High-liquid-absorption spunlace non-woven mask base cloth production process based on online detection

Abstract

The invention relates to the technical field of non-woven materials, and discloses a high-liquid-absorption spunlaced non-woven mask base cloth production process based on online detection, which comprises the steps of fiber component uniformity detection, fiber web defect detection, moisture content monitoring, AI dynamic feedback control, process parameter adjustment and finished product quality verification. Through integration of a near infrared spectrum, a CCD visual system and a microwave sensor, real-time monitoring and intelligent regulation and control in the production process are realized. According to the application, the uniformity of fiber components can be remarkably improved, defects of a fiber net are reduced, and the water content is accurately controlled, so that the liquid absorption performance and mechanical strength of a product are improved, the quality consistency of the product is ensured, and the application requirements in the fields of skin care products and medical dressings are met.

Description

Technical Field

[0001] The present invention relates to the technical field of nonwoven materials, in particular to a production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection. Background Art

[0002] With the continuous development of nonwoven material technology, highly absorbent spunlace nonwoven facial mask fabrics have been widely used in skin care products and medical dressings due to their excellent liquid absorption properties and comfort. However, existing production processes still have certain limitations in terms of real-time monitoring and dynamic control. For example, traditional detection methods for fiber composition uniformity are inefficient, making it difficult to promptly detect fluctuations in the production process; the identification of fiber web defects (such as clouding and holes) often relies on manual visual inspection, which has poor consistency; and the lack of precision in moisture content control affects the performance stability of the final product. These issues have restricted the quality improvement of highly absorbent spunlace nonwoven facial mask fabrics. Therefore, a new process combining multi-sensor fusion and AI dynamic control is urgently needed to optimize the production process. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a production process for highly absorbent spunlace nonwoven facial mask fabrics based on online detection. This process addresses issues such as fluctuations in fiber composition uniformity, inefficient web defect identification, and insufficient moisture content control accuracy during production. By integrating multi-sensor fusion with AI dynamic control technology, the matching between real-time monitoring and process parameter adjustment is optimized, significantly improving product quality and production stability.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: A production process for a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection, comprising the following steps: Step 1: Fiber composition uniformity testing uses a near-infrared spectroscopy (NIRS) analysis system to monitor fiber composition in real time. Using a near-infrared light source with a wavelength range of 1000-2500 nm, the reflectance spectrum of the fiber sample is collected. A chemometrics algorithm is then used to establish a correlation model between the cellulose degree of polymerization (DP) and spectral characteristics. The model output is used to assess fiber composition uniformity. If fluctuations in the DP exceed a preset threshold, subsequent process parameter adjustments are automatically triggered.

[0005] Step 2: Fiber web defect detection uses a high-resolution CCD vision system to scan and image the fiber web surface. The system has a resolution of 10 μm / pixel and is equipped with a ring-shaped light source to reduce shadow interference. An image processing algorithm extracts defect features in the fiber web, such as abnormal grayscale distribution in clouding areas or the edge contours of holes. The algorithm outputs the defect location coordinates and type information, and transmits this data to the central control system.

[0006] Step 3: Moisture content monitoring: A microwave sensor is used to measure the moisture content of the fiber web in real time. Operating in a frequency range of 1-3 GHz, the microwave sensor transmits a microwave signal, receives the reflected signal, and calculates the signal attenuation and phase change to derive the moisture content. The sensor probe is kept 5-10 mm from the fiber web surface to ensure a measurement accuracy of ±0.5%.

[0007] Step 4: AI dynamic feedback control. A deep learning-based CNN+LSTM model was developed to correlate inspection data with process parameters. The model inputs included cellulose degree of polymerization (DOP) detected by NIRS, defect information identified by the CCD vision system, and moisture content measured by microwave sensors. The outputs were specific adjustments for hydroentanglement pressure, line speed, and heating temperature. The model was trained using historical production data and integrated with an online update mechanism to ensure that prediction accuracy improved over time.

[0008] Step 5: Process parameter adjustment. Specific process parameter adjustment operations are performed based on the results output by the AI ​​model. For example, when NIRS detects that the degree of cellulose polymerization has dropped by more than 5%, the water injection energy density is automatically increased to 1.5-2.5J / cm² to compensate for the performance loss caused by the reduction in fiber strength. At the same time, if the CCD vision system detects an increase in the number of holes, the production line speed is reduced by 5%-10% to improve the density of the fiber web. In addition, when the microwave sensor detects that the moisture content exceeds the target range, the heating temperature is adjusted by ±5°C to restore the moisture content to the set value.

[0009] Step 6: Finished Product Quality Verification. After completing the above process flow, the final product undergoes comprehensive performance testing. Test items include liquid absorption rate, liquid retention capacity, and mechanical strength. The testing equipment used is a fully automatic liquid absorption tester and a tensile testing machine. The test results are compared with pre-set standards, and products that do not meet the requirements are eliminated.

[0010] Preferably, the NIRS analysis system in step 1 further includes a fiber optic probe assembly, the probe is kept at a distance of 10-20 mm from the surface of the fiber mesh, the fiber diameter is 200 μm, and the numerical aperture is 0.22, to ensure the stability and accuracy of the spectral signal acquisition.

[0011] Preferably, in step 2, the ring light source of the high-resolution CCD vision system uses an LED array with a light source color temperature of 6500K and a color rendering index Ra≥90 to reduce the impact of chromatic aberration on image acquisition. The image processing algorithm uses adaptive threshold segmentation technology combined with morphological filtering to remove noise interference.

[0012] Preferably, in step 3, the probe of the microwave sensor is designed as a dual-port structure, with the transmitting end and the receiving end located on both sides of the fiber mesh, and the measurement sensitivity is improved by differential signal processing technology. The sensor housing is made of polytetrafluoroethylene material, which is resistant to high temperature and corrosion.

[0013] Preferably, in step 4, the training process of the CNN+LSTM model is divided into two stages: the first stage uses historical data for initial training, and the second stage introduces online data for incremental learning. The model input data is normalized, and the output layer uses the Softmax function to implement multi-classification tasks.

[0014] Preferably, in step five, the adjustment range of the water injection energy density is further refined into three levels: when the cellulose polymerization degree fluctuates between 5%-10%, the energy density is adjusted to 1.5J / cm²; when it fluctuates between 10%-15%, the energy density is adjusted to 2.0J / cm²; when it fluctuates more than 15%, the energy density is adjusted to 2.5J / cm².

[0015] Preferably, the finished product quality verification step further includes optimizing the liquid absorption rate test method. The tester has a liquid absorption time of 10 seconds, uses deionized water as the liquid, records the change curve of liquid absorption over time during the test, and calculates the liquid absorption rate per unit area.

[0016] Preferably, the strategy for optimizing the cellulose polymerization degree control based on NIRS detection results further includes: collecting spectral data of different batches of fiber raw materials to construct a spectral database; extracting key characteristic variables through principal component analysis (PCA); establishing a linear regression model based on the relationship between characteristic variables and cellulose polymerization degree; when it is detected that the cellulose polymerization degree is lower than a preset lower limit, automatically starting the raw material mixing program and adding high-polymerization-degree fibers to the production line in proportion.

[0017] Preferably, the CCD vision system-based optimization strategy for fiber web defect detection further includes: analyzing image features of different types of defects and establishing a defect classification model; extracting deep image features through a convolutional neural network (CNN); combining a support vector machine (SVM) algorithm to achieve accurate identification of defect types; and triggering corresponding process parameter adjustment instructions when a specific defect type is detected.

[0018] Preferably, the moisture content control strategy based on microwave sensor optimization further includes: calibrating the temperature drift characteristics of the sensor, and determining the influence coefficient of temperature change on signal attenuation through experiments; in the actual measurement process, introducing a temperature compensation algorithm to eliminate the interference of ambient temperature on the moisture content measurement accuracy.

[0019] The present invention provides a production process for a highly absorbent spunlace nonwoven facial mask base fabric based on online detection. It has the following beneficial effects: 1. The present invention realizes real-time monitoring of fiber composition uniformity, fiber web defects and moisture content by integrating a near-infrared spectroscopy analysis system, a high-resolution CCD vision system and a microwave sensor, significantly improving the controllability of the production process.

[0020] 2. The present invention establishes a correlation between detection data and process parameters by developing an AI dynamic feedback control model based on deep learning, realizes intelligent adjustment of process parameters, and effectively reduces human intervention.

[0021] 3. The present invention solves the problems of fluctuations in cellulose polymerization degree, increased fiber web defects and unstable moisture content by precisely controlling the water injection energy density, production line speed and heating temperature, thereby improving the liquid absorption performance and mechanical strength of the product.

[0022] 4. The present invention ensures the performance consistency of the final product by optimizing the finished product quality verification method, meeting the application requirements of highly liquid-absorbent spunlace non-woven mask base fabric in the fields of skin care products and medical dressings. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a schematic diagram of the overall process flow in an embodiment of the present invention; Figure 2 Schematic diagram of the fiber composition uniformity detection process in an embodiment of the present invention; Figure 3 Schematic diagram of the fiber web defect detection process in an embodiment of the present invention; Figure 4 Schematic diagram of the water content monitoring process in an embodiment of the present invention; Figure 5 Schematic diagram of the flow of the AI ​​dynamic feedback control module in an embodiment of the present invention; Figure 6 Schematic diagram of the process parameter adjustment process in an embodiment of the present invention; Figure 7 This is a schematic diagram of the finished product quality verification process in an embodiment of the present invention; Figure 8 Schematic diagram of the connection relationship of devices in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solution of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] The present invention provides a production process of high liquid absorption spunlace nonwoven mask base fabric based on online detection, and its specific implementation method is combined with Figures 1-8 Provide detailed explanation. Figure 1 This is a process flow diagram that shows the complete process steps from fiber composition uniformity detection to finished product quality verification, including the collaborative workflow of the NIRS analysis system, CCD vision system, microwave sensor and AI dynamic feedback control module.

[0026] In this implementation, the entire process is divided into six main steps: fiber composition uniformity testing, web defect detection, moisture content monitoring, AI dynamic feedback control, process parameter adjustment, and finished product quality verification. The equipment and modules in each step work together through specific connections and positional relationships, ensuring a stable production process and consistent product quality.

[0027] In the fiber composition uniformity detection step, the NIRS analysis system is set at the entrance of the fiber raw material of the production line for real-time monitoring of the fiber composition. The NIRS analysis system includes a near-infrared light source, a fiber optic probe assembly, and a signal processing unit. The wavelength range of the near-infrared light source is set to 1000-2500nm, the fiber optic probe assembly maintains a distance of 10-20mm from the surface of the fiber web, the fiber diameter is 200μm, and the numerical aperture is 0.22 to ensure the stability of the spectral signal acquisition. The fiber optic probe assembly is installed above the production line through a fixed bracket and is connected to the signal processing unit through an optical fiber. The signal processing unit is further connected to the central control system through a data cable. The central control system receives the spectral signal from the NIRS analysis system and establishes a correlation model between the cellulose degree of polymerization and the spectral characteristics through a chemometric algorithm. When it is detected that the cellulose degree of polymerization fluctuates beyond a preset threshold, the central control system automatically triggers the subsequent process parameter adjustment instructions.

[0028] The fiber web defect detection step is completed by a CCD vision system, which is installed in the conveying section after the fiber web is formed. The CCD vision system includes a high-resolution CCD camera, a ring light source, and an image processing unit. The resolution of the CCD camera is 10μm / pixel. It is installed directly above the conveying section and maintains a fixed distance from the fiber web surface. The ring light source uses an LED array with a light source color temperature of 6500K and a color rendering index of Ra≥90 to reduce the impact of color difference on image acquisition. The ring light source is connected to the external power supply through a power cord and communicates with the image processing unit through a control line. The image processing unit is connected to the central control system through a data cable, receives image data collected by the CCD camera, and extracts defect features through adaptive threshold segmentation technology and morphological filtering. Defect feature information includes abnormal grayscale distribution in the cloud spot area or edge contour features of the hole. This information is transmitted to the central control system for subsequent processing.

[0029] The moisture content monitoring step is completed by a microwave sensor, which is installed in the drying section of the fiber web after the hydroentanglement treatment. The microwave sensor includes a transmitter, a receiver, and a signal processing unit. The transmitter and the receiver are located on both sides of the fiber web, forming a dual-port structure, and the measurement sensitivity is improved by differential signal processing technology. The transmitter and the receiver are connected to the signal processing unit via a coaxial cable, and the signal processing unit further communicates with the central control system via a data cable. The operating frequency range of the microwave sensor is 1-3GHz, and the probe maintains a distance of 5-10mm from the surface of the fiber web to ensure that the measurement accuracy reaches ±0.5%. The signal processing unit calculates the signal attenuation and phase change, derives the moisture content value, and transmits the result to the central control system.

[0030] The AI ​​dynamic feedback control module is the core of the entire process, responsible for associating detection data with process parameters and outputting adjustment instructions. The AI ​​dynamic feedback control module includes a deep learning model, a data storage unit, and an output interface. The deep learning model adopts a CNN+LSTM architecture. The model input data comes from the NIRS analysis system, CCD vision system, and microwave sensor. The output parameters include water needle energy density, production line speed, and heating temperature. The data storage unit is used to store historical production data and online update data. The output interface is connected to the actuator on the production line via a data cable. The training process of the deep learning model is divided into two stages. The first stage uses historical data for initial training, and the second stage introduces online data for incremental learning. The model input data is normalized, and the output layer uses the Softmax function to implement multi-classification tasks. The AI ​​dynamic feedback control module adjusts the output parameters in real time based on changes in input data to ensure dynamic optimization of the production process.

[0031] The process parameter adjustment steps are driven by the output instructions of the AI ​​dynamic feedback control module. The specific adjustment operations include the adjustment of the water needle energy density, production line speed and heating temperature. The adjustment range of the water needle energy density is refined into three levels: when the cellulose polymerization degree fluctuates between 5% and 10%, the energy density is adjusted to 1.5J / cm²; when it fluctuates between 10% and 15%, the energy density is adjusted to 2.0J / cm²; when it fluctuates more than 15%, the energy density is adjusted to 2.5J / cm². The adjustment range of the production line speed is 5%-10%. When the CCD vision system detects an increase in the number of holes, the production line speed is reduced to increase the density of the fiber web. The adjustment range of the heating temperature is ±5°C. When the microwave sensor detects that the moisture content exceeds the target range, the heating temperature is adjusted to restore the moisture content to the set value. The above adjustment operations are completed by sending instructions to the actuators on the production line through the central control system.

[0032] The finished product quality verification step is carried out after completing the above process flow, which mainly includes tests on liquid absorption rate, liquid retention capacity and mechanical strength. The test equipment uses a fully automatic liquid absorption tester and a tensile testing machine, and the test items are compared with the preset standards. The liquid absorption rate test method is optimized to record the curve of liquid absorption over time. The liquid absorption time is 10 seconds, deionized water is used as the liquid, and the liquid absorption rate per unit area is calculated. The liquid retention capacity test measures the weight change after liquid absorption by weighing, and the mechanical strength test measures the breaking strength and elongation at break by a tensile testing machine. The test results are compared with the preset standards, and products that do not meet the requirements are eliminated.

[0033] Throughout the entire process, the positional and connection relationships of the various devices and modules ensure efficient data transmission and precise adjustment of process parameters. The NIRS analysis system, CCD vision system, and microwave sensors are installed at different key locations on the production line and connected to the central control system via data cables. The central control system further communicates with the AI ​​dynamic feedback control module. The AI ​​dynamic feedback control module is connected to the actuators on the production line via an output interface, forming a closed-loop control system. This layout and connection method enables the entire process to be seamlessly connected from detection to adjustment to verification, significantly improving product quality and production stability.

[0034] In order to better enable relevant personnel in this technical field to fully understand and implement the present invention, the specific implementation principle of the present invention is supplemented below with reference to a specific application scenario.

[0035] In the production process of highly absorbent spunlace nonwoven face mask fabrics based on online detection, the NIRS analysis system is first used to complete real-time monitoring of fiber composition uniformity. The near-infrared light source emits light signals in the wavelength range of 1000-2500nm, and the optical fiber probe assembly maintains a distance of 10-20mm from the surface of the fiber web to ensure the collection of stable reflectance spectrum signals. After receiving the spectral data, the signal processing unit uses the partial least squares method (PLS) to establish a mathematical model between the cellulose degree of polymerization and the spectral characteristics. When it is detected that the cellulose degree of polymerization fluctuates beyond a preset threshold, for example, the fluctuation amplitude reaches 8%, the central control system generates adjustment instructions based on the model output results and transmits them to the AI ​​dynamic feedback control module.

[0036] The CCD vision system then inspects the fiber web surface for defects. A high-resolution CCD camera scans the web surface at a resolution of 10μm / pixel, while a ring-shaped light source provides uniform illumination, minimizing shadows and chromatic aberration. After receiving the captured image data, the image processing unit uses adaptive threshold segmentation to extract grayscale anomalies in the clouded area and morphological filtering to remove noise. If the number of holes detected exceeds the set standard, for example, more than five holes per square meter, the image processing unit transmits the defect location coordinates and type information to the central control system, triggering process parameter adjustments.

[0037] Moisture content is monitored using a microwave sensor, with its transmitter and receiver located on either side of the fiber web, forming a dual-port structure. The microwave signal attenuates and changes phase after passing through the fiber web. The signal processing unit calculates these changes and derives the moisture content. If the moisture content deviates by more than ±0.5% from the target value—for example, if the moisture content rises from the target value of 8% to 8.6%—the signal processing unit transmits the data to the central control system for subsequent process parameter adjustments.

[0038] The AI ​​dynamic feedback control module serves as the core component, receiving data from the NIRS analysis system, CCD vision system, and microwave sensors. The deep learning model utilizes a CNN+LSTM architecture, with input data including cellulose degree of polymerization, defect characteristics, and moisture content. By learning from historical and online data, the model predicts the optimal combination of process parameters. For example, when the cellulose degree of polymerization decreases by 8% and the moisture content increases by 0.6%, the model outputs adjustment instructions to increase the water injection energy density to 2.0 J / cm², reduce the production line speed to 95% of the standard speed, and increase the heating temperature by 5°C.

[0039] After process parameter adjustments are completed, the actuators on the production line operate according to the instructions of the central control system. For example, the energy density of the water injection system is adjusted to 2.0J / cm², the conveyor motor speed is reduced, and the heating device temperature is increased by 5°C. These adjustments are fed back in real time through the closed-loop control system, ensuring accurate and timely process parameter adjustments.

[0040] Finally, during the final product quality verification phase, a fully automated liquid absorption tester records the amount of liquid absorbed over time for 10 seconds using deionized water. Once the absorption rate per unit area reaches a preset standard, a tensile testing machine further measures the breaking strength and elongation at break. Products that show a lower-than-standard absorption rate or insufficient mechanical strength are rejected to ensure consistent performance in the final product.

[0041] Throughout the entire process, the positioning and connection of various devices and modules ensure efficient data transmission. The NIRS analysis system, CCD vision system, and microwave sensors are installed at key locations on the production line and connected to the central control system via data cables. The central control system communicates with the AI ​​dynamic feedback control module, which in turn connects to the actuators on the production line via output interfaces, forming a complete closed-loop control system. This layout and connection method achieves a seamless transition from detection to adjustment and verification, significantly improving product quality and production stability.

[0042] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A production process for highly absorbent spunlace nonwoven facial mask base fabric based on online detection, characterized in that: The following steps are involved: Step 1: Fiber composition uniformity detection: A near-infrared spectroscopy analysis system is used to monitor fiber composition in real time. A near-infrared light source with a wavelength range of 1000-2500nm is set to collect the reflectance spectrum signal of the fiber sample. A correlation model between the cellulose degree of polymerization and spectral characteristics is established in combination with a chemometric algorithm. When the fluctuation of the cellulose degree of polymerization exceeds a preset threshold, a process parameter adjustment instruction is triggered. Step 2: Fiber web defect detection uses a high-resolution CCD vision system with a resolution of 10μm / pixel to scan and image the fiber web surface. A ring light source is used to reduce shadow interference. The image processing algorithm extracts the defect features in the fiber web and outputs the defect location coordinates and type information. Step 3: Moisture content monitoring: A microwave sensor with an operating frequency range of 1-3 GHz is used to measure the moisture content of the fiber web in real time. By transmitting a microwave signal and receiving the reflected signal, the signal attenuation and phase change are calculated to derive the moisture content value. The probe is kept 5-10 mm away from the fiber web surface. Step 4: AI dynamic feedback control: Develop a deep learning-based CNN+LSTM model. The input data includes cellulose degree of polymerization, defect information, and moisture content data, and the output is the specific adjustment value of the water injection energy density, production line speed, and heating temperature. Step 5: Process parameter adjustment: perform specific process parameter adjustment operations based on the results output by the AI ​​model; Step 6: Finished product quality verification. After completing the above process flow, the final product is subjected to comprehensive performance testing. The test items include liquid absorption rate, liquid retention capacity and mechanical strength.

2. The production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection according to claim 1, characterized in that: The near-infrared spectroscopy analysis system in step 1 further includes a fiber optic probe assembly, the probe maintains a distance of 10-20 mm from the surface of the fiber web, the fiber diameter is 200 μm, and the numerical aperture is 0.

22.

3. The production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection according to claim 1, characterized in that: The annular light source in step 2 adopts an LED array, the color temperature of the light source is 6500K, and the color rendering index Ra≥90. The image processing algorithm adopts adaptive threshold segmentation technology combined with morphological filtering to remove noise interference.

4. The production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection according to claim 1, characterized in that: The microwave sensor probe in step three is designed as a dual-port structure, with the transmitting end and the receiving end located on both sides of the fiber mesh respectively. The measurement sensitivity is improved by differential signal processing technology, and the sensor housing is made of polytetrafluoroethylene material.

5. The production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection according to claim 1, characterized in that: The CNN+LSTM model training process in step 4 is divided into two stages. The first stage uses historical data for initial training, and the second stage introduces online data for incremental learning. The model input data is normalized, and the output layer uses the Softmax function to implement multi-classification tasks.

6. The production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection according to claim 1, characterized in that: The adjustment range of the water injection energy density in step five is divided into three levels. When the cellulose polymerization degree fluctuates between 5% and 10%, the energy density is adjusted to 1.5J / cm²; when it fluctuates between 10% and 15%, the energy density is adjusted to 2.0J / cm²; when it fluctuates more than 15%, the energy density is adjusted to 2.5J / cm².

7. The production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection according to claim 1, characterized in that: The liquid absorption rate test method in step six is ​​optimized to record the change curve of liquid absorption amount over time, the liquid absorption time is 10 seconds, deionized water is used as the liquid, and the liquid absorption rate per unit area is calculated.

8. The production process of a highly liquid-absorbent spunlace nonwoven facial mask base fabric based on online detection according to claim 1, characterized in that: The cellulose polymerization degree control strategy in step one further includes constructing a spectral database, extracting key characteristic variables through principal component analysis, and establishing a linear regression model. When it is detected that the cellulose polymerization degree is lower than a preset lower limit, a raw material mixing program is started and high-polymerization-degree fibers are added to the production line in proportion.

Images