A fluff-removing function status detection system and method for a snagging machine

Abstract

The present invention relates to the technical field of state detection, and specifically to a fluff removal function state detection system and method for a raising machine, including: a sensor detection unit that detects a first fluff feature vector through an infrared light sensor, detects a second fluff feature vector through a pressure sensor, and detects a third fluff feature vector through a level sensor; an adaptive adjustment unit that analyzes the fabric type and raising process parameters, and dynamically adjusts the feature baseline set and the evaluation threshold set; a state evaluation unit that respectively performs risk assessment on the fluff feature vectors according to the feature baseline set and the evaluation threshold set to obtain the fluff removal function state; a state prediction unit that predicts the remaining risk time according to the fluff feature vectors, the fluff removal function state, the fabric type, and the raising process parameters. The present invention can achieve forward-looking detection and evaluation of the fluff removal function state of the raising machine.

Description

Technical Field

[0001] The present invention relates to the technical field of state detection, and in particular to a system and method for detecting the state of a napping machine's deflocculation function. Background Art

[0002] In the field of industrial equipment condition monitoring, detection technology based on physical parameter analysis is a core means of achieving functional health assessment. For example, the deflocculation process of a napping machine relies on accurate perception of the dynamic accumulation of fiber flocs, involving the coordinated measurement of multiple physical quantities such as airflow disturbance, material distribution, and channel pressure drop.

[0003] With the advancement of sensing technology and data analysis algorithms, detection systems based on multimodal signal fusion are becoming increasingly popular in the field of industrial detection. For example, the optical scattering method is used to measure the concentration of suspended particles, or a flow channel resistance distribution model is constructed through a pressure differential sensor array. However, existing technologies have not yet been adapted and optimized for the particularities of the deflocculation scenario of a napping machine: the deflocculation function of existing napping machines generally lacks the ability to monitor the accumulation state of materials in the entire flocculent processing path in real time, continuously, and quantitatively, resulting in the inability to predict and judge the state of the channel, filter screen, and collection device in a timely manner. As a result, problems are not discovered until they have seriously affected the performance of the system, affecting the deflocculation efficiency, equipment operation stability, and the economic efficiency of maintenance management.

[0004] Therefore, a system and method for detecting the de-flocculation function status of a napping machine are proposed. Summary of the Invention

[0005] The present invention aims to provide a system and method for detecting the deflocculation function status of a napping machine, thereby enabling forward-looking detection and evaluation of the deflocculation function status of the napping machine. The present invention comprises: a sensor detection unit, which detects a first floc feature vector via an infrared sensor, a second floc feature vector via a pressure sensor, and a third floc feature vector via a level sensor; an adaptive adjustment unit, which analyzes the fabric type and napping process parameters and dynamically adjusts the feature baseline set and the evaluation threshold set; a state evaluation unit, which performs risk assessment on the floc feature vectors based on the feature baseline set and the evaluation threshold set to determine the deflocculation function status; and a state prediction unit, which predicts the remaining risk time based on the floc feature vectors, the deflocculation function status, the fabric type, and the napping process parameters.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A system for detecting the de-flocculation function of a napping machine, comprising:

[0008] a sensor detection unit, detecting a first flocculent feature vector via an infrared light sensor, detecting a second flocculent feature vector via a pressure sensor, and detecting a third flocculent feature vector via a level sensor;

[0009] An adaptive adjustment unit that analyzes the fabric type and napping process parameters, dynamically adjusts the characteristic baseline set and the evaluation threshold set;

[0010] A state evaluation unit that obtains the characteristic baseline set and the evaluation threshold set, performs a first risk assessment on the first floc feature vector to obtain a first deflocculation function state, performs a second risk assessment on the second floc feature vector to obtain a second deflocculation function state, and performs a third risk assessment on the third floc feature vector to obtain a third deflocculation function state;

[0011] A state prediction unit that predicts the remaining risk time based on the floc feature vector, the deflocculation function state, the fabric type, and the napping process parameters.

[0012] Preferably: The first floc feature vector includes the normalized infrared scattering intensity and the floc concentration change rate at each monitoring position; the second floc feature vector includes the real-time pressure difference and the pressure difference change rate; the third floc feature vector includes the floc cumulative height and the height cumulative rate.

[0013] Preferably: The original infrared optical signal at each detection position of the deflocculation pipeline of the napping machine is collected, amplified and filtered and then converted into the normalized infrared scattering intensity, and the floc concentration change rate is calculated according to the normalized infrared scattering intensity to obtain the first floc feature vector;

[0014] The original inlet pressure signal and the original outlet pressure signal at the deflocculation filter inlet of the napping machine are collected, amplified and filtered and then converted into the actual inlet pressure and the actual outlet pressure according to the calibration conversion relationship, and the real-time pressure difference and the pressure difference change rate are calculated according to the actual inlet pressure and the actual outlet pressure to obtain the second floc feature vector;

[0015] The original level sensor signal of the floc collection container is collected, amplified and filtered and then converted into the real-time distance from the sensor to the floc cumulative surface according to the calibration conversion relationship, and the floc cumulative height and the height cumulative rate are calculated according to the real-time distance to obtain the third floc feature vector.

[0016] Preferably, the adaptive adjustment unit includes: encoding the fabric type being processed into a feature vector, and normalizing the raising process parameters to generate a working condition feature vector; training a multi-layer perceptron mapping model based on historical data, receiving the working condition feature vector, and outputting a set of feature baselines and a set of evaluation thresholds corresponding to the current working condition through the forward propagation algorithm; the set of feature baselines includes a first floc feature baseline, a second floc feature baseline, and a third floc feature baseline; the first floc feature baseline includes an infrared scattering intensity baseline and a concentration change rate baseline; the second floc feature baseline includes a pressure difference baseline and a pressure difference change rate baseline; the third floc feature baseline includes a floc cumulative height baseline and a height cumulative rate baseline; the set of evaluation thresholds includes a first evaluation threshold, a second evaluation threshold, and a third evaluation threshold; the first evaluation threshold includes an infrared scattering intensity threshold and a concentration change rate threshold; the second evaluation threshold includes a pressure difference threshold and a pressure difference change rate threshold; the third evaluation threshold includes a floc cumulative height threshold and a height cumulative rate threshold.

[0017] Preferably, the state evaluation unit includes:

[0018] For the first floc feature vector, calculate the intensity risk score of each detection position of the deflocculation pipeline of the raising machine according to the infrared scattering intensity baseline and the infrared scattering intensity threshold; calculate the trend risk score of each detection position of the deflocculation pipeline of the raising machine according to the concentration change rate baseline and the concentration change rate threshold; take the maximum value of the intensity risk score and the trend risk score as the single-point risk score of the detection position; calculate the maximum value of the single-point risk scores of all detection positions at the same time as the first deflocculation function state of the deflocculation pipeline of the raising machine;

[0019] For the second floc feature vector, calculate the degree of filter blockage according to the pressure difference baseline and the pressure difference threshold; calculate the filter blockage adjustment factor according to the pressure difference change rate baseline and the pressure difference change rate threshold; multiply the degree of filter blockage and the filter blockage adjustment factor to obtain the second deflocculation function state;

[0020] For the third floc feature vector, calculate the saturation according to the floc cumulative height baseline and the floc cumulative height threshold; calculate the cumulative rate risk according to the height cumulative rate baseline and the height cumulative rate threshold; determine the cumulative rate risk coefficient according to the cumulative rate risk, and multiply the saturation and the cumulative rate risk coefficient to obtain the third deflocculation function state.

[0021] Preferably, the state prediction unit includes: constructing a multi-source input sequence based on the floc feature vectors, deflocculation function states, corresponding fabric types, and raising process parameters in the historical time series; using a pre-trained LSTM model to receive the multi-source input sequence and output a floc feature vector prediction sequence; obtaining a deflocculation function state prediction sequence through a state evaluation unit according to the floc feature vector prediction sequence; checking the time steps when each component in the deflocculation function state prediction sequence first exceeds the corresponding risk threshold to obtain the remaining time of pipeline blockage, the remaining time of filter blockage, and the remaining time of collection saturation respectively, and forming the remaining risk time.

[0022] A method for detecting the deflocculation function state of a raising machine includes:

[0023] Detecting a first floc feature vector through an infrared light sensor, detecting a second floc feature vector through a pressure sensor, and detecting a third floc feature vector through a level sensor;

[0024] Analyzing the fabric type and raising process parameters, and dynamically adjusting the feature baseline set and the evaluation threshold set;

[0025] Obtaining the feature baseline set and the evaluation threshold set, performing a first risk assessment on the first floc feature vector to obtain a first deflocculation function state, performing a second risk assessment on the second floc feature vector to obtain a second deflocculation function state, and performing a third risk assessment on the third floc feature vector to obtain a third deflocculation function state;

[0026] Predicting the remaining risk time according to the floc feature vector, deflocculation function state, fabric type, and raising process parameters.

[0027] Compared with the prior art, the beneficial effects of the present invention are:

[0028] 1. The first floc feature vector collects the scattering intensity and concentration change rate in the pipeline through an infrared light sensor, which not only reflects the static distribution of flocs but also captures their dynamic accumulation trend; the second floc feature vector measures the real-time pressure difference and pressure difference change rate through a pressure sensor, reflecting both the current blockage degree and blockage speed of the filtering device; the third floc feature vector measures the cumulative height and height accumulation rate of flocs through a level sensor to comprehensively monitor the saturation condition of the collection device. Based on these three feature vectors, corresponding risk assessment methods are designed. The combination of multi-level feature vectors and risk assessment methods can achieve real-time, continuous, and quantitative monitoring of the entire floc treatment path, timely detect problems such as pipeline blockage, filter blockage, and collection device saturation, and solve the problem of being unable to predict and judge the blockage degree of channels and filters and the saturation state of the collection device in a timely manner.

[0029] 2. Applying the LSTM neural network to the prediction of the defluffing function status of the raising machine effectively solves the problem of being unable to predict the changes in the system status. First, a multi-source input sequence is constructed by integrating historical fluff feature vectors, historical defluffing function status, and corresponding fabric types and raising process parameters to form input data containing rich time-series information. Then, the pre-trained LSTM model effectively captures long-term and short-term dependencies. Finally, the predicted sequence of fluff feature vectors obtained is converted into a predicted sequence of defluffing function status through a state evaluation unit, and the remaining time for each functional component to reach the risk threshold is calculated to achieve advance prediction. By providing a time window for predictive maintenance, intervention can be carried out before it seriously affects the production efficiency, avoiding the problem of reduced efficiency caused by passive discovery of functional failures and improving the defluffing efficiency and the operating stability of the equipment.

[0030] 3. Based on the dynamic parameter adjustment according to working conditions, it solves the problem of inaccurate monitoring caused by large differences in the characteristics of fluff generation under different fabrics and raising processes. First, a standardized working condition feature vector is generated according to the fabric type and raising process parameters to effectively capture the core features of different working conditions. Then, through a multi-layer perceptron mapping model, the mapping relationship between working conditions and optimal parameters is trained based on historical data to achieve an accurate conversion from the working condition feature vector to the feature baseline set and the evaluation threshold set. The monitoring parameters are automatically adjusted according to the fabric type and raising process parameters of the current processing to adapt to the different characteristics of fluff generation and accumulation under different working conditions, ensuring that the fluff status can be accurately monitored under various production conditions, avoiding false alarms or missed alarms caused by fixed parameter settings, improving the reliability and practicality of the detection of the defluffing function status, and providing guarantee for the economy of maintenance management. Brief Description of the Drawings

[0031] Figure 1 It is a schematic structural diagram of a defluffing function status detection system for a raising machine according to the present invention;

[0032] Figure 2 It is a schematic diagram of the first risk assessment process according to the present invention;

[0033] Figure 3 It is a schematic diagram of the second risk assessment process according to the present invention;

[0034] Figure 4 It is a schematic diagram of the third risk assessment process according to the present invention;

[0035] Figure 5 It is a schematic diagram of a defluffing function status detection method process for a raising machine according to the present invention. Detailed Embodiments

[0036] 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. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0037] Please refer to Figures 1 to 5 , the present invention provides a fluff removal function state detection system and method for a raising machine, and the technical solutions are as follows:

[0038] Embodiment 1:

[0039] This embodiment is applied to the raising process in the cotton and wool blended fabric production line of a certain textile processing enterprise. The production line of this enterprise needs to process different types of fabrics every day, including pure cotton, cotton-polyester blend, wool-polyester blend and other fabrics, and the raising process parameters of different fabrics are different, resulting in obvious differences in the generation amount and characteristics of fluff. To solve the problem of fluff removal function maintenance, this embodiment deploys a fluff removal function state detection system for a raising machine. A fluff removal function state detection system for a raising machine, see Figure 1 , including:

[0040] The sensor detection unit detects the first fluff characteristic vector through an infrared light sensor, detects the second fluff characteristic vector through a pressure sensor, and detects the third fluff characteristic vector through a level sensor;

[0041] The adaptive adjustment unit analyzes the fabric type and raising process parameters, and dynamically adjusts the characteristic baseline set and the evaluation threshold set;

[0042] The state evaluation unit obtains the characteristic baseline set and the evaluation threshold set, conducts a first risk assessment on the first fluff characteristic vector to obtain the first fluff removal function state, conducts a second risk assessment on the second fluff characteristic vector to obtain the second fluff removal function state, and conducts a third risk assessment on the third fluff characteristic vector to obtain the third fluff removal function state;

[0043] The state prediction unit predicts the remaining risk time according to the fluff characteristic vector, the fluff removal function state, the fabric type and the raising process parameters.

[0044] Furthermore, the first fluff characteristic vector includes the normalized infrared scattering intensity and the fluff concentration change rate at each monitoring position; the second fluff characteristic vector includes the real-time pressure difference and the pressure difference change rate; the third fluff characteristic vector includes the fluff accumulation height and the height accumulation rate.

[0045] By simultaneously monitoring the scattering intensity and the rate of change of concentration, the pressure difference and the rate of change of pressure difference, and the cumulative height and the height accumulation rate, the system can achieve real-time, continuous, and quantitative monitoring of the entire floc treatment path, timely detect changes in the accumulation state of flocs in the pipeline, filtration, and collection links, and provide a data basis for comprehensively monitoring the effective operation of the defluffing function of the brushing machine.

[0046] Furthermore, the original infrared optical signals at each detection position of the defluffing pipeline of the brushing machine are collected, amplified and filtered, and then converted into normalized infrared scattering intensities. The rate of change of floc concentration is calculated based on the normalized infrared scattering intensities to obtain the first floc feature vector; the normalized infrared scattering intensity can reflect the floc concentration, and the rate of change of floc concentration is obtained by calculating the first-order derivative of the normalized infrared scattering intensity within a preset time window.

[0047] The original inlet pressure signal and the original outlet pressure signal at the defluffing filter inlet of the brushing machine are collected, amplified and filtered, and then converted into the actual inlet pressure and the actual outlet pressure according to the calibration conversion relationship. The real-time pressure difference and the rate of change of pressure difference are calculated based on the actual inlet pressure and the actual outlet pressure to obtain the second floc feature vector.

[0048] The original level sensor signal of the floc collection container is collected, amplified and filtered, and then converted into the real-time distance from the sensor to the floc accumulation surface according to the calibration conversion relationship. The floc accumulation height and the height accumulation rate are calculated based on the real-time distance to obtain the third floc feature vector.

[0049] In the embodiment of the present application, a total of 6 infrared scattering sensors are evenly deployed along the defluffing pipeline, and the monitoring positions are the pipeline starting point, the main bend, and the front end of the filter. One pressure sensor is deployed at the inlet and outlet of the filter, and the selected model is a digital pressure sensor with an accuracy of 0.01 kPa. An ultrasonic level sensor is deployed above the collection device, with a measurement range of 0 - 2 m and an accuracy of ±1 cm.

[0050] Through the standardized signal amplification, filtering, and conversion processes, effective feature information can be extracted from the original signals, ensuring the accuracy and reliability of the monitoring data, providing high-quality basic data for accurately judging the blockage degree of the channel and the filter screen and the saturation state of the collection device, and avoiding the problem of misjudging the state caused by improper signal processing.

[0051] Further, the adaptive adjustment unit includes: encoding the type of fabric being processed into a feature vector, and normalizing the raising process parameters to generate a working condition feature vector; training a multi-layer perceptron mapping model based on historical data, receiving the working condition feature vector, and outputting a set of feature baselines and a set of evaluation thresholds corresponding to the current working condition through the forward propagation algorithm; the set of feature baselines includes a first floc feature baseline, a second floc feature baseline, and a third floc feature baseline; the first floc feature baseline includes an infrared scattering intensity baseline and a concentration change rate baseline; the second floc feature baseline includes a pressure difference baseline and a pressure difference change rate baseline; the third floc feature baseline includes a floc cumulative height baseline and a height cumulative rate baseline; the set of evaluation thresholds includes a first evaluation threshold, a second evaluation threshold, and a third evaluation threshold; the first evaluation threshold includes an infrared scattering intensity threshold and a concentration change rate threshold; the second evaluation threshold includes a pressure difference threshold and a pressure difference change rate threshold; the third evaluation threshold includes a floc cumulative height threshold and a height cumulative rate threshold.

[0052] The feature encoding method for different fabric types is characterized by their composition ratios (percentage of cotton, polyester, and wool content) and physical properties (fabric density and thickness). This encoding method enables the adaptive adjustment unit to identify the characteristic differences of different fabrics and generate monitoring parameters matching the fabric characteristics through the mapping model. The normalization of the raising process parameters adopts a linear mapping method to map each parameter to the range of 0 to 1. The process parameters include the rotation speed of the raising roller, the pressure of the pressure roller, the raising depth, the conveying speed, and the air suction intensity.

[0053] Table 1 shows the baseline and threshold settings of three typical fabrics under standard raising process parameters; the baseline represents the expected value under normal working conditions, and the threshold represents the risk level requiring intervention. It can be seen from Table 1 that there are obvious differences in the floc characteristics generated by different fabric types: the wool-polyester blended fabric generates more flocs, with higher baselines for infrared scattering intensity and cumulative rate, but its filtration resistance is smaller (low pressure difference baseline); the pure cotton fabric generates fewer flocs but is more likely to clog the filter (high pressure difference threshold). When the system is first deployed, the adaptive adjustment unit has not collected enough data, and these baselines and thresholds are set based on expert experience as the initial state.

[0054] Table 1 Dynamic Baselines and Thresholds Corresponding to Different Fabric Types

[0055]

[0056] Through the multi-layer perceptron mapping model, the system can automatically adjust the baseline and threshold according to different fabric types and process parameters, solving the problem of inaccurate monitoring caused by large differences in the characteristics of floc formation under different fabrics and processes, and ensuring accurate monitoring of the floc state under various production conditions. By dynamically adjusting the monitoring parameters, false alarms or missed alarms caused by fixed parameter settings are avoided, the reliability of the detection of the deflocculation function state is improved, and technical support is provided for the stable operation of the equipment and the economy of maintenance management.

[0057] Further, the state evaluation unit includes:

[0058] See Figure 2 , for the first floc feature vector, calculate the intensity risk score of each detection position of the deflocculation pipeline of the raising machine according to the infrared scattering intensity baseline and the infrared scattering intensity threshold; calculate the trend risk score of each detection position of the deflocculation pipeline of the raising machine according to the concentration change rate baseline and the concentration change rate threshold; take the maximum value of the intensity risk score and the trend risk score as the single-point risk score of the detection position; calculate the maximum value of the single-point risk scores of all detection positions at the same time as the first deflocculation function state of the deflocculation pipeline of the raising machine;

[0059] Specifically, the calculation method of the intensity risk score is: first calculate the first difference between the normalized infrared scattering intensity and the corresponding baseline, and then take the ratio of the first difference to the intensity threshold range (normalized infrared scattering intensity threshold minus normalized infrared scattering intensity baseline) as the intensity risk score. When the normalized infrared scattering intensity is less than the corresponding baseline, the intensity risk score is set to 0; when the normalized infrared scattering intensity exceeds the corresponding threshold, the intensity risk score will exceed 1. The calculation method of the trend risk score is: first calculate the second difference between the current concentration change rate and the corresponding baseline, and then take the ratio of the second difference to the concentration threshold range (concentration change rate threshold minus concentration change rate baseline) as the trend risk score; when the concentration change rate is less than the corresponding baseline, the trend risk score is set to 0; when the change rate exceeds the corresponding threshold, the trend risk score will exceed 1.

[0060] See Figure 3 , for the second floc feature vector, calculate the degree of filter blockage according to the differential pressure baseline and the differential pressure threshold; calculate the filter blockage adjustment factor according to the differential pressure change rate baseline and the differential pressure change rate threshold; multiply the degree of filter blockage and the filter blockage adjustment factor to obtain the second deflocculation function state;

[0061] Specifically, the calculation method of the filter blockage degree is as follows: First, calculate the third difference between the current pressure difference and the pressure difference baseline, and then multiply the ratio of the third difference to the pressure difference threshold range (the pressure difference threshold minus the pressure difference baseline) by 100%. When the calculation result of the blockage degree exceeds 100%, it is limited to 100%. The calculation method of the filter blockage adjustment factor is as follows: First, calculate the fourth difference between the pressure difference change rate and the pressure difference change rate baseline, then calculate the ratio of the fourth difference to the pressure difference change rate threshold range (the pressure difference change rate threshold minus the pressure difference change rate baseline), multiply it by a preset rate weight coefficient (set to 0.5 in this embodiment), and then add 1 to obtain the adjustment factor. When the pressure difference change rate is less than the pressure difference change rate baseline, it indicates that the blockage speed slows down, and the adjustment factor becomes a value less than 1.

[0062] See Figure 4 , for the third floc feature vector, calculate the saturation according to the floc cumulative height baseline and the floc cumulative height threshold; calculate the cumulative rate risk according to the height cumulative rate baseline and the height cumulative rate threshold; determine the cumulative rate risk coefficient according to the cumulative rate risk, and multiply the saturation and the cumulative rate risk coefficient to obtain the third deflocculation function state.

[0063] Specifically, multiply the ratio of the currently measured floc cumulative height to the floc cumulative height threshold by 100% to obtain the saturation. When the calculation result exceeds 100%, it is limited to 100%. The calculation method of the cumulative rate risk is as follows: First, calculate the fifth difference between the current height cumulative rate and the height cumulative rate baseline value, and then use the ratio of the fifth difference to the cumulative rate threshold range (the height cumulative rate value minus the height cumulative rate baseline) as the cumulative rate risk; when the height cumulative rate is less than the height cumulative rate baseline, the cumulative rate risk is set to 0; when the height cumulative rate exceeds the height cumulative rate threshold, the cumulative rate risk will exceed 1. According to the cumulative rate risk, determine the corresponding cumulative rate risk coefficient according to the preset grading standard. The specific grading is as follows: when the cumulative rate risk is less than 0.2, the cumulative rate risk coefficient is 1.0; when the cumulative rate risk is between 0.2 and 0.6, the cumulative rate risk coefficient is 1.1; when the cumulative rate risk is between 0.6 and 0.8, the cumulative rate risk coefficient is 1.3; when the cumulative rate risk is between 0.8 and 1.0, the cumulative rate risk coefficient is 1.5; when the cumulative rate risk is greater than 1.0, the cumulative rate risk coefficient is 2.0.

[0064] For the differential risk assessment method of the three floc feature vectors, adopting different assessment strategies according to the characteristics of different components can discover potential risks more comprehensively and earlier. The multi-dimensional risk assessment mechanism enables the system to avoid the problem of passive discovery of blockages, and at the same time, by evaluating the saturation state of the collection device, it prevents the phenomena of overflow or premature emptying.

[0065] Furthermore, the state prediction unit includes: constructing a multi-source input sequence based on the floc feature vectors, deflocculation function states, corresponding textile types, and raising process parameters in the historical time series; using a pre-trained LSTM (Long Short-Term Memory) model to receive the multi-source input sequence and output a floc feature vector prediction sequence; obtaining a deflocculation function state prediction sequence through a state evaluation unit according to the floc feature vector prediction sequence; checking the time steps at which each component in the deflocculation function state prediction sequence first exceeds the corresponding risk threshold to obtain the remaining time for pipeline blockage, filter blockage, and collection saturation respectively, which constitutes the remaining risk time.

[0066] Specifically, the floc feature vector prediction sequence generated by the LSTM model contains the predicted values of three types of floc feature vectors at multiple future time points. The predicted different floc feature vectors are sequentially input into the state evaluation unit, and the corresponding deflocculation function state prediction sequence is calculated using the same method as the current state evaluation to obtain three deflocculation function state prediction values at each future time point. A risk threshold is set for each deflocculation function state, and the time point at which each state component in the prediction sequence first exceeds its corresponding risk threshold is checked, and the interval between this time point and the current time is calculated to obtain the corresponding remaining time. Table 2 shows the verification results of the prediction effect within different time ranges on the actual production line, demonstrating the prediction accuracy of the LSTM model for different deflocculation function states:

[0067] Table 2 Prediction Effect of Remaining Time

[0068]

[0069] The prediction error represents the average deviation between the predicted value and the actual occurrence time, and the accurate warning rate represents the proportion of predicted events that actually occur. As can be seen from Table 2, the model has high accuracy in short-term prediction (within 30 minutes). As the prediction time range extends, the accuracy decreases but remains within an acceptable range. The prediction accuracy of the collection saturation state is the highest because the collection process is relatively stable, while pipeline and filter blockages are affected by more random factors.

[0070] By predicting the future system state through the LSTM model, it is possible to predict in advance the time of pipeline blockage, filter blockage, and saturation of the collection device. By calculating the remaining time for each functional component to reach the risk threshold, a time window for predictive maintenance is provided, enabling maintenance personnel to intervene at the best time, solving the problem of being unable to predict blockage and saturation states in a timely manner, providing a scientific basis for maintenance decisions, and significantly improving the deflocculation efficiency, equipment operation stability, and economic efficiency of maintenance management.

[0071] The fluff-removing function status detection system of this embodiment constructs a complete fluff monitoring, evaluation, prediction, and adaptive adjustment system through the organic combination of a sensor detection unit, a status evaluation unit, a status prediction unit, and an adaptive adjustment unit. This system can achieve real-time, continuous, and quantitative monitoring of the material accumulation status in the entire fluff processing path, timely judge the blockage degree of the channel and filter screen, as well as the saturation status of the collection device; predict the future system status through the LSTM model, and anticipate potential risks in advance; and automatically adjust the monitoring parameters according to the fabric type and process parameters to adapt to the characteristics of fluff under different working conditions. It enables blockages to be detected and handled before seriously affecting the system efficiency, avoids the problems of overflow or premature emptying of the collection device, and significantly improves the fluff-removing efficiency, equipment operation stability, and the economy of maintenance management. At the same time, the predictive maintenance ability of the system reduces the unplanned downtime, extends the service life of the equipment, reduces the maintenance cost, and provides a strong guarantee for the efficient and stable operation of the fluffing machine.

[0072] Embodiment 2:

[0073] This embodiment provides a method for detecting the fluff-removing function status of a fluffing machine, which is used to implement the system in Embodiment 1. Refer to Figure 5 , and the method steps include:

[0074] Step 1: Detect the first fluff feature vector through an infrared light sensor, detect the second fluff feature vector through a pressure sensor, and detect the third fluff feature vector through a level sensor;

[0075] The first fluff feature vector includes the normalized infrared scattering intensity and the fluff concentration change rate at each monitoring position; the second fluff feature vector includes the real-time pressure difference and the pressure difference change rate; the third fluff feature vector includes the fluff accumulation height and the height accumulation rate;

[0076] Collect the original infrared optical signals at each detection position of the fluff-removing pipeline of the fluffing machine, amplify and filter them, and then convert them into the normalized infrared scattering intensity. Calculate the fluff concentration change rate according to the normalized infrared scattering intensity to obtain the first fluff feature vector;

[0077] Collect the original inlet pressure signal and the original outlet pressure signal at the fluff-removing filter inlet of the fluffing machine, amplify and filter them, and then convert them into the actual inlet pressure and the actual outlet pressure according to the calibration conversion relationship. Calculate the real-time pressure difference and the pressure difference change rate according to the actual inlet pressure and the actual outlet pressure to obtain the second fluff feature vector;

[0078] Collect the original level sensor signal of the floc collection container, amplify and filter it, and then convert it into the real-time distance from the sensor to the cumulative surface of the flocs according to the calibration conversion relationship. Calculate the cumulative height and height accumulation rate of the flocs based on the real-time distance to obtain the third floc feature vector.

[0079] Step 2: Analyze the fabric type and raising process parameters, and dynamically adjust the feature baseline set and the evaluation threshold set;

[0080] Encode the fabric type being processed into a feature vector, normalize the raising process parameters, and generate a working condition feature vector; Train a multi-layer perceptron mapping model based on historical data, receive the working condition feature vector, and output the corresponding feature baseline set and evaluation threshold set for the current working condition through the forward propagation algorithm; The feature baseline set includes the first floc feature baseline, the second floc feature baseline, and the third floc feature baseline; The first floc feature baseline includes the infrared scattering intensity baseline and the concentration change rate baseline; The second floc feature baseline includes the pressure difference baseline and the pressure difference change rate baseline; The third floc feature baseline includes the floc cumulative height baseline and the height accumulation rate baseline; The evaluation threshold set includes the first evaluation threshold, the second evaluation threshold, and the third evaluation threshold; The first evaluation threshold includes the infrared scattering intensity threshold and the concentration change rate threshold; The second evaluation threshold includes the pressure difference threshold and the pressure difference change rate threshold; The third evaluation threshold includes the floc cumulative height threshold and the height accumulation rate threshold.

[0081] Step 3: Obtain the feature baseline set and the evaluation threshold set, perform the first risk assessment on the first floc feature vector to obtain the first deflocculation function state, perform the second risk assessment on the second floc feature vector to obtain the second deflocculation function state, and perform the third risk assessment on the third floc feature vector to obtain the third deflocculation function state;

[0082] For the first floc feature vector, calculate the intensity risk score of each detection position of the deflocculation pipeline of the raising machine according to the infrared scattering intensity baseline and the infrared scattering intensity threshold; Calculate the trend risk score of each detection position of the deflocculation pipeline of the raising machine according to the concentration change rate baseline and the concentration change rate threshold; Take the maximum value of the intensity risk score and the trend risk score as the single-point risk score of the detection position; Calculate the maximum value of the single-point risk scores of all detection positions at the same time as the first deflocculation function state of the deflocculation pipeline of the raising machine;

[0083] For the second floc feature vector, calculate the degree of filter blockage according to the pressure difference baseline and the pressure difference threshold; Calculate the filter blockage adjustment factor according to the pressure difference change rate baseline and the pressure difference change rate threshold; Multiply the degree of filter blockage and the filter blockage adjustment factor to obtain the second deflocculation function state;

[0084] For the third floc feature vector, calculate the saturation degree according to the floc cumulative height baseline and the floc cumulative height threshold; calculate the cumulative rate risk according to the height cumulative rate baseline and the height cumulative rate threshold; determine the cumulative rate risk coefficient according to the cumulative rate risk, and multiply the saturation degree by the cumulative rate risk coefficient to obtain the third deflocculation function state.

[0085] Step 4: Predict the remaining risk time according to the floc feature vector, the deflocculation function state, the fabric type, and the raising process parameters;

[0086] Construct a multi-source input sequence according to the floc feature vector, the deflocculation function state, the corresponding textile type, and the raising process parameters in the historical time series; use the pre-trained LSTM model to receive the multi-source input sequence and output the floc feature vector prediction sequence; obtain the deflocculation function state prediction sequence through the state evaluation unit according to the floc feature vector prediction sequence; check the time steps when each component in the deflocculation function state prediction sequence first exceeds the corresponding risk threshold, and respectively obtain the remaining time of pipeline blockage, the remaining time of filter blockage, and the remaining time of collection saturation, which constitute the remaining risk time.

[0087] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand 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. A fluff removal function status detection system for a brushing machine, characterized in that, Including: A sensor detection unit that detects a first floc feature vector through an infrared light sensor, a second floc feature vector through a pressure sensor, and a third floc feature vector through a level sensor; the first floc feature vector includes the normalized infrared scattering intensity and the floc concentration change rate at each monitoring position; the second floc feature vector includes the real-time pressure difference and the pressure difference change rate; the third floc feature vector includes the floc cumulative height and the height cumulative rate; An adaptive adjustment unit that analyzes the fabric type and the raising process parameters, and dynamically adjusts the feature baseline set and the evaluation threshold set; A state evaluation unit that obtains the feature baseline set and the evaluation threshold set, performs a first risk assessment on the first floc feature vector to obtain a first deflocculation function state, performs a second risk assessment on the second floc feature vector to obtain a second deflocculation function state, and performs a third risk assessment on the third floc feature vector to obtain a third deflocculation function state; A state prediction unit that predicts the remaining risk time according to the floc feature vector, the deflocculation function state, the fabric type, and the raising process parameters.

2. The deflocculation function state detection system for a raising machine according to claim 1, wherein: Collect the original infrared optical signals at each detection position of the deflocculation pipeline of the raising machine, amplify and filter them, and then convert them into normalized infrared scattering intensities. Calculate the floc concentration change rate according to the normalized infrared scattering intensity to obtain the first floc feature vector; Collect the original inlet pressure signal and the original outlet pressure signal at the deflocculation filter of the raising machine, amplify and filter them, and then convert them into the actual inlet pressure and the actual outlet pressure according to the calibration conversion relationship. Calculate the real-time pressure difference and the pressure difference change rate according to the actual inlet pressure and the actual outlet pressure to obtain the second floc feature vector; Collect the original level sensor signal of the floc collection container, amplify and filter it, and then convert it into the real-time distance from the sensor to the floc cumulative surface according to the calibration conversion relationship. Calculate the floc cumulative height and the height cumulative rate according to the real-time distance to obtain the third floc feature vector.

3. A fluff-removing function state detection system for a hair roughening machine according to claim 1, characterized in that, The adaptive adjustment unit includes: encoding the fabric type being processed into a feature vector, normalizing the raising process parameters, and generating a working condition feature vector; training a multi-layer perceptron mapping model based on historical data, receiving the working condition feature vector, and outputting a set of feature baselines and a set of evaluation thresholds corresponding to the current working condition through the forward propagation algorithm; the set of feature baselines includes a first floc feature baseline, a second floc feature baseline, and a third floc feature baseline; the first floc feature baseline includes an infrared scattering intensity baseline and a concentration change rate baseline; the second floc feature baseline includes a pressure difference baseline and a pressure difference change rate baseline; the third floc feature baseline includes a floc cumulative height baseline and a height accumulation rate baseline; the set of evaluation thresholds includes a first evaluation threshold, a second evaluation threshold, and a third evaluation threshold; the first evaluation threshold includes an infrared scattering intensity threshold and a concentration change rate threshold; the second evaluation threshold includes a pressure difference threshold and a pressure difference change rate threshold; the third evaluation threshold includes a floc cumulative height threshold and a height accumulation rate threshold.

4. A fluff-removing function state detection system for a raising machine according to claim 1, characterized in that, The state evaluation unit includes: For the first floc feature vector, calculate the intensity risk score of each detection position of the floc removal pipeline of the raising machine according to the infrared scattering intensity baseline and the infrared scattering intensity threshold; calculate the trend risk score of each detection position of the floc removal pipeline of the raising machine according to the concentration change rate baseline and the concentration change rate threshold; take the maximum value of the intensity risk score and the trend risk score as the single-point risk score of the detection position; calculate the maximum value of the single-point risk scores of all detection positions at the same time as the first floc removal function state of the floc removal pipeline of the raising machine; For the second floc feature vector, calculate the degree of filter blockage according to the pressure difference baseline and the pressure difference threshold; calculate the filter blockage adjustment factor according to the pressure difference change rate baseline and the pressure difference change rate threshold; multiply the degree of filter blockage and the filter blockage adjustment factor to obtain the second floc removal function state; For the third floc feature vector, calculate the saturation according to the floc cumulative height baseline and the floc cumulative height threshold; calculate the cumulative rate risk according to the height accumulation rate baseline and the height accumulation rate threshold; determine the cumulative rate risk coefficient according to the cumulative rate risk, and multiply the saturation and the cumulative rate risk coefficient to obtain the third floc removal function state.

5. A fluff-removing function state detection system for a hair-raising machine according to claim 1, characterized in that, The state prediction unit includes: constructing a multi-source input sequence according to the floc feature vector, the floc removal function state, and the corresponding textile type and raising process parameters in the historical time series; using a pre-trained LSTM model to receive the multi-source input sequence and output a floc feature vector prediction sequence; obtaining a floc removal function state prediction sequence through the state evaluation unit according to the floc feature vector prediction sequence; checking the time steps when each component in the floc removal function state prediction sequence first exceeds the corresponding risk threshold, and respectively obtaining the remaining time of pipeline blockage, the remaining time of filter blockage, and the remaining time of collection saturation, which constitute the remaining risk time.

6. A method for detecting the function state of defluffing of a flocking machine, characterized in that, Includes: Detect the first floc feature vector through an infrared light sensor, detect the second floc feature vector through a pressure sensor, and detect the third floc feature vector through a level sensor; the first floc feature vector includes the normalized infrared scattering intensity and the change rate of floc concentration at each monitoring position; the second floc feature vector includes the real-time pressure difference and the change rate of pressure difference; the third floc feature vector includes the cumulative height of flocs and the cumulative rate of height. Analyze the fabric type and napping process parameters, dynamically adjust the feature baseline set and the evaluation threshold set. Obtain the feature baseline set and the evaluation threshold set, conduct a first risk assessment on the first floc feature vector to obtain the first deflocculation function state, conduct a second risk assessment on the second floc feature vector to obtain the second deflocculation function state, and conduct a third risk assessment on the third floc feature vector to obtain the third deflocculation function state. Predict the remaining risk time according to the floc feature vector, deflocculation function state, fabric type, and napping process parameters.

Images