Nursing pants monitoring system and monitoring method based on multi-sensor fusion

Abstract

The invention provides a nursing pants monitoring system and monitoring method based on multi-sensor fusion, and relates to the technical field of nursing pants monitoring. Humidity, pressure, temperature and gas concentration data are cooperatively collected through a multi-mode sensor module integrated in nursing pants; a sensor data stream is converted into a track in a high-dimensional state space by an excretion phase diagram construction module to represent continuous changes of physiological states, and accurate prediction of excretion event types and development stages is realized by calculating state change vectors in real time and matching the state change vectors with a preset vector pattern library. According to the method, leg locking and hip supporting instructions are generated according to event tracks, graded nursing decision making is realized based on an excretion early warning index fusing event intensity, duration and chaos degree, and finally parameters of cleaning, dirt suction, drying and disinfection are dynamically adjusted according to event characteristics, so that adaptive, refined and low-energy-consumption intelligent nursing is realized.

Description

Technical Field

[0001] This invention relates to the field of nursing pants monitoring technology, specifically to a nursing pants monitoring system and method based on multi-sensor fusion. Background Technology

[0002] In the daily care of disabled people, excrement management is a crucial and extremely challenging task. Traditional care methods rely heavily on diapers and manual care, which suffer from poor breathability, easy leakage, and frequent changes. This not only easily leads to skin erythema, bedsores, and other infections in patients, but also places a heavy burden on caregivers and creates a poor care environment. With the development of the Internet of Things and smart sensing technology, the industry has begun to explore smart nursing pants with integrated sensors, attempting to achieve automatic monitoring and preliminary treatment of excrement through technological means, aiming to improve the quality and efficiency of care and enhance the dignity of life for disabled people.

[0003] In existing technologies, smart nursing pants with monitoring functions mostly use a single humidity sensor as the core detection unit. They trigger simple alarms or basic cleaning actions by determining whether the humidity exceeds a threshold. Some solutions attempt to add temperature sensors to distinguish between sweat and excrement, or integrate a communication module in a fixed part to send alarm information to the nursing end. They mostly rely on a single or a few types of sensors and determine the occurrence of events by setting simple static thresholds. For example, they only use a humidity sensor to detect liquids, or a gas sensor to detect the concentration of a specific gas. Once the reading exceeds a predetermined value, a unified nursing process is initiated.

[0004] Existing technologies in this area still have significant shortcomings. Due to the complexity of human activities and physiological processes, false alarms are easily generated. For example, sweat and ambient humidity can be mistaken for urination, leading to frequent false triggers of the system. Secondly, existing technologies cannot effectively distinguish the type, intensity, and process of excretion events, failing to differentiate between minor urination and severe defecation. Consequently, they can only implement a "one-size-fits-all" care procedure, resulting in either insufficient cleaning leading to hygiene risks or excessive cleaning causing resource waste, inefficiency, and a poor user experience. Most importantly, existing technologies lack the ability to dynamically perceive and predict the event process, only responding passively when excrement is detected, unable to predict the event before it begins and take adaptive measures to prevent its spread.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a nursing pants monitoring system and method based on multi-sensor fusion to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A nursing pants monitoring system based on multi-sensor fusion includes the following steps: The multimodal sensor module, integrated inside the nursing pants, includes a distributed impedance sensing array, a temperature sensor, a gas sensor, and a distributed micro-piezoresistive array, used to collect multimodal sensing data of the internal environment of the nursing pants. The discharge phase diagram construction module is used to divide multimodal sensing data into time windows of preset length, extract feature values ​​of multimodal sensing data within each time window, and construct a discharge phase diagram with the time window sequence as the horizontal axis and feature values ​​as the vertical axis, where each time window corresponds to a state point on the phase diagram; The dynamic vector analysis module is used to calculate the change vector between state points in a continuous time window on the discharge phase diagram in real time. It performs multi-dimensional similarity matching and weighted scoring between the current change vector and each vector template in the vector pattern library to dynamically predict the type and development stage of the current discharge event. The vector pattern library contains several vector templates that represent different discharge events and development stages. The nursing strategy decision-making module is used to generate leg locking control instructions when the start of an excretion event is predicted, and to generate hip lifting control instructions when the end of an excretion event is predicted. It also calculates a comprehensive excretion warning index based on the average intensity of all change vectors during the start to end of the excretion event, and generates categorized nursing action control instructions based on the high, medium and low ranges of this index. The dynamic optimization execution module receives nursing action control instructions and adjusts the execution parameters of cleaning, vacuuming, drying, and disinfection nursing actions based on the excretion warning index of the current excretion event.

[0008] Furthermore, the multimodal sensor module includes: A distributed impedance-capacitance sensor array is used to simultaneously measure the impedance and capacitance values ​​of different areas of the lining of nursing pants; A temperature sensor, installed near a node of a distributed impedance-capacitance sensor array, is used to monitor and output the temperature measurement value at the node location. A gas sensor, encapsulated within a protective diaphragm with a hydrophobic and breathable membrane, is used to quantitatively detect and output the concentration values ​​of ammonia and hydrogen sulfide in the internal environment of the nursing pants. A distributed micro-piezoresistive array is used to synchronously measure and output the pressure value and physical coordinates of each sensing unit; A distributed impedance-capacitance sensor array and a distributed micro-piezoresistive array are embedded in the crotch and hip lining of the nursing pants in a grid pattern; temperature and gas sensors are placed in the sensor cluster at the center of the crotch.

[0009] Furthermore, for each time window, the time-domain statistical characteristics and frequency-domain energy characteristics of each sensor data are calculated. The time-domain statistical characteristics include the mean, variance, and mean absolute value of the first difference. The frequency-domain energy characteristics are obtained by integrating the signal energy within a preset physiologically relevant frequency band after performing a fast Fourier transform on the sensor data. All feature values ​​extracted by all sensors are combined into a multidimensional vector, which is the coordinate of the state point in the high-dimensional space of the discharge phase diagram within the corresponding time window; In chronological order, the multidimensional vectors corresponding to each time window are sequentially depicted in a high-dimensional state space with the time window index as the horizontal axis and each dimension of the vector as the vertical axis, thus forming a trajectory composed of continuous state points. This trajectory is the excretion phase diagram that represents the continuous changes in the user's physiological state. For any two consecutive state points on the discharge phase diagram, calculate the difference between the latter state point and the former state point. This difference constitutes a change vector, which contains information on the direction and magnitude of the state change from the previous time window to the current time window.

[0010] Furthermore, the specific logic for matching analysis performed by the dynamic vector analysis module is as follows: For the current change vector and the vector template in the pattern library, the directional similarity and magnitude similarity are calculated and weighted and fused to obtain a matching score; The calculation logic for the directional similarity is as follows: calculate the cosine value of the angle between the current changing vector and the vector template to represent the consistency of their directions, and map the cosine value to the interval between 0 and 1 to obtain the normalized directional similarity score. The calculation logic of the amplitude similarity is as follows: calculate the Euclidean distance between the current change vector and the vector template to represent the difference in amplitude between the two, introduce a normalization constant to scale the Euclidean distance, and calculate a normalized distance similarity score between 0 and 1 accordingly. Based on preset weighting coefficients, the normalized directional similarity score and the normalized distance similarity score are weighted and summed to obtain the final matching score. The excretion event type and development stage labeled by the vector template with the highest matching score are used as the prediction result of the current change vector.

[0011] Furthermore, the operation of the nursing strategy decision-making module specifically includes: When the prediction result output by the dynamic vector analysis module indicates the start of an excretion event, a leg locking control command is immediately generated; when the prediction result indicates the end of an excretion event, a hip lifting control command is immediately generated. After the discharge event ends, a comprehensive discharge early warning index is calculated based on the average intensity of all change vectors during the discharge event from start to finish. The calculation logic of the discharge early warning index is as follows: calculate the average value of the magnitude of all change vectors in the entire discharge event, which is used to characterize the cumulative intensity of the event, and record the total duration of the event; calculate the variance of all change vector magnitudes, which is used to characterize the volatility or disorder of the event process; finally, multiply the average value, total duration and variance by a preset normalization weight coefficient and sum them up. The result is the discharge early warning index.

[0012] Furthermore, the nursing strategy decision-making module has preset high index thresholds and low index thresholds, and compares the calculated excretion warning index with the thresholds: When the excretion warning index is greater than the high index threshold, it is determined to be a high-risk level, and a high-intensity nursing instruction is generated accordingly; when the excretion warning index is between the low index threshold and the high index threshold, it is determined to be a medium-risk level, and a standard-intensity nursing instruction is generated accordingly; when the excretion warning index is less than the low index threshold, it is determined to be a low-risk level, and an energy-saving intensity nursing instruction is generated accordingly. For cleaning actions, the higher the discharge warning index, the higher the cleaning water temperature, the greater the water pressure, and the longer the cleaning time set in the command. For suction actions, the greater the impedance sensor characteristic value in the final state point coordinates, the greater the suction intensity and the longer the duration set in the command. For drying and disinfection actions, the greater the gas sensor characteristic value in the final state point coordinates, the higher the drying temperature, the greater the air volume, and the more disinfectant sprayed in the command.

[0013] The present invention also provides a method for monitoring nursing pants based on multi-sensor fusion, the method being used to execute the above-described nursing pants monitoring system based on multi-sensor fusion, comprising: Step 1: Multiple sensors are integrated inside the nursing pants body, including a distributed impedance sensing array, a temperature sensor, a gas sensor, and a distributed micro-piezoresistive array to collect multimodal sensing data of the internal environment of the nursing pants. Step 2: Divide the multimodal sensing data into time windows of a preset length, extract the feature values ​​of the multimodal sensing data in each time window, and construct a discharge phase diagram with the time window sequence as the horizontal axis and the feature values ​​as the vertical axis, where each time window corresponds to a state point on the phase diagram; Step 3: Calculate the change vector between state points in a continuous time window on the excretion phase diagram in real time, perform multi-dimensional similarity matching and weighted scoring between the current change vector and each vector template in the vector pattern library, and dynamically predict the type and development stage of the current excretion event. The vector pattern library contains several vector templates that represent different excretion events and development stages. Step 4: Generate a leg locking control command when the excretion event is predicted to begin, and a hip lifting control command when the excretion event is predicted to end. Calculate a comprehensive excretion warning index based on the average intensity of all change vectors during the excretion event from start to end, and generate categorized nursing action control commands based on the high, medium, and low ranges of this index. Step 5: Receive nursing action control instructions and adjust the execution parameters of cleaning, vacuuming, drying and disinfection nursing actions according to the excretion warning index of this excretion event.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention uses the multimodal sensor module to collaboratively collect physical and chemical signals, and the excretion phase diagram construction module to convert the time-series data stream into a trajectory in a high-dimensional state space, fundamentally solving the problem of high false alarm rate in the traditional single-sensor threshold method; different sensor signals corroborate each other in the phase diagram space, enabling the system to effectively distinguish different event sources such as sweat, urine, and feces, thereby greatly avoiding false triggering caused by environmental humidity or non-excretion activities of the human body; This invention uses the dynamic vector analysis module to perform real-time matching analysis of state change trajectories. This not only accurately identifies the start and end points of events, enabling predictive leg locking and timely buttock lifting, but more importantly, it can classify and assess events based on the excretion warning index calculated by the nursing strategy decision module. This index comprehensively considers the average intensity, duration, and process disorder of the event, allowing the system to generate differentiated nursing instructions according to the actual severity of the event. This completely changes the existing "one-size-fits-all" nursing model, significantly improving nursing efficiency and reducing resource consumption while ensuring cleaning effectiveness. The dynamic optimization execution module of this invention makes fine adjustments to the parameters of cleaning, vacuuming, drying and disinfection based on the sensor characteristic values ​​of the final state point, such as impedance value and gas concentration. This achieves precise resource allocation based on the amount of residual pollution, and ultimately forms a personalized closed-loop care system that becomes smarter with use. Attached Figure Description

[0015] Figure 1 This is a block diagram of the overall system module structure of the present invention; Figure 2 This is a probability diagram of the excretion early warning index of the present invention; Figure 3 This is a fitted curve of the total duration of the event and the discharge early warning index of the present invention; Figure 4 The fitted curve of the number of changing vectors versus the average magnitude of the changing vectors in this invention. Figure 5This is a grid diagram of the total duration of events in this invention, the average and variance of the magnitude of the change vector, and the discharge early warning index. Figure 6 This is a block diagram of the multimodal sensor module structure of the present invention; Figure 7 This is a schematic diagram of the overall method flow of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0017] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0018] Example: Please see Figures 1-6 The present invention provides a technical solution: A nursing pants monitoring system based on multi-sensor fusion includes the following steps: The multimodal sensor module, integrated inside the nursing pants, includes a distributed impedance sensing array, a temperature sensor, a gas sensor, and a distributed micro-piezoresistive array, used to collect multimodal sensing data of the internal environment of the nursing pants. Integrated into the body of the nursing pants: This defines the physical form and final application scenario of the entire module. It is not an external device, but to be seamlessly integrated with the nursing pants and become an internal part of them. This means that all sensors and cables must be ultra-thin, flexible, washable (or at least protective), and non-irritating to human skin. The technical challenges lie in flexible electronics technology and structural packaging technology. Multimodal is the core idea of ​​the entire system. The information provided by a single type of sensor is one-sided and prone to false alarms. For example, it is impossible to distinguish whether the increase in humidity is urine, sweat or external water leakage; it is impossible to distinguish whether the change in pressure is excretion or the user turning over. By fusing information from two completely different modes, physical (impedance, pressure, temperature) and chemical (gas), the system can cross-validate, which greatly improves the accuracy and reliability of event detection and classification. The multimodal sensor module consists of a distributed impedance-capacitance sensing array, a distributed micro piezoresistive array, a temperature sensor, and a gas sensor, and is managed by a single main control unit for unified power management, signal synchronization, and data aggregation. A distributed impedance-capacitance sensor array is used to simultaneously measure the impedance and capacitance values ​​of different areas of the lining of nursing pants; A temperature sensor, installed near a node of a distributed impedance-capacitance sensor array, is used to monitor and output the temperature measurement value at the node location. A gas sensor, encapsulated within a protective diaphragm with a hydrophobic and breathable membrane, is used to quantitatively detect and output the concentration values ​​of ammonia and hydrogen sulfide in the internal environment of the nursing pants. A distributed micro-piezoresistive array is used to synchronously measure and output the pressure value and physical coordinates of each sensing unit; A distributed impedance-capacitance sensor array and a distributed micro piezoresistive array are embedded in the crotch and hip lining of the nursing pants in a grid pattern; temperature and gas sensors are placed in the sensor cluster at the center of the crotch. The distributed impedance-capacitance sensor array consists of excitation electrodes and sensing electrodes arranged in a grid pattern in the crotch and buttock lining of the nursing pants. The output of each sensing node of the array is electrically connected to the input of a bioimpedance analog front end (specifically model AD5940, optimized for bioimpedance measurement and supporting multi-band scanning) and a high-precision capacitance-to-digital converter (specifically model AD7746, suitable for detecting minute capacitance changes) through a multi-channel analog switch array. The AD5940 and AD7746 are connected via SPI or I 2 The C-channel communication bus is electrically connected to the main control unit (specifically, the nRF52840, whose core is ARM Cortex-M4 and integrates multiple peripherals). The main control unit is configured to: control the multiplexer array to cyclically select each sensing node in a time-division multiplexing manner, and send synchronous measurement commands to AD5940 and AD7746, and then read the converted digital impedance and capacitance values. The gas sensor module specifically comprises an electrochemical ammonia sensor (such as SPEC Sensors NH3-A1) and a hydrogen sulfide sensor (such as H2S-A1). The weak current signal output terminal of its working electrode is electrically connected to the input terminal of a low-noise transimpedance amplifier circuit, which converts the current signal into a voltage signal that can be acquired. The output terminal of the transimpedance amplifier circuit is electrically connected to the dedicated analog input channel of the 12-bit precision analog-to-digital converter (ADC) inside the main control unit, or the signal can be routed to an external high-precision ADC (such as ADS1220) for sampling. In the temperature sensor cluster, at least one high-precision digital temperature sensor (such as DS18B20, accuracy ±0.5°C) is positioned adjacent to the gas sensor. Its data line is electrically connected to the GPIO pin of the main control unit via a pull-up resistor (single bus protocol) or via I... 2 The main control unit is configured to: read the temperature value in real time and, based on the temperature-sensitivity characteristic curve provided by the sensor manufacturer, perform real-time compensation and calibration of the gas concentration reading in the firmware to eliminate the influence of ambient temperature changes on the gas measurement accuracy. The distributed micro-piezoresistive array consists of multiple flexible piezoresistive thin-film pressure sensors (such as Interlink Electronics FSR400), forming a pressure-sensing grid in the lining of the nursing pants. Each FSR400 sensor is connected in series with a precision reference resistor (such as 10kΩ, 1%) to form a voltage divider circuit. The voltage of this series node serves as an analog signal characterizing the pressure magnitude. The outputs of all FSR400 voltage divider nodes are connected to the input channels of a 16-channel multiplexer analog switch (such as TI's CD74HC4067). The common output of this multiplexer is electrically connected to the analog input of a high-precision, low-noise 24-bit analog-to-digital converter (ADC) (such as TI's ADS1220). The ADS1220 is electrically connected to the main control unit via an SPI interface. The main control unit selects each pressure sensing unit sequentially by controlling the address selection line of the CD74HC4067, and synchronously triggers the ADS1220 to perform analog-to-digital conversion, thereby obtaining the pressure value of each unit and its physical coordinates in the array. To achieve strict timing alignment of multimodal data, the main control unit (nRF52840) uses its hardware timer to generate fixed-period interrupts. In this interrupt service routine, all signal paths (impedance, capacitance, pressure, gas) are synchronously triggered to perform a data acquisition. The main control unit adds a unified high-precision timestamp to all data acquired in this synchronous acquisition and packages them into a standard data frame. The main control unit sends the packaged multimodal data stream to the subsequent discharge phase diagram construction module through its integrated Bluetooth Low Energy (BLE) 5.0 wireless communication module. All sensors, discrete circuit components, and flexible printed circuit (FPC) interconnects are encapsulated in flexible, washable materials such as polyimide (PI) or thermoplastic polyurethane (TPU) to ensure wearability and system durability. The gas sensor module is sealed within a separate protective layer with an expanded polytetrafluoroethylene (ePTFE) hydrophobic and breathable membrane. This design allows gas molecules to diffuse freely while effectively blocking liquid water, water vapor, and solid contaminants, preventing physical contamination and chemical "poisoning" of the sensor's sensitive elements. Distributed impedance-capacitance sensor arrays mean that the sensors are not individual, but rather arranged in a grid-like, multi-node configuration within a region. The greatest significance of this is that it upgrades the detection from whether something has occurred to spatial perception of where it has occurred, its extent, and its trend. This is crucial for subsequent refined care, such as targeted rinsing. Impedance measurement focuses more on measuring conductivity and is very sensitive to liquids containing electrolytes, such as urine. Capacitance measurement focuses more on measuring dielectric constant. The dielectric constant of water is much higher than that of air and textile fibers, so the wetting of any liquid (including pure water and sweat) will cause a significant increase in capacitance. The readings of impedance and capacitance are affected by temperature. Placing the temperature sensor close to the impedance-capacitance node can obtain the most accurate local temperature, thereby enabling real-time compensation and calibration of the impedance and capacitance readings to ensure data accuracy. Fresh excrement carries body temperature. When a sudden increase in humidity is detected in a certain area, a small pulse-like increase in temperature also occurs at that point. This is a very strong correlation feature indicating that excretion is taking place and can be used to trigger high-priority alarms. Long-term monitoring of local body surface temperature can also provide auxiliary data for pressure ulcer early warning, etc. The core encapsulation technology that ensures the reliability and lifespan of the sensor is encapsulated in a protective layer with a hydrophobic and breathable membrane. The hydrophobic and breathable membrane allows gas molecules to pass through freely, but can effectively block liquid water, water vapor and solid contaminants. This prevents urine, feces and other substances from directly contaminating the sensor's sensitive elements and avoids sensor failure or poisoning. This is a prerequisite for pushing this technology from the laboratory environment to actual wearable applications. Ammonia and hydrogen sulfide are characteristic gases produced after the decomposition of excrement (especially feces). Detecting them is the gold standard for confirming defecation events at the chemical level and is a key basis for distinguishing urine from feces. Electrochemical sensors have high sensitivity and low power consumption, making them very suitable for battery-powered wearable devices. The change in pressure distribution caused by the gravity of excrement is a very clever auxiliary dimension for judgment. When excrement is discharged, it has a certain mass and impact force, which will act on the lining of the nursing pants, causing dynamic changes in local pressure distribution. This change is different from the pressure changes caused by body movement and turning over in terms of waveform and duration. Physical coordinates locate the core area of ​​excretion. The pressure array can act like a scale to sense the area with the most significant increase in weight. This information is then fused with the humidity distribution map drawn by the impedance-capacitance array to accurately pinpoint the core location of excretion. This is crucial for activating the targeted cleaning function, which allows for precise rinsing of the most heavily contaminated areas with only a small amount of cleaning fluid, improving efficiency and saving resources and time. The distributed impedance-capacitance sensor array and the distributed micro piezoresistive array clearly define the sensor arrangement as a regular row and column matrix in a grid-like form. This simplifies the design of the data acquisition circuit and facilitates the generation of clear two-dimensional spatial images. The groin and buttocks are the key areas with the highest probability of excretion. Dense sensing coverage in this area ensures monitoring without blind spots. The center of the groin area is the hotspot where excrement first accumulates and has the highest concentration. The most critical gas sensor and temperature sensor used for critical compensation are placed here to ensure that they can contact the target signal as quickly and sensitively as possible, reduce response delay, and improve the real-time performance of detection. A sensor cluster describes a highly integrated design that physically combines sensors with different functions (temperature, gas) to form a fully functional supersensing unit. The advantage of this is that it provides the strongest data correlation, the richest local information, and facilitates modular design and production assembly.

[0019] The discharge phase diagram construction module is used to divide multimodal sensing data into time windows of preset length, extract feature values ​​of multimodal sensing data within each time window, and construct a discharge phase diagram with the time window sequence as the horizontal axis and feature values ​​as the vertical axis, where each time window corresponds to a state point on the phase diagram; For each time window, the time-domain statistical characteristics and frequency-domain energy characteristics of each sensor data are calculated. The time-domain statistical characteristics include the mean, variance, and mean absolute value of the first difference. The frequency-domain energy characteristics are obtained by integrating the signal energy within a preset physiologically relevant frequency band after performing a fast Fourier transform on the sensor data. All feature values ​​extracted from all sensors are combined into a multidimensional vector, which is the coordinate of the state point in the high-dimensional space of the excretion phase diagram within the corresponding time window. In chronological order, the multidimensional vectors corresponding to each time window are sequentially depicted in a high-dimensional state space with the time window index as the horizontal axis and each dimension of the vector as the vertical axis, thus forming a trajectory composed of continuous state points. This trajectory is the excretion phase diagram that represents the continuous changes in the user's physiological state. For any two consecutive state points on the discharge phase diagram, calculate the difference between the latter state point and the former state point. This difference constitutes a change vector, which contains information on the direction and magnitude of the state change from the previous time window to the current time window. The change vector reflects the amount and direction of change in the system state between two adjacent time windows. The direction of the change vector indicates which state the system state is evolving towards, such as evolving towards "increased humidity + increased pressure" or evolving towards "increased gas concentration + stable temperature". The direction is used to determine the type of event. The magnitude of the change vector indicates the severity of the state change. A large magnitude means that a very drastic change has occurred, which is the start or end point of the event. The magnitude is used to determine the intensity of the event. The larger the change vector and the larger its magnitude, the more drastic and rapid the change in the user's physiological state occurs from time window t-1 to t; the current state point Compared to the previous state point The greater the difference across all dimensions, the larger the calculated change vector. The larger the mold length, the better. For the index of the time window; The raw sensor data is a continuous high-speed data stream that cannot be processed directly. Cutting it into data segments of fixed length, such as 1 second or 500 milliseconds, is the first step in all signal processing. The raw data points are huge in number and full of redundancy and noise. The purpose of calculating feature values ​​is to use a few highly condensed numerical indicators to represent the essential information of this segment of raw data. For example, using the mean to represent the average level of the signal within the window is far more meaningful than listing all the raw data points. Phase in excretion diagram Figure 1 The term borrows the concept from physics to describe the state of a system. Here, it is not a simple two-dimensional curve, but a visual representation of the evolution of the system state over time. The state point is the core of understanding the phase diagram. A state point is no longer a single sensor reading, but represents a "comprehensive snapshot" of the entire system (the internal environment of the nursing pants) in that one second. This snapshot is composed of multiple characteristic values. Time-domain features are used to analyze the pattern of signal amplitude changes over time, while frequency-domain features, through Fourier transform, convert the signal from a "time-amplitude" perspective to a "frequency-energy" perspective, answering the question of "whether the signal contains a periodic or rhythmic component of a specific frequency". This is crucial for identifying periodic physiological activities such as muscle contraction and body tremors. Time-domain statistical characteristics include at least the mean, variance, and mean absolute value of the first difference. The mean represents the average intensity level of the sensor signal within the given time window. 1) Impedance mean: the overall level of liquid / humidity; a higher value indicates higher average humidity. 2) Pressure mean: the average pressure level; an increased value indicates increased weight due to accumulated excrement or a change in user posture. 3) Gas concentration mean: the average concentration of the target gas; a higher value indicates more gas detected from excrement. 4) Temperature mean: the average temperature. A higher mean indicates a higher overall intensity of the physical / chemical quantity measured by the sensor. An increase in any raw reading within the time window will lead to an increase in the mean. Variance reflects the degree of fluctuation or stability of the sensor signal within a time window. 1) Impedance / pressure variance represents the degree of drastic change in liquid flow or pressure. A stable state has low variance. During discharge events, the impact of liquid or the release of gas will cause drastic fluctuations in the signal, and the variance will increase sharply. Variance is an excellent indicator for detecting the occurrence of events. 2) Gas variance represents the degree of drastic change in gas concentration. Sudden release will cause the variance to increase. The larger the variance, the more unstable the signal is, and the more drastic the change is, which may indicate that an event is occurring. The magnitude of variance depends on the degree of deviation of each raw reading from the mean. The greater the fluctuation of the reading, the greater the variance. It is not directly related to the magnitude of a single reading, but to the "dispersion" of the readings. The mean absolute value of the first-order difference reflects the average level of the instantaneous rate of change of the sensor signal within a time window, indicating the "rapidness" of the signal change. It is particularly sensitive to capturing the edges of the signal—that is, sudden rises or falls. At the moment of excrement discharge, humidity and pressure signals will have a rapid rising edge, which will make the mean absolute value of the first-order difference very large within a time window. The larger it is, the more rapidly the signal changes, and it is often used to detect the starting point of an event. The greater the difference between two adjacent raw readings, the larger the mean absolute value of the first-order difference. It describes the rate of change, not the signal strength itself. Frequency domain energy characteristics reflect the total energy of periodic components generated by specific physiological activities (such as trembling and muscle contraction) in the original signal. Micro-movements of the human body (such as shivering and straining during defecation) have certain frequency characteristics within the range of 0.5-5Hz. By setting this "physiologically relevant frequency band," low-frequency postural changes and high-frequency electronic noise can be filtered out, extracting only the signal components generated by real physiological activities. When a user feels discomfort or strains during defecation, it is accompanied by specific muscle activity, which is reflected in the frequency domain energy of the pressure or impedance signal. The larger the value, the more intense the periodic physiological activity related to defecation during that time period. The larger the amplitude of the periodic fluctuation component in the original signal that conforms to this frequency band, the larger the calculated frequency domain energy characteristics. High-dimensional space is the embodiment of multimodal fusion. Suppose there are 4 types of sensors, each of which extracts 4 features. Then each state point is a 16-dimensional vector. Each dimension of this high-dimensional vector describes the state of the system at that moment from a unique perspective (a feature of a sensor). It is the most comprehensive and digital definition of the excretion state. A single state point is like a "photograph," while a continuous trajectory of state points is like a "movie." The trajectory reveals the continuous evolution of the user's physiological state: normal state, the start of excretion, the continuation of excretion, and the end of excretion. These stages will draw distinctly different paths in this high-dimensional space. By analyzing the shape of the trajectory, such as sudden changes, rapid movements, or entering a loop, events can be accurately identified.

[0020] The dynamic vector analysis module is used to calculate the change vector between state points in a continuous time window on the discharge phase diagram in real time. It performs multi-dimensional similarity matching and weighted scoring between the current change vector and each vector template in the vector pattern library to dynamically predict the type and development stage of the current discharge event. The vector pattern library contains several vector templates that represent different discharge events and development stages. Real-time computing emphasizes the responsiveness of the system, requiring calculations to be completed and prediction results to be output within a very short time after data is generated, in order to meet the timing requirements of control execution. The change vector is the core input of this module. This vector quantifies the dynamic evolution of the system state within a time window and contains all the information about the direction and magnitude of change. The task of the module is to interpret this vector. The vector pattern library is the system's "knowledge base" or "memory." It is built up through extensive prior experiments and data collection. It collects data on various types of excretion events, such as urination, defecation, mixed, small amounts, and large amounts, as well as data on each stage (start, in progress, and end). It calculates the corresponding typical change vectors, labels them with their types and stages, and stores them in this pattern library. Event types include: urine, feces, mixed, sweat, and disturbances such as turning over. Development stages include: start, in progress, and end. Predicting the "start" stage is key to achieving "predictive care," such as locking the legs in advance. The logic for matching analysis performed by the dynamic vector analysis module is as follows: For the current change vector and the vector template in the pattern library, the directional similarity and magnitude similarity are calculated and weighted and fused to obtain a matching score; Includes the following steps: The calculation logic for the directional similarity is as follows: calculate the cosine value of the angle between the current changing vector and the vector template to represent the consistency of their directions, map this cosine value to the interval between 0 and 1 to obtain the normalized directional similarity score, and the specific formula is as follows: For the current change vector vector templates in the pattern library The directional similarity is taken as cosine similarity and mapped to :

[0021] in, For cosine similarity, The normalized directional similarity score; It reflects the degree of proximity of two vectors in direction, and this is a range of... The value; ,mean and The directions are almost identical, indicating that the current change pattern is highly consistent with the templates in the pattern library; This means that the two vectors are nearly orthogonal but have completely different directions; This means that the two vectors are in completely opposite directions; The larger the value, the better the trend and direction of the current state change matches the target template. For example, a template representing "humidity rising sharply" will have a high cosine similarity to a current vector that also displays "humidity rising sharply." The numerator is the dot product; the closer the directions of the two vectors are, the larger the dot product value (assuming a fixed modulus). Also bigger; The reflection will from Linear mapping to The score after the interval refers to converting "similarity" into a more intuitive "score". Originally, -1 (least similar) got 0 points, and 1 (most similar) got 1 point. The larger the value, the more similar the directions; The calculation logic for amplitude similarity is as follows: Calculate the Euclidean distance between the current changing vector and the vector template to characterize the difference in amplitude between the two. Introduce a normalization constant to scale this Euclidean distance, and calculate a normalized distance similarity score between 0 and 1. The specific formula is as follows: Calculate Euclidean distance : Then calculate the normalized distance similarity:

[0022] in, This is a normalization constant used to standardize the Euclidean distance scale. This is the normalized distance similarity score; This reflects the Euclidean distance between the current vector and the template vector. This value is always greater than or equal to 0, indicating the absolute difference between the two vectors in high-dimensional space, taking into account both differences in direction and magnitude. The smaller the value, the closer the two vectors are. The larger the value, the greater the overall difference between the current change vector and the target template; Reflects the Euclidean distance Convert to a The similarity score within the range is a very clever transformation, when hour, (Completely similar), with Increase Approaching 0 (completely dissimilar); A larger value indicates that the magnitude of the current change vector is closer to the magnitude of the template vector. For example, a template representing a drastic change will have a higher value than a current vector representing an equally drastic change. If a change is very slight, even if the direction is correct, It will also be very low; Used to adjust the sensitivity of distance calculation, if It's set to a very large size. It will be very small, making Insensitive to changes in distance; if It's set very small. It will be very large, making They are extremely sensitive to even minute changes in distance. The setting is based on the typical distance between vectors in historical data, with the aim of making It can vary within a discriminative range; Based on preset weighting coefficients, the normalized directional similarity score and the normalized distance similarity score are weighted and summed to obtain the final matching score; the specific formula is as follows: Weighted matching score :

[0023] in, These are the weighting coefficients, and This is used to adjust the weight of directional similarity and amplitude similarity in the total score; Reflects the current change vector With template vector The overall similarity score refers to the final quantitative evaluation of the matching degree from two dimensions: direction and magnitude. The higher, the more it means The more it resembles a vector template, the larger the value, indicating that the current physiological event is more likely to be the type and stage labeled by the vector template; This means that the system trusts the direction of change more because direction is more distinguishable between event types. For example, the initial gas response directions of urine and feces are different, while the amplitude is easily affected by individual differences or excretion volume. This means the system places more faith in the magnitude of change, which is used to distinguish the intensity level of events, such as whether there are many or few; the specific values ​​of the weights need to be determined through training and optimization with a large amount of experimental data to obtain the highest classification accuracy. Matching scores The type and development stage of the excretion event labeled by the highest vector template are used as the prediction result output at the current moment. This is the final decision mechanism. The system selects the "label" of the template that best matches as the current prediction result. The advantage of this method is that it is intuitive and computationally efficient. Its core idea is that any excretion event will draw a unique trajectory in the high-dimensional phase diagram, and the tangent (i.e. the change vector) at each point on the trajectory will also have characteristic direction and amplitude.

[0024] The nursing strategy decision-making module is used to generate leg locking control instructions when the start of an excretion event is predicted, and to generate hip lifting control instructions when the end of an excretion event is predicted. It also calculates a comprehensive excretion warning index based on the average intensity of all change vectors during the start to end of the excretion event, and generates categorized nursing action control instructions based on the high, medium and low ranges of this index. Prediction means that the system does not act only after detecting excrement, but rather issues instructions in advance when the dynamic vector analysis module judges that an event is about to begin or has just started, such as detecting a specific muscle pressure pattern or the initial release of gas. This achieves predictive care, with the aim of physical restraint before excrement causes large-scale pollution. The leg locking control command controls the actuators on the nursing pants, such as the airbag driven by a miniature air pump, to slightly tighten the pants at the root of the thighs, forming a "dam" to effectively prevent liquid from spreading to the legs and control the contamination to the core area of ​​the crotch. This is a key step in reducing the amount of subsequent cleaning work and improving the user experience. The buttock lift control command activates the actuators beneath the buttocks, such as a lifting platform or airbag, to gently lift the user's buttocks, separating them from the source of contamination (the lining of the nursing pants). This action serves two core purposes: 1) Avoid soaking to prevent prolonged skin contact with excrement; 2) Create operating space to provide the necessary mechanical operating space for subsequent cleaning, vacuuming, drying and other care actions, ensuring that these actions can be performed effectively; The average intensity of the change vector clarifies the calculation One of the core inputs is an indicator that reflects the severity of an event; Comprehensive excretion early warning index It is the core of the entire module's output. It is not a simple physical quantity, but a comprehensive evaluation index that integrates multi-dimensional information. Its purpose is to use a number to quantify the challenge that this excretion event poses to the nursing system, thereby determining how much resources (water, time, energy) need to be used to deal with it. The operation of the nursing strategy decision-making module specifically includes: When the prediction result output by the dynamic vector analysis module indicates the start of an excretion event, a leg locking control command is immediately generated; when the prediction result indicates the end of an excretion event, a hip lifting control command is immediately generated. After the discharge event ends, a comprehensive discharge early warning index is calculated based on the average intensity of all change vectors during the discharge event from start to finish. The calculation logic of the discharge early warning index is as follows: Calculate the average value of the magnitudes of all change vectors throughout the entire discharge event, which is used to characterize the cumulative intensity of the event, and record the total duration of the event; calculate the variance of the magnitudes of all change vectors, which is used to characterize the volatility or disorder of the event process; finally, multiply the average value, total duration, and variance by preset normalized weighting coefficients and sum them up. The result is the discharge early warning index. The calculation formula is as follows:

[0025] in, To discharge the early warning index, This represents the average magnitude of all change vectors throughout the entire event, used to characterize the cumulative intensity of the event. This represents the number of change vectors contained in the event. The total duration of the event. The variance of the magnitudes of all change vectors is used to characterize the disorder of the event process. These are the normalized weighting coefficients, and ; It reflects the overall severity level of this excretion incident, and its absolute value is only meaningful when compared with a preset threshold. The higher the number, the more "troublesome" the incident, requiring a more thorough and energy-intensive nursing process to handle. A higher numerical value indicates a higher overall severity of the event; Reflects the average drastic degree of state change throughout the entire event. A high value indicates a strong impact force and high speed during excretion, or physical properties of the excrement, such as viscosity, leading to drastic signal changes. This value is positively correlated with the amount and speed of excretion. The larger the value, the greater the average intensity of the event; the magnitude of each change vector. The larger they are, the higher their average value. The larger it is; It reflects the total time from the beginning to the end of the event. A longer duration means that the excretion process is intermittent or the total amount of excretion is large. A longer exposure time means that the skin is irritated for a longer period of time and requires a longer washing time to ensure cleanliness. The larger the value, the longer the event lasts; High variance reflects the degree of fluctuation or instability during an event. This is a mathematical manifestation of high "chaos," meaning that the excretion process is not smooth but intermittent; other activities of the user, such as turning over or coughing, are mixed in during this period, which interfere with the excretion signal itself; the excrement itself is complex in nature, such as a mixture of solid and liquid, which leads to irregular signal changes. These three coefficients determine the average intensity Duration and Chaos The relative importance of these three factors in the final evaluation needs to be optimized and adjusted through extensive experimental data and feedback on care effects. For example, if it is found that the cleaning effect mainly depends on the intensity, then the intensity should be increased. If prolonged soaking is found to be the main problem, then the dosage should be increased. If a chaotic process often leads to incomplete cleaning, then the cleaning intensity should be increased. ; Table 1 shows the specific data for some event numbers and discharge warning indices.

[0026] Table 1

[0027] Data analysis reveals a clear synergistic relationship among multiple characteristic parameters of excretion events. The data shows a significant positive correlation between the average magnitude of the change vector, event duration, and magnitude variance, and the excretion warning index. As the event number increases from 1 to 15, the average magnitude of the change vector gradually rises from 0.12 to 2.05, the event duration increases from 7 seconds to 48 seconds, and the magnitude variance also increases from 0.07 to 1.08. Correspondingly, the excretion warning index significantly increases from 2.60 to 31.52. This indicates that when the excretion event is more intense, longer-lasting, and exhibits more dramatic fluctuations, the system determines the event to be of higher severity, requiring a more robust nursing response. When analyzing the intrinsic relationships between different characteristic parameters, events with longer durations are often accompanied by larger mean and variance of modulus. For example, event number 14 lasted for 48 seconds, with a mean modulus of 2.05 and a variance of 1.08, all three parameters being at relatively high levels. Conversely, event number 1 lasted only 7 seconds, with its mean and variance of modulus remaining at relatively low levels. This coordinated change among parameters indicates that, in actual excretion processes, events with greater intensity tend to require longer durations, and their changes are more unstable. This reflects the combined influence of different excrement properties and excretion intensity on the sensing signal.

[0028] The nursing strategy decision-making module has a preset index threshold of high index threshold. and low exponential threshold and discharge warning index In comparison: When the excretion warning index is greater than the high index threshold, that is When the risk level is high, a high-intensity nursing instruction is generated; when the excretion warning index is between the low index threshold and the high index threshold, i.e. When the risk level is determined to be medium, a standard intensity nursing instruction is generated; when the excretion warning index is less than the low index threshold, i.e. When the risk level is determined to be low, an energy-saving intensity care instruction is generated; and These are system-defined decision boundaries, and they will be continuous. The values ​​are divided into three discrete care level intervals; these two thresholds are determined based on a combination of historical data, expert knowledge (caregiver experience), and product design goals, such as energy efficiency. It can be set to handle 95% of excretion events. value; Nursing instructions are high-level strategy commands rather than specific execution parameters. For example, a high-intensity nursing instruction contains the following information: {cleaning mode: strong, water consumption: high, drying temperature: high, duration: long}. These strategy instructions are sent to the final dynamic optimization execution module, which then translates them into specific, executable low-level control signals such as motor speed, water pump power, and heater temperature.

[0029] The dynamic optimization execution module is used to receive nursing action control instructions and adjust the execution parameters of cleaning, vacuuming, drying and disinfection nursing actions according to the excretion warning index of this excretion event. This module receives nursing action control commands. It receives high-level, categorized commands from the upper-level "Nursing Strategy Decision Module", such as "High-Intensity Nursing Command", which determines the overall intensity level of nursing care. The module's responsibility is to translate and refine these high-level strategy commands into specific, quantifiable control signals that drive the underlying actuators (pumps, motors, heaters, sprayers). Regarding the cleaning process, the excretion warning index... The higher the value, the higher the cleaning water temperature, the greater the water pressure, and the longer the cleaning time set in the instruction; Input: This is a comprehensive indicator that represents the overall severity of the incident. The output includes three parameters of the cleaning action: water temperature, water pressure, and duration. High → Indicates a serious incident (large volume, chaotic, prolonged) → necessitates a more thorough cleaning. Higher water temperatures better dissolve grease and kill bacteria, improving cleaning effectiveness and hygiene. Greater water pressure generates a more powerful water flow to wash away even the most stubborn fecal residue. A longer duration ensures sufficient time for the washing process to fully cover all contaminated areas. The cleaning strategy is based on a comprehensive assessment of the incident. Perform a global strength adjustment; For the suction action: the larger the impedance sensor characteristic value in the final state point coordinates, the greater the suction intensity and the longer the duration set in the command; input: the impedance sensor characteristic value (mean) of the final state point, which directly reflects the total amount of residual liquid in the nursing pants lining after the event ends; output: two parameters of the suction action: suction intensity (vacuum pump power) and duration. A high electrical impedance value means more residual liquid and higher humidity, requiring stronger and longer suction. Greater suction intensity increases instantaneous suction power, quickly removing most of the free liquid. Longer duration ensures that liquid embedded deeper in the textiles can be extracted, achieving a drier level and saving time and energy for subsequent drying. The suction strategy is precisely adjusted based on the final result of the contamination (liquid residue). For drying and disinfection operations: the larger the characteristic value of the gas sensor in the final state point coordinates, the higher the drying temperature, the larger the air volume, and the more disinfectant is sprayed in the command. Input: Gas sensor characteristic value (mean) at the final state point, which directly reflects the concentration of residual odor gases (ammonia, hydrogen sulfide) in the inner space of the nursing pants after the event, and this concentration is a direct indicator of the degree of fecal contamination; Output: Parameters of the drying action (temperature, air volume) and parameters of the disinfection action (disinfectant spray volume). High gas characteristic values ​​indicate severe fecal contamination, high levels of organic residue, and a high risk of bacterial growth, necessitating more thorough drying and stronger disinfection. Higher drying temperatures and larger air volumes effectively kill microorganisms, while large air volumes quickly remove moisture and residual gases, ensuring complete dryness of skin contact surfaces and preventing eczema and bedsores. Increased disinfectant spray volume enhances disinfection of severely contaminated areas, ensuring hygiene and safety.

[0030] Please see Figure 7 The present invention also provides a method for monitoring nursing pants based on multi-sensor fusion, the method being used to execute the above-described nursing pants monitoring system based on multi-sensor fusion, comprising: Step 1: Multiple sensors are integrated inside the nursing pants body, including a distributed impedance sensing array, a temperature sensor, a gas sensor, and a distributed micro-piezoresistive array to collect multimodal sensing data of the internal environment of the nursing pants. Step 2: Divide the multimodal sensing data into time windows of a preset length, extract the feature values ​​of the multimodal sensing data in each time window, and construct a discharge phase diagram with the time window sequence as the horizontal axis and the feature values ​​as the vertical axis, where each time window corresponds to a state point on the phase diagram; Step 3: Calculate the change vector between state points in a continuous time window on the excretion phase diagram in real time, perform multi-dimensional similarity matching and weighted scoring between the current change vector and each vector template in the vector pattern library, and dynamically predict the type and development stage of the current excretion event. The vector pattern library contains several vector templates that represent different excretion events and development stages. Step 4: Generate a leg locking control command when the excretion event is predicted to begin, and a hip lifting control command when the excretion event is predicted to end. Calculate a comprehensive excretion warning index based on the average intensity of all change vectors during the excretion event from start to end, and generate categorized nursing action control commands based on the high, medium, and low ranges of this index. Step 5: Receive nursing action control instructions and adjust the execution parameters of cleaning, vacuuming, drying and disinfection nursing actions according to the excretion warning index of this excretion event.

[0031] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0032] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0033] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0034] 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 multi-sensor fusion based care pant monitoring system, characterized in that, Specifically comprising: A multi-modal sensor module integrated inside the nursing pants body, including a distributed impedance sensor array, a temperature sensor, a gas sensor, and a distributed micro-piezoresistance array, for collecting multi-modal sensor data of the internal environment of the nursing pants; A discharge phase diagram construction module for dividing the multi-modal sensor data into time windows of a preset length, extracting feature values of the multi-modal sensor data in each time window, and constructing a discharge phase diagram with time window sequences as the horizontal axis and feature values as the vertical axis, wherein each time window corresponds to a state point on the phase diagram; A dynamic vector analysis module for real-time calculation of change vectors between consecutive time window state points on the discharge phase diagram, multi-dimensional similarity matching and weighted scoring of the current change vector with each vector template in the vector pattern library, dynamic prediction of the type and development stage of the current discharge event, and the vector pattern library containing a plurality of vector templates representing different discharge events and development stages; A nursing strategy decision module for generating leg locking control instructions when predicting the start of a discharge event, generating hip lifting control instructions when predicting the end of a discharge event, and calculating a comprehensive discharge warning index based on the average intensity of all change vectors during the start and end of the current discharge event, and generating classified nursing action control instructions based on the high, medium, and low intervals of the index; A dynamic optimization execution module for receiving nursing action control instructions and adjusting the execution parameters of cleaning, sewage absorption, drying, and disinfection nursing actions according to the discharge warning index of the current discharge event.

2. The multi-sensor fusion based monitoring system for care pants of claim 1, wherein: The multi-modal sensor module includes: A distributed impedance-capacitance sensor array for synchronous measurement of resistance and capacitance values of different regions of the nursing pants lining; A temperature sensor installed adjacent to the nodes of the distributed impedance-capacitance sensor array for monitoring and outputting temperature measurements at the node locations; A gas sensor encapsulated in a protective layer with a hydrophobic and breathable membrane for quantitative detection and output of ammonia and hydrogen sulfide concentration values in the internal environment of the nursing pants; A distributed micro-piezoresistance array for synchronous measurement and output of pressure values and physical coordinates of each sensing unit; The distributed impedance-capacitance sensor array and the distributed micro-piezoresistance array are embedded in the crotch and hip linings of the nursing pants in a grid pattern; the temperature sensor and the gas sensor are placed in the sensing cluster at the center of the crotch.

3. The multi-sensor fusion based monitoring system for care pants of claim 2, wherein: For each time window, the time-domain statistical features and frequency-domain energy features of each sensor data are calculated respectively, where the time-domain statistical features include mean, variance, and first-order difference absolute value mean; The frequency-domain energy features are obtained by integrating the signal energy in the preset physiological frequency band after fast Fourier transform of the sensor data; All feature values extracted from all sensors are combined into a multi-dimensional vector, which is the coordinate of the state point in the high-dimensional space of the discharge phase diagram for the corresponding time window. In time sequence, the multi-dimensional vector corresponding to each time window is sequentially depicted in a high-dimensional state space with time window index as the horizontal axis and each dimension of the vector as the vertical axis according to the time sequence of the vector, thereby forming a trajectory composed of continuous state points, which is a voiding phase diagram representing the continuous change of the physiological state of the user; For any two consecutive state points on the voiding phase diagram, the difference between the latter state point and the former state point is calculated, and the difference constitutes a change vector, which contains the direction and amplitude information of the state change from the previous time window to the current time window.

4. The multi-sensor fusion based monitoring system for care pants of claim 3, wherein: The specific logic of the dynamic vector analysis module for matching analysis is as follows: For the current change vector and the vector template in the pattern library, the direction similarity and amplitude similarity are calculated and weighted to obtain a matching score; The calculation logic of the direction similarity is as follows: the cosine value of the angle between the current change vector and the vector template is calculated to represent the consistency of the two directions, and the cosine value is mapped to the interval of 0 to 1 to obtain the normalized direction similarity score; The calculation logic of the amplitude similarity is as follows: the Euclidean distance between the current change vector and the vector template is calculated to represent the difference in amplitude between the two, and a normalization constant is introduced to scale the Euclidean distance, and a normalized distance similarity score between 0 and 1 is calculated accordingly; According to the preset weight coefficient, the normalized direction similarity score and the normalized distance similarity score are weighted and summed to obtain the final matching score; The voiding event type and development stage labeled by the vector template with the highest matching score are output as the prediction result of the current change vector.

5. The multi-sensor fusion based monitoring system for care pants of claim 4, wherein: The operation of the nursing strategy decision module specifically includes: When the prediction result output by the dynamic vector analysis module is the start of the voiding event, a leg locking control instruction is immediately generated; when the prediction result is the end of the voiding event, a hip lifting control instruction is immediately generated; After the voiding event ends, a comprehensive voiding warning index is calculated based on the average intensity of all change vectors in the process from the start to the end of the voiding event; The calculation logic of the voiding warning index is as follows: the average value of the lengths of all change vectors in the entire voiding event is calculated to represent the cumulative intensity of the event, and the total duration of the event is recorded; the variance of the lengths of all change vectors is calculated to represent the volatility or chaos degree of the event process, and finally the average value, total duration and variance are multiplied by the preset normalization weight coefficients respectively and summed, and the result is the voiding warning index.

6. The multi-sensor fusion based monitoring system for care pants of claim 5, wherein: The nursing strategy decision module has a high index threshold and a low index threshold, and compares the calculated voiding warning index with the thresholds: When the voiding warning index is greater than the high index threshold, it is determined to be a high-risk level, and a high-intensity nursing instruction is generated accordingly; when the voiding warning index is between the low index threshold and the high index threshold, it is determined to be a medium-risk level, and a standard-intensity nursing instruction is generated accordingly; when the voiding warning index is less than the low index threshold, it is determined to be a low-risk level, and an energy-saving-intensity nursing instruction is generated accordingly; For the cleaning action, the higher the excretion warning index, the higher the water temperature, the greater the water pressure, and the longer the cleaning time set in the instruction; for the suction action: the larger the impedance sensing characteristic value in the final state point coordinate, the greater the suction intensity and the longer the duration set in the instruction; for the drying and disinfection action: the larger the gas sensor characteristic value in the final state point coordinate, the higher the drying temperature, the greater the air volume, and the more disinfectant injection set in the instruction.

7. A multi-sensor fusion based care pant monitoring method, characterized by: The method is used to execute a multi-sensor fusion-based care pant monitoring system according to any one of claims 1-6, comprising: Step 1: multi-sensor integration in the care pant body, including a distributed impedance sensing array, a temperature sensor, a gas sensor, and a distributed micro-piezoresistive array to collect multi-modal sensing data of the internal environment of the care pant; Step 2: divide the multi-modal sensing data into time windows of a preset length, extract the characteristic values of the multi-modal sensing data in each time window, and construct a defecation phase diagram with time window sequences as the horizontal axis and characteristic values as the vertical axis, wherein each time window corresponds to a state point on the phase diagram; Step 3: real-time calculation of the change vector between consecutive time window state points on the defecation phase diagram, multi-dimensional similarity matching and weighted scoring of the current change vector with each vector template in the vector pattern library, dynamic prediction of the type and development stage of the current defecation event, the vector pattern library containing a plurality of vector templates representing different defecation events and development stages; Step 4: generating leg locking control instructions when predicting the start of the defecation event, generating hip lifting control instructions when predicting the end of the defecation event, and calculating a comprehensive excretion warning index based on the average intensity of all change vectors during the start and end of the current defecation event, generating classified care action control instructions based on the high, medium, and low intervals of the index; Step 5: receiving care action control instructions, adjusting the execution parameters of cleaning, suction, drying, and disinfection care actions according to the excretion warning index of the current defecation event.

Images