Textile fiber component qualitative online analysis system and method

Abstract

The invention relates to the technical field of textile fiber component analysis, and discloses a textile fiber component qualitative online analysis system and method.The system comprises a spectrum collection module, a data processing terminal and a component recognition center, and the component recognition center comprises a spectrum preprocessing unit, a feature matching unit and a component judgment unit. The spectrum acquisition module acquires multi-band spectrum data of a textile sample in real time, and the data processing terminal receives an instruction to perform spectrum data format conversion. The spectrum preprocessing unit determines an interference section by recording spectrum distortion and a baseline correction interval, and controls the processing terminal to generate a spectrum characteristic matrix to calibrate component distribution on the surface of a sample; a feature matching unit generates a component report according to matched fiber types; when the component determination unit detects an abnormal spectrum, re-measurement is controlled in an interference section. The system realizes online real-time qualitative analysis of the textile fiber components, enhances the anti-interference capability, and can be widely applied to the field of textile fiber component analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of textile fiber composition analysis, and specifically to an on-line qualitative analysis system and method for textile fiber composition. Background Art

[0002] In the development process of the textile industry, the accurate analysis of textile fiber composition is crucial. It is not only a key link to ensure the quality of textile products, but also an important foundation for multiple aspects such as textile product R & D, production, and quality control. With the continuous progress of the textile industry, the market's requirements for textile products are increasing day by day, which poses many challenges to textile fiber composition analysis technology.

[0003] Traditional methods for analyzing textile fiber composition have obvious limitations. For example, some methods require destructive treatment of textile samples, which will undoubtedly damage the samples and cannot meet some detection scenarios with requirements for sample integrity. Moreover, most of these traditional methods need to be analyzed offline in a laboratory environment, with a cumbersome analysis process and long time consumption, making it difficult to achieve real-time monitoring and quality control of the textile production process. In today's rapidly developing textile production, the production efficiency is constantly increasing, and the lag of traditional offline analysis methods seriously affects the timely adjustment and optimization in the production process, and cannot meet the requirements of modern textile industry for efficient and real-time detection.

[0004] In addition, there are also many problems with some existing on-line analysis technologies. The detection accuracy of some technologies is relatively low, making it difficult to accurately identify some fiber components with complex compositions or low contents. At the same time, when facing a complex textile production environment, these technologies have weak anti-interference ability and are easily affected by external factors, resulting in inaccurate detection results. For example, in the textile production process, there may be changes in environmental factors such as temperature, humidity, and light, as well as vibrations during the operation of production equipment, which may all have an adverse impact on the detection results of on-line analysis technologies.

[0005] Moreover, existing textile fiber composition analysis systems also have deficiencies in data processing and analysis. The data processing efficiency is low, and it is unable to process and analyze a large amount of detection data in a timely manner, making it difficult to quickly generate accurate composition analysis reports. At the same time, for abnormal situations occurring during the detection process, existing systems lack an effective processing mechanism and cannot conduct recheck and correction in a timely manner, further affecting the reliability of the detection results.

[0006] With the continuous innovation and diversification of textile materials, new types of textile fibers are emerging continuously, which puts forward higher requirements for textile fiber composition analysis technology. Traditional analysis methods and technologies have been difficult to meet the accurate analysis of new textile fiber compositions, and more advanced, efficient, and accurate on-line analysis systems and methods need to be developed to meet the development needs of the textile industry. Summary of the Invention

[0007] The object of the present invention is to provide a textile fiber component qualitative on-line analysis system and method to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: a textile fiber component qualitative on-line analysis system and method, the system includes: a spectrum acquisition module, a data processing terminal and a component identification center, the spectrum acquisition module and the data processing terminal are respectively communicatively connected to the component identification center, and the component identification center includes a spectrum preprocessing unit, a feature matching unit and a component determination unit; The spectrum acquisition module is used to obtain multi-band spectrum data of textile samples in real time; The data processing terminal is used to receive the processing instruction of the component identification center to perform spectrum data format conversion; The spectrum preprocessing unit is used to record the spectrum distortion interval and baseline correction interval during the processing of the data processing terminal, determine the interference section according to the spectrum distortion interval and baseline correction interval, control the data processing terminal to process the interference spectrum with a preset filtering threshold and record the correction parameters, generate a spectrum feature matrix according to the correction parameters and calibrate the component distribution on the sample surface; The feature matching unit is used to match the fiber types on the sample surface after the component distribution calibration is completed and generate a component report; The component determination unit includes a conventional analysis unit and an abnormal review unit. The conventional analysis unit is used to control the data processing terminal to execute the reference component mode when the data processing terminal is in the standard state; the abnormal review unit is used to control the data processing terminal to perform a retest operation in the interference section according to the component report when the data processing terminal detects abnormal spectra.

[0009] Preferably, the spectrum acquisition module includes an optical dark box and a multi-channel sensor, a spectrum analysis device, a quality evaluation module and a communication relay module arranged in the dark box; The multi-channel sensor is used to synchronously acquire spectrum signals in the near-infrared band and the mid-infrared band; The spectrum analysis device is used to extract the absorption peak characteristics of the spectrum data; The quality evaluation module is used to calculate the similarity between the current absorption peak characteristics and the standard spectrum library based on the support vector machine, judge the data signal-to-noise ratio level, and trigger a retest signal according to the calculation result; The communication relay module is used to upload the spectrum data to the component identification center through an optical fiber channel.

[0010] Preferably, the data processing terminal includes a cabinet housing and a signal conversion module, a baseline calibration module, an instruction execution module and a protocol adaptation module arranged in the housing.

[0011] Preferably, the spectral distortion interval and the baseline correction interval during the processing by the recording data processing terminal include: When the data processing terminal detects spectral distortion, it records the wavelength data of the distortion starting point, continuously monitors the correction end point through the baseline calibration module, and records the channel number where the correction is completed.

[0012] Preferably, the determination of the interference section based on the spectral distortion interval and the baseline correction interval includes: Generate distortion coordinates and correction coordinates based on the wavelength data and the channel number; Perform spatial association on the distortion coordinates, correction coordinates, and the position of the spectral acquisition module to form a continuous section, and mark the continuous section as the interference section.

[0013] Preferably, the control of the data processing terminal to maintain a preset filtering threshold for the interference spectrum and record the correction parameters, generate a spectral feature matrix based on the correction parameters, and perform component distribution calibration on the sample surface includes the following steps: Control the data processing terminal to adjust the filtering parameters of the interference section, and obtain the intensity value in real time through the baseline calibration module; Preset the filtering threshold range. When the intensity value exceeds the preset filtering threshold range, control the data processing terminal to maintain the preset filtering threshold processing; Continuously record the spectral correction values of the data processing terminal to form a set of correction parameters; According to the set of correction parameters, use the wavelet transform algorithm to generate a spectral feature matrix; Extract a number of feature vectors at equal wavelength intervals in the spectral feature matrix, and the number of the feature vectors is proportional to the matrix dimension; Perform component distribution calibration on the sample surface based on the extracted feature vectors.

[0014] Preferably, the fiber type matching for the sample surface after the component distribution calibration includes logically masking the preset detection path that overlaps with the high interference section on the sample surface after the component distribution calibration of the sample surface.

[0015] Preferably, the control of the data processing terminal to perform a retest operation in the interference section according to the component report when an abnormal spectrum is detected by the data processing terminal includes: S1. Select the starting position of the interference section with the shortest processing flow according to the next detection node to be scanned; S2. Control the data processing terminal to switch to the starting position and perform parameter initialization; S3. Control the data processing terminal to cut into the interference section from the starting position and perform spectral resampling according to the detection path in the component report; S4. Obtain the output spectrum of the data processing terminal in real time, and perform baseline fitting on the output spectrum using the least squares method; S5. After each continuous scanning operation in the interference section is completed, control the data processing terminal to pause data acquisition and exit the interference section, and then re-execute S1.

[0016] Preferably, S1 includes the following steps: Calculate the feasible starting positions of each interference section through the Floyd algorithm, and conduct a comprehensive evaluation based on the processing flow and wavelength interval parameters; According to the comprehensive evaluation results, select the starting position of the interference section with the shortest processing flow and the smallest wavelength interval; The comprehensive evaluation based on the processing flow and wavelength interval parameters includes: Establish a two-dimensional evaluation matrix including the number of steps in the processing flow and the wavelength interval value; After normalizing the two-dimensional evaluation matrix, use factor analysis to extract the first principal factor as the comprehensive evaluation index; The component determination unit is used to select the starting position of the interference section with the largest value of the comprehensive evaluation index.

[0017] Preferably, the present invention further includes a method for qualitative on-line analysis of textile fiber components, which is applied to the above-mentioned on-line analysis system for qualitative analysis of textile fiber components. The method includes the following steps: Step 1. Obtain the multi-band spectral data of the textile sample in real time through the spectral acquisition module; Step 2. The spectral preprocessing unit determines the interference section according to the spectral distortion interval and baseline correction interval recorded during the processing of the data processing terminal, controls the data processing terminal to process the interference spectrum with a preset filtering threshold and records the correction parameters, generates a spectral feature matrix according to the correction parameters, and calibrates the component distribution on the sample surface; Step 3. The feature matching unit matches the fiber types on the sample surface with the component distribution calibrated, and generates a component report; Step 4. The component determination unit judges the state of the data processing terminal. If it is in the standard state, it controls the data processing terminal to execute the reference component mode through the conventional analysis unit. If an abnormal spectrum is detected, the data processing terminal is controlled to perform a retest operation in the interference section through the abnormal review unit according to the component report. At the same time, the data processing terminal receives the processing instruction from the component identification center to perform spectral data format conversion.

[0018] Compared with the prior art, the beneficial effects of the present invention are: The textile fiber composition qualitative on-line analysis system and method provided by the present invention have significant advantages in many aspects. In the spectral acquisition link, the spectral acquisition module includes components such as an optical dark box and a multi-channel sensor. The multi-channel sensor can synchronously acquire spectral signals in the near-infrared and mid-infrared bands, extract the absorption peak characteristics in combination with the spectral analysis device, and then perform similarity calculation and signal-to-noise ratio level judgment based on the support vector machine through the quality evaluation module, which can ensure the accuracy and reliability of the acquired data and lay a good foundation for subsequent analysis. Components such as the signal conversion module of the data processing terminal work in coordination with each unit of the composition recognition center. By recording the spectral distortion interval and the baseline correction interval to determine the interference section, performing preset filtering threshold processing on the interference spectrum and recording the correction parameters, generating a spectral feature matrix and completing the composition distribution calibration. This precise preprocessing method effectively improves the data quality and reduces the influence of interference factors on the analysis result.

[0019] After the composition distribution calibration is completed, the feature matching unit logically shields the preset detection path overlapping with the high-interference section, avoiding the adverse effect of the interference section on the fiber type matching, improving the accuracy of the fiber type matching, and then generating a more reliable composition report. The conventional analysis unit and the abnormal review unit of the composition determination unit have clear division of labor. The conventional analysis unit executes the reference composition mode under standard conditions to ensure the analysis efficiency under normal circumstances; when the abnormal review unit detects abnormal spectra, it performs a retest operation within the interference section through a series of steps, such as using the Floyd algorithm to select the optimal starting position, performing parameter initialization, and using the least squares method for baseline fitting, etc., to ensure the accurate handling of abnormal situations and improve the reliability and stability of the system.

[0020] This system realizes the on-line analysis of textile fiber composition, without the need for destructive treatment of samples, can obtain spectral data in real time and perform analysis, greatly improving the analysis efficiency and meeting the requirements of real-time monitoring and quality control in the textile production process. In terms of data processing, a spectral feature matrix is generated through the wavelet transform algorithm, as well as a series of parameter processing and analysis methods, improving the efficiency and accuracy of data processing and being able to quickly generate an accurate composition analysis report. At the same time, the effective processing mechanism for abnormal situations of the system further ensures the reliability of the detection results. In addition, this system can adapt to complex textile production environments, has strong anti-interference ability, can operate stably under various environmental conditions, and can also accurately analyze new textile fiber compositions, having wide applicability and good development prospects, providing strong technical support for the development of the textile industry. Description of the Drawings

[0021] Figure 1 It is the working principle diagram of the textile fiber composition qualitative on-line analysis system described in the present invention; Figure 2It is the schematic diagram of the working principle of the spectral acquisition module; Figure 3 It is the design diagram of the spectral preprocessing process; Figure 4 It is the design diagram of the spectral preprocessing and interference section processing process. Specific implementation manners

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0023] Please refer to Figures 1 - 4 , a textile fiber component qualitative on-line analysis system involved in the present invention, the system includes: a spectral acquisition module, a data processing terminal and a component recognition center. The spectral acquisition module and the data processing terminal are respectively communicatively connected to the component recognition center. The component recognition center includes a spectral preprocessing unit, a feature matching unit and a component determination unit. Specific implementation manners: The spectral acquisition module is used to obtain multi-band spectral data of textile samples in real time. The data processing terminal is used to receive the processing instructions of the component recognition center to perform spectral data format conversion. The spectral preprocessing unit is used to record the spectral distortion interval and the baseline correction interval during the processing of the data processing terminal, determine the interference section according to the spectral distortion interval and the baseline correction interval, control the data processing terminal to process the interference spectrum with a preset filtering threshold and record the correction parameters, generate a spectral feature matrix according to the correction parameters and perform component distribution calibration on the sample surface. The feature matching unit is used to perform fiber type matching on the sample surface with the component distribution calibration completed and generate a component report. The component determination unit includes a conventional analysis unit and an abnormal review unit. The conventional analysis unit is used to control the data processing terminal to execute the reference component mode when the data processing terminal is in the standard state; the abnormal review unit is used to control the data processing terminal to perform a retest operation within the interference section according to the component report when the data processing terminal detects abnormal spectra.

[0024] Embodiment 1: In this embodiment, the spectral acquisition module, as the front-end data acquisition unit of the entire system, its structural design and function realization directly affect the accuracy and real-time performance of textile fiber component analysis. This module includes an optical dark box and a multi-channel sensor, a spectral analysis device, a quality assessment module and a communication relay module arranged inside the dark box. Each component works through precise integration and coordination to achieve efficient acquisition and preliminary processing of multi-band spectral data of textile samples.

[0025] The optical dark box provides a stable measurement environment for the internal optical components. Its box body is made of light-impermeable materials, and the inner wall is lined with a light-absorbing coating, which can effectively avoid the interference of ambient light on the spectral signal. The size of the dark box is designed according to the detection requirements of common textile samples. The internal space is sufficient to accommodate the sample placement platform and various functional components, while ensuring the stability of the optical path. A sample inlet is provided on one side of the dark box, and this inlet is equipped with a light-shielding curtain that can automatically close. When placing or replacing the sample, the light-shielding curtain opens and automatically closes after completion to maintain the optical sealed environment inside the dark box.

[0026] The multi-channel sensor is installed on a fixed bracket inside the dark box. Its detection end faces the sample placement platform and is used to synchronously acquire spectral signals in the near-infrared and mid-infrared bands. This sensor adopts an array detector design. The near-infrared channel covers a wavelength range of 780nm - 2500nm, and the mid-infrared channel covers a wavelength range of 2.5μm - 25μm. The sampling frequencies of the two channels are the same, ensuring the consistency of the synchronously collected spectral data in the time dimension. The optical lens group of the multi-channel sensor is precisely calibrated, which can accurately focus the light signal reflected or transmitted from the sample surface onto the detector chip. The detector chip converts the light signal into an electrical signal and converts it into a digital spectral signal through the internal analog-to-digital conversion circuit.

[0027] The spectral analysis device is connected to the multi-channel sensor through a high-speed data bus and is used to extract the absorption peak characteristics of the spectral data. This device is built with a digital signal processing chip and uses a preset algorithm model to preprocess the original spectral data, including operations such as smoothing and denoising, baseline correction, etc., to eliminate the influence of high-frequency noise and baseline drift. During the absorption peak feature extraction process, the spectral analysis device first performs peak detection on the preprocessed spectral data to identify the positions of all significant absorption peaks, and then calculates the characteristic parameters such as peak wavelength, peak height, peak area, and full width at half maximum of each absorption peak. These characteristic parameters constitute the key feature vector of the spectral data and provide basic data for subsequent fiber component analysis.

[0028] The quality assessment module calculates the similarity between the current absorption peak characteristics and the standard spectral library based on the support vector machine algorithm, judges the data signal-to-noise ratio level, and triggers a retest signal according to the calculation result. The standard spectral library stores the standard spectral data of various common textile fibers and their corresponding absorption peak characteristic parameters. The quality assessment module compares the absorption peak characteristics collected in real time with the data in the standard spectral library and calculates the similarity value between the two through the support vector machine algorithm. At the same time, the module also evaluates the noise level of the spectral data and calculates the signal-to-noise ratio index. When the similarity value is lower than the preset threshold or the signal-to-noise ratio is low, the quality assessment module determines that the quality of the currently collected data does not meet the analysis requirements and sends a retest signal to the multi-channel sensor to trigger the operation of re-collecting spectral data.

[0029] The communication relay module establishes a high-speed data transmission link with the composition identification center through an optical fiber channel for uploading spectral data. This module uses an optical fiber transceiver as the core device to package and encapsulate the absorption peak feature data processed by the spectral analysis device and the evaluation results of the quality assessment module, and transmits them to the composition identification center through the optical fiber channel. Optical fiber transmission has the advantages of strong anti-interference ability, high transmission rate, and long transmission distance, which can ensure that the spectral data is not affected by electromagnetic interference during transmission and guarantee the integrity and accuracy of the data. The communication relay module also has a data caching function. When the optical fiber channel is temporarily congested, the data to be transmitted can be temporarily stored in the buffer and continue to be transmitted after the channel returns to normal, avoiding data loss.

[0030] During the actual working process, the textile sample is placed on the sample placement platform inside the dark box, and the multi-channel sensor synchronously collects the spectral signals of the sample in the near-infrared and mid-infrared bands and transmits the signals to the spectral analysis device for feature extraction. The absorption peak feature data processed by the spectral analysis device and the evaluation results of the quality assessment module are uploaded to the composition identification center through the communication relay module. If the quality assessment module determines that the data quality is poor, it triggers the multi-channel sensor to re-collect the spectral data until the spectral signals that meet the quality requirements are obtained.

[0031] Embodiment 2: In the textile fiber composition qualitative online analysis system, the data processing terminal undertakes the key tasks of spectral data format conversion, baseline calibration, and executing the processing instructions of the composition identification center. Its structural design and function implementation are directly related to the accuracy and reliability of subsequent composition analysis. In this embodiment, the data processing terminal includes a cabinet housing and a signal conversion module, a baseline calibration module, an instruction execution module, and a protocol adaptation module arranged inside the housing. Each component works together to ensure that the spectral data can be effectively processed.

[0032] The cabinet housing is made of high-strength metal materials, has good electromagnetic shielding performance and heat dissipation performance, and can provide a stable working environment for the internal components. The front panel of the housing is provided with a display screen and operation buttons for displaying the system working status and setting parameters. The rear panel is equipped with various interfaces, including a power interface, a data input / output interface, etc., for easy connection with other devices.

[0033] The signal conversion module is one of the core components of the data processing terminal, mainly responsible for converting the received spectral data into a format that meets the requirements of subsequent processing. This module uses a high-performance digital signal processor and a dedicated conversion algorithm, and can quickly and accurately convert the original spectral data collected by multi-channel sensors into a unified data format. During the conversion process, the signal conversion module will perform operations such as sampling rate adjustment and bit depth conversion on the data to ensure data consistency and compatibility. At the same time, this module also has a data compression function, which can reduce the data transmission volume and storage volume without affecting the data quality.

[0034] The baseline calibration module is used to calibrate the baseline of spectral data and eliminate the influence of baseline drift on spectral analysis. Baseline drift is a common problem in spectral measurement, which will cause the overall shift of the spectral signal, thus affecting the accurate extraction of absorption peak characteristics. The baseline calibration module uses an adaptive filtering algorithm, which can monitor the baseline change of spectral data in real time and make dynamic adjustments according to the monitoring results. In actual work, this module will first analyze the original spectral data, identify the baseline area, then generate a smooth baseline curve through a fitting algorithm, and finally subtract the baseline curve from the original spectral data to obtain the corrected spectral data. This dynamic calibration method can effectively cope with the baseline drift problem under different samples and measurement environments and improve the quality of spectral data.

[0035] The instruction execution module is the bridge between the data processing terminal and the composition identification center, responsible for receiving the processing instructions from the composition identification center and performing corresponding operations. This module uses an embedded microcontroller as the core processor, with high real-time performance and reliability. The instruction execution module maintains a connection with the composition identification center through a dedicated communication interface and can receive and parse the instructions from the composition identification center in real time. After receiving the instructions, the instruction execution module will call the corresponding function module for processing according to the type and content of the instructions. For example, if a spectral data acquisition instruction is received, the instruction execution module will control the multi-channel sensor to perform data acquisition; if a data processing instruction is received, the instruction execution module will call the signal conversion module and the baseline calibration module to perform corresponding data processing operations.

[0036] The protocol adaptation module is used to implement the communication protocol adaptation between the data processing terminal and other devices, ensuring smooth communication between various parts of the system. In the on-line analysis system for textile fiber composition qualitative analysis, different devices may adopt different communication protocols, which requires the protocol adaptation module to perform protocol conversion and adaptation. This module supports a variety of common industrial communication protocols, such as Modbus, Profibus, CANopen, etc., and can be flexibly configured according to actual needs. The protocol adaptation module converts the instructions from the composition recognition center into a format that conforms to the communication protocol of the target device through a built-in protocol conversion engine, and at the same time converts the data returned by the target device into a format that the composition recognition center can understand. This protocol adaptation mechanism enables the data processing terminal to be seamlessly connected to various different types of devices, improving the compatibility and scalability of the system.

[0037] During the actual working process, when the data processing terminal detects spectral distortion, it will immediately start the abnormal handling process. First, the baseline calibration module will record the wavelength data of the distortion starting point, and at the same time continuously monitor the calibration termination point and record the channel number where the calibration is completed. These data are very important for subsequent analysis and processing, and they can help the system accurately locate the position and scope of the distortion. According to the recorded wavelength data and channel number, the system will generate distortion coordinates and calibration coordinates, which respectively represent the position where the distortion occurs and the position where the calibration is completed. Then, the system will perform spatial association on the distortion coordinates, calibration coordinates and the position of the spectral acquisition module to form a continuous section, and mark this continuous section as an interference section. This method of spatial association and marking can help the system quickly identify and process the interference section in the subsequent processing process, improving the analysis efficiency and accuracy.

[0038] Under the protection of the cabinet shell, the components of the data processing terminal work together through precise circuit connections and software control to achieve efficient processing of spectral data and timely response to abnormal situations. The signal conversion module ensures the consistency and compatibility of data formats, the baseline calibration module improves the quality of spectral data, the instruction execution module ensures the coordinated work between various parts of the system, and the protocol adaptation module enhances the compatibility and scalability of the system. In the face of abnormal situations such as spectral distortion, the data processing terminal can provide strong support for subsequent analysis and processing through a series of operations such as recording key data, generating coordinates, spatial association and marking interference sections.

[0039] Embodiment 3: In the on-line qualitative analysis system for textile fiber components, the spectral processing and component distribution calibration of the interference section are the key links to ensure the analysis accuracy. In this embodiment, the spectral preprocessing unit controls the data processing terminal to filter the interference spectrum, record the correction parameters, and generate the spectral feature matrix and component distribution calibration based on this. At the same time, when matching the fiber type, the overlapping path of the high-interference section is logically masked to form a complete processing flow.

[0040] When the system determines the interference section through the data processing terminal, first, the spectral preprocessing unit issues an instruction to control the data processing terminal to adjust the filtering parameters of the interference section. The adjustment of the filtering parameters is based on the spectral characteristics of the interference section. For example, for the noise type in a specific wavelength range, a band-pass filtering or low-pass filtering method is selected. The baseline calibration module of the data processing terminal will continuously obtain the spectral intensity value of this area. During this process, the baseline calibration module continuously monitors the baseline offset of the spectral signal to ensure the acquisition accuracy of the intensity value. The preset filtering threshold range is set according to the normal spectral noise level of the same type of fiber in the standard spectral library. For example, for the near-infrared band, the preset threshold may be set within the range of 2-3 times the standard deviation of the noise intensity. If the intensity value collected in real time exceeds this preset range, it indicates that the current filtering effect fails to effectively suppress the interference. The system will control the data processing terminal to maintain the preset filtering threshold processing to avoid spectral distortion caused by frequent parameter adjustment.

[0041] During the filtering process, the data processing terminal will continuously record the spectral correction values of each wavelength point. These correction values include the intensity adjustment amount after filtering, the baseline offset correction amount, etc. All the correction values form a correction parameter set in the order of wavelengths. This set is stored in the form of an array, and each element corresponds to the correction parameter of a specific wavelength point. For example, in the mid-infrared band range from 4000 cm⁻¹ to 400 cm⁻¹, the correction parameters are recorded at an interval of 1 cm⁻¹ to form an array containing 3601 elements. Based on the correction parameter set, the system uses the wavelet transform algorithm to perform multi-scale decomposition on the spectral data. The wavelet transform can decompose the spectral signal into different frequency components, retain the useful feature information while removing the high-frequency noise. By selecting an appropriate wavelet basis function (such as Daubechies wavelet or Symlet wavelet), the corrected spectral data is decomposed into multiple layers, and after extracting the coefficients of each layer, the spectral feature matrix is reconstructed. The rows of this matrix correspond to different wavelength ranges, and the columns correspond to the decomposed feature coefficients. For example, for the near-infrared band of 2500 nm - 780 nm, it is divided into 35 wavelength ranges at an interval of 50 nm, and 10 feature coefficients are generated for each range, forming a 35×10 feature matrix.

[0042] In the spectral feature matrix, a number of eigenvectors are extracted at equal wavelength intervals. The extraction interval is determined according to the spectral resolution. For example, in spectral data with a resolution of 10 nm, eigenvectors are extracted at intervals of 20 nm to ensure that the eigenvectors can reflect the overall spectral characteristics while avoiding redundancy. The number of eigenvectors is proportional to the matrix dimension. For the above 35×10 matrix, 15 eigenvectors can be extracted, and each vector contains a combination of characteristic coefficients in the corresponding wavelength range. These eigenvectors are further processed by dimensionality reduction algorithms such as principal component analysis (PCA) to remove highly correlated dimensions and retain the features that can best characterize the fiber components. Based on the extracted eigenvectors, the system calibrates the component distribution on the sample surface. The calibration process uses a spatial mapping algorithm to correspond the eigenvectors to the physical positions on the sample surface. For example, the sample placement platform is divided into 100×100 grid points, each grid point corresponding to a spectral acquisition position. The eigenvectors are filled into each grid point through an interpolation algorithm to form a two-dimensional heat map of the component distribution. The color depth of the heat map represents the relative content of different fiber components.

[0043] After completing the calibration of the component distribution on the sample surface, the feature matching unit performs fiber type matching on the sample surface. At this time, the system logically masks the preset detection paths that overlap with the high-interference sections on the sample surface. The preset detection paths are typical scanning routes set according to the general requirements of fiber component analysis, such as paths planned in a zigzag or grid pattern on the sample surface. The high-interference sections refer to areas where the interference degree exceeds the preset threshold. For example, in the component distribution calibration, areas with abnormal heat map colors and large fluctuations in correction parameters. The logical masking is achieved by modifying the coordinate points of the detection paths, removing the path points in the overlapping areas from the scanning sequence, and at the same time adjusting the connection method of adjacent path points to ensure the continuity of the scanning path. For example, if the original detection path contains coordinate points A(10,10), B(20,20), C(30,30), and point B is in the high-interference section, after masking point B, the path is adjusted to directly connect A(10,10) to C(30,30), and new intermediate points are inserted between A and C to ensure the scanning density.

[0044] Throughout the entire processing process, the data processing terminal maintains real-time communication with the spectral preprocessing unit. The adjustment of filtering parameters, the acquisition of intensity values, and the recording of correction parameters are all achieved through two-way data transmission to ensure the dynamic response of the processing process. The generation of the spectral feature matrix and the extraction of eigenvectors are completed in the computing unit of the component recognition center, using a high-performance processor for parallel computing to improve the processing efficiency. The spatial mapping algorithm for component distribution calibration combines the mechanical coordinate system of the sample platform and obtains the sample position information in real time through an encoder to ensure the spatial accuracy of the calibration results. The logical masking operation does not affect the normal scanning of non-interference areas, avoiding misjudgment of the fiber type matching results by the interference sections while improving the analysis efficiency.

[0045] Example 4: In the on-line qualitative analysis system for textile fiber components, when the data processing terminal detects abnormal spectra, it is necessary to control the retest operation within the interference section through the abnormal review unit to ensure the accuracy of the spectral data. The working process of this implementation method is described in detail below with specific examples.

[0046] Suppose that during the on-line analysis of a batch of cotton-linen blended fabrics, the system obtains multi-band spectral data of the fabric through the spectral acquisition module. During the data processing, the data processing terminal detects that there is an abnormality in a certain section of the spectral data. After analysis, it is determined that the interference section corresponding to the abnormal spectrum is the area with a wavelength range of 1500nm - 1700nm and a channel number of 3 - 5. At this time, the abnormal review unit starts the retest process, and the specific steps are as follows: First, perform step S1. Select the starting position of the interference section with the shortest processing flow according to the next detection node to be scanned. Suppose that the current detection process has completed the scanning of most areas on the surface of the sample, and the next detection node to be scanned is the coordinate point (50, 60). At this time, the system needs to determine which starting position in the interference section of 1500nm - 1700nm and channels 3 - 5 can start the retest to make the processing flow the shortest.

[0047] The system calculates the feasible starting positions of each interference section through the Floyd algorithm. The Floyd algorithm is used here to analyze the path length and processing steps from different starting positions within the interference section to the next detection node. For this interference section, the possible starting positions include the channel positions corresponding to wavelengths of 1500nm, 1550nm, 1600nm, 1650nm, and 1700nm. The system calculates the number of processing steps involved in the path from each starting position to the next detection node (50, 60), such as the number of filtering parameters that need to be adjusted when starting from 1500nm, the number of spectral points that need to be collected, etc., and at the same time considers the wavelength interval parameter, that is, the degree of proximity between the starting position and the required wavelength range of the next detection node.

[0048] Next, a comprehensive evaluation is carried out based on the processing flow and the wavelength interval parameter. The system establishes a two-dimensional evaluation matrix including the number of processing flow steps and the wavelength interval value. For example, for the starting position of 1500nm, the number of processing flow steps is 10, and the wavelength interval value is 200nm; for the starting position of 1550nm, the number of processing flow steps is 8, and the wavelength interval value is 150nm; for the starting position of 1600nm, the number of processing flow steps is 6, and the wavelength interval value is 100nm; for the starting position of 1650nm, the number of processing flow steps is 7, and the wavelength interval value is 50nm; for the starting position of 1700nm, the number of processing flow steps is 9, and the wavelength interval value is 0nm.

[0049] Then, normalize the two-dimensional evaluation matrix to convert the number of processing steps and the wavelength interval value into a unified dimension. For example, divide the number of processing steps by the maximum number of steps, which is 10, to obtain normalized values of 1, 0.8, 0.6, 0.7, and 0.9 respectively; divide the wavelength interval value by the maximum interval value of 200 nm to obtain normalized values of 1, 0.75, 0.5, 0.25, and 0 respectively. Then, use factor analysis to extract the first principal factor as the comprehensive evaluation index. Factor analysis synthesizes the number of processing steps and the wavelength interval value into a principal factor by analyzing their correlation, and this principal factor can reflect the information of these two variables to the greatest extent. After calculation, the comprehensive evaluation index value at the starting position of 1600 nm is the largest. Therefore, the component determination unit selects this position as the starting position for retesting.

[0050] After selecting the starting position, execute step S2 to control the data processing terminal to switch to the starting position of 1600 nm and perform parameter initialization. After receiving the instruction, the data processing terminal adjusts the internal wavelength control module to set the starting wavelength of spectral acquisition to 1600 nm, and at the same time initializes the sensor parameters of channels 3 - 5, including gain, integration time, etc., to ensure that the sensor is in the best working state.

[0051] Then execute step S3 to control the data processing terminal to cut into the interference section from the starting position of 1600 nm and perform spectral re-acquisition according to the detection path in the component report. The detection path for this sample is recorded in the component report. For example, scan in a grid pattern on the sample surface. The data processing terminal starts from the starting position of 1600 nm and performs spectral acquisition on the interference section of 1500 nm - 1700 nm according to this path, and strictly follows the preset acquisition parameters during the acquisition process, such as sampling interval, number of scans, etc.

[0052] During the spectral re-acquisition process, execute step S4 to obtain the output spectrum of the data processing terminal in real time and perform baseline fitting on the output spectrum using the least squares method. The least squares method finds an optimal baseline curve to minimize the sum of the squares of the errors between the original spectral data and the baseline curve. For example, for the re-acquired spectral data, the system takes the wavelength as the x-axis and the spectral intensity as the y-axis, fits a smooth baseline curve using the least squares method, and then subtracts the original spectral data from this baseline curve to obtain the corrected spectral data to eliminate the influence of baseline drift.

[0053] After completing a continuous scanning operation, step S5 is executed to control the data processing terminal to pause data acquisition and exit the interference section, and then step S1 is executed again. Assume that this scan has completed the spectral re-acquisition in the interference section of 1500nm - 1700nm. The system controls the data processing terminal to stop the acquisition operation, adjusts the wavelength outside the interference section, and exits this area. Then, the system again selects the starting position of the interference section with the shortest processing flow according to the next detection node to be scanned, and repeats the above steps S1 to S5 until all the interference sections that need to be re-measured are completed.

[0054] In this specific example, through the Floyd algorithm and factor analysis method, the system can select the optimal starting position from multiple feasible starting positions, reducing the processing steps and time consumption during the re-measurement process. Parameter initialization ensures that the data processing terminal is in an accurate working state during re-measurement, improving the quality of the re-acquired spectral data. The least squares baseline fitting further optimizes the re-acquired spectral data, making it more accurately reflect the true spectral characteristics of the sample.

[0055] During the entire re-measurement process, the system continuously adjusts the starting position dynamically according to the detection nodes and the situation of the interference sections to ensure the efficiency and accuracy of the re-measurement operation. At the same time, by logically masking the preset detection paths with overlapping high-interference sections, the influence of abnormal spectra on the component analysis results is avoided, ensuring the reliability of the final component report. This re-measurement mechanism can effectively handle the abnormal spectral situations that occur during the textile fiber component analysis process, improving the overall analysis performance and stability of the system.

[0056] Example 5: In the on-line qualitative analysis system for textile fiber components, the analysis method involved in Example 5 realizes the on-line qualitative analysis of textile sample components through the coordinated operation of each module and unit. The implementation process of this method is elaborated in detail below with specific examples.

[0057] Taking a batch of polyester fiber and cotton blended fabric produced by a certain garment factory as an example, the system conducts a component qualitative analysis on this fabric. First, step 1 is executed to obtain the multi-band spectral data of the textile sample in real time through the spectral acquisition module. Place this blended fabric on the sample placement platform inside the optical dark box of the spectral acquisition module, and the multi-channel sensor synchronously acquires the spectral signals in the near-infrared band and the mid-infrared band. The near-infrared channel covers the wavelength range of 780nm - 2500nm, and the mid-infrared channel covers the wavelength range of 2.5μm - 25μm. The two channels collect the optical signals reflected from the fabric surface at the same sampling frequency, convert the optical signals into electrical signals, and then convert them into digital spectral signals through the analog-to-digital conversion circuit. Finally, the absorption peak characteristics such as peak wavelength, peak height, and peak area are extracted by the spectral analysis device, and these spectral data are uploaded to the component recognition center through the optical fiber channel by the communication relay module.

[0058] Then enter Step 2. The spectral preprocessing unit determines the interference section by recording the spectral distortion interval and baseline correction interval during the processing of the data processing terminal, and processes the interference spectrum and calibrates the component distribution. Suppose that during the data processing, the data processing terminal detects that the spectrum is distorted near the wavelength of 1730 nm. The baseline calibration module records the wavelength data of the starting point of the distortion as 1730 nm, and continuously monitors the termination point of the correction. When the baseline correction is completed at the wavelength of 1750 nm, the channel number where the correction is completed is recorded as 4. According to the wavelength data 1730 nm - 1750 nm and the channel number 4, the distortion coordinates and correction coordinates are generated, and then the distortion coordinates, correction coordinates and the position of the spectral acquisition module are spatially associated to form a continuous interference section, that is, the fabric surface area corresponding to the wavelength 1730 nm - 1750 nm and channel 4.

[0059] After determining the interference section, the spectral preprocessing unit controls the data processing terminal to adjust the filtering parameters of this interference section, and the baseline calibration module obtains the intensity value in real time. The preset filtering threshold range is 2 - 3 times the standard deviation of the normal spectral noise intensity. If the intensity value obtained in real time exceeds this range, the system controls the data processing terminal to maintain the preset filtering threshold for processing. At the same time, continuously record the spectral correction values of the data processing terminal to form a set of correction parameters. Based on this set, the wavelet transform algorithm is used to generate a spectral feature matrix. For example, the spectral data in the mid-infrared band is decomposed into different frequency components, and the matrix is reconstructed after retaining the characteristic information. In the spectral feature matrix, feature vectors are extracted at equal wavelength intervals. For example, a feature vector is extracted every 5 nm, and the number of feature vectors is proportional to the matrix dimension. Finally, based on these feature vectors, the component distribution on the sample surface is calibrated. The fabric surface is divided into multiple grid points, each grid point corresponds to a feature vector, and a component distribution heat map is formed through the interpolation algorithm. Different colors in the heat map represent the relative distribution of different fiber components.

[0060] After completing Step 2, execute Step 3. The feature matching unit performs fiber type matching on the sample surface with the component distribution calibrated and generates a component report. On the basis of the component distribution calibration, the feature matching unit compares the spectral feature vectors on the sample surface with the standard features of various fibers in the standard spectral library. For example, the standard spectral features of polyester fiber and cotton are used as references. For the preset detection path on the sample surface that overlaps with the high-interference section (i.e., the area corresponding to the wavelength 1730 nm - 1750 nm and channel 4), if a certain scanning path originally planned in a grid shape passes through this area, then this part of the path is logically masked, the overlapping path points are removed, and the connection of adjacent path points is adjusted to avoid the influence of the interference section on the matching result. Through matching, the component distribution of polyester fiber and cotton in this blended fabric is determined, and a component report is generated. The report contains information such as the distribution area and relative content of each fiber component.

[0061] Finally, it enters step 4. The component determination unit determines the state of the data processing terminal and performs corresponding operations. At the same time, the data processing terminal receives a processing instruction to perform spectral data format conversion. Assume that at the initial processing, the data processing terminal is in the standard state. The conventional analysis unit controls it to execute the reference component mode, that is, processes and analyzes the spectral data according to a preset standard process to generate a preliminary component analysis result. If during subsequent processing, the data processing terminal detects that a certain section of spectral data is abnormal, such as the signal-to-noise ratio being too low in another wavelength region, the anomaly review unit controls the data processing terminal to perform a retest operation within the corresponding interference section according to the component report.

[0062] For example, assume that the spectral anomaly is detected in the region corresponding to wavelength 800nm - 850nm and channel 2. The anomaly review unit first calculates the feasible starting positions of this interference section, such as positions 800nm, 820nm, 850nm, etc., according to the next detection node to be scanned (such as the coordinate point (30, 40)) through the Floyd algorithm. A two-dimensional evaluation matrix is established based on the number of processing flow steps and the wavelength interval parameter. After normalization processing and factor analysis method, the first principal factor is extracted, and the starting position with the largest comprehensive evaluation index value (such as 820nm) is selected. Then it controls the data processing terminal to switch to the starting position of 820nm and perform parameter initialization, cut into the interference section from this position, perform spectral resampling according to the detection path in the component report, obtain the output spectrum in real time and perform baseline fitting using the least squares method. After each continuous scan is completed, the acquisition is paused and the interference section is exited, and a new starting position is selected until the retest is completed. Throughout the process, the data processing terminal continuously receives the processing instructions from the component recognition center to perform format conversion on the spectral data to ensure the compatibility and transmission accuracy of the data among various modules.

[0063] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0064] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An on-line qualitative analysis system for textile fiber components, characterized in that, Including: A spectrum acquisition module, a data processing terminal, and a component identification center. The spectrum acquisition module and the data processing terminal are respectively communicatively connected to the component identification center. The component identification center includes a spectrum preprocessing unit, a feature matching unit, and a component determination unit; The spectrum acquisition module is used to obtain multi-band spectrum data of textile samples in real time; The data processing terminal is used to receive a processing instruction from the component identification center to perform spectrum data format conversion; The spectrum preprocessing unit is used to record the spectrum distortion interval and the baseline correction interval during the processing of the data processing terminal, determine the interference section according to the spectrum distortion interval and the baseline correction interval, control the data processing terminal to process the interference spectrum with a preset filtering threshold and record the correction parameters, generate a spectrum feature matrix according to the correction parameters, and calibrate the component distribution on the sample surface; The feature matching unit is used to perform fiber type matching on the sample surface with the component distribution calibrated, and generate a component report; The component determination unit includes a conventional analysis unit and an abnormal review unit. The conventional analysis unit is used to control the data processing terminal to execute the reference component mode when the data processing terminal is in the standard state; the abnormal review unit is used to control the data processing terminal to perform a retest operation within the interference section according to the component report when the data processing terminal detects abnormal spectra.

2. The on-line qualitative analysis system for textile fiber components according to claim 1, characterized in that The spectrum acquisition module includes an optical dark box and a multi-channel sensor, a spectrum analysis device, a quality assessment module, and a communication relay module arranged in the dark box; The multi-channel sensor is used to synchronously obtain spectrum signals in the near-infrared band and the mid-infrared band; The spectrum analysis device is used to extract the absorption peak characteristics of the spectrum data; The quality assessment module is used to calculate the similarity between the current absorption peak characteristics and the standard spectrum library based on the support vector machine, judge the data signal-to-noise ratio level, and trigger a retest signal according to the calculation result; The communication relay module is used to upload the spectrum data to the component identification center through an optical fiber channel.

3. The on-line qualitative analysis system for textile fiber components according to claim 1, characterized in that, The data processing terminal includes a cabinet housing and a signal conversion module, a baseline calibration module, an instruction execution module, and a protocol adaptation module arranged in the housing.

4. An on-line qualitative analysis system for textile fiber components according to claim 3, characterized in that, The recording of the spectrum distortion interval and the baseline correction interval during the processing of the data processing terminal includes: When the data processing terminal detects spectrum distortion, record the wavelength data of the distortion starting point, continuously monitor the correction end point through the baseline calibration module, and record the channel number where the correction is completed.

5. The on-line qualitative analysis system for textile fiber components according to claim 4, characterized in that, The determination of the interference section according to the spectrum distortion interval and the baseline correction interval includes: Generate distortion coordinates and correction coordinates according to the wavelength data and the channel number; Perform spatial association on the distortion coordinates, the correction coordinates, and the position of the spectrum acquisition module to form a continuous section, and mark the continuous section as the interference section.

6. The on-line qualitative analysis system for textile fiber components according to claim 5, characterized in that, The control of the data processing terminal to process the interference spectrum with a preset filtering threshold and record the correction parameters, generate a spectrum feature matrix according to the correction parameters, and calibrate the component distribution on the sample surface includes the following steps: Control the data processing terminal to adjust the filtering parameters of the interference section, and obtain the intensity value in real time through the baseline calibration module; Preset the filtering threshold range. If the intensity value exceeds the preset filtering threshold range, control the data processing terminal to maintain the preset filtering threshold processing; Continuously record the spectral correction values of the data processing terminal to form a set of correction parameters; According to the set of correction parameters, use the wavelet transform algorithm to generate a spectral feature matrix; Extract a number of feature vectors at equal wavelength intervals in the spectral feature matrix, and the number of the feature vectors is proportional to the matrix dimension; Calibrate the composition distribution of the sample surface based on the extracted feature vectors.

7. The on-line qualitative analysis system for textile fiber components according to claim 6, characterized in that, The fiber type matching for the sample surface after the composition distribution calibration is completed includes, after calibrating the composition distribution of the sample surface, logically masking the preset detection path overlapping with the high-interference section on the sample surface.

8. The on-line qualitative analysis system for textile fiber components according to claim 1, characterized in that The control of the data processing terminal to perform a retest operation within the interference section according to the composition report when an abnormal spectrum is detected by the data processing terminal includes: S1. Select the starting position of the interference section with the shortest processing flow according to the next detection node to be scanned; S2. Control the data processing terminal to switch to the starting position and perform parameter initialization; S3. Control the data processing terminal to cut into the interference section from the starting position and perform spectral resampling according to the detection path in the composition report; S4. Real-time obtain the output spectrum of the data processing terminal, and perform baseline fitting on the output spectrum using the least squares method; S5. After each continuous scanning operation in the interference section is completed, control the data processing terminal to pause the acquisition and exit the interference section, and re-execute S1.

9. The online analysis system for qualitative analysis of textile fiber components according to claim 8, characterized in that The S1 includes the following steps: Calculate the feasible starting positions of each interference section through the Floyd algorithm, and perform comprehensive evaluation based on the processing flow and wavelength interval parameters; According to the comprehensive evaluation results, select the starting position of the interference section with the shortest processing flow and the smallest wavelength interval; The comprehensive evaluation based on the processing flow and wavelength interval parameters includes: Establish a two-dimensional evaluation matrix including the number of steps of the processing flow and the wavelength interval value; After normalizing the two-dimensional evaluation matrix, use the factor analysis method to extract the first principal factor as the comprehensive evaluation index; The composition determination unit is used to select the starting position of the interference section with the largest value of the comprehensive evaluation index.

10. A method for on-line qualitative analysis of textile fiber components, applied to a system for on-line qualitative analysis of textile fiber components as described in any one of claims 1 to 9, characterized in that, It includes the following steps: Step 1, real-time obtain the multi-band spectral data of the textile sample through the spectral acquisition module; Step 2, the spectral preprocessing unit records the spectral distortion interval and baseline correction interval during the processing of the data processing terminal, determines the interference section according to the spectral distortion interval and baseline correction interval, controls the data processing terminal to perform processing with a preset filtering threshold on the interference spectrum and records the correction parameters, generates a spectral feature matrix according to the correction parameters and calibrates the composition distribution of the sample surface; Step 3, the feature matching unit performs fiber type matching on the sample surface after the composition distribution calibration is completed to generate a composition report; Step 4, the composition determination unit judges the state of the data processing terminal. If it is in the standard state, control it to execute the reference composition mode through the conventional analysis unit. If an abnormal spectrum is detected, control the data processing terminal to perform a retest operation within the interference section according to the composition report through the abnormal review unit. At the same time, the data processing terminal receives the processing instruction of the composition identification center to perform spectral data format conversion.

Images