A diaper leakage prevention intelligent detection system and method

Abstract

The present application relates to the technical field of intelligent detection of sanitary products, and discloses a paper diaper leakage prevention intelligent detection system and method. The method comprises the following steps: synchronously collecting liquid infiltration, material deformation, heat conduction and other multi-physical field signals, fusing and space-time calibrating, and constructing a multi-dimensional seepage trend chart reflecting the liquid dynamic diffusion process. Based on the seepage dynamics knowledge graph, the trend chart is deduced to predict the seepage breakthrough probability and potential leakage channel network. The probability is corrected by behavior disturbance in combination with user body posture data, and an intervention scheme of internal absorbent structure reorganization and external barrier layer topology optimization is deduced reversely according to the predicted leakage channel. The method realizes early warning of leakage events and mechanism tracing of leakage sources, thereby promoting the product design to change from passive detection to active warning and precise optimization.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection technology for hygiene products, specifically to an intelligent detection system and method for leak-proof diapers. Background Technology

[0002] Current diaper leak detection technologies typically rely on built-in, single-type humidity sensors distributed at specific locations in the absorbent core or leak-proof side panels. They work by triggering a signal when liquid reaches the sensor's location, thus initiating a leak alarm or status indication. This method is essentially a discrete point-based monitoring and static threshold judgment, acquiring information limited to whether liquid has reached a few preset discrete locations.

[0003] This technical solution has shortcomings. Because it relies solely on a single, delayed liquid presence signal, the system cannot acquire the continuous path, velocity, and pattern of liquid diffusion within complex three-dimensional porous materials. It can only indicate the outcome of a leak, but cannot reveal the dynamic seepage process from wetting to breaching, nor can it capture early signs and potential pathways before a leak occurs. This lack of information prevents existing technology from providing early warning of leakage risks, and further hinders the provision of precise, mechanistic guidance for fundamental improvements to product structures.

[0004] The key challenge currently facing this field is how to achieve continuous, global, and dynamic perception and characterization of liquid seepage behavior inside diapers, and on this basis, to deduce the mechanism of leakage events and make early predictions. This requires moving beyond the existing judgment paradigm based on single-point thresholds and shifting towards a deep characterization and forward-looking intelligent analysis of the seepage physical process. Summary of the Invention

[0005] The purpose of this invention is to provide a smart leak detection system and method for diapers to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a smart leak detection method based on diapers, the method comprising:

[0007] Multiple distributed sensing elements are used to synchronously capture physical field change signals in the internal and edge areas of the target diaper. These physical field change signals include liquid wetting trajectory signals, material fiber structure deformation signals, and heat conduction change signals.

[0008] The captured physical field change signals are synchronously fused and spatiotemporally calibrated to construct a multidimensional seepage situation map that reflects the dynamic existence state of the liquid. The multidimensional seepage situation map embeds the liquid flow front position sequence, the internal saturation region outline and the micro-leakage situation of the external contact edge.

[0009] Based on the preset seepage dynamics knowledge graph, the seepage state evolution of the multi-dimensional seepage state graph is deduced, and the occurrence probability of the seepage breakthrough event and the potential leakage channel network are deduced;

[0010] The occurrence probability of the seepage breakthrough event is corrected by combining the target user body motion posture data, and a behavior-related leakage probability is obtained. The behavior disturbance correction is completed according to the corresponding relationship between liquid inertia migration law and body motion posture;

[0011] According to the deduced potential leakage channel network, a structure intervention scheme for inhibiting the formation of the leakage channel network is reversely deduced, and the structure intervention scheme includes an internal absorber structure reorganization strategy and an external barrier layer topology optimization strategy.

[0012] Preferably, the captured physical field change signal is synchronously fused and space-time calibrated to construct a multi-dimensional seepage state graph reflecting the dynamic existence state of the liquid, which is specifically realized by the following steps:

[0013] A plurality of independent liquid flow branch signals are separated from the liquid infiltration trajectory signal, and each liquid flow branch signal is marked with its starting time and propagation vector;

[0014] The material fiber structure deformation signal is mapped to a pore connectivity change field in three-dimensional space, and the pore connectivity change field is used to describe the real-time opening and closing state of the liquid passable path;

[0015] The heat conduction change signal is analyzed into a heat field gradient change associated with the liquid distribution, and the high-temperature region in the heat field gradient change has spatial correspondence with the liquid accumulation region;

[0016] A unified space-time coordinate system is established, and the propagation vector of the liquid flow branch signal, the pore connectivity change field, and the heat field gradient change are superimposed and associated matched;

[0017] In the space-time coordinate system, according to the result of the associated matching, the liquid flow front position sequence is dynamically drawn, and the internal saturation region profile after the liquid is absorbed is outlined;

[0018] The energy attenuation mode of the liquid infiltration trajectory signal near the boundary of the paper diaper is continuously tracked, and the micro-leakage state of the external contact edge is comprehensively judged in combination with the mutation characteristics of the pore connectivity change field at the boundary.

[0019] Preferably, the material fiber structure deformation signal is mapped to a pore connectivity change field in three-dimensional space, which includes:

[0020] performing multi-scale decomposition on the material fiber structure deformation signal to separate a macroscopic overall bending component and a microscopic fiber network displacement component;

[0021] for the macroscopic overall bending component, calculating a compression rate change in the thickness direction of the absorber caused by the macroscopic overall bending component, and generating a compression-dominated porosity change distribution map according to a compression rate and porosity relationship model;

[0022] for the microscopic fiber network displacement component, analyzing the displacement vector of the intersection point between the fibers, calculating the equivalent hydraulic diameter change of the capillary channel between the fibers according to the displacement vector, and generating a capillary channel dynamic adjustment map;

[0023] fusing the porosity change distribution map and the capillary channel dynamic adjustment map to generate a comprehensive porosity connectivity change field, wherein the value of each spatial point in the porosity connectivity change field represents the real-time conduction ability of each spatial point to liquid flow.

[0024] Preferably, the state evolution deduction of the multi-dimensional seepage state pattern based on the preset seepage dynamics knowledge graph comprises:

[0025] matching a historical seepage mode corresponding to the current liquid flow front position sequence from the knowledge graph, and extracting a critical condition for the historical seepage mode to successfully break through the barrier;

[0026] comparing the geometric features of the internal saturated area contour with the saturated area instability criterion stored in the knowledge graph, and calculating a risk measurement value of the overall lateral overflow of the current saturated area;

[0027] analyzing the evolution trend of the micro-leakage state of the external contact edge, and predicting the possibility of connecting micro-leakage points into a through leakage path by using the edge failure propagation model in the knowledge graph;

[0028] comprehensively considering the critical condition, the risk measurement value and the possibility, and calculating the occurrence probability of the seepage breakthrough event by using the probability fusion rule in the knowledge graph;

[0029] Meanwhile, according to the advancing direction of the liquid flow front position sequence and the dominant conduction path presented in the porosity connectivity change field, all possible paths of liquid flow formed inside the material are traced back to form a network of potential leakage channels.

[0030] Preferably, the behavior disturbance correction of the occurrence probability of the seepage breakthrough event combined with the target user body motion posture data obtains a behavior-related leakage probability, comprising:

[0031] analyzing the target user's body gesture data, and identifying periodic gesture actions and non-periodic strong disturbance actions;

[0032] calculating a periodic shaking model of the liquid inside the diaper caused by the periodic gesture actions, applying a periodic deformation disturbance to the internal saturation region profile according to the periodic shaking model, and updating the fluctuation state of the liquid distribution;

[0033] evaluating an inertial impact force vector generated by the non-periodic strong disturbance actions, converting the inertial impact force vector into an additional driving force for the liquid flow front position sequence, and updating the instantaneous position and velocity of the liquid flow front;

[0034] re-inputting the updated fluctuation state of the liquid distribution and the instantaneous position and velocity of the liquid flow front into the seepage dynamics knowledge graph for rapid local deduction;

[0035] obtaining an instant occurrence probability obtained after rapid local deduction, and taking the instant occurrence probability as the behavior-related leakage probability.

[0036] Preferably, the calculation of the periodic shaking model of the liquid inside the diaper caused by the periodic gesture actions comprises:

[0037] establishing a fluid equivalent model that discretizes the internal space of the diaper into a plurality of interconnected micro-cavities;

[0038] decomposing the periodic gesture actions into rotational components and translational components around different body axes;

[0039] applying each rotational component and translational component as a driving input to the corresponding boundary of the fluid equivalent model, and solving the liquid exchange flow rate between each micro-cavity generated by the driving input;

[0040] According to the liquid exchange flow rate of all micro-cavities, the periodic movement trajectory of the liquid mass center in space and the periodic fluctuation amplitude of the liquid free surface are calculated, which together constitute the periodic shaking model.

[0041] Preferably, the structure intervention scheme for inhibiting the formation of the potential leakage channel network is reversely deduced according to the deduced potential leakage channel network, comprising:

[0042] Performing key node and key path analysis on the potential leakage channel network to identify a few core channel segments that play a decisive role in the connectivity of the entire network;

[0043] For each of the core channel segments, locate its corresponding spatial region in the multi-dimensional seepage state diagram, and extract the historical conduction ability data of the spatial region in the pore connectivity change field;

[0044] designing a targeted local densification scheme or a flow blocking structure to weaken the conductive capacity of the core channel segment, the local densification scheme being achieved by changing the fiber arrangement density of the absorbent body, and the flow blocking structure being achieved by implanting micro-hydrophobic units;

[0045] from the perspective of the overall structure of the material, according to the overall topology of the leakage channel network, the internal absorbent structure reorganization strategy of adjusting the density gradient distribution between different regions of the absorbent body is proposed;

[0046] from the perspective of boundary protection, according to the spatial distribution of the micro-leakage situation, the external barrier layer topology optimization strategy of enhancing the sealing performance and flow guiding capacity of specific edge regions is proposed.

[0047] Preferably, the targeted local densification scheme or the flow blocking structure includes:

[0048] If the historical conductive capacity data shows that the formation of the core channel segment is mainly caused by the connection of high porosity regions, a local densification scheme is developed, specifically: implanting clusters of superfine fibers with higher specific surface area in the high porosity region, the implantation density of the superfine fiber clusters being positively correlated with the peak value of the historical conductive capacity data;

[0049] If the historical conductive capacity data shows that the formation of the core channel segment is mainly dominated by a specific capillary dominant path, a flow blocking structure is designed, specifically: along the trend of the capillary dominant path, micro-hydrophobic units are set at intervals, and the spacing of the micro-hydrophobic units is adaptively adjusted according to the capillary pressure gradient of the path;

[0050] Trigger threshold values are defined for the local densification scheme and the flow blocking structure, and when the behavior-associated leakage probability exceeds the trigger threshold value, the corresponding scheme or structure is implemented.

[0051] Preferably, it further includes a self-evolution process of the seepage flow dynamics knowledge graph:

[0052] In actual application scenarios, the full-cycle physical field change signal sequence and the corresponding user behavior data sequence of the target diaper from the start of use to the occurrence of an observable leakage event are continuously collected;

[0053] When an observable leakage event occurs, key situation characteristics before the event are extracted from the full-cycle physical field change signal sequence, and the key situation characteristics are associated with the final leakage result to form a new situation-result pair;

[0054] The new situation-result is used to correct the historical infiltration mode, saturation area instability criterion and parameter in the edge failure propagation model in the knowledge graph;

[0055] The modified knowledge graph is used as the basis for subsequent situation evolution deduction, and the seepage dynamics knowledge graph is iteratively optimized.

[0056] Preferably, the application also includes a paper diaper leak-proof intelligent detection system based on the paper diaper leak-proof intelligent detection method, the system comprising a memory, a processor and a computer program stored in the memory and running on the processor, and the processor implements the steps of the paper diaper leak-proof intelligent detection method when executing the computer program.

[0057] Compared with the prior art, the application has the following beneficial effects:

[0058] By synchronously capturing multi-physical field signals such as liquid infiltration, material deformation and heat conduction, and fusing and spatiotemporal calibration, a dynamic multi-dimensional seepage situation graph is constructed. This method discards the mode of relying on the state of individual points for judgment and changes to continuous field description of the seepage front moving track, three-dimensional contour of the saturated area and edge contact state. This change enables the system to capture early diffusion behavior and weak edge contact signals of the liquid before it reaches the final leakage point, realizes global and real-time visualization monitoring of the seepage process, and provides rich information dimensions for early identification of leakage.

[0059] A preset seepage dynamics knowledge graph is introduced to deduce the multi-dimensional situation graph, and simple signal perception is upgraded to intelligent prediction driven by physical mechanism. The knowledge graph contains prior knowledge such as seepage law of porous media, material properties and interface effects, can deduce the current seepage situation, calculate the probability of liquid breaking through the leak-proof boundary, and inversely outline the potential main seepage channel network leading to breakthrough. This enables detection to progress from reporting "whether it has occurred" to predicting "when and where it may occur", and clearly points out the key path leading to the risk of leakage, surpassing the judgment logic of traditional statistical or threshold-based models.

[0060] Based on the deduced potential leakage channel network, a structure intervention scheme is reversely deduced. This process establishes a direct closed loop from "detection-prediction" to "design-optimization". The prediction result of the leakage channel network directly points out the weak link in the current product structure that is easy to be broken by liquid, so that specific strategies for absorbent structure reorganization or barrier layer topology optimization can be generated. This makes product improvement no longer dependent on trial and error or experience, but driven by specific seepage failure mechanisms and simulation results, improving the accuracy and efficiency of structure optimization. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1A working principle diagram of the diaper leakage prevention intelligent detection method according to the present application;

[0062] Figure 2 A flowchart for constructing a multi-dimensional seepage situation map;

[0063] Figure 3 A diagram of the relationship between the micro-leakage point distance and the connection probability of the through passage;

[0064] Figure 4 A flowchart for calculating a periodic shaking model;

[0065] Figure 5 A diagram of the relationship between the intervention scheme triggering threshold and the efficiency. DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0067] Please refer to Figure 1 The present application provides a diaper leakage prevention intelligent detection method, which comprises: using a plurality of sensing elements distributed in the diaper to synchronously capture liquid infiltration trajectory signals, material fiber structure deformation signals and heat conduction change signals in the target diaper and the edge area. The captured physical field change signals are transmitted to a processing unit for synchronous fusion and space-time calibration. The fusion process integrates the correlation of different signal sources in space and time, thereby constructing a multi-dimensional seepage situation map reflecting the dynamic existence state of the liquid. The multi-dimensional seepage situation map embeds the liquid flow front position sequence, the internal saturation area contour and the micro-leakage situation information of the external contact edge. Subsequently, the system calls a preset seepage dynamics knowledge graph to perform situation evolution deduction on the multi-dimensional seepage situation map. The knowledge graph stores priori knowledge such as material permeation characteristics and historical leakage patterns, and the deduction process calculates the occurrence probability of the seepage breakthrough event and predicts the potential leakage channel network based on these knowledge. To achieve more accurate prediction, the method combines the target user body motion posture data obtained through the external sensor, and according to the corresponding relationship between the liquid inertia migration law and the body motion posture, the behavior disturbance correction is performed on the aforementioned occurrence probability, so as to obtain the behavior-associated leakage probability. Finally, for the potential leakage channel network deduced, the method reversely analyzes the key path and node formed by the network, and deduces a structure intervention scheme capable of inhibiting the formation of the network. The scheme includes specific suggestions for the internal absorbent structure reorganization strategy and the external barrier layer topology optimization strategy of the diaper.

[0068] Embodiment 1:Figure 2 Synchronous fusion and space-time calibration are performed on the captured physical field change signals to construct a multi-dimensional seepage state diagram reflecting the dynamic existence state of the liquid. The specific implementation includes the following steps. A plurality of independent liquid flow branch signals are separated from the liquid infiltration trajectory signal, and each liquid flow branch signal is labeled with its starting time and propagation vector. The material fiber structure deformation signal is mapped to a pore connectivity change field in three-dimensional space, which is used to describe the real-time opening and closing state of the liquid passable path. The heat conduction change signal is analyzed into a heat field gradient change associated with the liquid distribution, and the high temperature region in the heat field gradient change has spatial correspondence with the liquid accumulation region. A unified space-time coordinate system is established to superimpose and correlate match the propagation vector of the liquid flow branch signal, the pore connectivity change field and the heat field gradient change. In the space-time coordinate system, the liquid flow front position sequence is dynamically drawn according to the correlation matching result, and the internal saturation region profile after the liquid is absorbed is outlined. The energy attenuation mode of the liquid infiltration trajectory signal near the boundary of the diaper is continuously tracked, and the mutation characteristics of the pore connectivity change field at the boundary are combined to comprehensively determine the micro-leakage state of the external contact edge. In the process of mapping the pore connectivity change field, the material fiber structure deformation signal is subjected to multi-scale decomposition to separate the macroscopic overall bending component and the microscopic fiber network displacement component. For the macroscopic overall bending component, the compression rate change in the thickness direction of the absorbent body caused thereby is calculated, and according to the relationship model between the compression rate and the porosity, a porosity change distribution graph dominated by compression is generated. For the microscopic fiber network displacement component, the displacement vector of the intersection point between the fibers is analyzed, the equivalent hydraulic diameter change of the capillary channel between the fibers is calculated according to the displacement vector, and a capillary channel dynamic adjustment graph is generated. Fusion of the porosity change distribution graph and the capillary channel dynamic adjustment graph generates a comprehensive pore connectivity change field, in which the value of each spatial point represents the real-time conduction ability of the spatial point to the liquid flow.

[0069] In practical implementation, the synchronous fusion and spatiotemporal calibration of physical field change signals to construct a multidimensional seepage situation map relies on the processing and integration of three types of sensor signals. The original liquid wetting trajectory signal captured by distributed sensing elements manifests as a time-varying electrical characteristic distribution sequence. In practical implementation, the processing unit separates multiple spatially independent liquid flow branch signals from the liquid wetting trajectory signal and marks the initial detection time and propagation direction vector calculated based on the spatial sequence for each liquid flow branch signal. The material fiber structure deformation signal is acquired by a flexible strain sensing network. In practical implementation, the material fiber structure deformation signal is mapped to a pore connectivity change field in three-dimensional space. This process first involves multi-scale decomposition of the material fiber structure deformation signal, separating the macroscopic overall bending component reflecting the overall bending of the absorber and the microscopic fiber network displacement component reflecting the relative displacement between fibers. For the macroscopic overall bending component, the resulting change in compressibility in the thickness direction of the absorber is calculated, and based on the relationship model between compressibility and porosity, a porosity change distribution map dominated by compression is generated. It can be understood that the relationship between compressibility and porosity can be expressed as:

[0070]

[0071] in: Indicates spatial location and time Porosity at that location Indicates the initial porosity. It is the material compression-porosity coupling coefficient. This represents the change in compressibility in the thickness direction. For the displacement components of the microfiber network, the displacement vectors at fiber intersections are analyzed. Based on these displacement vectors, the equivalent hydraulic diameter change of the capillary channels formed between fibers is calculated, thus generating a dynamic adjustment map of the capillary channels. The porosity change distribution map and the dynamic adjustment map of the capillary channels are fused to generate a comprehensive pore connectivity change field. The value of each spatial point in the pore connectivity change field represents the real-time conductivity of that point for liquid flow; conductivity is a function of porosity and capillary hydraulic diameter. The heat conduction change signal is captured by a distributed temperature sensor array. In practice, the heat conduction change signal is interpreted as a heat field gradient change associated with the liquid distribution. The processing unit calculates the heat field gradient change by comparing the current temperature field with the initial background temperature field. In liquid accumulation areas, due to the high heat capacity of the liquid, the heat field gradient change typically presents as a high-temperature region.

[0072] In some embodiments, establishing a unified space-time coordinate system is the basis for signal fusion, and the space-time coordinate system takes the geometric center of the diaper as the origin. In the space-time coordinate system, the liquid flow branch signal marked with the starting time and the propagation direction vector, the pore connectivity change field, and the heat field gradient change are superimposed and correlated. The correlation is based on the synchronization of the time stamp and the consistency of the spatial coordinates, for example, the liquid flow branch signal propagation vector, the pore connectivity change field conduction ability value, and the heat field gradient change temperature value at the same time and the same spatial coordinates are correlated. In the space-time coordinate system, according to the results of correlation matching, a liquid flow front position sequence is dynamically drawn, which is composed of the frontmost coordinate points of the liquid flow branch signal in time sequence connection, and simultaneously outlines the internal saturation region profile after the liquid is absorbed, which is jointly defined by the high-temperature region boundary in the heat field gradient change and the high-conduction ability pore connectivity change field region. In some embodiments, the energy attenuation mode of the liquid infiltration trajectory signal near the diaper boundary is continuously tracked, which is characterized by an unexpected sharp decrease in signal amplitude in the edge region. Combined with the conduction ability mutation feature of the pore connectivity change field at the boundary, the micro-leakage trend of the external contact edge is comprehensively determined, which describes whether the liquid forms a potential leakage point at the edge.

[0073] In some embodiments, the knowledge graph is used to analyze the evolution of the multi-dimensional seepage state trend. The specific implementation process is as follows. From the knowledge graph, the historical seepage mode corresponding to the current liquid flow front position sequence is matched, and the critical condition for the historical seepage mode to successfully break through the barrier is extracted. The geometric characteristics of the internal saturation region profile are compared with the saturation region instability criterion stored in the knowledge graph, and the risk measurement value of the current saturation region is calculated. The evolution trend of the micro-leakage trend of the external contact edge is analyzed, and the possibility of connecting the micro-leakage points into a through leakage path is predicted using the edge failure propagation model in the knowledge graph. By comprehensively considering the extracted critical condition, the calculated risk measurement value, and the predicted possibility, the occurrence probability of the seepage breakthrough event is calculated through the probability fusion rule in the knowledge graph. At the same time, according to the advancing direction of the liquid flow front position sequence and the dominant conduction path in the pore connectivity change field, all the paths through which the liquid flows in the material are traced back to form a potential leakage channel network.

[0074] In practical implementation, the evolution of the multi-dimensional seepage situation map is deduced based on a pre-set seepage dynamics knowledge graph. This deduction process relies on the structured prior knowledge stored in the knowledge graph. Firstly, historical seepage patterns corresponding to the current flow front position sequence are matched from the seepage dynamics knowledge graph. These historical seepage patterns are stored as graphical sequences, recording similar flow front advancement paths and their final results in the past. The processing unit calculates the spatiotemporal similarity between the current flow front position sequence and each historical seepage pattern sequence, selects the historical seepage pattern with the highest similarity, and extracts the critical conditions for its successful breakthrough of the barrier. These critical conditions include saturation thresholds, pressure gradient thresholds, or time thresholds at specific locations. Simultaneously, the geometric features of the internal saturated region contour are compared with the saturated region instability criteria stored in the seepage dynamics knowledge graph. These criteria include geometric or morphological parameters such as area-to-perimeter ratio, contour compactness, and centroid offset. The risk metric for the overall lateral overflow of the current saturated region is calculated. This risk metric... The calculation can be expressed as:

[0075]

[0076] in: This represents a measure of instantaneous instability risk. This indicates the number of instability criterion features included in the comparison. Indicates the first The weight coefficients of each feature, The first part representing the current internal saturation region contour The actual value of each feature, The first term defined in the knowledge graph of seepage dynamics represents the... Critical thresholds for each feature.

[0077] In some embodiments, the evolution trend of the micro-leakage tendency of the external contact edge is analyzed, the possibility of the micro-leakage points connecting into a through leakage path is predicted by using an edge failure propagation model in the seepage mechanics knowledge graph, and the edge failure propagation model describes the relationship between the probability of connection of adjacent micro-leakage points and distance and time under the coupling effect of liquid surface tension, pressure and material deformation. In some embodiments, the specific implementation of the edge failure propagation model includes: the model takes the Euclidean distance between adjacent micro-leakage points and the time length of the existence of the micro-leakage points as core input variables according to the pre-stored historical edge failure cases in the knowledge graph; the model simultaneously couples the additional capillary pressure generated by the liquid surface tension in the micro-scale channel, the liquid static pressure gradient formed by the internal saturated area, and the boundary pore geometry change caused by the material fiber structure deformation, and establishes a functional relationship between the connection probability and distance and time; the functional relationship is that the probability decays nonlinearly as the distance increases, and the probability gradually increases as the time accumulates, so as to realize the dynamic quantitative evaluation of the connection possibility between the micro-leakage points. Optionally, by comprehensively matching the obtained critical condition, the calculated risk measurement value and the predicted possibility, the occurrence probability of the seepage breakthrough event is calculated by using a probability fusion rule in the seepage mechanics knowledge graph, and the probability fusion rule maps the above-mentioned multiple source inputs into a unified probability value.

[0078] In some embodiments, an example scenario involves a simulated liquid pouring event, and the current liquid flow front position sequence shows that two main branches of the front leg are advancing in the direction of the leak-proof separation edge. The historical seepage mode matched from the seepage mechanics knowledge graph indicates that when the liquid flow front reaches a certain speed threshold in a specific material density gradient area, the probability of subsequent breakthrough of the separation edge significantly increases. At the same time, the area-perimeter ratio of the internal saturated area profile has exceeded the instability criterion threshold recorded in the knowledge graph, and the calculated risk measurement value is high. The edge failure propagation model predicts that the possibility of the connection of the current three adjacent micro-leakage points within the next five seconds is medium according to the distance reduction trend of the three adjacent micro-leakage points. The probability fusion rule comprehensively considers the critical speed condition, the high risk measurement value and the medium connection possibility, and finally calculates a high seepage breakthrough event occurrence probability. In the specific implementation, according to the advancing direction of the liquid flow front position sequence and the dominant conduction path presented in the pore connectivity change field, all the paths of the liquid flow formed in the material are traced back, the processing unit regards the pore connectivity change field as a weighted graph network, the nodes represent the material areas, and the weights of the edges represent the conduction ability. The graph search algorithm is used to find the high-weight paths converging into the saturated area from the current liquid flow front position, and the network of these found paths is the potential leakage channel network.

[0079] Referring to Figure 3, which is based on the core quantitative results of the seepage dynamics knowledge graph deduction situation link, uses the edge failure propagation model in the knowledge graph to predict the possibility of connecting micro leakage points into a through leakage path. The horizontal axis of the graph is the distance of the micro leakage point, and the vertical axis is the probability of connecting into a through path; the gray dots are the actual monitoring connection probability data, and the black curve is the fitted correlation trend line. This graph provides key data basis for calculating seepage breakthrough event probability by integrating critical conditions, risk measurement values, and micro leakage connection possibilities. It clearly defines the path risk of micro leakage points at different distances and is one of the core reference indicators for deducing potential leakage channel networks and quantifying leakage risks.

[0080] Embodiment 3: Refer to Figure 4 , the behavior disturbance correction of the seepage breakthrough event probability is combined with the target user body motion posture data to obtain the behavior correlation leakage probability. The correction process includes the following steps. Analyze the target user body motion posture data and identify periodic posture actions and non-periodic strong disturbance actions. Calculate the liquid periodic shaking model inside the diaper caused by periodic posture actions, apply periodic deformation disturbance to the internal saturated area profile according to the periodic shaking model, and update the fluctuation state of the liquid distribution. Evaluate the inertial impact force vector generated by non-periodic strong disturbance actions, convert the inertial impact force vector into additional driving amount for the liquid flow front position sequence, and update the instantaneous position and velocity of the liquid flow front. Re-input the updated fluctuation state of the liquid distribution and the instantaneous position and velocity of the liquid flow front into the seepage dynamics knowledge graph for fast local deduction. Obtain the immediate occurrence probability obtained after fast local deduction, and take this immediate occurrence probability as the behavior correlation leakage probability. When calculating the periodic shaking model, a fluid equivalent model is established to discretize the diaper internal space into multiple interconnected micro cavities. The periodic posture action is decomposed into rotation components and translation components around different body axes. Each rotation component and translation component is applied to the corresponding boundary of the fluid equivalent model as a driving input, and the liquid exchange flow rate between each micro cavity generated by the driving input is solved. According to the liquid exchange flow rate of all micro cavities, the periodic movement trajectory of the liquid mass center in space and the periodic fluctuation amplitude of the liquid free surface are counted, which together constitute the periodic shaking model.

[0081] In specific implementations, the behavior-dependent leakage probability is obtained by modifying the occurrence probability of the seepage breakthrough event in combination with the target user's body motion posture data. The modification process begins with the analysis of the posture data. In specific implementations, the target user's body motion posture data is analyzed, which is obtained from an inertial measurement unit attached to the user's clothing or body. The processing unit performs time-frequency analysis and pattern recognition on the continuous posture data stream, thereby separating and identifying periodic posture actions and non-periodic strong disturbance actions. It can be understood that periodic posture actions include regular lower limb swings during walking and running, while non-periodic strong disturbance actions include sudden sitting, turning around, or jumping. In specific implementations, a model of the periodic shaking of liquid inside the diaper caused by periodic posture actions is calculated. The calculation process first establishes a fluid equivalent model that discretizes the diaper internal space into a plurality of interconnected micro-cavities. Each micro-cavity has a fluid capacity and a flow resistance coefficient defined according to the material properties of the absorbent body. In specific implementations, the identified periodic posture actions are decomposed into rotational components around the body's sagittal axis, coronal axis, and vertical axis, and translational components in three-dimensional space. Each rotational component and translational component is used as a driving input, which is converted into a corresponding acceleration field or displacement boundary condition applied to the corresponding geometric boundary of the fluid equivalent model. By solving the simplified fluid dynamics equation, the liquid exchange flow between each micro-cavity generated by the driving input is obtained. It can be understood that the liquid exchange flow describes the periodic exchange from a micro-cavity to an adjacent cavity , which can be calculated as:

[0082]

[0083] where: represents the periodic liquid exchange flow from cavity to cavity at time , represents the flow conductance coefficient connecting cavities and , and represent the liquid pressures inside cavities and , is the liquid density, is the direction of cavity connection after decomposition of the periodic posture action The acceleration components generated above. According to the liquid exchange flow rate between all microcavities, the periodic movement trajectory of the liquid mass center in space and the periodic fluctuation amplitude of the liquid free surface are counted, and the movement trajectory of the liquid mass center and the fluctuation amplitude of the liquid free surface jointly constitute the periodic sloshing model. In some embodiments, according to the periodic sloshing model, periodic deformation disturbances are applied to the internal saturated region profile in the multi-dimensional percolation state diagram, and the fluctuation state of the liquid distribution is updated by simulating the movement of the liquid mass center and the fluctuation of the free surface. The boundary of the internal saturated region profile will present regular expansion and contraction with the same frequency as the sloshing model.

[0084] In specific implementations, the inertial impact force vector generated by the non-periodic strong disturbance action is evaluated, which is determined by the angular acceleration and linear acceleration calculated from the attitude data and the estimated liquid mass. The inertial impact force vector is converted into additional pushing amount on the liquid flow front position sequence. The additional pushing amount is calculated according to the momentum theorem and acts on the liquid motion direction indicated by the liquid flow front position sequence, thereby updating the instantaneous position and velocity of the liquid flow front. Optionally, in an example scenario, when the system identifies a non-periodic strong disturbance action of a user suddenly sitting down, a downward inertial impact force vector is calculated, which is converted into additional pushing amount on the liquid flow front in the downward direction, resulting in the instantaneous position of the liquid flow front in this direction moving downward and the velocity temporarily increasing. In some embodiments, the updated fluctuation state of the liquid distribution and the updated instantaneous position and velocity of the liquid flow front are re-input into the percolation dynamics knowledge graph as new input conditions for fast local deduction. Fast local deduction focuses on the local area and short time window directly affected by behavior disturbance, and reuses the logic in the knowledge graph but uses updated boundary conditions for calculation. In specific implementations, the instant occurrence probability obtained after fast local deduction is obtained, which integrates the dynamic influence brought by behavior disturbance, and this instant occurrence probability is output as the final behavior-related leakage probability.

[0085] Example 4: According to the deduced potential leakage channel network, the structure intervention scheme for inhibiting the formation of leakage channel network is deduced reversely, and the process is as follows. Key node and key path analysis is performed on the potential leakage channel network to identify a few core channel segments that play a decisive role in the connectivity of the entire network. For each core channel segment, its corresponding spatial region is located in the multi-dimensional seepage state diagram, and the historical connectivity data of the spatial region in the pore connectivity change field is extracted. Based on the historical connectivity data, a targeted local densification scheme or flow blocking structure is designed to weaken the connectivity of the core channel segment, and the local densification scheme is achieved by changing the fiber arrangement density of the absorber, and the flow blocking structure is achieved by implanting a micro-hydrophobic unit. From the perspective of the overall structure of the material, according to the overall topology of the leakage channel network, an internal absorber structure reorganization strategy is proposed to adjust the density gradient distribution between different regions of the absorber. From the perspective of boundary protection, according to the spatial distribution of micro-leakage state, an external barrier layer topology optimization strategy is proposed to enhance the sealing and drainage capacity of specific edge regions. When designing the local densification scheme or flow blocking structure, if the historical connectivity data shows that the formation of the core channel segment is mainly caused by the continuous connection of the high porosity region, then the local densification scheme is developed, which is to implant a cluster of ultra-fine fibers with higher specific surface area in the high porosity region, and the implantation density of the ultra-fine fiber cluster is positively correlated with the peak value of the historical connectivity data. If the historical connectivity data shows that the formation of the core channel segment is mainly dominated by the capillary dominant path, then the flow blocking structure is designed, which is to set micro-hydrophobic units along the direction of the capillary dominant path at intervals, and the spacing of the micro-hydrophobic units is adaptively adjusted according to the capillary pressure gradient of the path. Trigger thresholds are defined for the local densification scheme and the flow blocking structure, and when the behavior-related leakage probability exceeds the trigger threshold, the corresponding scheme or structure is implemented.

[0086] In practical implementation, a structural intervention scheme to suppress the formation of the potential leakage channel network is derived in reverse from the deduced potential leakage channel network. This derivation process begins with a topological analysis of the potential leakage channel network. In practice, critical node and critical path analysis is performed on the potential leakage channel network. The processing unit uses graph theory algorithms to calculate the betweenness centrality of each node and the flow load of each path in the network, identifying a few core channel segments that play a decisive role in the connectivity of the entire network. Core channel segments typically have high betweenness centrality or carry the main flow in the simulation prediction. For each identified core channel segment, its corresponding spatial region is located in the multidimensional seepage situation map, and the historical conductivity data of this spatial region in the pore connectivity change field is extracted. The historical conductivity data includes the sequence, statistical peak, and trend of conductivity changes of the region over time within the observation time window. In some embodiments, the leakage channel network analysis of the example scenario identifies three core channel segments, as shown in Table 1, which shows the historical conductivity data extraction results for their corresponding spatial regions.

[0087] Table 1. Characteristics and Historical Conductivity Data of Core Channel Segments

[0088]

[0089] Based on historical conductivity data, targeted local densification schemes or flow-blocking structures are designed to weaken the conductivity of the core channel segment. Local densification schemes are achieved by altering the fiber arrangement density of the absorber, while flow-blocking structures are implemented by implanting micro-hydrophobic units. In practice, if historical conductivity data shows that the formation of the core channel segment is mainly due to the confluence of high-porosity regions, a local densification scheme is formulated. Specifically, this involves implanting ultrafine fiber clusters with a higher specific surface area into these high-porosity regions. The implantation density of these ultrafine fiber clusters is positively correlated with the peak value of the historical conductivity data. For example, for channel segment Alpha in Table 1, its historical conductivity peak value is 0.92, and the dominant factor is the confluence of high-porosity regions. Therefore, a higher density of ultrafine fiber clusters is designed to be implanted in this region. In some embodiments, if historical conductivity data shows that the formation of the core channel segment is mainly dominated by capillary dominant paths, a flow-guiding and blocking structure is designed. Specifically, along this capillary dominant path, micro-hydrophobic units are intermittently arranged. The spacing of the micro-hydrophobic units is adaptively adjusted according to the capillary pressure gradient of the path. The adaptive adjustment formula can be expressed as:

[0090]

[0091] in: Indicates the first The spacing between each micro hydrophobic unit and the next unit It is the basic spacing constant. is the attenuation coefficient, represents the modulus of the capillary pressure gradient calculated along the path at the unit position. The area with a large capillary pressure gradient value has a small spacing and the spacing is automatically reduced to arrange more dense blocking units. For example, for the channel segment Beta in Table 1, which is dominated by the capillary dominant path, the array of micro-hydrophobic units is non-uniformly arranged along its path according to the real-time capillary pressure gradient distribution. Optionally, a trigger threshold is defined for the local densification scheme and the flow guiding blocking structure, which is a preset threshold compared with the behavior-associated leakage probability value. When the behavior-associated leakage probability exceeds the trigger threshold, the implementation of the corresponding scheme or structure is started, which can refer to sending structure parameter instructions to the manufacturing system or activating the pre-set response mechanism in the material.

[0092] In specific implementations, from the perspective of the overall structure of the material, according to the overall topology of the leakage channel network, an internal absorbent structure reorganization strategy is proposed to adjust the density gradient distribution between different regions of the absorbent. The reorganization strategy changes the mixing ratio of superabsorbent polymer and fluff pulp or fiber orientation in different regions to achieve a preset density gradient. From the perspective of boundary protection, according to the spatial distribution of the micro-leakage situation, an external barrier layer topology optimization strategy is proposed to enhance the sealing and guiding ability of specific edge regions. The optimization strategy includes adjusting the height and elasticity of the leak-proof edge, optimizing the fitting curve of the waist pad area, or increasing the distribution density of hydrophobic flow guiding grooves in the high micro-leakage situation area.

[0093] Referring to Figure 5 , the figure is a core quantitative analysis diagram of the structural intervention scheme for inhibiting leakage channels in reverse derivation. The horizontal axis is the leakage probability trigger threshold, and the vertical axis is the intervention scheme efficiency. The dashed line in the figure indicates the recommended trigger threshold, which is a key node for balancing the intervention timing and effect: a too low threshold is easy to start intervention too early, and a too high threshold is easy to lag intervention. This threshold not only ensures that intervention is started when the leakage risk is high, but also avoids false triggering, providing a quantitative basis for the dynamic start mechanism of the local densification / flow guiding blocking scheme, and is one of the core reference indexes for the landing of the structural intervention scheme.

[0094] In a specific implementation, the self-evolution process of the seepage dynamics knowledge graph is driven by observable leakage events occurring in actual application scenarios. In a specific implementation, the full-cycle physical field change signal sequence of the target diaper from the start of use to the occurrence of an observable leakage event and the corresponding user behavior data sequence are continuously collected. The full-cycle physical field change signal sequence is recorded by distributed sensing elements at a preset sampling frequency, and the user behavior data sequence is recorded synchronously by an associated inertial measurement unit. These data are time-stamped and stored in a buffer queue. When the system confirms the occurrence of an observable leakage event through an edge signal threshold or visual assistance, the self-evolution process is triggered. The key situation features within a preset time window before the occurrence of the event are extracted from the stored full-cycle physical field change signal sequence. The key situation features include, but are not limited to, the final position sequence of the liquid flow front, the morphological parameters of the internal saturated region before the contour breaks, the evolution curve of the connectivity change field near the leakage point, and the final spatial distribution pattern of the micro-leakage situation. In a specific implementation, the key situation features extracted by backtracking are associated with the final leakage result, which records the specific location, amount, and mode of the leakage. Thus, a new situation-result pair is formed, which is used as a new observation sample for knowledge updating.

[0095] In a specific implementation, the self-evolution process of the seepage dynamics knowledge graph is driven by observable leakage events occurring in actual application scenarios. In a specific implementation, the full-cycle physical field change signal sequence of the target diaper from the start of use to the occurrence of an observable leakage event and the corresponding user behavior data sequence are continuously collected. The full-cycle physical field change signal sequence is recorded by distributed sensing elements at a preset sampling frequency, and the user behavior data sequence is recorded synchronously by an associated inertial measurement unit. These data are time-stamped and stored in a buffer queue. When the system confirms the occurrence of an observable leakage event through an edge signal threshold or visual assistance, the self-evolution process is triggered. The key situation features within a preset time window before the occurrence of the event are extracted from the stored full-cycle physical field change signal sequence. The key situation features include, but are not limited to, the final position sequence of the liquid flow front, the morphological parameters of the internal saturated region before the contour breaks, the evolution curve of the connectivity change field near the leakage point, and the final spatial distribution pattern of the micro-leakage situation. In a specific implementation, the key situation features extracted by backtracking are associated with the final leakage result, which records the specific location, amount, and mode of the leakage. Thus, a new situation-result pair is formed, which is used as a new observation sample for knowledge updating.

[0096] In some embodiments, the example scenario records a lateral leakage event occurred after a user runs, the system backtracks to extract the key situation features within three seconds before the leakage occurs, the liquid flow front position sequence shows a continuous impact on the left side of the leakage barrier, the tightness index of the internal saturation area profile continues to decrease before the event, and the pore connectivity change field shows a high conductivity belt inside the barrier. Optionally, these key situation features are bound with the final result of "left side barrier micro-leakage developing into a through leakage" to form a new situation-result pair. By using the new situation-result pair to modify the historical seepage patterns, saturation area instability criteria and edge failure propagation model parameters in the seepage dynamics knowledge graph, it can be understood that the parameter modification follows the learning rule based on Bayesian update. For the historical seepage pattern, the matching weight or critical condition parameter of successful breakthrough will be adjusted according to the new sample, and the specific modification can be expressed as:

[0097]

[0098] wherein: and respectively represent the parameter vectors of the th historical seepage pattern or criterion model in the seepage dynamics knowledge graph before and after modification, represents the learning rate, is the actual result code observed in the new situation-result pair, is the result code predicted by the knowledge graph based on the original parameters for the key situation features extracted by backtracking, is the feature contribution vector of the th pattern or model to the new sample. In some embodiments, for the saturation area instability criterion, the geometric feature threshold parameter will be fine-tuned according to the morphological data before the profile breaks in the new sample; for the edge failure propagation model, the connection probability and distance, time relationship parameters will be calibrated according to the new micro-leakage situation space distribution pattern and the final leakage path. The modified seepage dynamics knowledge graph is used as the basis for subsequent situation evolution deduction of any new situation, thereby realizing the iterative optimization of the seepage dynamics knowledge graph, and the optimized knowledge graph can output more accurate deduction results when dealing with similar situations.

[0099] It should be noted that, in this document, relationship terms such as first and second are used only to distinguish one entity or action from another, and do not necessarily require or imply that these entities or actions have any such actual relationship or order. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0100] While embodiments of the application have been shown and described, it is to be understood that the application is not limited to the details of the embodiments described, since numerous changes, modifications, substitutions and variations can be made thereto without departing from the spirit and scope of the application as defined by the appended claims and their equivalents.

Claims

1. A smart leak detection method based on diapers, characterized in that, This includes the following collaborative processes: Multiple distributed sensing elements are used to synchronously capture physical field change signals in the internal and edge areas of the target diaper. These physical field change signals include liquid wetting trajectory signals, material fiber structure deformation signals, and heat conduction change signals. The captured physical field change signals are synchronously fused and spatiotemporally calibrated to construct a multidimensional seepage situation map that reflects the dynamic existence state of the liquid. The multidimensional seepage situation map embeds the liquid flow front position sequence, the internal saturation region outline and the micro-leakage situation of the external contact edge. Based on a pre-defined knowledge graph of seepage dynamics, the multidimensional seepage situation map is used to perform situation evolution deduction, and the probability of seepage breakthrough events and potential leakage channel networks are deduced. The probability of the seepage breakthrough event is corrected by behavioral perturbation based on the target user's body posture data to obtain the behavior-related leakage probability. The behavioral perturbation correction is completed based on the correspondence between the liquid inertial migration law and body posture. Based on the deduced potential leakage channel network, a structural intervention scheme to suppress the formation of the leakage channel network is derived in reverse. The structural intervention scheme includes an internal absorber structure reorganization strategy and an external barrier layer topology optimization strategy. The deformation signal of the material fiber structure is mapped into a pore connectivity change field in three-dimensional space. The pore connectivity change field is used to describe the real-time opening and closing state of the liquid passageway. The process of performing situation evolution deduction on the multidimensional seepage situation map based on a preset seepage dynamics knowledge graph includes: Match historical penetration patterns from the knowledge graph that correspond to the current sequence of fluid flow front positions, and extract the critical conditions under which the historical penetration patterns successfully overcome the barrier. The geometric features of the internal saturated region contour are compared with the saturated region instability criteria stored in the knowledge graph to calculate the risk metric value of the current saturated region causing overall lateral overflow. The evolution trend of the micro-leakage state at the external contact edge is analyzed, and the edge failure propagation model in the knowledge graph is used to predict the possibility of micro-leakage points connecting to form a through-leakage path. Based on the critical conditions, the risk metric, and the probability, the probability of the seepage breakthrough event is calculated using the probability fusion rules in the knowledge graph. Simultaneously, based on the advancing direction of the liquid flow front position sequence and the dominant conduction path presented in the pore connectivity change field, all possible paths for the liquid to form and converge within the material are traced in reverse, and these possible paths are networked into the potential leakage channel network.

2. The intelligent leak detection method for diapers according to claim 1, characterized in that, The process of synchronously fusing and spatiotemporally calibrating the captured physical field change signals to construct a multidimensional seepage situation map reflecting the dynamic state of the liquid is achieved through the following steps: Multiple independent fluid flow branch signals are separated from the fluid wetting trajectory signal, and the start time and propagation vector of each fluid flow branch signal are marked. The deformation signal of the material's fiber structure is mapped into a pore connectivity change field in three-dimensional space. The heat conduction change signal is analyzed as a heat field gradient change associated with the liquid distribution, and the high temperature region and the liquid accumulation region have a spatial correspondence in the heat field gradient change; Establish a unified spatiotemporal coordinate system, and overlay and correlate the propagation vector of the fluid flow branch signal, the pore connectivity change field, and the thermal field gradient change. In the spatiotemporal coordinate system, based on the results of the correlation matching, the sequence of liquid flow front positions is dynamically drawn, and the outline of the internal saturation region after the liquid is absorbed is delineated. By continuously tracking the energy decay pattern of the liquid wetting trajectory signal near the diaper boundary and combining it with the abrupt change characteristics of the pore connectivity change field at the boundary, the micro-leakage status of the external contact edge is comprehensively determined.

3. The intelligent leak detection method for diapers according to claim 2, characterized in that, The process of mapping the material fiber structure deformation signal into a pore connectivity change field in three-dimensional space includes: The deformation signal of the fiber structure of the material is decomposed into macroscopic overall bending components and microscopic fiber network displacement components. For the macroscopic overall bending component, the resulting change in compressibility in the thickness direction of the absorber is calculated, and a porosity change distribution map dominated by compression is generated based on the relationship model between compressibility and porosity. For the displacement components of the microfiber network, the displacement vectors at the intersections between fibers are analyzed, and the equivalent hydraulic diameter changes of the capillary channels between fibers are calculated based on the displacement vectors to generate a dynamic adjustment diagram of the capillary channels. By integrating the porosity variation distribution map and the capillary channel dynamic adjustment map, a comprehensive pore connectivity variation field is generated. The value of each spatial point in the pore connectivity variation field represents the real-time conductivity of each spatial point for liquid flow.

4. The intelligent detection method for diaper leak prevention according to claim 1, characterized in that, The step of combining target user body posture data to perform behavioral perturbation correction on the probability of the seepage breakthrough event to obtain the behavior-related leakage probability includes: The target user's body posture data is analyzed to identify periodic posture movements and non-periodic strong disturbance movements; Calculate the periodic sloshing model of the liquid inside the diaper caused by the periodic posture movement, and apply periodic deformation perturbation to the contour of the internal saturated region based on the periodic sloshing model to update the fluctuation state of the liquid distribution. The inertial impact force vector generated by the non-periodic strong disturbance is evaluated, and the inertial impact force vector is converted into an additional driving force on the position sequence of the fluid flow front, and the instantaneous position and velocity of the fluid flow front are updated. The updated fluctuation state of the liquid distribution and the instantaneous position and velocity of the liquid flow front are re-input into the seepage dynamics knowledge graph for rapid local extrapolation; The instantaneous occurrence probability obtained after rapid local extrapolation is used as the probability of leakage associated with the behavior.

5. The intelligent leak detection method for diapers according to claim 4, characterized in that, The calculation model of the periodic sloshing of liquid inside the diaper caused by the periodic posture movement includes: A fluid equivalent model is established that discretizes the internal space of a diaper into multiple interconnected microcavities; The periodic posture motion is decomposed into rotational and translational components about different body axes. Each rotational and translational component is used as a driving input and applied to the boundary corresponding to the fluid equivalent model to solve the liquid exchange flow rate between each micro-cavity generated by the driving input. Based on the liquid exchange flow rate of all microcavities, the periodic movement trajectory of the liquid mass center in space and the periodic fluctuation amplitude of the liquid free surface are statistically determined. These two together constitute the periodic swaying model.

6. The intelligent leak detection method for diapers according to claim 2, characterized in that, The method of reverse-deriving structural intervention schemes to suppress the formation of the potential leakage channel network based on the deduced network includes: Critical node and critical path analysis was performed on the potential leakage channel network to identify a few core channel segments that play a decisive role in the overall network connectivity. For each core channel segment, its corresponding spatial region is located in the multidimensional seepage situation map, and the historical conductivity data of the spatial region in the pore connectivity change field is extracted. Based on the historical conductivity data, a targeted local densification scheme or a flow-blocking structure is designed to weaken the conductivity of the core channel segment. The local densification scheme is achieved by changing the fiber arrangement density of the absorber, and the flow-blocking structure is achieved by implanting micro hydrophobic units. From the perspective of the overall material structure, based on the overall topology of the leakage channel network, a structural reorganization strategy for the internal absorber is proposed to adjust the density gradient distribution between different regions of the absorber. From the perspective of boundary protection, based on the spatial distribution of the micro-leakage situation, a topology optimization strategy for the external barrier layer is proposed to enhance the sealing and drainage capabilities of specific edge areas.

7. The intelligent leak detection method for diapers according to claim 6, characterized in that, The targeted local densification scheme or flow-blocking structure includes: If the historical conductivity data shows that the formation of the core channel segment is mainly caused by the confluence of high porosity regions, then a local densification scheme is formulated, specifically: implanting ultrafine fiber clusters with a higher specific surface area in the high porosity regions, and the implantation density of the ultrafine fiber clusters is positively correlated with the peak value of the historical conductivity data; If the historical conductivity data shows that the formation of the core channel segment is mainly dominated by a specific capillary dominant path, then a flow-guiding and blocking structure is designed, specifically: along the direction of the capillary dominant path, micro hydrophobic units are set at intervals, and the spacing of the micro hydrophobic units is adaptively adjusted according to the capillary pressure gradient of the path. A trigger threshold is defined for the local densification scheme and the flow blocking structure. When the probability of leakage associated with the behavior exceeds the trigger threshold, the implementation of the corresponding scheme or structure is initiated.

8. The intelligent leak detection method for diapers according to claim 1, characterized in that, It also includes a process of self-evolution of the aforementioned seepage dynamics knowledge graph: In practical application scenarios, the physical field change signal sequence and corresponding user behavior data sequence of the target diaper are continuously collected throughout the entire cycle from the start of use to the occurrence of an observable leakage event. When an observable leakage event occurs, key situational features prior to the event are extracted from the full-cycle physical field change signal sequence, and these key situational features are correlated with the final leakage result to form a new situation-result pair. The new situation-results are used to revise the parameters in the historical penetration pattern, saturation region instability criterion, and edge failure propagation model of the knowledge graph. The revised knowledge graph will serve as the basis for subsequent situational evolution deduction, thereby achieving iterative optimization of the seepage dynamics knowledge graph.

9. A smart leak detection system based on diapers, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent diaper leak-proof detection method according to any one of claims 1 to 8.

Images