Method for measuring impurity rate of cotton fiber based on dual-band monitoring

Abstract

The invention relates to a cotton fiber impurity rate measuring method based on dual-band monitoring, and the method specifically comprises the steps: building a pipeline experiment platform, and simulating the dynamic change of the fiber impurity rate of an impurity falling region in actual cotton carding production; respectively collecting the reflectivity of the cotton fibers and the various falling impurities in different wavelength ranges, and obtaining the reflection spectrum data of the fibers and the falling impurities; based on the reflection spectrum data, selecting two wavebands with the maximum reflectivity difference as monitoring wavebands, and screening out first monitoring light and second monitoring light from the two wavebands; irradiating the same fiber impurity sample by using the two monitoring lights, and collecting reflected light electric signals of samples with different impurity rates to obtain dual-band reflected light voltage signals and impurity rate data; the cotton fiber impurity rate measuring method is established by fitting the function relationship between the dual-band voltage signal and the fiber impurity rate. According to the invention, high-efficiency, accurate and nondestructive measurement of the falling impurity content of the fiber under the working condition can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent manufacturing spinning equipment monitoring, and particularly to a method for measuring the impurity content rate of cotton fibers based on dual-band monitoring. Background Art

[0002] As one of the important natural fibers globally, the quality of cotton fibers directly affects the production efficiency and product quality of the textile industry. The impurity content rate of cotton fibers, that is, the ratio of non-cotton fiber components (such as impurities like cotton husks, cotton leaves, cotton stems, etc.) contained in cotton, is one of the key indicators affecting the quality assessment of cotton. Traditionally, the measurement of the impurity content rate of cotton mainly relies on methods such as manual selection, mechanical screening, and chemical analysis. Although these methods can relatively accurately determine the impurity content, they all have certain limitations, such as time-consuming, high cost, cumbersome operation, and difficulty in realizing on-line detection. Therefore, how to achieve rapid, accurate, and non-destructive detection of the impurity content rate of cotton has become a major challenge in the field of cotton quality detection.

[0003] In recent years, with the development of image recognition technology, researchers have attempted to combine machine learning algorithms and image processing technology to distinguish the content ratio of cotton fibers and impurities, and relevant patented technologies have emerged one after another, including CN112767367B, CN109738436A, etc. These technologies use image classification algorithms to extract the characteristics of different fiber impurities, and have proposed various optimization schemes, further promoting the innovation and development of cotton quality detection technology. However, image recognition technology requires a large amount of computing power support, with high cost, limited speed, and low accuracy. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for measuring the impurity content rate of cotton fibers based on dual-band monitoring, which can efficiently, accurately, and non-destructively detect the impurity content in the fiber during the cotton processing in real time.

[0005] The technical solution adopted by the present invention to solve its technical problems is: to provide a method for measuring the impurity content rate of cotton fibers based on dual-band monitoring, including the following steps:

[0006] Build a pipeline experimental platform to simulate the dynamic change of the impurity content rate of fiber impurity removal in the waste area during carding production;

[0007] Collect the reflectance of cotton fiber samples and different types of impurity samples under the irradiation of detection light with different wavelengths respectively to obtain reflectance spectral data;

[0008] Based on the reflectance spectral data, extract the two bands with the largest difference in reflectance, and respectively screen out the first monitoring light and the second monitoring light from them;

[0009] The first monitoring light and the second monitoring light were used to irradiate the same fiber trash sample, and the reflected photoelectric signals of the two monitoring lights of the fiber trash samples with different trash contents were collected to obtain dual-band voltage data. Then, the functional relationship between the dual-band voltage and the trash content was fitted, and a cotton fiber trash content measurement model was established.

[0010] Furthermore, the fitting of the functional relationship between the dual-band voltage and the impurity content includes:

[0011] Based on the basic principles of reflectivity and sensor photoelectric conversion, as well as the physical definition of trash content, a mathematical relationship model between the ratio K of two reflected photoelectric signals in the same cotton fiber trash sample and the cotton fiber trash content under the time-sharing irradiation of the first monitoring light and the second monitoring light is derived;

[0012] The statistical characteristics of dual-band voltage data and cotton fiber trash content data were calculated respectively, and data preprocessing was performed;

[0013] According to the mathematical relationship model, the functional relationship between the ratio K and the impurity content of cotton fibers is fitted by a polynomial fitting method using the pre-processed statistical characteristics.

[0014] Furthermore, the method also includes the step of verifying the obtained cotton fiber trash content measurement model using the reflectance spectrum data.

[0015] Furthermore, the reflectance of the cotton fiber samples and the different types of impurity samples under different wavelengths of detection light is collected separately, including:

[0016] Use the pipeline experimental platform to adjust different types of fiber debris samples in the debris area;

[0017] For these samples, different wavelengths of detection light are used to illuminate the impurity area, and the reflected light signal is collected in real time through a photometer with an integrating sphere to calculate the corresponding reflectivity.

[0018] Furthermore, the photometer is an ultraviolet-visible-near infrared spectrophotometer.

[0019] Furthermore, the method of using the first monitoring light and the second monitoring light to illuminate the same fiber trash sample and collecting the reflected photoelectric signals of the fiber trash samples with different trash contents to the two monitoring lights respectively includes:

[0020] The first monitoring light and the second monitoring light are used to irradiate the impurity-removing area in a time-sharing manner;

[0021] The pipeline experimental platform is used to adjust the impurity content of the fiber impurity sample in the impurity area, and the reflected photoelectric signals of the first monitoring light and the second monitoring light are respectively collected by the photoelectric sensor in real time.

[0022] Further, the statistical feature of the dual-band voltage data is the mean or standard deviation of the voltage signal, or a feature obtained by filtering and smoothing the voltage signal.

[0023] Further, the photoelectric sensor uses lens focusing to accurately focus the detection light on the cotton fiber trash sample.

[0024] Further, based on the reflection spectral data, extracting the two bands with the largest reflectance difference as the monitoring bands includes:

[0025] Based on the reflection spectral data, calculating the reflectance of each sample to the detection light of different wavelengths;

[0026] Selecting the wavelength range with a large change in reflectance as the alternative band, selecting several wavelengths from the alternative band as the key wavelengths, and then setting several target bands with the key wavelengths as the boundary wavelengths;

[0027] Calculating the mean reflectance of each sample in each target band, and then selecting the two target bands with the largest standard deviation or the largest maximum difference value of the mean reflectance as the monitoring bands.

[0028] Further, the monitoring bands include a visible light band with a wavelength of 680 nm - 850 nm and a near-infrared light band with a wavelength of 850 nm - 1100 nm.

[0029] Beneficial effects

[0030] Due to the adoption of the above technical solution, compared with the prior art, the present invention has the following advantages and positive effects: By using a photometer in combination with an integrating sphere to measure the reflection spectrum of fiber impurities, and establishing a complete reflection spectrum database, then analyzing the relationship between the fiber impurity content and the reflection spectrum through this database, extracting the visible light and near-infrared light with the largest reflection difference for dual-band monitoring, and then through joint analysis with the fiber impurity content ratio, constructing a functional relationship between the fiber impurity rate and the reflected optoelectronic signal, obtaining a cotton fiber impurity rate measurement model. After the deployment of this model, it can not only improve the measurement accuracy, but also monitor the fiber impurity content in the spinning production process in real time; The present invention simulates the change of fiber impurity content and position movement in the actual working condition through a pipeline simulation experimental platform, so as to collect sample data with different impurity rates, increasing the data diversity and helping the model better learn the relationship between the fiber impurity rate and the optoelectronic signal; By calculating the statistical characteristics of the dual-band voltage data, the input data of the model can more accurately represent the fiber impurity rate. In addition, by performing processing such as smoothing, denoising, normalization or standardization on the statistical characteristics, the noise and data inconsistency are eliminated, thereby improving the stability and robustness of the model; By adjusting the model parameters (such as weights and biases), the mapping relationship between the optoelectronic signal and the fiber impurity rate is optimized. The best model parameters are found by minimizing the sum of squared errors, and a regularization method is used to prevent the model from overfitting and improve the generalization ability of the model; By matching the optoelectronic sensor with a focusing lens and a filter amplification circuit, the optical path system is optimized, the reflected signal is enhanced, and the stability and accuracy of the collected signal are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a flowchart of an embodiment of the present invention;

[0032] Figure 2 is a diagram showing the difference in reflectivity of different fiber impurity components to detection light of different bands in an embodiment of the present invention;

[0033] Figure 3 is a fitting effect diagram of the cotton fiber impurity rate measurement model in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] The following further elaborates the present invention in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of this application.

[0035] An embodiment of the present invention relates to a method for measuring the cotton fiber impurity rate based on dual-band monitoring, including the following steps:

[0036] Build a pipeline experimental platform to simulate the dynamic change of the impurity content rate of fiber impurity removal in carding production;

[0037] Respectively collect the reflectance of cotton fiber samples and different types of impurity samples under the irradiation of detection light with different wavelengths to obtain reflectance spectral data;

[0038] Based on the reflectance spectral data, extract the two bands with the largest difference in reflectance, and respectively screen out the first monitoring light and the second monitoring light from them;

[0039] Use the first monitoring light and the second monitoring light to irradiate the same fiber impurity removal sample, collect the reflected photoelectric signals of the fiber impurity removal samples with different impurity content rates for the two monitoring lights respectively to obtain dual-band voltage data, and then fit the functional relationship between the dual-band voltage and the impurity content rate to establish a cotton fiber impurity content rate measurement model.

[0040] The flow block diagram is as Figure 1 shown, where the pipeline experimental platform is used to simulate the working conditions in actual production. This platform can simulate the change of fiber impurity content and the movement of position, and test the change of voltage signal under different conditions by precisely controlling the state of the mixed sample. During the experiment, the change of fiber impurity content and the movement of position are controlled as independent variables, so as to obtain the change of voltage signal reflecting different impurity-containing states.

[0041] In this embodiment, a UV-Visible-Near Infrared Spectrophotometer (UV-3600i-Plus) is selected and equipped with an integrating sphere (ISR-603) to measure the reflectance spectrum. This device can obtain reflectance spectral data in the range from visible light (680nm–850nm) to near infrared light (850nm–1100nm). By this means, a fiber impurity reflectance spectral database is established, and the difference in reflectance characteristics between fibers and impurities in different bands is analyzed. On the other hand, a photoelectric sensor is used to test the reflected light of fiber impurities to obtain a voltage signal. The optical path design of the photoelectric sensor adopts a focusing lens structure to ensure that the light is accurately focused on the fiber sample. The focusing lens is connected to the sensor. By optimizing the optical path structure and the illumination light intensity distribution, the illumination uniformity of the detection area can be improved, and the reflected signals of fibers and impurities can be enhanced.

[0042] To ensure the accuracy of the photoelectric sensor signal, filtering and amplification technologies are adopted in the signal processing process of the present invention. Through the circuit simulation of the preamplifier and the tuned amplifier, the amplitude-frequency characteristics of the signal are optimized to ensure the stable response of the photoelectric sensor to modulated light with different frequencies; at the same time, the frequency selection performance of the tuned amplifier is strengthened, so as to effectively improve the signal-to-noise ratio of the signal and enhance the stability of the photoelectric signal.

[0043] As Figure 2As shown, a UV-Vis-NIR spectrophotometer is used to perform multiple reflection spectroscopy measurements on the samples. By collecting the reflection spectroscopy data, a reflection spectroscopy database of fiber impurities is established to further analyze the reflectance differences in different wavelength bands.

[0044] The reflection spectroscopy database is a pre-established mapping relationship library of impurity content-spectral characteristics through experiments. Its core role is to provide data support for the selection of monitoring wavelength bands. It can be found in the figure that the reflectance of the fiber is close to 90% in the wavelength range of 680 nm - 1000 nm, higher than the reflectance of the dropped impurities, and the change amplitude is extremely small. Multiple reflection spectroscopy measurements are carried out, and the results are consistent. However, through the quantitative analysis of the spectral characteristics and impurity content in different wavelength bands, the characteristic wavelength bands that respond significantly to impurities (the reflectance difference of impurities in specific wavelength bands is the largest) can be screened out, so as to determine the optimal monitoring wavelength band combination. Specifically, the reflection spectroscopy database stores the complete reflection spectroscopy curves of samples with different impurity contents in the full wavelength band (such as 250 - 2500 nm), but by observing Figure 2 It is possible to quantitatively analyze and identify the reflection wavelength bands of the target impurities. Finally, it is determined that there are obvious larger reflectance differences in the visible light (680 nm - 850 nm) and near-infrared light (850 nm - 1100 nm) wavelength bands. This characteristic provides an effective signal basis for the subsequent measurement of fiber impurity content. Therefore, these two wavelength bands are selected as the monitoring wavelength bands.

[0045] In some embodiments, the monitoring wavelength bands can be obtained through the following quantitative analysis method.

[0046] Figure 2 In it, the horizontal axis represents the wavelength (from 250 to 2500 nm), and the vertical axis represents the reflectance (from 0% to 100%). By observing the wavelength band range, it can be found that the change in reflectance of different samples is particularly significant in a specific wavelength range. The change in reflectance mainly occurs in the wavelength band of about 680 nm to 1100 nm. Especially in these wavelength ranges, the reflectance differences of impurity samples such as yellow jute fiber, cotton leaves, cotton stalks, and cotton seeds are relatively large. Especially near 680 nm, the change in reflectance is particularly obvious, especially between the cotton samples (green, blue, and purple triangle data points) and the jute fiber samples (black squares). In addition, in the wavelength ranges of 850 nm and 1100 nm, the reflectance differences are also relatively significant.

[0047] For further analysis, key wavelengths such as 680 nm, 850 nm, 1100 nm, etc. are selected, and multiple target bands are set with each key wavelength as the boundary wavelength. The mean reflectance of the cotton impurity samples in these target bands is extracted. By calculating the reflectance differences of different samples in these target bands, as well as calculating the standard deviation or the maximum difference value of the reflectance, the bands with the most significant reflectance changes can be identified. From the figure, it can be observed that in the wavelength range from 680 nm to 1100 nm, especially between 850 nm and 1100 nm, the reflectance difference is the most obvious. Therefore, it can be considered that the reflectance characteristics of these two bands have the largest difference and are suitable as the key bands for quantitative analysis and classification detection.

[0048] The database contains multi-dimensional sample data (combinations of different cotton varieties and impurity types), which can provide an adequate sample space to cover the complexity of the real scenario. During model training, feature engineering can be used to extract feature parameters (dual-band reflectance intensity ratio and spectral slope) strongly related to the impurity content from the database spectra, avoiding the overfitting risk of directly using the original optoelectronic signals.

[0049] The spectral data in the database reflect the optical property differences between cotton fibers and impurities, which provides a physical basis for optoelectronic signal processing. Through the joint analysis of signal processing and mathematical models, the reflectance spectral data of fiber impurities and voltage signals are fused to establish a functional relationship between the reflectance and the fiber impurity content. The following are the specific steps for the fusion process, as well as the data and methods to be used:

[0050] 1. Data Preparation

[0051] Reflectance spectral data: The reflectance data of different samples (such as cotton fibers, cotton leaves, cotton husks, etc.) in multiple bands are obtained through an ultraviolet-visible-near-infrared spectrophotometer. These reflectance data should cover the target wavelength range, Figure 2 from 680 nm to 1100 nm in

[0052] Dual-wavelength voltage signal value data: The dual-wavelength voltage signal values associated with the reflectance spectral data are obtained through optoelectronic sensors. These signal values are obtained by optoelectronic conversion of the optoelectronic sensors according to the reflected light intensity of the fiber impurity ratio.

[0053] Fiber impurity content data: By calibrating samples (fiber content from 0% to 100%) through standard tests, the fiber impurity content data of each sample are obtained using the weighing method as calibration data.

[0054] 2. Fusion Processing Method

[0055] Key information was extracted from the reflection spectral data. Statistics such as the average value, maximum value, and minimum value of the reflectance at different bands were selected, and principal component analysis was used to reduce the data dimension and retain the band characteristics that can best reflect the fiber impurity information. The dual-band voltage signal was processed to calculate its statistical features, such as the signal mean, standard deviation, etc., or the features obtained after signal filtering and smoothing. Data preprocessing was performed on the obtained features so that two different types of data can be fused on the same scale, avoiding the influence caused by dimensional differences.

[0056] A method for measuring the impurity content of fibers is provided, using voltage signal data as input and the fiber impurity content as output. The processed results are input into this mathematical method, and a deterministic functional relationship between the fiber impurity content and the dual-band voltage signal is established through polynomial fitting, so as to be used to predict fiber samples with unknown impurity content. The specific steps are as follows:

[0057] (1) Data acquisition: Obtain reflection spectral data, voltage signals, and fiber impurity content from different samples.

[0058] (2) Feature extraction: Extract the reflectance features and voltage signal features of each sample.

[0059] (3) Model construction: Assume that the weighted sum of the reflectance and the voltage signal can predict the impurity content, and establish a mathematical relationship model. According to the mathematical relationship model, regression analysis is performed using the polynomial fitting method (the least squares method is selected in this embodiment), and the best polynomial parameters are found by minimizing the sum of squared errors to obtain the relationship between the reflectance, voltage signal, and fiber impurity content.

[0060] More specifically, first, a mathematical relationship model between the ratio K of the two reflected light signals of the cotton fiber impurity samples irradiated by the first monitoring light and the second monitoring light separately and the fiber impurity content needs to be derived:

[0061]

[0062] Among them, U λ1 and U λ2 are the voltage signal data after conversion of the reflected light of the sample to the first monitoring light with wavelength λ1 and the second monitoring light with wavelength λ2 respectively, K is the ratio of the two, n λ1 and n λ2 are the quantum efficiencies of the detection light with wavelengths λ1 and λ2 respectively, v λ1 and v λ2 are the detection light frequencies with wavelengths λ1 and λ2 respectively, R 杂,λ1 and R 杂,λ2 are the reflectances of the pure fibers with wavelengths λ1 and λ2 respectively, R 棉 is the reflectance of the pure cotton fiber with dual-band wavelengths, ρ棉 and ρ 杂 where ρ is the density of cotton fibers and the density of impurities, and α is the impurity content rate of the fibers.

[0063] From the derived mathematical relationship model, the functional relationship between the ratio K and the impurity content rate of the fibers can be expressed in the following form:

[0064] α = β2K 2 + β1K + β0

[0065] where β0, β1, and β2 are polynomial parameters. By using the regression analysis method to solve for the optimal polynomial parameters that minimize the sum of squared errors, a measurement model for the impurity content rate of cotton fibers is obtained.

[0066] (4) Model verification and optimization: Based on the reflection spectrum data and the impurity content rate data of the fibers, methods such as cross-validation and residual analysis are used to evaluate the prediction accuracy of the model, and the model parameters are adjusted as needed or a more suitable method is selected.

[0067] More specifically, the model can be verified and optimized through the following methods.

[0068] (1) Optimize the mapping relationship between the optoelectronic signal and the impurity content rate of the fibers by adjusting the model parameters (such as weights and biases). Find the optimal model parameters by minimizing the sum of squared errors, and use the regularization method to prevent the model from overfitting and improve the generalization ability of the model.

[0069] (2) Extract key features (such as the average reflectance, standard deviation, maximum value, etc. of the wavelength bands) from the reflection spectrum data and the voltage signal. These features can more accurately represent the impurity content rate of the fibers. Adopt the principal component analysis (PCA) dimensionality reduction method to reduce unnecessary feature dimensions, improve the model training efficiency, and prevent overfitting. By fusing the reflectance data of different wavelength bands with the voltage signal data, comprehensively characterize the optoelectronic properties of the fibers, thereby enhancing the prediction ability of the model.

[0070] (3) Through the pipeline simulation experimental platform, collect sample data with different impurity content rates to increase data diversity and help the model better learn the relationship between the impurity content rate of the fibers and the optoelectronic signal. Perform processing such as smoothing, denoising, normalization, or standardization on the optoelectronic signal and the reflection spectrum data to eliminate noise and data inconsistencies, thereby improving the stability and robustness of the model.

[0071] (4) Through methods such as residual analysis and bias-variance analysis, evaluate the error distribution of the model at different wavelength bands and impurity contents, and further adjust the model structure and parameters to optimize the model performance.

[0072] The finally obtained mathematical model is based on the difference in the reflectance of two wavelength bands and uses algorithms such as regression analysis to accurately calculate the proportion of the content of fiber impurities, such asFigure 3 As shown. In practical applications, this model can predict the impurity content of fibers in real time, providing decision-making support for quality management in the carding production process.

Claims

1. A method for measuring the impurity content rate of cotton fibers based on dual-band monitoring, characterized in that, The following steps are involved: Build a pipeline experimental platform to simulate the dynamic changes of the trash content of fibers in the trash area during carding production; The reflectance of cotton fiber samples and different types of impurity samples under different wavelengths of detection light are collected to obtain reflectance spectrum data; Based on the reflectance spectrum data, two bands with the largest reflectance difference are extracted, and the first monitoring light and the second monitoring light are respectively selected therefrom; The first monitoring light and the second monitoring light are used to irradiate the same fiber trash sample, and the reflected photoelectric signals of the fiber trash samples with different trash contents to the two monitoring lights are collected to obtain dual-band voltage data. Then, the functional relationship between the dual-band voltage and the trash content is fitted to obtain a cotton fiber trash content measurement method.

2. The method according to claim 1, wherein The fitting of the functional relationship between the dual-band voltage and the impurity content includes: Based on the basic principles of reflectivity and sensor photoelectric conversion, as well as the physical definition of trash content, a mathematical relationship model between the ratio K of two reflected photoelectric signals in the same cotton fiber trash sample and the cotton fiber trash content under the time-sharing irradiation of the first monitoring light and the second monitoring light is derived; The statistical characteristics of dual-band voltage data and cotton fiber trash content data were calculated respectively, and data preprocessing was performed; According to the mathematical relationship model, the functional relationship between the ratio K and the impurity content of cotton fibers is fitted by a polynomial fitting method using the pre-processed statistical characteristics.

3. The method according to claim 2, characterized in that, The method also includes a step of verifying the obtained cotton fiber trash content measurement method using the reflectance spectrum data.

4. The method according to claim 2, wherein The method of collecting the reflectance of the cotton fiber samples and the different types of impurity samples under different wavelengths of detection light comprises: The pipeline experimental platform is used to adjust the fiber impurities in the impurity area to a sample containing only cotton fibers or any impurities; For each sample, different wavelengths of detection light are used to illuminate the impurity area, and the reflected light signal is collected in real time through a photometer with an integrating sphere to calculate the corresponding reflectivity.

5. The method according to claim 2, wherein The first monitoring light and the second monitoring light are used to illuminate the same fiber trash sample, and the fiber trash samples with different trash contents are collected to obtain the reflected photoelectric signals of the two monitoring lights respectively. include: The first monitoring light and the second monitoring light are used to irradiate the impurity-removing area in a time-sharing manner; The pipeline experimental platform is used to adjust the impurity content of the fiber impurity sample in the impurity area, and the reflected photoelectric signals of the first monitoring light and the second monitoring light are respectively collected by the photoelectric sensor in real time.

6. The method according to claim 5, wherein The statistical feature of the dual-band voltage data is the mean value or standard deviation of the voltage signal, or is a feature obtained after filtering and smoothing the voltage signal.

7. The method according to claim 5, wherein The photoelectric sensor uses a lens to focus so that the detection light is accurately focused on the cotton fiber trash sample.

8. The method according to claim 1, wherein The method of extracting two bands with the largest reflectivity difference as monitoring bands based on the reflectance spectrum data includes: Based on the reflectance spectrum data, the reflectance of each sample to the detection light of different wavelengths is calculated; Selecting a wavelength range with a large reflectivity change as a candidate band, selecting a number of wavelengths from the candidate band as key wavelengths, and then setting a number of target bands with the key wavelengths as boundary wavelengths; The mean reflectivity of each sample in each target band is calculated, and then the two target bands with the largest standard deviation of the mean reflectivity or the largest difference value are selected as monitoring bands.

9. The method according to claim 1, wherein The monitoring bands include a visible light band with a wavelength of 680 nm - 850 nm and a near-infrared light band with a wavelength of 850 nm - 1100 nm.

Images