Methods for monitoring the production environment of nonwoven products

Abstract

This invention discloses a method for monitoring the production environment of nonwoven products, specifically relating to the field of production environment monitoring technology. The method involves collecting temperature, humidity, electrostatic intensity, and suspended particle concentration data from multiple areas of the production workshop to construct an environmental state sequence matrix; performing anomaly detection based on an environmental anomaly threshold model to extract a set of regional feature points; calculating dynamic risk factors by combining historical anomaly frequency and spatial proximity to generate a set of high-priority monitoring areas; constructing a regional environmental evolution model to predict parameter trends and identify potential secondary out-of-control areas; merging high-risk and secondary out-of-control areas to form a set of microenvironmental anomaly zones and implementing targeted environmental control; collecting a new round of data after control and performing differential analysis to evaluate the control effect. This method enables continuous identification and dynamic intervention of the workshop microenvironment, improving the production quality and environmental stability of nonwoven products.

Description

Technical Field

[0001] This invention relates to the field of production environment monitoring technology, specifically to a method for monitoring the production environment of nonwoven products. Background Technology

[0002] Nonwoven fabrics are widely used in medical, hygiene, protective, and filtration fields. Their production processes often involve meltblowing, spunbonding, and hot rolling, which have extremely high requirements for sensitive parameters such as temperature, humidity, static electricity concentration, and particulate contaminant content in the production environment. However, in actual production, due to the high-temperature operation of production equipment and intense airflow, non-uniform microclimate zones can easily form inside the workshop. Some areas may be in abnormal states for a long time (such as high static electricity, high humidity, or particle aggregation). These abnormal changes are often difficult for conventional sensor networks to detect or respond to in a timely manner, resulting in product defects such as reduced fiber adhesion, abnormal porosity, decreased breaking strength, and increased fiber scattering rate. Especially in medical protective product production lines, products may fail to meet standards due to uncontrolled microenvironment.

[0003] In existing technologies, environmental monitoring is often based on the deployment of sensor nodes at fixed locations. It lacks the ability to identify abnormal and correlated changes at multiple points, and cannot dynamically mark and adjust potential risk areas, which can easily create "blind spots" and ultimately lead to batch scrapping of products, creating irreversible quality hazards. Summary of the Invention

[0004] The purpose of this invention is to provide a method for monitoring the production environment of nonwoven products, so as to overcome the shortcomings of the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring the production environment of nonwoven fabric products, comprising:

[0006] S100. Obtain environmental parameters from multiple areas within the nonwoven fabric production workshop, including temperature, humidity, electrostatic intensity, and suspended particle concentration, and construct an environmental state sequence matrix E based on the collection time of each area's parameters.

[0007] S200. Based on the preset environmental anomaly threshold model, perform point-by-point anomaly detection on the environmental state sequence matrix E, and extract the set of regional feature points P as a candidate set of potential micro-environment out-of-control points.

[0008] S300. Perform weighted calculations on the historical anomaly frequency and spatial proximity of each feature point in the set of regional feature points P, construct the dynamic risk factor R of the feature points, and sort the set P according to the size of the risk factor to obtain the set of high-priority monitoring areas Q.

[0009] S400: Deploy multiple sensor nodes in each area of ​​the high-priority monitoring area set Q, and construct an area environment evolution model M based on the correlation of parameter changes between nodes.

[0010] S500. Based on the prediction results of the regional environmental evolution model M, identify potential secondary out-of-control areas S′ and add them to the next round of dynamic monitoring targets.

[0011] S600: Merge the high-risk areas and potential secondary loss-of-control areas S′ identified in the current round of monitoring to form a set of micro-environmental anomaly zones;

[0012] S700. Based on the risk type of each region in the set of microenvironmental anomaly zones, conduct targeted environmental regulation of the target region;

[0013] S800: Collect the environmental status of each area after regulation and regenerate matrix E′. Perform differential analysis with the environmental status sequence matrix E of the previous round. If the difference rate is lower than the set value, maintain the current strategy; otherwise, enter S200 and start a new round of monitoring cycle.

[0014] Preferably, the dynamic risk factor R for constructing feature points includes:

[0015] Statistical analysis is performed on the historical abnormal records of each feature point in the set of regional feature points P within a preset time window, and the frequency of abnormal occurrence of the feature point F is calculated.

[0016] Based on the workshop spatial layout map, the Euclidean distance between a feature point and a feature point in its adjacent area is obtained, and the spatial proximity factor D is calculated in combination with the risk status of the adjacent points.

[0017] The dynamic risk factor R of the feature points is obtained by using a weighting function;

[0018] Feature points whose risk factor R is higher than the set risk threshold are marked as high-priority monitoring targets.

[0019] Preferably, based on the correlation of parameter changes between nodes, a regional environmental evolution model M is constructed, including:

[0020] Multiple environmental parameter acquisition nodes are deployed within the high-priority monitoring area set Q to acquire multi-dimensional time-series data, including temperature, humidity, electrostatic intensity, and suspended particulate concentration.

[0021] The parameter change trends of each node in the same region are synchronized over time, and the Pearson correlation coefficient of each parameter sequence between any two nodes is calculated to form a node correlation matrix.

[0022] Based on the node correlation matrix, a regional environmental parameter prediction model is constructed using a weighted linear regression algorithm, which is defined as the regional environmental evolution model M.

[0023] Preferably, identifying the potential secondary runaway region S′ includes:

[0024] Using the regional environmental evolution model M, the environmental parameters of each monitoring point in a high-priority monitoring area are continuously predicted over multiple future sampling periods to generate a prediction sequence.

[0025] For each parameter dimension in the predicted sequence, based on the threshold range set by the environmental anomaly threshold model, it is determined whether there is a trend of continuously approaching the critical value or about to cross the boundary.

[0026] For regions exhibiting multi-parameter trend anomalies, calculate their prediction deviation index, defined as the normalized difference between the predicted value and the corresponding threshold;

[0027] When a region experiences a prediction deviation exceeding a set warning threshold across multiple consecutive sampling periods in multiple parameter dimensions, the region is marked as a potential secondary out-of-control region S′.

[0028] Preferably, the formation of the set of microenvironmental anomaly zones includes:

[0029] Extract all regions in the high-priority monitoring region set Q obtained from the risk factor calculation in the current round that have risk factors greater than the risk threshold, and form a high-risk region set.

[0030] Extract the set of potential sub-runaway regions S′ predicted by the regional environmental evolution model M;

[0031] Perform a union operation on the set of high-risk areas and the set of potential sub-out-of-control areas S′ to form the initial set of microenvironmental anomaly zones;

[0032] Regions in the initial set of microenvironmental anomaly regions that have spatial overlap or are less than a set proximity threshold are fused and identified to generate the final set of microenvironmental anomaly regions.

[0033] Preferably, the targeted environmental control of the target area includes: for each target area in the set of microenvironmental anomaly areas, identifying the anomaly type based on the composition characteristics of its corresponding risk factors, including temperature anomaly, humidity anomaly, electrostatic intensity anomaly, and particle concentration anomaly; matching a preset environmental control strategy according to the identified anomaly type and selecting the corresponding control equipment; and adjusting the equipment's activation power, duration, and direction of action according to the spatial range and anomaly intensity of the target area.

[0034] Preferably, the step of collecting and regulating the environmental status of each region and regenerating matrix E′, and performing difference analysis with the previous round of environmental status sequence matrix E, includes:

[0035] After completing the targeted regulation of the target area, the temperature, humidity, electrostatic intensity and suspended particulate concentration data of each area were collected again to construct the environmental state sequence matrix E′.

[0036] The environmental state sequence matrix E before regulation and the matrix E′ after regulation are compared parameter by parameter in the corresponding time frame and region number dimension. The absolute difference of each parameter is calculated and normalized.

[0037] The weighted average of the parameter differences for each region is used to obtain the regional environmental change difference rate Δ.

[0038] The Δ value of each region is compared with the pre-standard value. If the difference rate of all regions is lower than the standard threshold, the regulation is deemed effective and the current regulation strategy remains unchanged.

[0039] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0040] 1. This invention, by constructing a multi-dimensional environmental parameter acquisition mechanism and combining risk factor modeling, trend prediction, and regional fusion analysis, achieves for the first time a full-process, dynamic, and closed-loop monitoring and control of the microenvironment in non-woven fabric production workshops. Unlike traditional methods relying on single-point monitoring or static threshold alarms, this invention can identify potential risk areas that are "not yet out of control but already exhibit abnormal trends," and intervene in advance accordingly, effectively improving the stability of the production process and the consistency of product quality.

[0041] 2. This invention introduces environmental state sequence matrix difference analysis technology to quantitatively evaluate the control effect, enabling adaptive updates and fine adjustments of the control strategy. By setting normalized deviation rates and risk weight models for multi-dimensional parameters, the system can execute targeted responses for different risk types, thereby reducing energy consumption while maximizing environmental safety during the nonwoven fiber forming, cooling, and lamination processes. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0043] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] For examples, please refer to Figure 1 As shown in this embodiment, the method for monitoring the production environment of nonwoven fabric products includes:

[0046] S100. Obtain environmental parameters from multiple areas within the nonwoven fabric production workshop, including temperature, humidity, electrostatic intensity, and suspended particle concentration, and construct an environmental state sequence matrix E based on the collection time of each area's parameters.

[0047] In this embodiment, a nonwoven fabric production workshop was selected as the monitoring object. The workshop is divided into 12 functional areas, including a raw material storage area, a fiber meltblown area, a hot rolling composite area, a cooling and shaping area, a primary testing area, and a packaging area. Each area undertakes different process steps in the nonwoven fabric production process, and their sensitivity and variation characteristics to environmental parameters differ.

[0048] In step S100, to achieve comprehensive perception of the environmental status of each area, four types of miniature environmental sensor nodes are deployed in each area, specifically including:

[0049] Temperature sensors (such as the DS18B20) are used to measure the local air temperature in the workshop in real time.

[0050] Humidity sensors (such as the DHT22) are used to obtain relative humidity;

[0051] Electrostatic potential sensors (such as electrostatic field sensing modules) are used to monitor the level of static electricity accumulation in an area;

[0052] Suspended particulate concentration sensors (such as the PMS5003 laser particle sensor) are used to detect the content of micro-particulate matter such as PM2.5 and PM10 in the air.

[0053] The sampling period was set to once every 30 seconds, and all sampled data were categorized and summarized according to region number. Each round of sampling constituted one frame of data, and the four types of parameters formed a four-dimensional vector, which was stored in the time series database. The environmental state sequence matrix E was constructed as follows:

[0054] The row dimension represents the sampling time points t1, t2, ..., tn;

[0055] The column dimensions represent the four types of environmental parameters corresponding to each region R1, R2, ..., Rm;

[0056] Each cell Eij represents a set of multidimensional parameter values ​​(Tij, Hij, Sij, Pij) in region Rj at time ti, corresponding to temperature, humidity, electrostatic intensity, and particle concentration, respectively.

[0057] S200. Based on the preset environmental anomaly threshold model, perform point-by-point anomaly detection on the environmental state sequence matrix E, and extract the set of regional feature points P as a candidate set of potential micro-environment out-of-control points.

[0058] To identify abnormal environmental conditions within the production workshop, an environmental anomaly threshold model is first constructed based on historical data and industry standards. This model defines the permissible range for each type of environmental parameter. The model is constructed as follows:

[0059] Temperature threshold range Tmin~Tmax: Based on the thermal stability requirements in the meltblown nonwoven fabric process, the reasonable temperature range is set to 18℃ to 28℃. If the temperature value in a certain area exceeds this range, it is considered abnormal.

[0060] Humidity threshold range Hmin~Hmax: Considering the effects of fiber cooling rate and static electricity, the reasonable range of relative humidity is set to 40% to 55%.

[0061] Electrostatic intensity threshold Smax: To prevent uncontrolled fiber adsorption and fluff accumulation, the upper limit of electrostatic potential is set at ±2 kV.

[0062] Suspended particulate concentration threshold Pmax: Based on the cleanliness standards of the workshop, the upper limit of PM2.5 concentration is set at 50 micrograms per cubic meter.

[0063] The aforementioned thresholds, serving as boundary conditions for determining whether the environment is abnormal, are collectively referred to as the environmental anomaly threshold model V, which can be expressed as a set of four value intervals:

[0064] Model V = {[18℃, 28℃], [40%, 55%], [–2kV, +2kV], [0, 50μg / m³]}; Dimensional comparison and detection are performed on each data unit (i.e., the parameter set of a certain region at a certain time) in the environmental state sequence matrix E. The detection process is as follows:

[0065] Iterate through each time step ti; for each workshop area Rj, the parameter set (Tij, Hij, Sij, Pij) is as follows: if the temperature Tij is not within the range of 18℃ to 28℃, record a temperature anomaly; if the humidity Hij is not within the range of 40% to 55%, record a humidity anomaly; if the absolute value of the electrostatic intensity Sij is greater than 2 kV, record an electrostatic anomaly; if the particle concentration Pij exceeds 50 micrograms per cubic meter, record a particle anomaly. Each anomaly type is assigned a Boolean flag value. If at least two or more types of parameter anomalies exist in a certain area at the same time, the area is determined to be in a potentially out-of-control state.

[0066] Extract the indices of all regions Rj that meet the above anomaly criteria within a given sampling time ti, constructing an anomalous region set Pi for that time. By traversing all time points, accumulate to form the total set of regional feature points P. The final set of regional feature points P is defined as: set P = {(ti, Rj) | Rj at time ti has at least two types of parameters that exceed the range of model V}; this set P serves as a candidate set of potential microenvironment out-of-control points, which will be passed to subsequent steps for calculating risk factors and prioritizing and controlling responses.

[0067] For example, in a certain sampling, region R7 was detected to have the following parameter set at time t35: temperature: 30℃ (exceeding the upper limit), humidity: 60% (exceeding the upper limit), electrostatic intensity: 1.5 kV, and particle concentration: 45 μg / m³. Since both temperature and humidity are abnormal, satisfying the condition of "at least two types of parameters being abnormal", the point (t35, R7) is added to the region feature point set P.

[0068] S300. Perform weighted calculations on the historical anomaly frequency and spatial proximity of each feature point in the set of regional feature points P to construct the dynamic risk factor R of the feature points. Sort the set P according to the size of the risk factor to obtain the set of high-priority monitoring areas Q.

[0069] For each feature point (ti, Rj) in the set of regional feature points P, within a fixed time window (e.g., the past 72 hours), the total number of abnormal records in the environmental state sequence matrix where "at least two types of environmental parameters simultaneously exceed the threshold" in that region Rj is retrieved, denoted as the frequency Fj.

[0070] Assuming a sampling interval of 30 seconds, a total of 8640 frames of data will be collected within 72 hours. If n of these frames Rj meet the abnormal conditions, then Fj = n / 8640, which is used to measure the abnormal density of region Rj within a specified time window.

[0071] Based on the physical spatial layout of the nonwoven fabric production workshop, the geometric coordinate information between all feature points is extracted. For any feature point Rj, the Euclidean distance Ljk between it and other surrounding feature points Rk is calculated, and adjacent feature points with a distance less than a set threshold (e.g., 3 meters) are selected.

[0072] For these adjacent feature points Rk, their calculated historical anomaly frequencies Fk are read to construct a set of proximity risk contribution factors. Let the total number of adjacent points be m, then the proximity factor Dj is calculated as follows: k = 1 to m. This calculation method reflects that the closer the neighborhood and the more frequent the anomalies, the greater the influence of the neighborhood on Rj.

[0073] Based on the historical anomaly frequency Fj and spatial proximity factor Dj obtained in the first two steps, the dynamic risk factor Rj is calculated using a weighted model, and the calculation formula is as follows: Here, α and β are empirical weighting coefficients, and their values ​​need to be optimized through field experiments. In this embodiment, α is set to 0.6 and β is set to 0.4, reflecting a higher weighting of the region's own historical anomaly frequency. The larger the Rj value, the more likely the region is to develop into a stable out-of-control zone, and the higher the risk.

[0074] Based on a preset risk factor threshold Rthreshold, for example, set to 0.08, when the dynamic risk factor Rj of a certain feature point exceeds this threshold, it is marked as a high-priority monitoring target. All feature points that meet the condition form a high-priority monitoring candidate set.

[0075] Finally, the set P is sorted from high to low according to the value of risk factor Rj, and the top few feature points (such as the top 10%) are selected to form the final high-priority monitoring area set Q. This set serves as the main basis for subsequent monitoring encryption and response control.

[0076] S400: Deploy multiple sensor nodes in each area of ​​the high-priority monitoring area set Q, and construct an area environment evolution model M based on the correlation of parameter changes between nodes.

[0077] Within each high-priority monitoring area Rj, select at least three physically dispersed monitoring points and install environmental parameter acquisition nodes at each location to obtain multi-dimensional time-series data, including temperature, humidity, electrostatic intensity, and suspended particulate concentration. Each type of node collects a complete set of parameters at fixed time intervals (e.g., every 30 seconds) and stores the data as a timestamp sequence.

[0078] The collected data is represented in the form of a quintuple: (Tt,k,Ht,k,St,k,Pt,k,t); where T, H, S, and P represent the temperature, humidity, electrostatic intensity, and particle concentration recorded at the k-th node at time t, respectively.

[0079] To analyze the consistency of environmental response among nodes within the same region, for each type of parameter, the data of different nodes are aligned in the time dimension to ensure that the calculations are compared based on the same sampling time.

[0080] For any two nodes A and B, after forming two time series on the same parameter dimension (such as humidity), the Pearson correlation coefficient is used to calculate their linear correlation. The above correlation calculation is performed on four parameter dimensions, and finally a node correlation matrix is ​​formed. Its elements represent the degree of trend consistency between nodes on each parameter, and the value ranges from -1 to 1.

[0081] After obtaining the node correlation matrix, node pairs with correlation values ​​significantly higher than 0 are selected, indicating that their parameter changes show a similar trend. Using these highly correlated nodes as feature inputs, a prediction model based on weighted linear regression is constructed.

[0082] In weighted linear regression modeling, a certain parameter of the target prediction node N (such as particle concentration P) is selected as the predicted variable Y, and the current and historical parameter values ​​of neighboring highly correlated nodes are used as explanatory variables X. The weight W of each explanatory variable is set as the corresponding absolute value of Pearson correlation.

[0083] The model expression is: Y(t+1) = W1×X1 + W2×X2 + … + Wn×Xn + ε; where Y(t+1) is the predicted value at the next time step, Xn is the parameter value of the nth neighboring node at the current time step, Wn is its corresponding weight, and ε is the residual term.

[0084] By establishing four types of prediction sub-models in this manner, a set of regional environmental evolution models M that can be used for short-term prediction is finally formed. This model has the ability to predict environmental trends within the target area over several future sampling periods, providing a quantitative basis for subsequent identification of potential abnormal trends.

[0085] S500. Based on the prediction results of the regional environmental evolution model M, identify potential secondary out-of-control areas S′ and add them to the next round of dynamic monitoring targets.

[0086] Using the aforementioned regional environmental evolution model M, continuous predictions are made for four types of environmental parameters—temperature, humidity, electrostatic intensity, and suspended particulate concentration—at each monitoring node in each high-priority monitoring area Rj. The prediction period covers several future sampling cycles, such as the next 30 minutes, and includes a total of 60 frames of prediction data.

[0087] The prediction results constitute four independent time series, each representing the changing trend of the parameters within a future time window, in the following form: Where ^ represents the predicted value, and t is the current time point.

[0088] For each of the above-mentioned predicted parameter sequences, a comparative analysis is performed according to the pre-built environmental anomaly threshold model V to determine whether the parameter values ​​show a trend of approaching the upper or lower limit of the threshold.

[0089] Taking temperature as an example, the allowable range is set to 18℃ to 28℃. When several consecutive predicted values ​​show a monotonically increasing trend, and at least one frame in k consecutive frames is less than 1℃ away from the upper limit, the trend is marked as an approaching anomaly. Other parameters are judged using the same rules.

[0090] For regions Rj exhibiting trend anomalies, the normalized distance between each predicted value and its corresponding threshold is further calculated, defined as the prediction deviation Dj. The calculation method is as follows:

[0091] For the predicted value of a certain parameter X If its threshold interval is [xmin, xmax], then the prediction deviation d is: If ,but ;like ,but ;like If d = 0, then d = 0; for each region Rj, the average of the maximum prediction deviations across all outlier parameter dimensions is taken as the overall deviation index Dj for that region.

[0092] Set a warning threshold Dthreshold, for example, 0.1, which means that the deviation of the predicted value from the threshold exceeds 10%. If the prediction deviation Dj of a certain region Rj exceeds the set value for a number of consecutive frames (e.g., no less than 5 consecutive frames) in at least two parameter dimensions, then the region is determined to be a potential secondary out-of-control region S′.

[0093] All identified areas Rj are included in the next round of environmental monitoring targets, and data collection and model update operations are performed at a higher frequency, thereby enabling early intervention and dynamic control of sub-out-of-control areas.

[0094] S600: Merge the high-risk areas and potential secondary out-of-control areas S′ identified in the current round of monitoring to form a set of microenvironmental anomaly areas.

[0095] Based on the high-priority monitoring area set Q obtained in the preceding steps, the dynamic risk factor Rj corresponding to each area Rj is filtered. If Rj is greater than the preset risk judgment threshold Rthreshold, then the area is included in the high-risk area set Qhigh. For example, if Rthreshold is 0.1, and R12 = 0.14, then area R12 belongs to Qhigh.

[0096] All potential sub-runaway regions predicted and identified by the regional environmental evolution model M constitute a set S′, which represents regions that may enter an abnormal state in the near future. The composition of S′ is based on the result of continuous prediction deviations exceeding the warning threshold across multiple parameter dimensions, as detailed in the "Identification of Potential Sub-Runaway Regions" section of the aforementioned embodiment.

[0097] The set of high-risk areas Qhigh and the set of potential sub-out-of-control areas S′ are combined to form the initial set of micro-environmental anomalies EF′ for the current round of monitoring, i.e.: EF′=Qhigh∪S′; this set contains all areas that have actual or trend risks in the current or prediction period.

[0098] Since there may be spatially similar or overlapping regions in Qhigh and S′, the physical distance between each region in F′ is calculated to avoid redundant adjustments. If the distance between the geometric center points of any two regions Rm and Rn is less than the set proximity fusion threshold Lthreshold (e.g., 1.5 meters), then the two regions are considered to have spatial overlap.

[0099] For regions that meet the overlapping conditions, a fusion identification process is performed to merge them into a single logical monitoring unit, with the smallest number representing the unit, and its risk status and control response parameters are recorded uniformly.

[0100] The resulting set of regions after fusion is defined as the final set of microenvironmental anomalies, which is used as the target set in subsequent response control steps to achieve targeted environmental regulation and dynamic updates.

[0101] S700. Based on the risk type of each region in the set of microenvironmental anomaly zones, conduct targeted environmental regulation of the target area.

[0102] For each target region Rj in the set of microenvironmental anomalies, the corresponding risk factor composition feature vector is analyzed. This feature vector consists of risk indicators of four environmental parameters: temperature, humidity, electrostatic intensity, and particle concentration. Each risk indicator is represented by the normalized deviation of its current value from its anomaly threshold range, with a value ranging from 0 to 1. Let the risk feature vector be: Vj=(rT,rH,rS,rP); where: rT represents temperature deviation; rH represents humidity deviation; rS represents electrostatic intensity deviation; and rP represents particle concentration deviation.

[0103] If the deviation of a certain dimension exceeds the set judgment threshold (e.g., 0.1), then the parameter of that dimension is judged as abnormal, and a set of abnormal type labels is formed. For example, if rH=0.23, rS=0.19, and the rest are 0, then the area is identified as "abnormal humidity + abnormal static electricity".

[0104] Based on the identified anomaly types, rules are matched against a pre-defined control strategy library. This library is built upon expert experience and historical control results, with each type corresponding to a specific handling method and equipment type. For example:

[0105] For abnormal temperatures, use a matching cooling fan or air conditioning cold air module;

[0106] For abnormal humidity levels, use a matching dehumidification unit;

[0107] For static electricity abnormalities, use an ion bar or static neutralizer.

[0108] For particulate abnormalities, a directional high-efficiency particulate air filter is used.

[0109] After selecting the equipment type, proceed to the specific control parameter configuration stage.

[0110] Based on the spatial coordinate range of the target area Rj (whose geometric boundaries are determined by the workshop layout diagram) and the degree of deviation from the abnormal parameters, the activation power Pj, duration Tj, and direction θj of the selected equipment are jointly set. The specific control parameter calculation rules are as follows:

[0111] Enable power , where rmax is the maximum risk deviation and Pbase is the base power value;

[0112] The duration Tj is determined by both rmax and the area of ​​the region; the larger the area or the stronger the anomaly, the longer the duration.

[0113] The direction of action θj is calculated based on the angle between the equipment and the center point of the target area, and offset correction is made according to the layout of obstacles in the workshop.

[0114] Through the above methods, precise control of the target area during abnormal response processes can be achieved, ensuring that environmental parameters can return to normal ranges in a short period of time, providing strong support for ensuring the quality stability of nonwoven products.

[0115] S800: Collect the environmental status of each area after regulation and regenerate matrix E′. Perform differential analysis with the environmental status sequence matrix E of the previous round. If the difference rate is lower than the set value, maintain the current strategy; otherwise, enter S200 and start a new round of monitoring cycle.

[0116] After the implementation of targeted control measures, environmental parameters including temperature, humidity, electrostatic intensity and suspended particulate concentration will be collected again in all monitoring areas, maintaining the same sampling cycle and format as before the control measures.

[0117] The newly collected data were organized according to time frames and region numbers to construct the environmental state sequence matrix E′ after regulation. Its structure is consistent with the matrix E before regulation, and it is used for subsequent comparative analysis.

[0118] Matrix E and matrix E′ are compared parameter-by-parameter across the same time frame and the same region numbering dimension. For each environmental parameter X, the difference between its pre-regulation value x and its post-regulation value x′ is calculated: Difference. To ensure comparability between different parameters, each difference δ is normalized according to its normal range in the abnormal threshold model V: Normalized difference Where xmax and xmin are the normal upper and lower limits defined for this parameter in model V.

[0119] The environmental change difference rate Δj for the region is obtained by weighted averaging the normalized differences of four types of parameters within the same region. The calculation formula is as follows: Where d1 to d4 are the normalized differences of temperature, humidity, electrostatic intensity, and particle concentration, respectively, and w1 to w4 are empirical weighting coefficients, satisfying w1 + w2 + w3 + w4 = 1. In this embodiment, the influence of particles and electrostatics on product quality is given priority, and w3 and w4 are set to be relatively large, for example, w1 = 0.2, w2 = 0.2, w3 = 0.3, w4 = 0.3.

[0120] A standard threshold Δthreshold is set for the difference rate, for example, 0.05, indicating that an overall change of less than 5% in regional environmental parameters can be considered as stabilizing. If Δj ≤ Δthreshold for all regions, the current round of control strategy is deemed effective, and the established parameter configuration is maintained; if Δj for any region exceeds this threshold, the process enters S200, triggering a new round of monitoring and control cycle.

[0121] This differential analysis process enables the dynamic evaluation of control effects in a quantitative manner, achieving closed-loop control and strategy optimization of the nonwoven fabric production environment.

[0122] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for monitoring the production environment of nonwoven fabric products, characterized in that: include: S100. Obtain environmental parameters from multiple areas within the nonwoven fabric production workshop, including temperature, humidity, electrostatic intensity, and suspended particle concentration, and construct an environmental state sequence matrix E based on the collection time of each area's parameters. S200. Based on the preset environmental anomaly threshold model, perform point-by-point anomaly detection on the environmental state sequence matrix E, and extract the set of regional feature points P as a candidate set of potential microenvironment out-of-control points. Specifically, this includes: traversing each time ti, for each workshop area parameter group (Tij, Hij, Sij, Pij): if the temperature Tij is not within the range of 18℃ to 28℃, record the temperature anomaly; if the humidity Hij is not within the range of 40% to 55%, record the humidity anomaly; if the absolute value of the electrostatic intensity Sij is greater than 2 kV, record the electrostatic anomaly; if the particle concentration Pij exceeds 50 micrograms per cubic meter, record the particle anomaly. Each anomaly type is assigned a Boolean label value. If at least two or more types of parameter anomalies exist in a certain area at the same time, the area is determined to be in a potential out-of-control state. Extract the numbers of all areas that meet the above anomaly judgment conditions within a certain sampling time ti, and construct the set of abnormal areas Pi for that time. By traversing all times, the total set of regional feature points P is accumulated. S300. For each feature point in the set P of regional feature points, a weighted calculation is performed based on the historical anomaly frequency and spatial proximity to construct a dynamic risk factor R for the feature points. The set P is then sorted according to the magnitude of the risk factor to obtain a high-priority monitoring area set Q, specifically including: Based on the physical spatial layout of the nonwoven fabric production workshop, the geometric coordinate information between all feature points is extracted. For any feature point Rj, the Euclidean distance Ljk between it and its surrounding feature points Rk is calculated, and adjacent feature points Rh with a distance less than a set threshold are selected. For adjacent feature points Rh, their calculated historical anomaly frequencies Fk are read, and a set of proximity risk contribution factors is constructed. Let the total number of adjacent points be m, then the spatial proximity factor Dj is calculated as follows: Based on the obtained historical anomaly frequency Fk and spatial proximity factor Dj, the dynamic risk factor R is calculated by weighted average summation; S400. Deploy multiple sensor nodes in each area of ​​the high-priority monitoring area set Q, and construct an area environmental evolution model M based on the correlation of parameter changes between nodes. The area environmental evolution model M includes: Multiple environmental parameter acquisition nodes are deployed within the high-priority monitoring area set Q to acquire multi-dimensional time-series data, including temperature, humidity, electrostatic intensity, and suspended particulate concentration. The parameter change trends of each node in the same region are synchronized over time, and the Pearson correlation coefficient of each parameter sequence between any two nodes is calculated to form a node correlation matrix. Based on the node correlation matrix, a regional environmental parameter prediction model is constructed using a weighted linear regression algorithm, which is defined as the regional environmental evolution model M. S500. Based on the prediction results of the regional environmental evolution model M, identify potential secondary runaway areas S′ and add them to the next round of dynamic monitoring targets; the identification of potential secondary runaway areas S′ includes: using the regional environmental evolution model M to continuously predict the environmental parameters of each monitoring point in the high-priority monitoring area in multiple future sampling periods, and generating a prediction sequence. For each parameter dimension in the predicted sequence, based on the threshold range set by the environmental anomaly threshold model, it is determined whether there is a trend of continuously approaching the critical value or about to cross the boundary. For regions exhibiting multi-parameter trend anomalies, a prediction deviation index is calculated, defined as the normalized difference between the predicted value and the corresponding threshold. The prediction deviation is calculated as follows: For the predicted value of parameter X If its threshold interval is [xmin, xmax], then the prediction deviation d is: If ,but ;like ,but ;like If d = 0, then d = 0; within each region, the average of the maximum prediction deviations across all outlier parameter dimensions is taken as the prediction deviation for that region. When a region experiences a prediction deviation exceeding a set warning threshold for multiple consecutive sampling periods across multiple parameter dimensions, the region is marked as a potential secondary out-of-control region S′. S600: Merge the high-risk areas and potential secondary loss-of-control areas S′ identified in the current round of monitoring to form a set of micro-environmental anomaly zones, including: S700. Based on the risk type of each region in the set of microenvironmental anomaly zones, conduct targeted environmental regulation of the target region; S800: Collect the environmental status of each area after regulation and regenerate matrix E′. Perform differential analysis with the environmental status sequence matrix E of the previous round. If the difference rate is lower than the set value, maintain the current strategy; otherwise, enter S200 and start a new round of monitoring cycle.

2. The method for monitoring the production environment of nonwoven fabric products according to claim 1, characterized in that: The dynamic risk factor R for constructing feature points includes: Statistical analysis is performed on the historical abnormal records of each feature point in the set of regional feature points P within a preset time window, and the frequency of abnormal occurrence of the feature point F is calculated. Based on the workshop spatial layout map, the Euclidean distance between a feature point and a feature point in its adjacent area is obtained, and the spatial proximity factor Dj is calculated in combination with the risk status of the adjacent points. The dynamic risk factor R of the feature points is obtained by using a weighting function; Feature points whose risk factor R is higher than the set risk threshold are marked as high-priority monitoring targets.

3. The method for monitoring the production environment of nonwoven fabric products according to claim 1, characterized in that: The targeted environmental control of the target area includes: for each target area in the set of microenvironmental anomaly areas, identifying the anomaly type based on the composition characteristics of its corresponding risk factors, including temperature anomaly, humidity anomaly, electrostatic intensity anomaly, and particle concentration anomaly; matching the pre-set environmental control strategy according to the identified anomaly type and selecting the corresponding control equipment; and adjusting the equipment's activation power, duration, and direction of action according to the spatial range and anomaly intensity of the target area.

4. The method for monitoring the production environment of nonwoven fabric products according to claim 1, characterized in that: The set of microenvironmental anomaly zones includes: Extract all regions in the high-priority monitoring region set Q obtained from the risk factor calculation in the current round that have risk factors greater than the risk threshold, and form a high-risk region set. Extract the set of potential sub-runaway regions S′ predicted by the regional environmental evolution model M; Perform a union operation on the set of high-risk areas and the set of potential sub-out-of-control areas S′ to form the initial set of microenvironmental anomaly zones; Regions in the initial set of microenvironmental anomaly regions that have spatial overlap or are less than a set proximity threshold are fused and identified to generate the final set of microenvironmental anomaly regions.

5. The method for monitoring the production environment of nonwoven fabric products according to claim 1, characterized in that: The process involves collecting and regulating the environmental status of each region and regenerating a matrix E′, then performing a difference analysis with the previous environmental status sequence matrix E, including: After completing the targeted regulation of the target area, the temperature, humidity, electrostatic intensity and suspended particulate concentration data of each area were collected again to construct the environmental state sequence matrix E′. The environmental state sequence matrix E before regulation and the matrix E′ after regulation are compared parameter by parameter in the corresponding time frame and region number dimension. The absolute difference of each parameter is calculated and normalized. The weighted average of the parameter differences for each region is used to obtain the regional environmental change difference rate Δ. The Δ value of each region is compared with the pre-standard value. If the difference rate of all regions is lower than the standard threshold, the regulation is deemed effective and the current regulation strategy remains unchanged.

Images