the broken yarn characteristic image data is subjected to pixel matrix analysis, the image is converted into a matrix, the numerical value and the position of each pixel are analyzed, then linear transformation such as matrix multiplication can be adopted to transform the broken yarn matrix, transformed broken yarn matrix pixel data is generated, the transformed broken yarn matrix pixel data is synthesized, the synthesized broken yarn matrix pixel data is subjected to abnormality calculation by the broken yarn abnormality detection algorithm, the position and degree of a broken yarn problem can be identified for a broken yarn area, and classification processing is carried out according to the abnormality degree, so that a broken yarn detection system can automatically identify and locate potential broken yarn defects.

In the embodiment of the invention, a vortex spun yarn breakage detection model is constructed according to a mathematical model modeling for vortex spun yarn breakage detection by using classified yarn breakage image data, for example, a vortex spun yarn breakage detection model is constructed by using a deep learning method, a basic model is selected based on a Convolutional Neural Network (CNN), pre-trained weight of the model is reserved by utilizing the classified yarn breakage image data to carry out migration learning, fine adjustment is carried out on a full-connection layer to adapt to yarn breakage detection tasks, a discarding layer is introduced for preventing overfitting, a training set is expanded by using a data enhancement technology, so that generalization capability of the model is improved, and intelligent recognition range of yarn breakage images is expanded.

In the embodiment of the invention, the vortex spinning broken yarn detection model is utilized to detect broken yarns of the turbine spinning equipment, for example, a real-time spinning image is obtained, the real-time image data is input into the pre-trained vortex spinning broken yarn detection model to detect broken yarns, model parameters including the number of convolution layers, the size of a filter, an activation function and the like and a threshold value of broken yarn detection are set, and after the broken yarns are detected by the model, the real-time broken yarn image data is obtained, wherein the real-time broken yarn image data contains information about broken yarn positions, broken yarn degrees and the like, for example, a broken yarn region located at coordinates (500,300) exists in the image, and the broken yarn degree is medium. And transmitting the yarn breakage detection data to a terminal so as to execute yarn breakage intelligent feedback operation, and carrying out intelligent analysis and decision by the terminal according to the received yarn breakage detection data. For example, according to the detected yarn breaking position and degree, the turbo spinning equipment and the yarn breaking are adjusted, and the real-time application of the vortex spinning yarn breaking detection model is adopted, so that the yarn breaking condition of the turbo spinning equipment can be timely found and processed, and the production efficiency and the equipment stability are improved.

Preferably, step S1 comprises the steps of:

As an example of the present invention, referring to fig. 2, a detailed implementation step flow diagram of step S1 in fig. 1 is shown, where step S1 includes:

In the embodiment of the invention, the high-resolution camera sensor is used for collecting the spinning image data of the turbo-spinning yarn-breaking device, for example, an industrial camera with the resolution of 1920x1080 pixels is arranged at a proper position of the turbo-spinning yarn-breaking device so as to ensure that the image in the spinning process can be comprehensively and clearly captured, the collection frequency of the sensor is set so as to ensure that the real-time data collection can be realized, the high-resolution sensor is used for collecting the image data of the spinning device in real time so as to ensure that the data has enough details and definition, and the reliable initial spinning image data is provided for the subsequent image processing and yarn-breaking detection.

In the embodiment of the invention, the gray processing is carried out on the initial spinning image data, the colored initial spinning image data is converted into the gray image, the standard gray processing algorithm can be adopted, the brightness and the contrast of the spinning image are reserved when the initial spinning image data is converted into the gray image, and the intelligent recognition speed of the subsequent broken yarn is improved when the characteristics are reserved through the proper gray processing algorithm.

In the embodiment of the invention, the image enhancement processing is carried out on the gray spun yarn image data, for example, for a certain pixel value in the gray spun yarn image, the gray spun yarn image data can be mapped to a new value through histogram equalization so as to ensure that the gray distribution of the whole image is more uniform, the brightness and the contrast of the image can be adjusted through the histogram equalization, and more favorable image characteristics are provided for subsequent broken yarn detection, so that details in the image are clearer and more suitable for subsequent broken yarn detection.

Step S14: and carrying out noise reduction processing on the enhanced spinning image data by utilizing Gaussian filtering to generate spinning image data.

In the embodiment of the invention, the convolution operation is performed on the enhanced spun yarn image by using the Gaussian filter, for example, a 3x3 Gaussian filter is considered, the weight coefficient is determined by the Gaussian function, the weight coefficient is in a discrete form of the Gaussian function, the central pixel point is ensured to obtain larger weight, the contribution of surrounding pixels is gradually reduced, and the noise in the image is reduced to a certain extent, so that the attention of broken yarn pixel points is better paid.

In the embodiment of the invention, the spinning image data is subjected to image optimization adjustment, such as adjustment of the brightness of the spinning image, so that details in the image can be enhanced by adjusting the contrast of the image, the broken yarn area is more prominent, and details in the image can be enhanced by adjusting the contrast of the image, so that the broken yarn area is more prominent.

Preferably, step S15 comprises the steps of:

The invention separates the pixel frequency of the spinning image data, can better analyze the image information under different frequencies to make the image information more targeted in the subsequent processing, the pixel interpolation processing is helpful for eliminating the image defects possibly caused by insufficient sampling rate or sensor distortion and other reasons, the integrity of the image is improved, and finally, the high-frequency and low-frequency information is integrated through the pixel frequency reconstruction, so as to generate finer and optimized spinning image data.

In the embodiment of the invention, pixel frequency separation is firstly carried out, the spinning image data can be converted into a frequency domain by adopting a frequency domain processing technology such as Fourier transform, the image is decomposed into components with different frequencies, such as a low-frequency component and a high-frequency component, then pixel interpolation processing is carried out according to the broken yarn identification requirement so as to highlight broken yarn area images, such as bilinear interpolation or bicubic interpolation, so as to increase the spatial resolution of the images, pixel frequency reconstruction is carried out, the spinning image data after interpolation processing can be reconstructed into the spatial domain by adopting technologies such as inverse Fourier transform, and the spinning image data is optimally regulated by the operations such as frequency separation, interpolation processing and frequency reconstruction, so that the spatial resolution of the images is improved and more detailed information is highlighted.

Preferably, step S2 comprises the steps of:

As an example of the present invention, referring to fig. 3, a detailed implementation step flow diagram of step S2 in fig. 1 is shown, where step S2 includes:

In the embodiment of the invention, the edge detection algorithm is utilized to carry out region division, such as a Canny edge detection algorithm, has better edge positioning and noise suppression capabilities, carries out edge processing on optimized spinning image data, and carries out edge detection on the optimized spinning image data, wherein the characteristic edge region is clearly identified, and the region division is favorable for subsequent fine recognition and dynamic yarn breakage detection of the yarn breakage region.

in the embodiment of the invention, spinning characteristic extraction is performed on regional spinning image data, the texture characteristics can be used for describing texture and detail information in an image, further analysis and recognition of broken yarn areas are facilitated, for example, a gray level co-occurrence matrix (GLCM) is taken as an example, various texture characteristics such as energy, contrast, correlation and the like are calculated from the gray level co-occurrence matrix, the calculated pixel points comprise the corresponding texture characteristics, the texture characteristics are mapped into the image, spinning characteristic image data are obtained, the characteristic image is helpful for distinguishing different spinning areas, and data with more information is provided for detecting broken yarn later.

In the embodiment of the invention, the yarn breakage feature extraction is performed according to the yarn feature image data, taking gray scale feature as an example, for each pixel point, the gray scale value of the yarn breakage feature can be directly used as the yarn breakage feature, the change of the gray scale value can reflect the brightness difference of the yarn breakage region, for example, the contour information of the image breakage region can be obtained by calculating the gray scale gradient of each pixel point, the yarn breakage feature image data is extracted from the spinning feature image data, and the obtained yarn breakage feature image data contains various feature information, thereby being beneficial to the subsequent targeted identification of the yarn breakage region.

Preferably, step S23 comprises the steps of:

The invention analyzes the spinning texture characteristics of the spinning characteristic image data, specifically comprises the steps of analyzing the shape of the fiber, including the length, the width, the curvature and the like of the fiber, so as to comprehensively understand the state of the fiber and more precisely understand all aspects of the spinning image. And carrying out spinning texture feature abnormal comparison analysis according to the texture feature image data, wherein the spinning texture feature abnormal comparison analysis comprises comparison analysis of spinning texture features of different time points or areas, and distinguishing abnormal changes of spinning area textures, such as abrupt changes of texture details or unusual shapes. The yarn breakage texture feature image extraction is carried out on the yarn breakage feature image data by using the texture abnormality data, so that abnormal texture features in the yarn breakage image can be accurately positioned, and the effective positioning and feature extraction of abnormal yarn breakage are realized.

In the embodiment of the invention, spinning texture feature analysis is carried out, different directions and frequencies of textures in an image are effectively captured, then spinning texture feature abnormal comparison analysis is carried out, the abnormal condition in texture feature image data can be measured by adopting a statistical method such as a mean value and a standard deviation, yarn breakage texture feature image extraction is carried out on the spinning feature image data according to the texture abnormal data, for example, the region with the median value of the texture abnormal data larger than a threshold value is marked as yarn breakage feature image data in an original image, and the detection precision and the reliability of spinning yarn breakage are improved.

Preferably, step S3 comprises the steps of:

According to the invention, through carrying out pixel matrix deep analysis on the broken yarn characteristic image data, the numerical value, the color, the brightness and the like of each pixel can be analyzed in detail, so that the position and the characteristics of broken yarns are understood in depth, and further, the broken yarn matrix pixel data is subjected to matrix transformation according to preset matrix transformation parameters, including rotation, translation, scaling and other operations, so that the shape, the size and the position of an image are dynamically adjusted, different actual broken yarn conditions are better adapted, and the flexible transformation mode is helpful for capturing the diversity of broken yarns, so that the analysis is more targeted. For the transformed broken yarn matrix pixel data, a comprehensive matrix fusion technology is adopted to integrate the information of different matrix transformations, and the reasonable integration of the characteristics of the different matrix transformations is ensured through weighted fusion or other mathematical operations to form a more comprehensive and rich broken yarn characteristic image, so that the accuracy and the representativeness of the finally generated transformed broken yarn characteristic image data are improved. The yarn breakage abnormal detection algorithm is utilized to perform yarn breakage region abnormal calculation on the converted yarn breakage characteristic image data, the converted yarn breakage characteristic image is analyzed through the yarn breakage abnormal detection algorithm, abnormal conditions of the yarn breakage region are positioned and calculated, the fine analysis on the texture, the color, the shape and the like of the yarn breakage region of the image is included, the degree or the confidence of the abnormality is calculated, and the generated yarn breakage image abnormal data provides important information for subsequent decision and feedback. The yarn breakage classification processing is carried out on the converted yarn breakage characteristic image data according to the yarn breakage image abnormal data, the operation of carefully marking, dividing or marking different yarn breakage types is carried out on the abnormal region, the yarn breakage region can be divided into different categories, thereby realizing the accurate classification of the yarn breakage condition, and the generated classified yarn breakage image data provides specific and meaningful references for the subsequent intelligent yarn breakage detection.

As an example of the present invention, referring to fig. 4, a detailed implementation step flow diagram of step S3 in fig. 1 is shown, where step S3 includes:

In the embodiment of the invention, pixel matrix analysis is performed on the broken yarn characteristic image data, for example, a matrix analysis method, such as Principal Component Analysis (PCA) or wavelet transformation, is adopted to extract important characteristics in the image, and the PCA is taken as an example, the pixel matrix D of the broken yarn characteristic image can be subjected to principal component decomposition to obtain a principal component matrix P, and the first several principal components of the principal component matrix P are selected and reserved, wherein the principal components contain the most obvious broken yarn characteristic information in the image, so that further broken yarn characteristic analysis and processing are facilitated.

In the embodiment of the invention, the pixel data of the broken yarn matrix is subjected to matrix transformation according to the preset matrix transformation parameters, and is assumed to be represented as T, wherein each element T (i, j) represents the pixel value of the ith row and the jth column in the broken yarn matrix before transformation, meanwhile, a matrix transformation parameter matrix M is preset for describing the transformation mode, the transformed broken yarn matrix pixel data matrix T_n is obtained by multiplying the broken yarn matrix pixel data matrix T with the transformation parameter matrix M, and different matrix transformations such as translation, rotation, scaling and the like can be realized by adjusting the value of the transformation parameter matrix M, so that the broken yarn matrix pixel data is transformed, and more flexibility and variability methods are provided for the subsequent broken yarn image processing.

In the embodiment of the invention, the comprehensive matrix fusion is carried out according to the pixel data of the transformed yarn breaking matrix, the yarn breaking information of different parts is weighted and integrated to generate a more comprehensive and representative yarn breaking characteristic image, for example, a matrix fusion technology is adopted, the contribution of each pixel position in the comprehensive image is effectively considered by setting the weight matrix, and corresponding weight values can be distributed in the weight matrix according to the importance of the yarn breaking characteristic, so that the space distribution and the form of the yarn breaking characteristic are better presented, the key characteristic of the yarn breaking is more accurately captured and presented, and the analysis and understanding effects of the yarn breaking problem are improved.

In the embodiment of the invention, the broken yarn abnormal detection algorithm is utilized to calculate the abnormal state of the broken yarn region of the converted broken yarn characteristic image data, the abnormal degree of each pixel position is calculated, whether the gray level abnormal degree of each pixel exceeds a set threshold value is judged, if the gray level abnormal degree of a certain pixel exceeds the threshold value or the gray level change trend of the pixel of a certain region is abnormal, the pixel position or region is marked as the abnormal state of the broken yarn region, and the abnormal region in the image can be sensitively captured by setting a proper threshold value and an abnormal degree calculation formula, so that powerful support is provided for the subsequent analysis and further processing of broken yarns.

In the embodiment of the invention, the yarn breakage classification processing is performed on the converted yarn breakage characteristic image data according to the yarn breakage image abnormal data, for example, the degree of abnormality is processed as a continuous value, the degree of abnormality is directly used for representing the degree of abnormality, and the generated classified yarn breakage image data presents more continuous abnormality information, thereby being beneficial to knowing the change of the yarn breakage degree in more detail, being capable of leading the yarn breakage classification to be more flexible and adapting to different scenes and requirements.

Preferably, step S4 comprises the steps of:

As an example of the present invention, referring to fig. 5, a detailed implementation step flow diagram of step S4 in fig. 1 is shown, where step S4 includes:

In the embodiment of the invention, the broken yarn area refining identification is performed on the classified broken yarn image data, the broken yarn area is refined and identified, the image processing technology can be utilized, the communication area analysis or morphological operation is performed, for example, the classified broken yarn image data is performed on the communication area analysis, different communication areas are marked, the area of each communication area is determined to be the size and the shape, for example, the area threshold of the broken yarn area is 100 pixels, namely, only the communication area with the area larger than 100 pixels is reserved, a more accurate and clear broken yarn area image can be obtained, and each pixel position represents whether the position belongs to the broken yarn area or not, thereby being beneficial to further positioning and refining the broken yarn problem.

In the embodiment of the invention, the dynamic detection of broken yarn edges is performed, such as an inter-frame difference or optical flow method, for example, the dynamic change of the edges is captured through inter-frame difference, a proper time interval can be set, the inter-frame difference can detect the motion through comparing the differences between adjacent frames, and the broken yarn image is dynamically monitored by applying a certain color or gray value to the edge pixels to highlight the dynamic broken yarn area.

In the embodiment of the invention, dynamic yarn breakage image data are used and divided into a training set and a testing set, each image is used as input, a corresponding vortex spun yarn breakage label is used as output, a training set is used for training a CNN model, model parameters are updated through a back propagation algorithm and an optimizer (such as an Adam optimizer), the testing set is used for evaluating the trained CNN model, indexes such as accuracy, precision, recall rate and the like of the model are calculated, the training process can be iterated for a plurality of times, model parameters are updated through back propagation in each iteration, automatic detection of vortex spun yarn breakage is realized, and the accuracy and efficiency of detection are improved.

Preferably, step S5 comprises the steps of:

The invention monitors the state of the spinning equipment in real time and updates the image data, and a production manager can acquire the working state, spinning condition and potential yarn breakage problem of the current turbo-spinning equipment at any time. The real-time spinning image data is detected by using the trained vortex spinning broken yarn detection model, so that accurate real-time broken yarn image data is generated, and meanwhile, model parameters of the vortex spinning broken yarn detection model are updated in real time by using the real-time spinning image data, so that different spinning conditions and environmental changes can be more accurately dealt with. By analyzing and processing the real-time spinning image data, the yarn breakage condition in the turbo spinning equipment can be accurately detected, and detailed and accurate yarn breakage detection data are generated. The yarn breakage detection data are transmitted to the terminal to execute yarn breakage intelligent feedback operation, and through real-time yarn breakage information feedback, a manager can quickly respond to abnormal conditions on a production line and timely adjust, so that the production efficiency is improved to the greatest extent, and the production loss is reduced.

As an example of the present invention, referring to fig. 6, a detailed implementation step flow diagram of step S5 in fig. 1 is shown, where step S5 includes:

In the embodiment of the invention, the turbo spinning equipment is subjected to real-time spinning image data updating, the corresponding sensor or camera equipment is deployed on the turbo spinning equipment to capture the image data in the spinning process in real time, the normal connection of the equipment and the data acquisition system is ensured, the frequency of updating once per second is set, the image data of the turbo spinning equipment is captured in real time, a data transmission channel is set, the spinning image data captured in real time is transmitted to the central processing system, so that the vortex spinning broken yarn detection model is used for intelligent broken yarn monitoring, and the real-time spinning image data updating flow can timely reflect the running state and broken yarn monitoring detailed condition of the turbo spinning equipment.

In the embodiment of the invention, the real-time spinning image data is input into the pre-trained vortex spinning broken yarn detection model, the model parameters of the vortex spinning broken yarn detection model are updated in real time by utilizing the real-time broken yarn image data, an updating period can be set, for example, the model parameters are updated once per hour in real time, and the real-time broken yarn detection requirements of the model under different conditions can be better met by utilizing the real-time broken yarn image data monitored in the period, and the updated vortex spinning broken yarn detection model is generated by utilizing the real-time updated parameters, so that broken yarn detection is more intelligent.

In the embodiment of the invention, the result of yarn breakage detection processing is generated into yarn breakage detection data, which comprises the accurate position, degree and affected area of yarn breakage, and the data provides detailed yarn breakage condition description for production management, so that operators can clearly know the yarn breakage condition of the current vortex spinning equipment, and corrective measures can be timely taken in the production process by recording the yarn breakage position in real time, thereby improving the production efficiency and spinning quality. The information on the degree of yarn breakage helps to evaluate the severity of yarn breakage and helps the operator to determine the treatment priority. The affected area data provides an important reference for subsequent production optimization and equipment adjustment, and is helpful for timely finding potential problems.

In the embodiment of the invention, the yarn breakage detection data are transmitted to the terminal to execute the yarn breakage intelligent feedback operation, the generated yarn breakage detection data are obtained, the data comprise detailed information such as the position, the degree and the area of yarn breakage, the yarn breakage detection data are transmitted to the terminal equipment such as a production monitoring system, an operator control interface and the like through a communication protocol, the timely and reliable transmission of the data is ensured through network transmission by using protocols such as HTTP, MQTT and the like, and the yarn breakage detection data are displayed on the operator control interface or the production monitoring system, including the information such as the position, the degree and the like of yarn breakage, so that an operator can intuitively know the yarn breakage condition, intelligent feedback control is realized, and the response speed and the processing efficiency to the yarn breakage problem are improved.

The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. The intelligent vortex spun yarn breakage detection method based on image analysis is characterized by comprising the following steps of:

2. The intelligent detection method of vortex spun yarn breakage based on image analysis according to claim 1, wherein step S1 comprises the steps of:

3. The intelligent detection method of vortex spun yarn breakage based on image analysis according to claim 2, wherein step S15 comprises the steps of:

4. The intelligent detection method of vortex spun yarn breakage based on image analysis according to claim 1, wherein step S2 comprises the steps of:

5. The intelligent detection method for vortex spun yarn breakage based on image analysis according to claim 4, wherein step S23 comprises the steps of:

6. The intelligent detection method of vortex spun yarn breakage based on image analysis according to claim 1, wherein step S3 comprises the steps of:

The yarn breakage abnormality detection algorithm is as follows:

7. The intelligent detection method of vortex spun yarn breakage based on image analysis according to claim 1, wherein step S4 comprises the steps of:

8. The intelligent detection method for vortex spun yarn breakage based on image analysis according to claim 7, wherein step S41 comprises the steps of:

9. The intelligent detection method of vortex spun yarn breakage based on image analysis according to claim 1, wherein step S5 comprises the steps of:

10. An image analysis-based vortex spun yarn breakage intelligent detection system for performing the image analysis-based vortex spun yarn breakage intelligent detection method according to claims 1 to 9, comprising: