Fiber separation method and device based on computer vision recognition

Abstract

The present invention provides a fiber separation method and device based on computer vision recognition, the method comprising collecting fiber image data to form a typical fiber database; dividing the fiber into modules according to the physical properties of the fiber, and identifying the influencing factors of fiber separation for different modules of the fiber; predicting the fiber separation effect based on the influencing factors of fiber separation to obtain a reference target for the fiber separation effect; setting key influencing factors of the same type of fiber based on the influencing factors of fiber separation, identifying the same type of historical fiber information based on the key influencing factors, and then obtaining the same type of historical fiber separation effect reference data through data analysis; and determining the fiber separation effect prediction target based on the fiber separation effect reference target and the fiber separation effect historical reference data. The present invention can improve the accuracy and efficiency of fiber separation and optimize the fiber processing process.

Description

Technical Field

[0001] The present invention relates to the technical field of computer vision, image processing and intelligent analysis, and more specifically, to a fiber separation method and device based on computer vision recognition. Background Art

[0002] In the textile industry, fiber separation is a crucial process link, which is directly related to the quality of the final product. Traditional fiber separation methods mainly rely on manual operation and empirical judgment, which is not only inefficient, but also difficult to ensure accuracy and consistency. With the development of computer vision technology, fiber separation methods based on computer vision recognition have gradually become a research hotspot. This method uses image processing and pattern recognition technology in an automated way to more accurately identify and separate fibers, thereby improving production efficiency and product quality.

[0003] However, existing fiber separation technology still has some problems and defects. First, it is highly dependent on manual labor. The traditional fiber separation method relies on the experience and skills of the operator, which is not only inefficient but also easily affected by human factors, resulting in unstable separation effect. Secondly, the accuracy and consistency are poor. Due to the subjectivity of manual operation, the accuracy and consistency of fiber separation are difficult to guarantee, which directly affects the quality of the final product. In addition, data is insufficiently utilized. In existing technologies, fiber image data and attribute data are often not fully utilized, lacking systematic analysis and application, resulting in waste of resources.

[0004] In terms of the prediction of fiber separation effect, the existing methods are relatively simple and lack in-depth analysis of fiber physical properties and influencing factors. The accuracy and reliability of the prediction results need to be improved. The low degree of automation is also a problem. Although some automated equipment has been introduced into the fiber separation process, the overall degree of automation is still low and many links still require manual intervention.

[0005] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: the collection and analysis of fiber image data is not systematic enough. In the prior art, the collection and analysis of fiber image data is often scattered, lacking unified standards and methods, resulting in low data utilization efficiency. The identification of factors affecting fiber separation is not comprehensive enough. In the prior art, the identification of factors affecting fiber separation is often one-sided, without fully considering the physical properties of the fiber and external environmental factors, resulting in unsatisfactory separation effect. The prediction method of fiber separation effect is not scientific enough. The existing prediction methods mostly rely on empirical formulas or simple statistical analysis, lack scientificity and systematicity, and the accuracy and reliability of the prediction results are limited. The utilization of historical data is not sufficient. In the prior art, the utilization of historical data is often limited to simple statistical analysis, and the information in the historical data is not fully utilized, resulting in insufficient scientificity in prediction and decision-making. The integration and application of automated equipment are not perfect. Although some automated equipment has been introduced into the fiber separation process, the integration and application between these equipment are not perfect enough, and the overall automation and efficiency need to be improved. Summary of the invention

[0006] The invention provides a fiber separation method and device based on computer vision recognition.

[0007] In a first aspect of the present invention, a fiber separation method based on computer vision recognition is provided, comprising:

[0008] Collecting fiber image data to form a typical fiber database, wherein the fiber image data includes fiber photos and corresponding fiber attribute data;

[0009] Divide the fibers into modules according to their physical properties, and identify factors affecting fiber separation for different modules of the fibers;

[0010] Predicting the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect;

[0011] Based on the influencing factors of the fiber separation, key influencing factors of the same type of fibers are set, based on the key influencing factors, historical fiber information of the same type is identified, and then historical fiber separation effect reference data of the same type is obtained through data analysis; and based on the fiber separation effect reference target and the fiber separation effect historical reference data, a fiber separation effect prediction target is determined.

[0012] Furthermore, after determining the fiber separation effect prediction target, it further includes: combining module decomposition, comparing the fiber separation design effects of different modules with the fiber separation effect prediction target, judging their deviations and influence ratios, and preferentially adjusting modules with greater separation effect elasticity to control the deviation within a preset error range; and using the fiber separation method based on computer vision recognition to predict the separation effects of different types of fibers to be processed.

[0013] Further, dividing the fiber into modules according to the physical properties of the fiber, and identifying the factors affecting the fiber separation for different modules of the fiber includes: determining the physical properties of the fiber according to the fiber photo;

[0014] The image of the fiber is partially decomposed into fiber bundles, fiber clusters, fiber filaments and fiber impurities according to the physical properties of the fiber; and the influencing factors of the fiber separation are identified for the fiber bundles, fiber clusters, fiber filaments and fiber impurities, wherein the influencing factors of the fiber separation include fiber length, fiber diameter, fiber strength, fiber color, fiber curvature, fiber surface roughness, fiber moisture content, fiber impurity content and fiber density.

[0015] Furthermore, the fiber separation effect is predicted based on the influencing factors of fiber separation to obtain a reference target for the fiber separation effect, including: separately predicting the separation effects of the fiber bundles, fiber clusters, fiber filaments and fiber impurities to obtain the fiber separation effects of different modules; and superimposing the fiber separation effects of the different modules to obtain the total separation effect of the fibers as the reference target for the fiber separation effect.

[0016] Further, identifying the same type of historical fiber information based on the key influencing factors includes: searching all sample data based on the key influencing factors, applying a particle swarm improvement algorithm, and finding the number of clusters with the smallest error square sum J value and the corresponding initial cluster center;

[0017] After the initial clustering center is determined, the category of the sample data and the center of each category are finally determined according to the improved fuzzy clustering algorithm, wherein the key influencing factors of the fiber bundle include fiber length, fiber diameter and fiber strength;

[0018] The key influencing factors of the fiber cluster include fiber curvature, fiber surface roughness and fiber moisture content;

[0019] The key influencing factors of the fiber filaments include fiber color, fiber impurity content and fiber density;

[0020] The key influencing factors of the fiber impurities include fiber length, fiber diameter and fiber impurity content.

[0021] Furthermore, the categories of sample data and the centers of each category are finally determined according to the improved fuzzy clustering algorithm, further including:

[0022] Calculate the density J of each type of data in the iterative process com and the inter-class separation J sep , and find J, and finally get the optimal number of clusters, cluster prototypes and the type of sample data by obtaining the maximum value of J;

[0023] Given the number of cluster categories c, 0≤c≤n, n is the number of data, set the iteration stop threshold ∈ and the fuzzy index m, and initialize the cluster prototype pattern v (0) , set the iteration counter b = 0;

[0024] Determine the best classification tree C m , calculate the value of J(C), and when the conditions are met, it is considered that C m is the optimal number of classifications, and the output partition matrix U and the cluster prototype V are obtained.

[0025] Further, obtaining historical fiber separation effect reference data of the same type through data analysis includes: calculating the median, upper quartile, lower quartile, maximum value, minimum value, sample mean and sample variance of the sample data;

[0026] The outliers are removed by box plot to perform an outlier analysis on the sample data, where the outliers are those less than the lower quartile Q. L Subtract 1.5 times the interquartile range IQR, greater than the upper quartile Q U plus 1.5 times the interquartile range (IQR);

[0027] According to Chebyshev’s inequality principle, at least 80% of the statistics fall in the interval φ;

[0028] The reasonable range of fiber separation effect is finally obtained by calculating the historical fiber separation effect reference data CKLS of the same type.

[0029] Further, determining the fiber separation effect prediction target according to the fiber separation effect reference target and the fiber separation effect historical reference data includes: by determining the fiber separation effect prediction target.

[0030] Further, based on the influencing factors of the fiber separation, the fiber separation effect is predicted to obtain a reference target of the fiber separation effect, including: based on the influencing factors of the fiber separation, a support vector machine model optimized based on a genetic algorithm is used to predict the fiber separation effect to obtain a reference target of the fiber separation effect, wherein the

[0031] The support vector machine model maps the input vector to a high-dimensional feature space through nonlinear mapping, and constructs an optimal classification hyperplane in the high-dimensional feature space. The classification function of the support vector machine is similar to that of a neural network, and the output is a linear combination of intermediate nodes, where each intermediate node corresponds to a support vector.

[0032] In a second aspect of the present invention, a fiber separation device based on computer vision recognition is provided, comprising:

[0033] A data acquisition module, used for collecting fiber image data to form a typical fiber database, wherein the fiber image data includes fiber photos and corresponding fiber attribute data;

[0034] An influencing factor identification module, used to divide the fiber into modules according to the physical properties of the fiber, and identify the influencing factors of fiber separation for different modules of the fiber;

[0035] A target prediction module predicts the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect;

[0036] A historical effect acquisition module, used for setting key influencing factors of the same type of fibers based on the influencing factors of the fiber separation, identifying the same type of historical fiber information based on the key influencing factors, and then obtaining the same type of historical fiber separation effect reference data through data analysis;

[0037] A prediction target acquisition module is used to determine a fiber separation effect prediction target based on the fiber separation effect reference target and the fiber separation effect historical reference data.

[0038] According to the above-mentioned embodiments of the present invention, at least the following beneficial effects are achieved: the present invention can accurately identify the key factors affecting the fiber separation effect by collecting fiber image data and constructing a typical fiber database, and combining the physical properties of the fiber for modular processing. This method can not only improve the accuracy of fiber separation, but also provide an important reference for the production process by predicting the fiber separation effect, thereby optimizing the fiber processing technology and improving product quality. In addition, the present invention can make a detailed comparison and adjustment of the fiber separation design effects of different modules by combining the module decomposition technology to ensure that the deviation of the separation effect is controlled within an acceptable error range. At the same time, by using the particle swarm improved algorithm and the fuzzy clustering algorithm, the sample data can be efficiently classified and clustered, which can further improve the accuracy of the fiber separation effect prediction. The application of the present invention can not only improve the efficiency of fiber processing and reduce production costs, but also help to achieve efficient utilization of fiber resources and sustainable development of the environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0040] Figure 1 A schematic flow chart of a fiber separation method based on computer vision recognition provided by one embodiment of the present invention;

[0041] Figure 2 A schematic structural diagram of a fiber separation device based on computer vision recognition provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0042] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0043] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, apparatus, method or computer program product. Therefore, the present invention can be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0044] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0045] Reference below Figure 1 , Figure 1 FIG. 1 is a flow chart of a fiber separation method based on computer vision recognition provided by an embodiment of the present invention. Figure 1 As shown, a fiber separation method 100 based on computer vision recognition includes:

[0046] Step 101, collecting fiber image data to form a typical fiber database, wherein the fiber image data includes fiber photos and corresponding fiber attribute data;

[0047] Step 102, dividing the fiber into modules according to the physical properties of the fiber, and identifying the factors affecting the fiber separation for different modules of the fiber;

[0048] Step 103, predicting the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect;

[0049] Step 104, setting key influencing factors for the same type of fibers based on the influencing factors of the fiber separation, identifying historical fiber information of the same type based on the key influencing factors, and then obtaining historical fiber separation effect reference data of the same type through data analysis; and determining a fiber separation effect prediction target based on the fiber separation effect reference target and the fiber separation effect historical reference data.

[0050] It should be noted that the fiber separation method based on computer vision recognition first involves the collection of fiber image data, which includes not only intuitive photos of the fibers, but also various attribute data related to the fibers, such as fiber length, diameter, strength, etc. With these data, a comprehensive typical fiber database can be constructed, providing a basis for subsequent fiber separation effect prediction and module division.

[0051] Specifically, the process of collecting fiber image data can be achieved through an automated image acquisition system that can capture high-definition images of fibers and record the attribute data of each fiber. These attribute data can be obtained through sensor measurements, such as using optical sensors to measure fiber diameter and mechanical sensors to measure fiber strength. The construction of the database requires the classification and labeling of the collected data to facilitate subsequent data processing and analysis.

[0052] Preferably, in order to improve the accuracy and practicality of the database, a machine learning algorithm can be used to preprocess and extract features from the collected fiber image data. For example, a convolutional neural network (CNN) is used to identify and classify different fiber types in fiber images, and a support vector machine (SVM) is used to identify key properties of fibers. In addition, the construction of the database can also include enhanced processing of fiber images, such as denoising, contrast adjustment, etc., to improve the accuracy of image recognition.

[0053] In some embodiments, after determining the fiber separation effect prediction target, it further includes: combining module decomposition, comparing the fiber separation design effects of different modules with the fiber separation effect prediction target, judging their deviations and influence ratios, and preferentially adjusting modules with greater separation effect elasticity to control the deviation within a preset error range; and using the fiber separation method based on computer vision recognition to predict the separation effects of different types of fibers to be processed.

[0054] It should be noted that after determining the prediction target of the fiber separation effect, it is necessary to combine the module decomposition technology to compare the fiber separation design effects of different modules with the prediction target and evaluate the deviation and impact ratio between them. The purpose of this step is to identify those modules that have a greater impact on the separation effect and have a larger adjustment space, so as to prioritize optimization.

[0055] Specifically, the module decomposition technology can use multidimensional feature analysis to take the physical properties of the fiber, such as length and diameter, as different dimensions to conduct a detailed analysis of the fiber separation process. During the comparison process, a deviation threshold can be set, and the modules exceeding this threshold will be considered as objects that need to be adjusted first. The evaluation of the impact ratio can be achieved by calculating the contribution of the module adjustment to the overall separation effect. In addition, statistical methods, such as analysis of variance (ANOVA), can be used to quantify the impact of different modules on the separation effect.

[0056] Preferably, in order to more accurately control the deviation, a real-time monitoring system can be introduced, which can dynamically track the real-time data in the fiber separation process and compare it with the predicted target in real time. If the deviation is found to be beyond the preset range, the system will automatically adjust the parameters of the relevant modules, such as changing the fiber processing speed or adjusting the settings of the separation equipment. In addition, optimization techniques such as simulated annealing algorithms or genetic algorithms can be used to find the optimal module adjustment scheme to minimize the deviation and maximize the separation effect.

[0057] In some embodiments, dividing the fiber into modules according to the physical properties of the fiber, and identifying the factors affecting the fiber separation for different modules of the fiber includes: determining the physical properties of the fiber according to the fiber photograph;

[0058] The image of the fiber is partially decomposed into fiber bundles, fiber clusters, fiber filaments and fiber impurities according to the physical properties of the fiber; and the influencing factors of the fiber separation are identified for the fiber bundles, fiber clusters, fiber filaments and fiber impurities, wherein the influencing factors of the fiber separation include fiber length, fiber diameter, fiber strength, fiber color, fiber curvature, fiber surface roughness, fiber moisture content, fiber impurity content and fiber density.

[0059] It should be noted that the module division and influencing factor identification steps in the fiber separation method based on computer vision recognition first involve decomposing the fiber image into different modules, such as fiber bundles, fiber clusters, fiber filaments and fiber impurities, according to the physical properties of the fibers. Then, for these different modules, the key factors affecting the fiber separation effect are identified, such as the physical properties of the fiber, such as length, diameter, strength, etc.

[0060] Specifically, module division can be achieved through image processing technology, using computer vision algorithms to identify and segment fiber structures in images. For example, edge detection algorithms are used to identify fiber outlines, and clustering algorithms are used to distinguish fiber bundles from fiber clusters. The identification of influencing factors involves further analysis of fiber images, determining the length and diameter of the fibers by measuring the geometric features of the fibers in the image, and evaluating the strength of the fibers through stress-strain tests. In addition, machine learning models can be used to analyze the relationship between fiber images and separation effects, thereby identifying key factors that affect the separation effect.

[0061] Preferably, in order to improve the accuracy of module division and influencing factor identification, deep learning techniques, such as convolutional neural networks (CNN), can be used to automatically extract the features of fiber images. In addition, expert systems can be combined to use the knowledge of domain experts to assist the recognition process and improve the accuracy of recognition. In terms of parameter setting, the parameters of the algorithm, such as the size of the convolution kernel, the step size, etc., can be adjusted according to the actual fiber type and production requirements to achieve the best segmentation effect. Alternatives may include using different image segmentation algorithms or optimization algorithms to adapt to different fiber characteristics and production environments.

[0062] In some embodiments, the fiber separation effect is predicted based on the influencing factors of fiber separation to obtain a reference target for the fiber separation effect, including: separately predicting the separation effects of the fiber bundles, fiber clusters, fiber filaments and fiber impurities to obtain the fiber separation effects of different modules; and superimposing the fiber separation effects of the different modules to obtain the total separation effect of the fibers as the reference target for the fiber separation effect.

[0063] It should be noted that the method for predicting the fiber separation effect based on the factors affecting fiber separation includes separately predicting the separation effects of different modules such as fiber bundles, fiber clusters, fiber filaments and fiber impurities, and then obtaining the total fiber separation effect by superimposing these prediction results. This method allows for a detailed evaluation of the separation effect of each module and a comprehensive consideration of their contribution to the overall separation effect.

[0064] Specifically, when predicting the fiber separation effect, a machine learning algorithm can be used to model the separation effect of each module. For example, regression analysis is used to predict the separation effect of fiber bundles, and classification algorithms are used to evaluate the separation quality of fiber clusters. The prediction model of each module can be trained according to its specific influencing factors, such as fiber length, diameter, etc. Parameter settings may include learning rate, number of iterations, etc., which can be adjusted according to the performance of the model to optimize the prediction accuracy. In addition, techniques such as cross-validation can be used to evaluate the generalization ability of the model.

[0065] Preferably, in order to improve the accuracy and reliability of the prediction, an ensemble learning method, such as random forest or gradient boosting machine (GBM), can be used to combine the prediction results of multiple prediction models. In addition, the knowledge of domain experts can be introduced to adjust the weights of the model to reflect the actual impact of different modules on the overall separation effect. In terms of parameter selection, automated techniques such as grid search or Bayesian optimization can be used to find the optimal parameter combination. Alternatives may include using different machine learning frameworks or algorithms to adapt to specific data characteristics and prediction needs.

[0066] In some embodiments, identifying the same type of historical fiber information based on the key influencing factors includes: searching all sample data based on the key influencing factors, applying a particle swarm improvement algorithm, and finding the number of clusters with the smallest error square sum J value and the corresponding initial cluster centers;

[0067] After the initial clustering center is determined, the category of the sample data and the center of each category are finally determined according to the improved fuzzy clustering algorithm, wherein the key influencing factors of the fiber bundle include fiber length, fiber diameter and fiber strength;

[0068] The key influencing factors of the fiber cluster include fiber curvature, fiber surface roughness and fiber moisture content;

[0069] The key influencing factors of the fiber filaments include fiber color, fiber impurity content and fiber density;

[0070] The key influencing factors of the fiber impurities include fiber length, fiber diameter and fiber impurity content.

[0071] It should be noted that the process of identifying the same type of historical fiber information based on key influencing factors involves using the particle swarm algorithm and fuzzy clustering algorithm to analyze and classify sample data. Specifically, the particle swarm algorithm is first used to determine the number of clusters and the initial cluster center, and then the fuzzy clustering algorithm is applied to finally determine the category and category center of the sample data. This method can effectively identify fiber groups with similar characteristics and provide accurate historical reference data for the prediction of fiber separation effect.

[0072] Specifically, the improved particle swarm algorithm can optimize the clustering process by setting parameters such as the number of particles, speed, and position. The goal of the algorithm is to minimize the square error and J value, which can be achieved through an iterative process. The fuzzy clustering algorithm involves the construction of a fuzzy matrix, where μ ik Indicates the probability that the kth sample belongs to the ith class. Fuzzy index m, distance d from the sample to the class center ik The distance between the cluster centers and the distance between the cluster centers are the key parameters to be considered in the algorithm. The setting of these parameters will directly affect the accuracy and reliability of the clustering results.

[0073] Preferably, in order to improve the performance and adaptability of the clustering algorithm, an adaptive parameter adjustment mechanism can be introduced to dynamically adjust the parameters in the particle swarm algorithm and the fuzzy clustering algorithm according to the execution of the algorithm. For example, a feedback loop can be designed to adjust the number or speed of particles in the particle swarm algorithm according to the quality of the clustering results. In addition, other clustering algorithms, such as K-means or DBSCAN, can be considered as an alternative or supplement to fuzzy clustering to enhance the classification ability of different types of fiber data. In practical applications, the algorithm can also be fine-tuned in combination with actual production data and expert experience to better adapt to specific industrial application scenarios.

[0074] In some embodiments, finally determining the category of sample data and the center of each category according to the improved fuzzy clustering algorithm further includes: the objective function of the improved fuzzy clustering algorithm is: Where J = J com -J sep ,

[0075]

[0076]

[0077] μ ik ∈U,U is a c×n fuzzy matrix;

[0078] The fuzzy matrix satisfies the following conditions:

[0079]

[0080] Among them, μ ik represents the probability that the kth sample belongs to the ith class, m is the fuzzy index, d ik is the distance from the kth sample to the center of the i-th class, d ij is the distance between the centers of the i-th and j-th classes.

[0081] Calculate the density J of each type of data in the iterative process com and the inter-class separation J sep , and find J = J com -J sep , and then finally get the optimal number of clusters, cluster prototypes and the type of sample data by taking the maximum value of J;

[0082] Given the number of cluster categories c, 0≤c≤n, n is the number of data, set the iteration stop threshold ∈ and the fuzzy index m, and initialize the cluster prototype pattern v (0) , set the iteration counter b = 0,

[0083]

[0084] If there exists i,k such that d ik >0, then

[0085]

[0086] If there exists i,k such that d ik = 0, then μ ik =1, and for j≠i, μ ik =0: If J (b) If there is no change, the algorithm stops and outputs the partition matrix U and cluster prototype V. Otherwise, let b = b + 1 and go to the previous step.

[0087] Determine the best classification tree C m , calculate the value of J(C), when the condition is met: J(C m )=max{J(C)}, then we think that C m is the optimal number of classifications, and the output partition matrix U and the cluster prototype V are obtained.

[0088] It should be noted that the objective function of the improved fuzzy clustering algorithm and its optimization algorithm determine the category and category center of the sample data by minimizing the objective function. The objective function takes into account the compactness and inter-class separation of the sample data, and optimizes the clustering results through iterative calculation. This optimization method helps to improve the accuracy of clustering analysis, thereby more accurately predicting the fiber separation effect.

[0089] Specifically, the objective function in the improved fuzzy clustering algorithm consists of two parts: the compactness U of the sample data d and the inter-class separation V d . Tightness U d reflects the degree of aggregation of sample data within their respective categories, while the separation V d Measures the distinction between different categories. Parameter μ ik represents the probability that the kth sample belongs to the ith class, m is the fuzzy index, d ik is the distance from the sample to the center of the class. The settings of these parameters directly affect the quality of the clustering results. For example, the fuzzy index m is usually set between 1.5 and 2.5 to ensure the rationality of the clustering results.

[0090] Preferably, in order to improve the efficiency and accuracy of the clustering algorithm, some strategies can be used to optimize the calculation process of the objective function. For example, parallel computing technology can be used to accelerate the iterative calculation of the objective function. In addition, a regularization term can be introduced to prevent overfitting and improve the generalization ability of the model. In terms of parameter setting, methods such as grid search or Bayesian optimization can be used to automatically adjust the fuzzy index m and other key parameters to obtain the best clustering effect. Alternatives may include using different optimization algorithms, such as gradient descent or Newton's method, to solve the minimum problem of the objective function.

[0091] In some embodiments, obtaining the same type of historical fiber separation effect reference data through data analysis includes: calculating the median, upper quartile, lower quartile, maximum value, minimum value, sample mean, and sample variance of the sample data;

[0092] The outliers are removed by box plot to perform an outlier analysis on the sample data, where the outliers are those less than the lower quartile Q. L Subtract 1.5 times the interquartile range IQR, greater than the upper quartile Q U plus 1.5 times the interquartile range (IQR);

[0093] According to Chebyshev’s inequality principle, at least 80% of the statistics fall in the interval φ;

[0094] φ={x|μ-kσ≤x≤μ+kσ}

[0095] Wherein, μ is the mean value of the sample data, σ is the standard deviation of the sample data, and k is the interval coefficient;

[0096] Finally, the reasonable range of fiber separation effect is μ-kσ1≤x≤μ+kσ1. The historical fiber separation effect reference data CKLS of the same type is calculated by the following formula:

[0097]

[0098] Among them, μ is the mean of the sample data, σ1 is the standard deviation of the sample data, and k is the interval coefficient of Chebyshev's inequality.

[0099] It should be noted that the process of obtaining historical fiber separation effect reference data of the same type through data analysis includes calculating the statistical characteristics of the sample data, such as the median, quartile, maximum, minimum, sample mean and sample variance. These statistical characteristics provide a quantitative reference for evaluating the fiber separation effect, help identify and eliminate outliers, and thus improve the accuracy of the prediction.

[0100] Specifically, the data analysis process can be implemented using statistical software or statistical libraries in programming languages. For example, Python's NumPy and SciPy libraries can be used to calculate statistics such as the median and quartiles of sample data. The elimination of outliers can be identified by the box plot method, where 1.5 times the interquartile range (IQR) is used to determine the range of outliers. The Chebyshev inequality principle can be used to estimate the reasonable interval of the data, where the interval coefficient can be adjusted according to the distribution characteristics of the actual data.

[0101] Preferably, in order to further improve the accuracy and robustness of data analysis, more advanced statistical methods such as Z-score or Mahalanobis distance can be used to identify outliers. In addition, machine learning techniques such as outlier detection algorithms can be combined to automatically identify and process outliers. When determining a reasonable interval, more complex distribution assumptions such as normal distribution or lognormal distribution can be considered to more accurately estimate the distribution characteristics of the data. Alternatives may include using other statistical software or libraries, or developing custom data analysis tools to meet specific needs.

[0102] In some embodiments, determining the fiber separation effect prediction target according to the fiber separation effect reference target and the fiber separation effect historical reference data includes: determining the fiber separation effect prediction target by the following formula:

[0103]

[0104] Among them, PG is the predicted target of the fiber to be processed, CKLS is the historical reference data of the fiber separation effect, and CKYC is the reference target of the fiber separation effect constructed in combination with influencing factors.

[0105] It should be noted that determining the prediction target of the fiber separation effect based on the reference target of the fiber separation effect and the historical reference data involves combining the historical separation effect data with the reference target constructed by combining the influencing factors to calculate a comprehensive prediction target. This method can provide more accurate predictions for the fiber separation process, help optimize the production process and improve product quality.

[0106] Specifically, determining the prediction target of the fiber separation effect can be achieved through the following steps: First, collect and analyze historical separation effect data, including but not limited to statistical data such as median, quartile, maximum, minimum, mean and variance. Then, based on the physical properties of the fiber and known influencing factors, construct a reference target, which may involve using a machine learning model to predict the separation effect under specific conditions. Next, combine the historical reference data with the reference target and calculate the prediction target through a specific formula. For example, a weighted average method can be used, in which the historical data and the reference target are each assigned different weights to reflect their importance in the prediction.

[0107] Preferably, in order to improve the accuracy and adaptability of the forecast target, a method of dynamically adjusting weights can be used to adjust the weights of historical data and reference targets based on the latest production data and feedback. In addition, the concept of confidence intervals can be introduced to provide a range for the forecast target instead of a single value, which can better reflect the uncertainty of the forecast. Alternatives may include using different statistical methods or machine learning algorithms to construct reference targets, or using methods such as Monte Carlo simulation to generate distributions of forecast targets, thereby providing more comprehensive forecast results.

[0108] In some embodiments, predicting the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect includes: predicting the fiber separation effect based on the influencing factors of the fiber separation using a support vector machine model optimized based on a genetic algorithm to obtain a reference target for the fiber separation effect, wherein the support vector machine model maps the input vector to a high-dimensional feature space through nonlinear mapping, and constructs an optimal classification hyperplane in the high-dimensional feature space, and the classification function of the support vector machine is similar to a neural network, and the output is a linear combination of intermediate nodes, wherein each intermediate node corresponds to a support vector.

[0109] It should be noted that the prediction method based on fiber separation influencing factors uses a support vector machine (SVM) model optimized by genetic algorithm to predict the fiber separation effect. This method maps the input vector to a high-dimensional feature space through nonlinear mapping, and constructs an optimal classification hyperplane in the space to achieve accurate prediction of the fiber separation effect. This prediction model can not only handle complex nonlinear relationships, but also adapt to the characteristics of different types of fibers, thereby improving the accuracy and reliability of the prediction.

[0110] Specifically, the support vector machine model based on genetic algorithm optimization involves multiple key steps and parameter settings. First, it is necessary to define the parameters of the genetic algorithm, such as population size, crossover rate, mutation rate, and number of iterations. These parameters determine the algorithm's search capability and convergence speed. Secondly, it is necessary to select appropriate support vector machine model parameters, such as kernel function type (such as linear kernel, polynomial kernel, radial basis function kernel, etc.), penalty parameter C, and kernel function parameters. These parameters affect the complexity and generalization ability of the model. By optimizing these parameters through genetic algorithms, the best model configuration can be found, thereby improving the accuracy of predictions.

[0111] Preferably, in order to further improve the performance of the prediction model, a multi-objective optimization strategy can be used to simultaneously consider the accuracy and generalization ability of the model. For example, multiple evaluation indicators such as prediction accuracy, model complexity, and computational time can be set and used to guide the search process of the genetic algorithm. In addition, it is possible to consider using ensemble learning methods such as bagging or random forests to combine the prediction results of multiple support vector machine models to improve the stability and reliability of the overall prediction. Alternatives may include using other types of machine learning models, such as neural networks or decision trees, or using different optimization algorithms, such as particle swarm optimization or simulated annealing, to optimize model parameters.

[0112] The above-mentioned embodiments of the present invention have the following beneficial effects: The present invention provides an efficient and accurate fiber processing technology. This method first collects and analyzes fiber image data to build a typical fiber database containing fiber photos and attribute data. Computer vision technology is used to divide the fibers into detailed modules and identify the key factors affecting the fiber separation effect. This method can not only improve the accuracy and efficiency of fiber separation, but also provide an important reference for the production process by predicting the fiber separation effect, optimize the fiber processing technology, and improve product quality. Furthermore, this technology uses advanced technologies such as module decomposition and particle swarm improvement algorithm to make detailed predictions and adjustments to the fiber separation effect. The support vector machine model optimized by fuzzy clustering algorithm and genetic algorithm can further improve the accuracy and reliability of the prediction. The application of this method can not only improve the efficiency of fiber processing and reduce production costs, but also help to achieve efficient utilization of fiber resources and sustainable development of the environment.

[0113] like Figure 2 As shown, in some embodiments, a fiber separation device 200 based on computer vision recognition includes:

[0114] A data acquisition module, used for collecting fiber image data to form a typical fiber database, wherein the fiber image data includes fiber photos and corresponding fiber attribute data;

[0115] An influencing factor identification module, used to divide the fiber into modules according to the physical properties of the fiber, and identify the influencing factors of fiber separation for different modules of the fiber;

[0116] A target prediction module predicts the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect;

[0117] A historical effect acquisition module, used for setting key influencing factors of the same type of fibers based on the influencing factors of the fiber separation, identifying the same type of historical fiber information based on the key influencing factors, and then obtaining the same type of historical fiber separation effect reference data through data analysis;

[0118] A prediction target acquisition module is used to determine a fiber separation effect prediction target based on the fiber separation effect reference target and the fiber separation effect historical reference data.

[0119] It can be understood that the modules described in the fiber separation device 200 based on computer vision recognition are similar to those described in the reference Figure 1 The various steps in the described fiber separation method based on computer vision recognition correspond to each other. Therefore, the operations, features and beneficial effects described above for the fiber separation method based on computer vision recognition are also applicable to the fiber separation device 200 based on computer vision recognition and the modules contained therein, and will not be repeated here.

[0120] It should be noted that some embodiments of the present disclosure may also include a computer-readable medium, and the computer-readable storage medium may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device. In some embodiments of the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0121] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0122] The computer-readable medium may be included in the electronic device; or it may exist independently without being assembled into the electronic device. The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: collects fiber image data to form a typical fiber database, wherein the fiber image data includes fiber photos and corresponding fiber attribute data; divides the fiber into modules according to the physical properties of the fiber, and identifies the influencing factors of fiber separation for different modules of the fiber; predicts the fiber separation effect based on the influencing factors of fiber separation to obtain a reference target for the fiber separation effect; sets key influencing factors for the same type of fiber based on the influencing factors of fiber separation, identifies the same type of historical fiber information based on the key influencing factors, and then obtains the same type of historical fiber separation effect reference data through data analysis; and determines the fiber separation effect prediction target based on the fiber separation effect reference target and the fiber separation effect historical reference data. Computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0123] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0124] The units described in some embodiments of the present disclosure may be implemented in software or in hardware. The units described may also be provided in a processor, and the functions described above herein may be performed at least in part by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0125] The above description is only some preferred embodiments of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) to form a technical solution. The above description is only some preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present invention (but not limited to) to form a technical solution.

Claims

1. A fiber separation method based on computer vision recognition, characterized in that: include: Collecting fiber image data to form a typical fiber database, wherein the fiber image data includes fiber photos and corresponding fiber attribute data; Divide the fibers into modules according to their physical properties, and identify factors affecting fiber separation for different modules of the fibers; Predicting the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect; Setting key influencing factors for the same type of fibers based on the influencing factors of fiber separation, identifying historical fiber information of the same type based on the key influencing factors, and then obtaining historical reference data of fiber separation effects of the same type through data analysis; and determining a fiber separation effect prediction target based on the fiber separation effect reference target and the fiber separation effect historical reference data; According to the physical properties of the fibers, the fibers are divided into modules, and at the same time, factors affecting fiber separation are identified for different modules of the fibers, including: determining the physical properties of the fibers according to the fiber photos; Decomposing the image of the fiber into fiber bundles, fiber clusters, fiber filaments and fiber impurities according to the physical properties of the fiber; and identifying the influencing factors of the fiber separation for the fiber bundles, fiber clusters, fiber filaments and fiber impurities, wherein the influencing factors of the fiber separation include fiber length, fiber diameter, fiber strength, fiber color, fiber curvature, fiber surface roughness, fiber moisture content, fiber impurity content and fiber density; Identifying the same type of historical fiber information based on the key influencing factors includes: searching all sample data based on the key influencing factors, applying a particle swarm improvement algorithm, and finding the number of clusters with the smallest error square sum J value and the corresponding initial cluster center; After the initial clustering center is determined, the category of the sample data and the center of each category are finally determined according to the improved fuzzy clustering algorithm, wherein the key influencing factors of the fiber bundle include fiber length, fiber diameter and fiber strength; The key influencing factors of the fiber cluster include fiber curvature, fiber surface roughness and fiber moisture content; The key influencing factors of the fiber filaments include fiber color, fiber impurity content and fiber density; The key influencing factors of the fiber impurities include fiber length, fiber diameter and fiber impurity content.

2. The fiber separation method based on computer vision recognition according to claim 1, characterized in that: After determining the fiber separation effect prediction target, the method further includes: combining module decomposition, comparing the fiber separation design effects of different modules with the fiber separation effect prediction target, judging their deviations and impact ratios, and preferentially adjusting modules with large separation effect elasticity to control the deviation within a preset error range; The fiber separation method based on computer vision recognition is used to predict the separation effects of different types of fibers to be processed.

3. The fiber separation method based on computer vision recognition according to claim 2, characterized in that: The fiber separation effect is predicted based on the influencing factors of fiber separation to obtain a reference target for the fiber separation effect, including: separately predicting the separation effects of the fiber bundles, fiber clusters, fiber filaments and fiber impurities to obtain the fiber separation effects of different modules; and superimposing the fiber separation effects of the different modules to obtain the total separation effect of the fibers as the reference target for the fiber separation effect.

4. The fiber separation method based on computer vision recognition according to claim 3, characterized in that: The categories of sample data and the centers of each category are finally determined according to the improved fuzzy clustering algorithm, which further includes: Calculate the closeness of each type of data in the iterative process and inter-class separation , and obtain , in order to obtain the maximum value Finally, the optimal number of clusters, cluster prototypes, and the type of sample data are obtained; Given the number of cluster categories , , Set the iteration stop threshold for the number of data and the fuzzy index , initialize the clustering prototype mode , set the iteration counter ; Determining the best classification tree ,calculate When the condition is met, it is considered is the optimal number of classifications, and the output partition matrix is ​​obtained and the cluster prototype .

5. The fiber separation method based on computer vision recognition according to claim 1, characterized in that: Obtaining historical reference data of the same type of fiber separation effect through data analysis includes: calculating the median, upper quartile, lower quartile, maximum value, minimum value, sample mean and sample variance of the sample data; The outliers are removed by box plot to perform an outlier analysis on the sample data, where the outliers are those less than the lower quartile. Subtract 1.5 times the interquartile range , greater than the upper quartile Add 1.5 times the interquartile range ; According to Chebyshev's inequality principle, at least 80% of the statistics fall within the interval middle; The reasonable range of fiber separation effect is finally obtained by calculating the historical reference data of fiber separation effect of the same type. .

6. The fiber separation method based on computer vision recognition according to claim 5, characterized in that: Determining a fiber separation effect prediction target according to the fiber separation effect reference target and the fiber separation effect historical reference data includes: determining the fiber separation effect prediction target.

7. The fiber separation method based on computer vision recognition according to claim 1, characterized in that: Predicting the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect includes: predicting the fiber separation effect based on the influencing factors of the fiber separation using a support vector machine model optimized based on a genetic algorithm to obtain a reference target for the fiber separation effect, wherein The support vector machine model maps the input vector to a high-dimensional feature space through nonlinear mapping, constructs an optimal classification hyperplane in the high-dimensional feature space, and the output is a linear combination of intermediate nodes, wherein each intermediate node corresponds to a support vector.

8. A fiber separation device based on computer vision recognition, the device implementing the method according to claim 1, characterized in that: include: A data acquisition module, used for collecting fiber image data to form a typical fiber database, wherein the fiber image data includes fiber photos and corresponding fiber attribute data; An influencing factor identification module, used to divide the fiber into modules according to the physical properties of the fiber, and identify the influencing factors of fiber separation for different modules of the fiber; A target prediction module predicts the fiber separation effect based on the influencing factors of the fiber separation to obtain a reference target for the fiber separation effect; A historical effect acquisition module, used for setting key influencing factors of the same type of fibers based on the influencing factors of the fiber separation, identifying the same type of historical fiber information based on the key influencing factors, and then obtaining the same type of fiber separation effect historical reference data through data analysis; A prediction target acquisition module is used to determine a fiber separation effect prediction target based on the fiber separation effect reference target and the fiber separation effect historical reference data.

Images