Intelligent alarm paper diaper based on AI technology

Abstract

The invention relates to an intelligent alarm paper diaper based on an AI technology, and relates to the technical field of paper diapers, the intelligent alarm paper diaper comprises a paper diaper main body comprising an outer layer and an absorption layer; the data acquisition assembly at least comprises a plurality of temperature sensors, a plurality of humidity sensors and a plurality of piezoelectric sensors which are arranged at a plurality of positions of the absorption layer; the data processing assembly and the wireless communication assembly are detachably arranged on the paper diaper body, the data processing assembly is detachably and electrically connected with the data acquisition assembly, and the data processing assembly is used for receiving sensing data acquired by the data acquisition assembly and sending the sensing data to the wireless communication assembly. The sensing data collected by the data collection assembly is uploaded to the cloud platform through the wireless communication assembly, and the cloud platform is used for determining the state information of the paper diaper based on the sensing data collected by the data collection assembly through the AI technology, generating alarm prompt information based on the state information of the paper diaper and sending the alarm prompt information to the guardian terminal. The intelligent level of the paper diaper is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of alarm paper, and in particular to a smart alarm diaper based on AI technology. Background Art

[0002] Diapers are nursing products designed for infants, middle-aged and elderly patients with urinary incontinence and paralysis. This product is easy to use and change, which reduces the burden on nursing staff. Diapers mainly include an outer layer and an absorbent layer. The outer layer is a non-permeable layer, while the absorbent layer can absorb liquids such as urine and body fluids. Diapers are made of non-woven fabrics, toilet paper, fluff pulp, polymer absorbent resin, PE film, rubber bands and other materials. There are two types of diapers: one for infants and one for adults.

[0003] Currently, most diapers on the market have poor leakage prevention effects and are very inconvenient to use. For example, when used for babies, guardians are required to open the diapers from time to time to understand the usage of the diapers. Guardians are also unable to understand the relevant situation and cannot implement monitoring reminders.

[0004] Therefore, a smart alarm diaper based on AI technology is used to improve the intelligence level of diapers. Summary of the invention

[0005] The present invention provides an intelligent alarm diaper based on AI technology, comprising: a diaper main body, wherein the diaper main body includes an outer layer and an absorption layer; a data acquisition component arranged on the absorption layer, wherein the data acquisition component at least includes multiple temperature sensors, multiple humidity sensors and multiple piezoelectric sensors arranged at multiple positions of the absorption layer; a data processing component and a wireless communication component detachably arranged on the diaper main body, wherein the data processing component is detachably electrically connected to the data acquisition component, the data processing component is used to receive the sensor data collected by the data acquisition component, and upload the sensor data collected by the data acquisition component to a cloud platform through the wireless communication component, the cloud platform is used to determine the diaper status information based on the sensor data collected by the data acquisition component through AI technology, and generate alarm prompt information based on the diaper status information, and send the alarm prompt information to the guardian terminal.

[0006] Furthermore, the cloud platform is also used to: obtain multiple groups of test data, wherein the test data include temperature data collected by the multiple temperature sensors at multiple test time points, humidity data collected by the multiple humidity sensors at multiple test time points, and pressure data collected by the multiple piezoelectric sensors at multiple test time points; based on the multiple groups of test data, establish an association relationship map, wherein the association relationship map is used to record the association relationship between any two of the temperature sensors, the association relationship between any two of the humidity sensors, the association relationship between any two of the piezoelectric sensors, the association relationship between any temperature sensor and any humidity sensor, the association relationship between any temperature sensor and any piezoelectric sensor, and the association relationship between any humidity sensor and any piezoelectric sensor.

[0007] Furthermore, the cloud platform is also used to: establish an AI model; establish multiple training samples based on the multiple groups of test data, wherein the labels of the training samples include diaper status information corresponding to the test data, wherein the diaper status information at least includes the type of excrement, the saturation of the absorption layer and the wetting area; train the AI ​​model through the multiple training samples and the association relationship map to obtain a trained AI model.

[0008] Furthermore, the cloud platform determines the diaper status information based on the sensor data collected by the data acquisition component through AI technology, including: based on the association relationship map, correcting the temperature data collected by the multiple temperature sensors at multiple time points, the humidity data collected by the multiple humidity sensors at multiple time points, and the pressure data collected by the multiple piezoelectric sensors at multiple time points to obtain the corrected temperature data, humidity data and pressure data of multiple positions of the absorption layer at multiple time points; determining the diaper status information based on the trained AI model based on the corrected temperature data, humidity data and pressure data of multiple positions of the absorption layer at multiple time points.

[0009] Furthermore, the cloud platform corrects the temperature data collected by the multiple temperature sensors at multiple time points, the humidity data collected by the multiple humidity sensors at multiple time points, and the pressure data collected by the multiple piezoelectric sensors at multiple time points based on the association relationship map, including: for each of the temperature sensors, based on the temperature data collected by the temperature sensor at multiple time points, determining the size of the mean filter window corresponding to the temperature sensor, and based on the mean filter window corresponding to the temperature sensor, performing a first correction on the temperature data collected by the temperature sensor at multiple time points to generate the temperature data of multiple time points corresponding to the temperature sensor after the first correction; for each of the humidity sensors, based on the humidity data collected by the humidity sensor at multiple time points, determining the size of the mean filter window corresponding to the humidity sensor, and based on the mean filter window corresponding to the humidity sensor, performing a first correction on the humidity data collected by the humidity sensor at multiple time points Perform a first correction to generate humidity data at multiple time points after the first correction corresponding to the humidity sensor; for each of the piezoelectric sensors, determine the size of the mean filter window corresponding to the piezoelectric sensor based on the pressure data collected by the piezoelectric sensor at multiple time points, and perform a first correction on the pressure data collected by the piezoelectric sensor at multiple time points based on the mean filter window corresponding to the piezoelectric sensor to generate pressure data at multiple time points after the first correction corresponding to the piezoelectric sensor; through a multi-correction model, based on the association relationship map, perform a second correction on the temperature data at multiple time points after the first correction corresponding to each temperature sensor, the humidity data at multiple time points after the first correction corresponding to each humidity sensor, and the pressure data at multiple time points after the first correction corresponding to each piezoelectric sensor, to obtain the corrected temperature data, humidity data, and pressure data at multiple positions of the absorption layer at multiple time points.

[0010] Furthermore, the cloud platform determines the size of the mean filter window corresponding to the temperature sensor, including: calculating the overall temperature variance corresponding to the temperature sensor based on the temperature data collected by the temperature sensor at multiple time points; and determining the size of the mean filter window corresponding to the temperature sensor based on the overall temperature variance corresponding to the temperature sensor.

[0011] Furthermore, the cloud platform performs a first correction on the temperature data collected by the temperature sensor at multiple time points based on the mean filter window corresponding to the temperature sensor, including: dividing the temperature data collected by the temperature sensor at multiple time points into multiple temperature data units based on the mean filter window corresponding to the temperature sensor; for each of the temperature data units, calculating the unit temperature variance of the temperature data unit based on the temperature data at multiple time points included in the temperature data unit, and judging whether to perform the first correction on the temperature data unit based on the unit temperature variance of the temperature data unit; if so, calculating the unit temperature mean of the temperature data unit based on the temperature data at multiple time points included in the temperature data unit, and performing the first correction on the temperature data unit based on the unit temperature mean of the temperature data unit.

[0012] Furthermore, the cloud platform uses a multi-correction model and, based on the association relationship map, performs a second correction on the temperature data of multiple time points after the first correction corresponding to each temperature sensor, the humidity data of multiple time points after the first correction corresponding to each humidity sensor, and the pressure data of multiple time points after the first correction corresponding to each piezoelectric sensor, to obtain the corrected temperature data, humidity data, and pressure data of multiple positions of the absorption layer at multiple time points, including: for each temperature sensor, performing variational modal decomposition on the temperature data of multiple time points after the first correction corresponding to the temperature sensor to generate the temperature modal function and temperature residual corresponding to the temperature sensor; for each humidity sensor, performing variational modal decomposition on the humidity data of multiple time points after the first correction corresponding to the humidity sensor to generate the humidity modal function and temperature residual corresponding to the humidity sensor. number and humidity residual; for each of the piezoelectric sensors, the pressure data at multiple time points after the first correction corresponding to the piezoelectric sensor are subjected to variational mode decomposition to generate the pressure mode function and pressure residual corresponding to the piezoelectric sensor; through the multi-correction model, based on the correlation relationship map, the temperature mode function and temperature residual corresponding to each of the temperature sensors, the humidity mode function and humidity residual corresponding to each of the humidity sensors, and the pressure mode function and pressure residual corresponding to each of the piezoelectric sensors, the temperature data at multiple time points after the first correction corresponding to each of the temperature sensors, the humidity data at multiple time points after the first correction corresponding to each of the humidity sensors, and the pressure data at multiple time points after the first correction corresponding to each of the piezoelectric sensors are subjected to a second correction to obtain the corrected temperature data, humidity data and pressure data at multiple positions of the absorption layer at multiple time points.

[0013] Furthermore, the cloud platform is also used to determine the user's excretion health assessment information based on diaper status information at multiple time points based on AI technology.

[0014] Furthermore, the cloud platform is also used to determine the user's bedsore risk factor based on diaper status information at multiple time points based on AI technology.

[0015] Compared with the prior art, the smart alarm diaper based on AI technology provided in this specification has at least the following beneficial effects:

[0016] 1. Through the arrangement of multiple temperature sensors, humidity sensors and piezoelectric sensors at different positions of the absorption layer, the humidity, temperature and pressure changes of diapers can be monitored in real time. The cloud platform can quickly analyze and generate alarm prompt information through AI technology. Once the need to change diapers is detected, the guardian will be notified in time to effectively prevent skin problems such as diaper rash and bedsores, and timely detect the user's excretion needs. For babies who cannot express their needs independently or elderly people who need special care, this smart diaper can greatly reduce the burden on guardians and improve the efficiency and quality of care. Guardians do not need to pay attention to the status of diapers at all times to ensure the comfort and health of users.

[0017] 2. The first correction of sensor data through the mean filter window can effectively reduce the noise data caused by factors such as sensor errors and environmental interference, and improve the accuracy and reliability of the data. The multi-sensor correction model combines the data of multiple sensors and conducts a comprehensive analysis through the correlation map, which can more comprehensively reflect the actual state of the diaper's absorbent layer and further improve the accuracy of the data. The size of the mean filter window can be dynamically adjusted according to the data characteristics of different sensors at different time points, so that the system can adapt to different usage scenarios and conditions. When a sensor fails or the data is abnormal, the system can supplement and correct it through the data of other sensors, reducing the impact of a single sensor failure on the overall system and enhancing the system's fault tolerance. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0019] Figure 1 It is a module diagram of a smart alarm diaper based on AI technology shown in an embodiment of the present application;

[0020] Figure 2 It is a schematic diagram of an association relationship map shown in an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly introduces the drawings required for describing the embodiments.

[0022] Figure 1 is a module diagram of a smart alarm diaper based on AI technology shown in an embodiment of the present application, such as Figure 1 As shown, a smart alarm diaper based on AI technology may include a diaper body, a data acquisition component, a data processing component and a wireless communication component.

[0023] The diaper body includes an outer layer and an absorption layer. The data acquisition component is arranged on the absorption layer, wherein the data acquisition component at least includes a plurality of temperature sensors, a plurality of humidity sensors and a plurality of piezoelectric sensors arranged at a plurality of positions of the absorption layer. The data processing component and the wireless communication component are detachably arranged on the diaper body, wherein the data processing component is detachably electrically connected to the data acquisition component.

[0024] It can be understood that when the diaper needs to be changed, the data processing component and the wireless communication component can be removed from the diaper body, and then the diaper is changed, and then the data processing component and the wireless communication component are installed on the new diaper, and the data processing component is electrically connected to the data acquisition component on the new diaper.

[0025] The data processing component is used to receive the sensor data collected by the data acquisition component, and upload the sensor data collected by the data acquisition component to the cloud platform through the wireless communication component. The cloud platform is used to determine the diaper status information based on the sensor data collected by the data acquisition component through AI technology, and generate alarm prompt information based on the diaper status information, and send the alarm prompt information to the guardian terminal.

[0026] In some embodiments, the cloud platform is also used to:

[0027] Acquire multiple sets of test data, wherein the test data include temperature data collected by multiple temperature sensors at multiple test time points, humidity data collected by multiple humidity sensors at multiple test time points, and pressure data collected by multiple piezoelectric sensors at multiple test time points, the multiple sets of test data are acquired by multiple test operations, and the multiple test operations are used to simulate various excretion situations of the user;

[0028] Based on multiple groups of test data, an association relationship map is established, wherein the association relationship map is used to record the association relationship between any two temperature sensors, the association relationship between any two humidity sensors, the association relationship between any two piezoelectric sensors, the association relationship between any temperature sensor and any humidity sensor, the association relationship between any temperature sensor and any piezoelectric sensor, and the association relationship between any humidity sensor and any piezoelectric sensor.

[0029] Specifically, the correlation coefficient between any two temperature sensors, the correlation coefficient between any two humidity sensors, the correlation coefficient between any two piezoelectric sensors, the correlation coefficient between any temperature sensor and any humidity sensor, the correlation coefficient between any temperature sensor and any piezoelectric sensor, and the correlation coefficient between any humidity sensor and any piezoelectric sensor can be calculated.

[0030] Preferably, the correlation coefficient of any two temperature sensors can be calculated based on the following formula:

[0031]

[0032] Among them, r (i,j) is the correlation coefficient between the i-th temperature sensor and the j-th temperature sensor, is the temperature sensor at the nth 1 The temperature data collected at each time point, T (i,average) is the mean value of the temperature data collected by the ith temperature sensor at multiple test time points, N 1 is the total number of test time points, The jth temperature sensor is at the nth 1 The temperature data collected at each time point, T (j,average) is the average value of the temperature data collected by the jth temperature sensor at multiple test time points.

[0033] The calculation method of the correlation coefficient between any two temperature sensors, the correlation coefficient between any two piezoelectric sensors, the correlation coefficient between any temperature sensor and any humidity sensor, the correlation coefficient between any temperature sensor and any piezoelectric sensor, and the correlation coefficient between any humidity sensor and any piezoelectric sensor is similar to the calculation method of the correlation coefficient between any two temperature sensors, and will not be repeated here.

[0034] When the correlation coefficient of two temperature sensors is greater than the temperature correlation coefficient threshold, the two temperature sensors are associated. When the correlation coefficient of two humidity sensors is greater than the temperature correlation coefficient threshold, the two humidity sensors are associated. When the correlation coefficient of two piezoelectric sensors is greater than the piezoelectric correlation coefficient threshold, the two piezoelectric sensors are associated. When the correlation coefficient of a temperature sensor and a humidity sensor is greater than the temperature-humidity correlation coefficient threshold, the temperature sensor and the humidity sensor are associated. When the correlation coefficient of a temperature sensor and a piezoelectric sensor is greater than the temperature-pressure correlation coefficient threshold, the temperature sensor and the piezoelectric sensor are associated. When the correlation coefficient of a humidity sensor and a piezoelectric sensor is greater than the humidity-pressure correlation coefficient threshold, the humidity sensor and the piezoelectric sensor are associated.

[0035] Figure 2 is a schematic diagram of an association relationship map shown in an embodiment of the present application, such as Figure 2 As shown, the association relationship graph may include three types of nodes. The first type of node represents a temperature sensor, the second type of node represents a humidity sensor, and the third type of sensor represents a piezoelectric sensor. The nodes corresponding to two temperature sensors, two humidity sensors, two piezoelectric sensors, a temperature sensor and a humidity sensor, a temperature sensor and a piezoelectric sensor, and a humidity sensor and a piezoelectric sensor that have an association relationship can be connected by edges.

[0036] In some embodiments, the cloud platform is also used to:

[0037] Establishing an AI model, wherein the AI ​​model may be a Long Short-Term Memory (LSTM) model;

[0038] Based on multiple groups of test data, multiple training samples are established, wherein the labels of the training samples include diaper status information corresponding to the test data, wherein the diaper status information at least includes the type of excrement, the saturation of the absorption layer, and the soaking area;

[0039] The AI ​​model is trained through multiple training samples and association relationship maps to obtain a trained AI model.

[0040] Specifically, the training samples are input into the AI ​​model for training. During the training process, the AI ​​model continuously updates the weights through optimization algorithms such as gradient descent to minimize the loss function. The training process may require a lot of computing resources and time. Regularly monitor indicators such as the loss function value and accuracy during the training process to evaluate the training effect of the AI ​​model. Use the validation set to evaluate the performance of the AI ​​model on unseen data. Evaluation indicators may include accuracy, recall, F1 score, etc. Adjust the AI ​​model structure and training strategy based on the evaluation results. It may be necessary to adjust hyperparameters, add regularization terms, use early stopping methods and other strategies to prevent overfitting and improve AI model performance. Perform multiple iterations of training and evaluation as needed until the AI ​​model achieves satisfactory results.

[0041] Preferably, the loss function used to train the AI ​​model is:

[0042]

[0043] Among them, L is the loss of the AI ​​model, β 11 , β 12 and β 13 are all preset weights, and β 11 , β 12 and β 13 are greater than 0, β 11 +β 12 +β 13 =1, N is the total number of training samples in a training sample group, J is the total number of types of excrement, P (i,j) is a binary indicator (0 or 1). If the true excrement type of the i-th training sample is the j-th excrement type, then P (i,j) =1, otherwise P (i,j) =0,Q (i,j) is the probability that the type of excrement of the i-th training sample predicted by the AI ​​model belongs to the j-th type of excrement, D (i,1) is the absorption layer saturation of the i-th training sample predicted by the AI ​​model, D (i,0) is the true absorption layer saturation of the i-th training sample, S (i,1) is the wetted area of ​​the i-th training sample predicted by the AI ​​model, S (i,0) is the actual wetted area of ​​the i-th training sample.

[0044] In some embodiments, the cloud platform determines the diaper status information based on the sensor data collected by the data collection component through AI technology, including:

[0045] Based on the correlation graph, the temperature data collected by multiple temperature sensors at multiple time points, the humidity data collected by multiple humidity sensors at multiple time points, and the pressure data collected by multiple piezoelectric sensors at multiple time points are corrected to obtain the corrected temperature data, humidity data, and pressure data of multiple positions of the absorption layer at multiple time points;

[0046] The trained AI model determines the diaper status information based on the temperature data, humidity data and pressure data of multiple positions of the corrected absorption layer at multiple time points.

[0047] In some embodiments, the cloud platform corrects the temperature data collected by multiple temperature sensors at multiple time points, the humidity data collected by multiple humidity sensors at multiple time points, and the pressure data collected by multiple piezoelectric sensors at multiple time points based on the association relationship graph, including:

[0048] For each temperature sensor, based on the temperature data collected by the temperature sensor at multiple time points, determine the size of the mean filter window corresponding to the temperature sensor, perform a first correction on the temperature data collected by the temperature sensor at multiple time points based on the mean filter window corresponding to the temperature sensor, and generate temperature data at multiple time points corresponding to the temperature sensor after the first correction;

[0049] For each humidity sensor, based on the humidity data collected by the humidity sensor at multiple time points, determine the size of the mean filter window corresponding to the humidity sensor, perform a first correction on the humidity data collected by the humidity sensor at multiple time points based on the mean filter window corresponding to the humidity sensor, and generate humidity data at multiple time points corresponding to the humidity sensor after the first correction;

[0050] For each piezoelectric sensor, based on the pressure data collected by the piezoelectric sensor at multiple time points, determine the size of the mean filter window corresponding to the piezoelectric sensor, and based on the mean filter window corresponding to the piezoelectric sensor, perform a first correction on the pressure data collected by the piezoelectric sensor at multiple time points to generate pressure data at multiple time points corresponding to the piezoelectric sensor after the first correction;

[0051] Through the multi-repair correction model, based on the association relationship map, the temperature data at multiple time points after the first correction corresponding to each temperature sensor, the humidity data at multiple time points after the first correction corresponding to each humidity sensor, and the pressure data at multiple time points after the first correction corresponding to each piezoelectric sensor are corrected for a second time to obtain the corrected temperature data, humidity data and pressure data at multiple positions of the absorption layer at multiple time points, wherein the multi-repair correction model can be a long short-term memory network (LSTM, Long Short-Term Memory) model.

[0052] In some embodiments, the cloud platform determines the size of the mean filter window corresponding to the temperature sensor, including:

[0053] Based on the temperature data collected by the temperature sensor at multiple time points, the overall temperature variance corresponding to the temperature sensor is calculated;

[0054] Based on the overall temperature variance corresponding to the temperature sensor, a size of a mean filter window corresponding to the temperature sensor is determined.

[0055] Preferably, the size of the mean filter window corresponding to the temperature sensor may be calculated based on the following formula:

[0056]

[0057] Among them, σ (i,temperature) is the overall temperature variance corresponding to the i-th temperature sensor, is the nth temperature sensor after the first correction corresponding to the i-th temperature sensor 2 Temperature data at each time point, N 2 is the total number of time points, N i is the size of the mean filter window corresponding to the i-th temperature sensor, σ 0 is the preset overall temperature variance, N 0 is the size of the preset mean filter window.

[0058] The method of determining the size of the mean filter window corresponding to the humidity sensor and the size of the mean filter window corresponding to the piezoelectric sensor is similar to the method of determining the size of the mean filter window corresponding to the temperature sensor, and will not be repeated here.

[0059] In some embodiments, the cloud platform performs a first correction on the temperature data collected by the temperature sensor at multiple time points based on the mean filter window corresponding to the temperature sensor, including:

[0060] Based on the mean filter window corresponding to the temperature sensor, the temperature data collected by the temperature sensor at multiple time points are divided into multiple temperature data units;

[0061] For each temperature data unit, the unit temperature variance of the temperature data unit is calculated based on the temperature data of multiple time points included in the temperature data unit. Based on the unit temperature variance of the temperature data unit, it is determined whether to perform a first correction on the temperature data unit. If so, the unit temperature mean of the temperature data unit is calculated based on the temperature data of multiple time points included in the temperature data unit. Based on the unit temperature mean of the temperature data unit, the temperature data unit is corrected for the first time.

[0062] Specifically, when the unit temperature variance of the temperature data unit is greater than the unit temperature variance threshold, it is determined that the temperature data unit is to be corrected for the first time, and the unit temperature mean of the temperature data unit is used to replace the temperature values ​​of all time points in the temperature data unit.

[0063] In some embodiments, the cloud platform uses a multi-correction model and based on the association relationship map to perform a second correction on the temperature data at multiple time points after the first correction corresponding to each temperature sensor, the humidity data at multiple time points after the first correction corresponding to each humidity sensor, and the pressure data at multiple time points after the first correction corresponding to each piezoelectric sensor, to obtain the corrected temperature data, humidity data, and pressure data at multiple positions of the absorption layer at multiple time points, including:

[0064] For each temperature sensor, performing variational mode decomposition on the temperature data of multiple time points after the first correction corresponding to the temperature sensor, and generating a temperature mode function and a temperature residual corresponding to the temperature sensor;

[0065] For each humidity sensor, performing variational mode decomposition on the humidity data of multiple time points after the first correction corresponding to the humidity sensor, and generating the humidity mode function and humidity residual corresponding to the humidity sensor;

[0066] For each piezoelectric sensor, performing variational mode decomposition on the pressure data of multiple time points after the first correction corresponding to the piezoelectric sensor to generate a pressure mode function and a pressure residual corresponding to the piezoelectric sensor;

[0067] Through the multi-correction model, based on the correlation relationship map, the temperature modal function and temperature residual corresponding to each temperature sensor, the humidity modal function and humidity residual corresponding to each humidity sensor, and the pressure modal function and pressure residual corresponding to each piezoelectric sensor, the temperature data at multiple time points after the first correction corresponding to each temperature sensor, the humidity data at multiple time points after the first correction corresponding to each humidity sensor, and the pressure data at multiple time points after the first correction corresponding to each piezoelectric sensor are corrected for a second time to obtain the corrected temperature data, humidity data, and pressure data at multiple positions of the absorption layer at multiple time points.

[0068] In some embodiments, the cloud platform is also used to:

[0069] Based on AI technology, the user's excretion health assessment information is determined based on the diaper status information at multiple time points.

[0070] Specifically, a health assessment model can be established on the cloud platform. The health assessment model determines the user's excretion health assessment information based on the diaper status information at multiple time points, wherein the excretion health assessment information may include the evaluation results of the user's excretion frequency, urine volume, etc. The health assessment model can be a long short-term memory network (LSTM, Long Short-Term Memory) model.

[0071] In some embodiments, the cloud platform is also used to:

[0072] Based on AI technology, the user's bedsore risk factor can be determined based on diaper status information at multiple time points.

[0073] Specifically, the cloud platform extracts key features related to the risk of bedsores from diaper status information at multiple time points, such as the frequency of excretion, the changing trend of the saturation of the absorption layer, the size of the wetted area and the duration, etc.

[0074] The cloud platform can build a bedsore risk assessment model based on machine learning or deep learning algorithms. The bedsore risk assessment model can learn patterns in historical data and predict the risk of bedsores for users in the future. The bedsore risk assessment model can be a long short-term memory network (LSTM, Long Short-Term Memory) model. The bedsore risk assessment model determines the user's bedsore risk coefficient based on the diaper status information at multiple time points.

[0075] When the cloud platform determines that diapers need to be changed based on the diaper status information, an alarm prompt message is sent to the guardian terminal to remind the guardian to change diapers. When the cloud platform determines that the user's excretion is abnormal based on the user's excretion health assessment information, an alarm prompt message is sent to the guardian terminal to remind the guardian that the user's excretion is abnormal. When the user's bedsore risk factor is greater than the bedsore risk factor threshold, an alarm prompt message is sent to the guardian terminal to remind the guardian.

[0076] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A smart alarm diaper based on AI technology, characterized in that: include: A diaper body, wherein the diaper body comprises an outer layer and an absorbent layer; A data acquisition component disposed on the absorption layer, wherein the data acquisition component at least includes a plurality of temperature sensors, a plurality of humidity sensors, and a plurality of piezoelectric sensors disposed at a plurality of positions on the absorption layer; A data processing component and a wireless communication component are detachably provided on the diaper body, wherein the data processing component is detachably electrically connected to the data acquisition component, the data processing component is used to receive the sensor data collected by the data acquisition component, and upload the sensor data collected by the data acquisition component to a cloud platform through the wireless communication component, the cloud platform is used to determine the diaper status information based on the sensor data collected by the data acquisition component through AI technology, and generate alarm prompt information based on the diaper status information, and send the alarm prompt information to the guardian terminal.

2. The AI-based smart alarm diaper according to claim 1, characterized in that: The cloud platform is also used for: Acquire multiple groups of test data, wherein the test data include temperature data collected by the multiple temperature sensors at multiple test time points, humidity data collected by the multiple humidity sensors at multiple test time points, and pressure data collected by the multiple piezoelectric sensors at multiple test time points; Based on the multiple groups of test data, an association relationship map is established, wherein the association relationship map is used to record the association relationship between any two of the temperature sensors, the association relationship between any two of the humidity sensors, the association relationship between any two of the piezoelectric sensors, the association relationship between any temperature sensor and any humidity sensor, the association relationship between any temperature sensor and any piezoelectric sensor, and the association relationship between any humidity sensor and any piezoelectric sensor.

3. The AI-based intelligent alarm diaper according to claim 2, characterized in that: The cloud platform is also used for: Building AI models; Based on the multiple groups of test data, multiple training samples are established, wherein the labels of the training samples include diaper status information corresponding to the test data, wherein the diaper status information at least includes the type of excrement, the saturation of the absorption layer, and the soaking area; The AI ​​model is trained using the multiple training samples and the association relationship map to obtain a trained AI model.

4. The AI-based intelligent alarm diaper according to claim 3, characterized in that: The cloud platform determines the diaper status information based on the sensor data collected by the data collection component through AI technology, including: Based on the association relationship map, the temperature data collected by the multiple temperature sensors at multiple time points, the humidity data collected by the multiple humidity sensors at multiple time points, and the pressure data collected by the multiple piezoelectric sensors at multiple time points are corrected to obtain the corrected temperature data, humidity data, and pressure data of multiple positions of the absorption layer at multiple time points; The trained AI model is used to determine the diaper status information based on the temperature data, humidity data and pressure data of multiple positions of the corrected absorption layer at multiple time points.

5. The AI-based intelligent alarm diaper according to claim 4, characterized in that: The cloud platform corrects the temperature data collected by the multiple temperature sensors at multiple time points, the humidity data collected by the multiple humidity sensors at multiple time points, and the pressure data collected by the multiple piezoelectric sensors at multiple time points based on the association relationship map, including: For each of the temperature sensors, based on the temperature data collected by the temperature sensor at multiple time points, determine the size of the mean filter window corresponding to the temperature sensor, and based on the mean filter window corresponding to the temperature sensor, perform a first correction on the temperature data collected by the temperature sensor at multiple time points to generate the temperature data of the multiple time points corresponding to the temperature sensor after the first correction; For each of the humidity sensors, based on the humidity data collected by the humidity sensor at multiple time points, determine the size of the mean filter window corresponding to the humidity sensor, perform a first correction on the humidity data collected by the humidity sensor at multiple time points based on the mean filter window corresponding to the humidity sensor, and generate humidity data at multiple time points corresponding to the humidity sensor after the first correction; For each of the piezoelectric sensors, based on the pressure data collected by the piezoelectric sensor at multiple time points, determine the size of the mean filter window corresponding to the piezoelectric sensor, and based on the mean filter window corresponding to the piezoelectric sensor, perform a first correction on the pressure data collected by the piezoelectric sensor at multiple time points to generate pressure data at multiple time points corresponding to the piezoelectric sensor after the first correction; Through the multi-correction model, based on the association relationship map, a second correction is performed on the temperature data of multiple time points after the first correction corresponding to each temperature sensor, the humidity data of multiple time points after the first correction corresponding to each humidity sensor, and the pressure data of multiple time points after the first correction corresponding to each piezoelectric sensor, to obtain the corrected temperature data, humidity data and pressure data of multiple positions of the absorption layer at multiple time points.

6. The AI-based intelligent alarm diaper according to claim 5, characterized in that: The cloud platform determines the size of the mean filter window corresponding to the temperature sensor, including: Calculating the overall temperature variance corresponding to the temperature sensor based on the temperature data collected by the temperature sensor at multiple time points; Based on the overall temperature variance corresponding to the temperature sensor, a size of a mean filter window corresponding to the temperature sensor is determined.

7. The AI-based intelligent alarm diaper according to claim 5, characterized in that: The cloud platform performs a first correction on the temperature data collected by the temperature sensor at multiple time points based on the mean filter window corresponding to the temperature sensor, including: Based on the mean filter window corresponding to the temperature sensor, the temperature data collected by the temperature sensor at multiple time points are divided into multiple temperature data units; For each of the temperature data units, the unit temperature variance of the temperature data unit is calculated based on the temperature data at multiple time points included in the temperature data unit, and based on the unit temperature variance of the temperature data unit, it is determined whether to perform a first correction on the temperature data unit; if so, the unit temperature mean of the temperature data unit is calculated based on the temperature data at multiple time points included in the temperature data unit, and the temperature data unit is corrected for the first time based on the unit temperature mean of the temperature data unit.

8. The AI-based intelligent alarm diaper according to claim 5, characterized in that: The cloud platform uses a multi-repair correction model and based on the association relationship map to perform a second correction on the temperature data at multiple time points after the first correction corresponding to each temperature sensor, the humidity data at multiple time points after the first correction corresponding to each humidity sensor, and the pressure data at multiple time points after the first correction corresponding to each piezoelectric sensor, to obtain the corrected temperature data, humidity data, and pressure data at multiple positions of the absorption layer at multiple time points, including: For each of the temperature sensors, performing variational mode decomposition on the temperature data of a plurality of time points after the first correction corresponding to the temperature sensor to generate a temperature mode function and a temperature residual corresponding to the temperature sensor; For each of the humidity sensors, performing variational mode decomposition on the humidity data of a plurality of time points after the first correction corresponding to the humidity sensor, to generate a humidity mode function and a humidity residual corresponding to the humidity sensor; For each of the piezoelectric sensors, performing variational modal decomposition on the pressure data of a plurality of time points after the first correction corresponding to the piezoelectric sensor to generate a pressure modal function and a pressure residual corresponding to the piezoelectric sensor; Through the multi-correction model, based on the correlation relationship map, the temperature modal function and temperature residual corresponding to each temperature sensor, the humidity modal function and humidity residual corresponding to each humidity sensor, and the pressure modal function and pressure residual corresponding to each piezoelectric sensor, the temperature data at multiple time points after the first correction corresponding to each temperature sensor, the humidity data at multiple time points after the first correction corresponding to each humidity sensor, and the pressure data at multiple time points after the first correction corresponding to each piezoelectric sensor are corrected for a second time to obtain the corrected temperature data, humidity data, and pressure data at multiple positions of the absorption layer at multiple time points.

9. The AI-based smart alarm diaper according to any one of claims 1 to 8, characterized in that: The cloud platform is also used for: Based on AI technology, the user's excretion health assessment information is determined based on the diaper status information at multiple time points.

10. The AI-based smart alarm diaper according to any one of claims 1 to 8, characterized in that: The cloud platform is also used for: Based on AI technology, the user's bedsore risk factor can be determined based on diaper status information at multiple time points.

Images