An all-weather unmanned infant monitoring method and system based on artificial intelligence

Abstract

An artificial intelligence-based all-weather unmanned infant monitoring method and system includes a spatial environment information recognition module, an infant behavior analysis module, and an infant behavior disposal module. Aiming at the problem of infant monitoring in special circumstances, the present invention uses an intelligent camera and sensor system to collect and identify the infant's living environment and the infant's physiological state, analyzes the infant's behavior according to an artificial intelligence algorithm, and takes corresponding monitoring measures.

Description

Technical Field

[0001] The present invention relates to the field of housekeeping, and in particular to an artificial intelligence-based all-weather unmanned baby monitoring method and system. Background Art

[0002] The all-weather unmanned infant monitoring method is based on artificial intelligence technology and aims to provide parents or caregivers with real-time and reliable infant monitoring. Using computer vision technology, the system can monitor the status of the baby in real time. This may include identifying the baby's posture, expression, and whether it is awake or sleeping. Audio sensors and voice recognition technology are used to monitor the baby's voice. The system can identify crying, laughing, etc. to assess the baby's emotions and needs. Some systems may integrate physiological sensors to monitor the baby's physiological parameters, such as heart rate, breathing, etc. Artificial intelligence-based algorithms and models are used to analyze the collected data. The development of the all-weather unmanned infant monitoring method has relied on the integration of these technologies to provide parents with a more convenient, reliable and real-time infant monitoring experience. The design goal of these systems is to improve the safety of infants and provide parents with more convenience and peace of mind. The present invention has a wide range of applications and can be extended to social applications, bringing good housekeeping benefits. Summary of the invention

[0003] The purpose of the present invention is to provide an artificial intelligence-based all-weather unmanned infant monitoring method and system to solve the problems raised in the above-mentioned background technology.

[0004] To achieve the above-mentioned purpose, a method and system for unmanned all-weather infant monitoring based on artificial intelligence are provided, including a spatial environment information recognition module, an infant behavior analysis module and an infant behavior treatment module; the specific process is described as follows:

[0005] S1. Arrange the first type of sensor network according to the conditions of the infant's living space, collect the environmental information of the infant's living space based on this, propose the infant environment suitability index according to the range value suitable for the infant's living environment, analyze and judge the current environment of the infant, propose a preference optimization algorithm, optimize the infant's environmental information, and automatically and mechanically adjust the current environment based on the analysis and judgment results to adapt to the infant's living needs;

[0006] S2. Under the premise that the environment meets the needs of infants, a second type of sensor network is deployed according to the physiological conditions of infants, a data storage database is constructed, and the physiological data of infants is collected in real time;

[0007] S3. Construct a chained dataset of infant behaviors. Since infants’ language systems are not fully developed, it is necessary to mark the key value matching between infant behaviors and guardians’ responses in the dataset. This dataset is used as training data for infant health status analysis and prediction algorithms.

[0008] S4. Based on infant physiological monitoring data, propose an infant health status analysis and prediction algorithm, build an infant behavior status analysis model, and analyze the current infant's expression behavior;

[0009] S5. Provide information feedback based on the baby’s behavior and adjust the resources needed for baby monitoring to ensure the baby’s life and health.

[0010] According to the artificial intelligence-based all-weather unmanned infant monitoring method and system, it is characterized in that in S1, a first type of sensor network is arranged according to the conditions of the infant's living space. The first type of sensor network includes a thermometer, a hygrometer, an air quality monitor, and a light intensity monitor, and each type of sensor is arranged in a linear distance with the crib as the center.

[0011] Furthermore, in S1, based on the environmental information of the baby's living space and the range value suitable for the baby's living environment, the baby's environment suitability index is proposed, the current environment of the baby is analyzed and judged, and the current environment is automatically and mechanically adjusted based on the analysis and judgment results to adapt to the baby's living needs. The detailed process is as follows:

[0012] (1) According to the arrangement of the first type of sensor network, the real-time data of the sensor is obtained, and the habitable threshold of the sensor and the baby's living environment is compared. Based on the judgment function, whether the current environment is suitable for the baby's life is determined as follows:

[0013]

[0014]

[0015]

[0016]

[0017]

[0018] in, represents a judgment function. When the variable is within the range, the value is 1, otherwise the value is 0. The range is set according to the specific situation of the user. I represents a comprehensive judgment of the baby's environment, judging whether the baby's current environment is a suitable environment. When , it means the current environment is suitable, T represents the temperature, It is the temperature range suitable for the baby's living environment. H represents humidity. It is the humidity range suitable for the baby's living environment. A represents the air quality. It is the air quality range suitable for infants’ living environment. L represents the light intensity. It is the light intensity range suitable for the baby's living environment;

[0019] (2) Under the condition that the environmental data are consistent with the infant's living conditions, the infant environment suitability index E is obtained;

[0020]

[0021] in, represents the temperature weight of the baby's environment, represents the humidity weight of the baby's environment, represents the air quality weight of the infant's environment, It represents the light intensity weight of the infant's environment, and E is used to measure the overall environmental conditions;

[0022] (3) Different infants have different preferences for the temperature, humidity, air quality, and light intensity of the environment. A preference optimization algorithm is proposed to adaptively optimize the temperature weight, humidity weight, air quality weight, and light intensity weight of the infant environment. Based on the weight vector, an optimized weight gene library is constructed. Each weight vector consists of the temperature weight, humidity weight, air quality weight, and light intensity weight of the infant environment. The optimized weight gene library is expressed as follows:

[0023]

[0024] Where D represents the optimized weighted gene pool, They represent the first optimized weight genome, the second optimized weight genome, the i-th optimized weight genome, and the n-th optimized weight genome respectively. Based on the optimized weight gene library, a weight optimization network is constructed. Each node in the weight optimization network represents an optimized weight-based group. An objective function is constructed to perform weight adaptive optimization. The objective function is expressed as follows:

[0025]

[0026] in, represents the comprehensive environmental index under the i-th optimized weight genome, It represents the comfort of the baby under the i-th optimized weight genome. The objective function represents finding the optimal weight that maximizes the comfort of the baby. The nodes in the weight optimization network are searched. During the node search process, different nodes have different fitness. In each round of search process, the node fitness determines the direction of the node search. The node fitness calculation formula is as follows:

[0027]

[0028] in, represents the node fitness of the node represented by the i-th optimized weight genome, They represent the first fitness parameter and the second fitness parameter respectively. Nodes with high node fitness have a high probability of being searched, and the weight genome is searched and optimized to perform weight adaptive optimization and adaptive adjustment for different babies.

[0029] (4) When the environmental data does not meet the living conditions of the baby, the temperature and humidity data are fed back to the connected air conditioner and air humidifier, and the air quality and light intensity data are fed back to the air purifier and ambient lighting. The environmental factors required for the baby's life are intelligently adjusted based on the standard environmental data range. After 10 to 15 minutes, the updated baby's living environment is determined based on the monitoring data of the first type of sensor network. The process is repeated until the environment meets the requirements.

[0030] Furthermore, in S2, a second type of sensor network is arranged according to the physiological condition of the infant. The second type of sensor network includes heart rate monitoring, body temperature monitoring and sound monitoring. These sensors are all wearable flexible sensors. Heart rate monitoring is used for the infant's heart health, body temperature monitoring is used for the infant's metabolic health, and sound detection is used for the infant's crying and laughing, which is used to analyze the infant's psychological state.

[0031] Furthermore, the data storage database constructed in S2 adopts MongoDB database.

[0032] Furthermore, the infant behavior chain data set is constructed in S3, and the detailed process is as follows:

[0033] (1) Since the infant's language system is not fully developed, it is necessary to mark the infant's behavior and the key value matching of the guardian's response measures in the data set. This data set is used as training data for the infant health status analysis and prediction algorithm. The present invention collects the physiological behavior of infants in different growth months within 20 days. The data of the first 10 days are used for information collection, and the data of the last 10 days are used as correction data;

[0034] (2) Based on the monitoring data types of the second type of sensor network and the attributes of the MongoDB database, the present invention stores the monitoring behaviors of infants in the form of a linked list, and defines the main nodes of the linked data set as heart rate, body temperature, and sound respectively; the data storage of each node adopts the form of a key-value pair, where the key represents the attribute description and the value represents the monitoring data; each key-value pair corresponds to an assessment of the health status of the infant;

[0035] Furthermore, the infant health status analysis and prediction algorithm is proposed in S4, and an infant behavior status analysis model is constructed to analyze the current infant's expression behavior. The detailed process is as follows:

[0036] (1) Construct a convolutional neural network with an attention mechanism to analyze and predict the health status of infants, including convolutional layers, pooling layers, and fully connected layers. The physiological health monitoring signals of infants include heart rate, body temperature, and sound. The time series data is used as input. The input data is a time series matrix with a dimension of 3, denoted as ;

[0037] (2) Use the convolution kernel to perform convolution operation on the convolution layer. Let the weight of the convolution kernel be W. m,n , with a bias of b m,n , the convolution kernel is represented as Conv3D, the neurons in different convolutional layers are connected with the neurons in the local windows in other layers, and the convolution operation is as follows:

[0038]

[0039]

[0040] in, is the convolution operation of the mth layer and the nth connection in the convolution layer, , () is the activation function, Represents the input data of the convolution of the mth layer and the nth connection, represents the weight of the nth connection in the mth layer, Represents the bias of the mth layer and the nth connection;

[0041] Define the pooling layer, use the maximum pooling operation, set the pooling window size to P, and consider the correlation between the heart rate, body temperature and sound of the infant monitoring data. Therefore, the pooling operation is as follows:

[0042]

[0043] in, It is the probability of the combination of different temperatures, humidity, light intensity, and air quality in the infant's living environment. The probability of the combination of different temperatures, humidity, light intensity, and air quality represents the probability of the environment formed under different temperature, humidity, light intensity, and air quality conditions. is the correlation coefficient of the three physiological monitoring parameters of the infant;

[0044] (3) Define the attention module,

[0045]

[0046]

[0047] in, is the attention weight in the prediction process of infant health status, is the attention output in the process of predicting the infant's health status, is the weight matrix, which is used to map the output of the convolutional layer; k represents the number of input data, represents the element-by-element convolution calculation of the matrix, e represents a natural constant, Represent the i-th and j-th convolution kernels respectively;

[0048] (4) The network structure adopts a fully connected operation. The output of the fully connected layer passes through the Flatten function. The loss function is expressed by linear calculation. The baby's heart rate, body temperature, and sound data are collected to form a data image. The baby's physiological state is predicted based on the slope of the data image at each collection point.

[0049]

[0050]

[0051] in, is the overall loss function in the process of predicting infant health status, is the weight parameter matrix that affects the heart rate loss, is the slope vector of the heart rate data image, is the weight parameter matrix affecting body temperature loss, is the slope vector of the temperature data image, is the weight parameter matrix that affects the sound loss, is the sound data image slope vector, is the final output of the infant health status prediction process. An input representing information about the infant's health status;

[0052] (5) Inferring the infant’s current physiological needs based on the predicted probability of the infant’s physiological state, which is obtained after parsing the chained data set.

[0053] Furthermore, in S5, information feedback is provided according to the baby's behavior, and the resources required for baby monitoring are adjusted to ensure the baby's life and health. The detailed process is as follows:

[0054] (1) On the premise that the current environmental data is consistent with the baby's life and the baby's body functions are normal, the information is pushed to the guardian and the furniture equipment with automated mechanical control is iterated, including air conditioners, air humidifiers, air purifiers and ambient lighting;

[0055] (2) If the current environmental data does not conform to the infant's living conditions or the infant's body functions are not normal, the information will be pushed to the guardian, who will be set to level 3. The push methods include mobile phones, computer user interfaces, smart phones and text messages. If the guardian cannot be contacted, other guardians will be contacted step by step, and the entire process will be encrypted.

[0056] Beneficial effects of the present invention: The present invention describes an all-weather unmanned baby monitoring method and system based on artificial intelligence, including a spatial environment information recognition module, an infant behavior analysis module and an infant behavior disposal module. First, the spatial environment information recognition module captures and identifies the spatial environment of the baby in real time through sensor network technology and algorithms. This covers key factors such as indoor temperature, humidity, and light. Based on the infant environmental suitability index, the comprehensive indicators of temperature, humidity, air quality, and light intensity in the environment where the baby is located are analyzed to perform environmental regulation to ensure the comfort and safety of the baby. Secondly, the infant behavior analysis module uses a chain data set for storage, and based on this, the relationship between infant behavior and needs for 60 days is collected to analyze the infant's body expression behavior, and it is used as a training data set for the algorithm. The module can identify the baby's heart rate, body temperature, and voice to fully understand the baby's physiological and emotional state, and describe the potential entanglement relationship of the monitored infant's physiological data through quantum entanglement calculation, propose an infant health status analysis and prediction algorithm, construct an infant behavior state analysis model, parse the chain data set, infer the current infant's physiological needs through the predicted probability of the infant's physiological state, and parse the current infant's expression behavior. Finally, the infant behavior handling module processes the analyzed data through a convolutional neural network that introduces an attention mechanism. Depending on the situation, the module can take corresponding measures, such as sending an alarm to notify parents, adjusting environmental conditions, or providing specific comfort measures to meet the needs of the infant. The entire system is connected via the Internet, allowing parents or guardians to access and monitor the infant's status in real time through smart devices. In addition, the system design takes security and privacy issues into consideration, and adopts strict data encryption and security measures to ensure that the user's information is fully protected. This AI-based all-weather unmanned infant monitoring method and system provides parents with a comprehensive and reliable infant monitoring solution, emphasizing the innovation and practicality of smart technology in improving the infant care experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The invention is further illustrated by the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0058] Figure 1 It is a schematic diagram of the structure of the present invention. DETAILED DESCRIPTION

[0059] The present invention is further described in conjunction with the following examples.

[0060] See also Figure 1The present invention aims to provide an artificial intelligence-based all-weather unmanned infant monitoring method and system to solve the problems raised in the above-mentioned background technology.

[0061] To achieve the above purpose, a method and system for unmanned all-weather infant monitoring based on artificial intelligence is provided, including a spatial environment information recognition module, an infant behavior analysis module and an infant behavior treatment module; the specific process is described as follows:

[0062] S1. Arrange the first type of sensor network according to the conditions of the baby's living space, and collect the environmental information of the baby's living space based on this. The first type of sensor network includes a thermometer, a hygrometer, an air quality monitor, and a light intensity monitor, and each type of sensor is arranged in a linear distance with the baby's crib as the center. Judge the current environment based on the range value suitable for the baby's living environment and adjust the current environment based on the automatic machinery to adapt to the baby's living needs. Compare the livable thresholds of the sensor and the baby's living environment, and judge whether the current environment is suitable for the baby's life based on the judgment function. The judgment method is as follows:

[0063]

[0064]

[0065]

[0066]

[0067]

[0068] in, represents a judgment function. When the variable is within the range, the value is 1, otherwise the value is 0. The range is set according to the specific situation of the user. I represents a comprehensive judgment of the baby's environment, judging whether the baby's current environment is a suitable environment. When , it means the current environment is suitable, T represents the temperature, It is the temperature range suitable for the baby's living environment. H represents humidity. It is the humidity range suitable for the baby's living environment. A represents the air quality. It is the air quality range suitable for infants’ living environment. L represents the light intensity. It is the light intensity range suitable for the baby's living environment;

[0069] Under the condition that the environmental data are all suitable for infants’ living, the infant environment suitability index E is obtained;

[0070]

[0071] in, represents the temperature weight of the baby's environment, represents the humidity weight of the baby's environment, represents the air quality weight of the infant's environment, It represents the light intensity weight of the infant's environment, and E is used to measure the overall environmental conditions;

[0072] Different babies have different preferences for the temperature, humidity, air quality, and light intensity of the environment. A preference optimization algorithm is proposed to adaptively optimize the temperature weight, humidity weight, air quality weight, and light intensity weight of the baby environment. Based on the weight vector, an optimized weight gene library is constructed. Each weight vector consists of the temperature weight, humidity weight, air quality weight, and light intensity weight of the baby environment. The optimized weight gene library is expressed as follows:

[0073]

[0074] Where D represents the optimized weighted gene pool, They represent the first optimized weight genome, the second optimized weight genome, the i-th optimized weight genome, and the n-th optimized weight genome respectively. Based on the optimized weight gene library, a weight optimization network is constructed. Each node in the weight optimization network represents an optimized weight-based group. An objective function is constructed to perform weight adaptive optimization. The objective function is expressed as follows:

[0075]

[0076] in, represents the comprehensive environmental index under the i-th optimized weight genome, It represents the comfort of the baby under the i-th optimized weight genome. The objective function represents finding the optimal weight that maximizes the comfort of the baby. The nodes in the weight optimization network are searched. During the node search process, different nodes have different fitness. In each round of search process, the node fitness determines the direction of the node search. The node fitness calculation formula is as follows:

[0077]

[0078] in, represents the node fitness of the node represented by the i-th optimized weight genome, They represent the first fitness parameter and the second fitness parameter respectively. Nodes with high node fitness have a high probability of being searched to search for the optimized weight genome, perform weight adaptive optimization, and perform adaptive adjustment for different babies.

[0079] If the environmental data does not meet the living conditions of the baby, the temperature and humidity data are fed back to the connected air conditioner and air humidifier, and the air quality and light intensity data are fed back to the air purifier and ambient lighting. The environmental factors required for the baby's life are intelligently adjusted based on the standard environmental data range. After 10 to 15 minutes, the updated baby's living environment is determined based on the monitoring data of the first type of sensor network, and the process is repeated until the environment meets the requirements;

[0080] S2. On the premise that the environment meets the needs of infants, a second type of sensor network is arranged according to the infant's physiological condition, and a data storage database is constructed to collect the infant's physiological data in real time. The second type of sensor network includes heart rate monitoring, body temperature monitoring and sound monitoring. These sensors are all wearable flexible sensors. Heart rate monitoring is used for infant heart health, body temperature monitoring is used for infant metabolic health, and sound detection is used for infant crying and laughing, which is used to analyze the infant's psychological state. The data storage database uses MongoDB database.

[0081] S3. Construct a chained dataset of infant behaviors. Since infants’ language systems are not fully developed, it is necessary to mark the key value matching between infant behaviors and guardians’ responses in the dataset. This dataset is used as training data for infant health status analysis and prediction algorithms. The detailed process is as follows:

[0082] (1) Since the infant's language system is not fully developed, it is necessary to mark the infant's behavior and the key value matching of the guardian's response measures in the data set. This data set is used as training data for the infant health status analysis and prediction algorithm. The present invention collects the physiological behavior of infants in different growth months within 20 days. The data of the first 10 days are used for information collection, and the data of the last 10 days are used as correction data;

[0083] (2) Based on the monitoring data types of the second type of sensor network and the attributes of the MongoDB database, the present invention stores the monitoring behaviors of infants in the form of a linked list, and defines the main nodes of the linked data set as heart rate, body temperature, and sound respectively; the data storage of each node adopts the form of a key-value pair, where the key represents the attribute description and the value represents the monitoring data; each key-value pair corresponds to an assessment of the health status of the infant;

[0084] S4. Based on the infant physiological monitoring data, an infant health status analysis and prediction algorithm is proposed, an infant behavior status analysis model is constructed, and the current infant expression behavior is analyzed. The detailed process is as follows:

[0085] (1) Construct a convolutional neural network with an attention mechanism to analyze and predict the health status of infants, including convolutional layers, pooling layers, and fully connected layers. The physiological health monitoring signals of infants include heart rate, body temperature, and sound. The time series data is used as input. The input data is a time series matrix with a dimension of 3, denoted as ;

[0086] (2) Use the convolution kernel to perform convolution operation on the convolution layer. Let the weight of the convolution kernel be W. m,n , with a bias of b m,n , the convolution kernel is represented as Conv3D, the neurons in different convolutional layers are connected with the neurons in the local windows in other layers, and the convolution operation is as follows:

[0087]

[0088]

[0089] in, is the convolution operation of the mth layer and the nth connection in the convolution layer, , () is the activation function, Represents the input data of the convolution of the mth layer and the nth connection, represents the weight of the nth connection in the mth layer, Represents the bias of the mth layer and the nth connection;

[0090] Define the pooling layer, use the maximum pooling operation, set the pooling window size to P, and consider the correlation between the heart rate, body temperature and sound of the infant monitoring data. Therefore, the pooling operation is as follows:

[0091]

[0092] in, It is the probability of the combination of different temperatures, humidity, light intensity, and air quality in the infant's living environment. The probability of the combination of different temperatures, humidity, light intensity, and air quality represents the probability of the environment formed under different temperature, humidity, light intensity, and air quality conditions. is the correlation coefficient of the three physiological monitoring parameters of the infant;

[0093] (3) Define the attention module,

[0094]

[0095]

[0096] in, is the attention weight in the prediction process of infant health status, is the attention output in the process of predicting the infant's health status, is the weight matrix, which is used to map the output of the convolutional layer; k represents the number of input data, represents the element-by-element convolution calculation of the matrix, e represents a natural constant, Represent the i-th and j-th convolution kernels respectively;

[0097] (4) The network structure adopts a fully connected operation. The output of the fully connected layer passes through the Flatten function. The loss function is expressed by linear calculation. The baby's heart rate, body temperature, and sound data are collected to form a data image. The baby's physiological state is predicted based on the slope of the data image at each collection point.

[0098]

[0099]

[0100] in, is the overall loss function in the process of predicting infant health status, is the weight parameter matrix affecting the heart rate loss, is the slope vector of the heart rate data image, is the weight parameter matrix affecting body temperature loss, is the slope vector of the temperature data image, is the weight parameter matrix that affects the sound loss, is the sound data image slope vector, is the final output of the infant health status prediction process. An input representing information about the infant's health status;

[0101] (5) Inferring the infant’s current physiological needs based on the predicted probability of the infant’s physiological state, which is obtained after parsing the chained data set.

[0102] S5. Provide information feedback based on the baby's behavior and adjust the resources needed for baby monitoring to ensure the baby's life and health. The detailed process is as follows:

[0103] (1) On the premise that the current environmental data is consistent with the baby's life and the baby's body functions are normal, the information is pushed to the guardian and the furniture equipment with automated mechanical control is iterated, including air conditioners, air humidifiers, air purifiers and ambient lighting;

[0104] (2) If the current environmental data does not conform to the infant's living conditions or the infant's body functions are not normal, the information will be pushed to the guardian, who will be set to level 3. The push methods include mobile phones, computer user interfaces, smart phones and text messages. If the guardian cannot be contacted, other guardians will be contacted step by step, and the entire process will be encrypted.

[0105] In a specific embodiment, the user first collects environmental information of the baby's living space by arranging thermometers, hygrometers, air quality monitors, and light intensity monitors around the baby, and proposes a baby environment suitability index based on a range of values ​​suitable for the baby's living environment, analyzes and judges the current environment in which the baby is located, and adaptively adjusts the baby's living needs; then, by arranging a second type of sensor network, a data storage database is constructed, and based on the chain data set of the baby's physiological state, a chain relationship of the baby's physiological behavior is constructed, and through the baby's behavioral state analysis model, the chain data set of the baby's physiological state is analyzed to infer the current physiological needs of the baby.

[0106] Effect of this embodiment: The present invention describes a method and system for unmanned baby care based on artificial intelligence around the clock, including a spatial environment information recognition module, a baby behavior analysis module and a baby behavior disposal module. First, the spatial environment information recognition module captures and identifies the spatial environment of the baby in real time through sensor network technology and algorithms. This covers key factors such as indoor temperature, humidity, and light. Based on the baby environment suitability index, the comprehensive indicators of temperature, humidity, air quality, and light intensity in the baby's environment are analyzed to perform environmental regulation to ensure the comfort and safety of the baby. Secondly, the baby behavior analysis module uses a chain data set for storage, and analyzes the baby's body expression behavior based on the relationship between the baby's behavior and needs collected for 60 days, and serves as a training data set for the algorithm. The module can identify the baby's heart rate, body temperature, and voice to fully understand the baby's physiological and emotional state, and describe the potential entanglement relationship of the monitored baby's physiological data through quantum entanglement calculation, propose an analysis and prediction algorithm for the baby's health status, construct a baby behavior state analysis model, parse the chain data set, and infer the current baby's physiological needs through the predicted probability of the baby's physiological state, and parse the current baby's expression behavior. Finally, the infant behavior handling module processes the analyzed data through a convolutional neural network that introduces an attention mechanism. Depending on the situation, the module can take corresponding measures, such as sending an alarm to notify parents, adjusting environmental conditions, or providing specific comfort measures to meet the needs of the infant. The entire system is connected via the Internet, allowing parents or guardians to access and monitor the infant's status in real time through smart devices. In addition, the system design takes security and privacy issues into consideration, and adopts strict data encryption and security measures to ensure that the user's information is fully protected. This AI-based all-weather unmanned infant monitoring method and system provides parents with a comprehensive and reliable infant monitoring solution, emphasizing the innovation and practicality of smart technology in improving the infant care experience.

[0107] The present invention also provides a computer-readable storage medium, in which at least one instruction is stored, and the instruction is loaded and executed by a processor to implement the above method. The computer-readable storage medium may be a ROM, a random access memory, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc. The instructions stored therein may be loaded by a processor in a terminal to execute the above method.

[0108] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0109] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0110] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An all-weather unmanned infant monitoring system based on artificial intelligence, including a spatial environment information recognition module, an infant behavior analysis module and an infant behavior treatment module; the specific operation process of the system is described as follows: S1. Arrange the first type of sensor network according to the conditions of the baby's living space, collect the environmental information of the baby's living space based on this, propose the baby environment suitability index according to the range value suitable for the baby's living environment, analyze and judge the current environment of the baby, propose a preference optimization algorithm, optimize the preference of the baby's environmental information, and automatically adjust the current environment to adapt to the baby's living needs; Based on the first type of sensor network, real-time data is obtained, and the habitable thresholds of the sensors and the baby's living environment are compared. Based on the judgment function, it is analyzed whether the current environment is suitable for the baby's life. The judgment method is as follows: in, represents a judgment function. When the variable is within the range, the value is 1, otherwise the value is 0. The range is set according to the specific situation of the user. I represents a comprehensive judgment of the baby's environment, judging whether the baby's current environment is a suitable environment. When , it means that the current environment is suitable. T represents temperature, H represents humidity, A represents air quality, and L represents light intensity. The baby environment suitability index E is calculated as follows: in, represents the temperature weight of the baby's environment, represents the humidity weight of the baby's environment, represents the air quality weight of the infant's environment, represents the light intensity weight of the baby's environment, and E is used to measure the overall environmental conditions. Different babies have different preferences for temperature, humidity, air quality, and light intensity in the environment. A preference optimization algorithm is proposed to adaptively optimize the temperature weight, humidity weight, air quality weight, and light intensity weight of the baby's environment, and construct a weight vector. Each weight vector consists of the temperature weight, humidity weight, air quality weight, and light intensity weight of the baby's environment. A weight optimization network is constructed, and the nodes in the weight optimization network represent the optimization weight based on the group. The constructed objective function is expressed as follows: in, The i-th optimized weight gene set, represents the comprehensive environmental index under the i-th optimized weight genome, It represents the comfort of the baby under the i-th optimized weight genome. The objective function represents finding the optimal weight that maximizes the comfort of the baby. The nodes in the weight optimization network are searched. During the node search process, different nodes have different fitness. In each round of search process, the node fitness determines the direction of the node search. The node fitness calculation formula is as follows: in, represents the node fitness of the node represented by the i-th optimized weight genome, They represent the first fitness parameter and the second fitness parameter respectively. Nodes with high node fitness have a high probability of being searched. By searching and optimizing the weight genome, weight adaptive optimization is performed to adaptively adjust different babies. When the environmental data does not meet the conditions for the baby's life, the temperature and humidity data are fed back to the connected air conditioner and air humidifier, and the air quality and light intensity data are fed back to the air purifier and the ambient light. Based on the standard environmental data range, the environmental factors required for the baby's life are intelligently adjusted. S2. Arrange the second type of sensor network, build a data storage database, and collect the physiological data of the baby; S3. Construct a chained dataset of infant behaviors, in which the key values ​​of infant behaviors and guardian responses are marked, and use this dataset as training data for infant health status analysis and prediction algorithms; S4. Based on the infant physiological monitoring data, an infant health status analysis and prediction algorithm is proposed, and an infant behavior status analysis model is constructed to analyze the current infant's expression behavior; a convolutional neural network based on the attention mechanism is used to analyze and predict the infant's health status. The convolutional neural network with the attention mechanism includes a convolutional layer, a pooling layer, and a fully connected layer. The infant's physiological health monitoring signals include heart rate, body temperature, and sound. Its time series data is used as input. The input data is a time series matrix, denoted as ; In the convolution layer, a convolution kernel is used to perform a convolution operation on the convolution layer. The convolution kernel is represented as Conv3D. There are multiple convolution layers. Neurons in different convolution layers are connected to neurons in local windows in other layers. The convolution operation is as follows: in, is the convolution operation of the mth layer and the nth connection in the convolution layer, , They represent the error gradient of the mth layer and the error gradient of the m-1th layer respectively. () is the activation function, Represents the input data of the convolution of the mth layer and the nth connection, represents the weight of the nth connection in the mth layer, Represents the bias of the nth connection in the mth layer. In the pooling layer, the maximum pooling operation is used, and the pooling window size is set to P. The pooling operation is as follows: in, It is the probability of the combination of different temperatures, humidity, light intensity, and air quality in the infant's living environment. The probability of the combination of different temperatures, humidity, light intensity, and air quality represents the probability of the environment formed under different conditions of temperature, humidity, light intensity, and air quality. is the correlation coefficient of the three physiological monitoring parameters of the infant; the attention module is defined as: in, is the attention weight in the prediction process of infant health status, is the attention output in the process of predicting the infant's health status, is the weight matrix, which is used to map the output of the convolutional layer; k represents the number of input data, represents the element-by-element dot multiplication of the matrix, e represents a natural constant, Represent the i-th and j-th convolution kernels respectively; the network structure adopts full connection operation, the output of the full connection layer passes through the Flatten function, and the loss function is expressed by linear calculation. The baby's heart rate, body temperature, and sound data are collected to form a data image. Based on the slope of the data image at each collection point, the baby's physiological state is predicted: in, is the overall loss function in the process of predicting infant health status, is the weight parameter matrix affecting the heart rate loss, is the slope vector of the heart rate data image, is the weight parameter matrix affecting body temperature loss, is the slope vector of the temperature data image, is the weight parameter matrix that affects the sound loss, is the sound data image slope vector, is the final output of the infant health status prediction process. An input representing information about the infant's health status; S5. Provide feedback based on infant behavior and adjust resources needed for infant monitoring.

2. According to claim 1, an artificial intelligence-based all-weather infant unmanned monitoring system is characterized in that: In S1, a first type of sensor network is arranged according to the conditions of the baby's living space, and the first type of sensor network includes a thermometer, a hygrometer, an air quality monitor, and a light intensity monitor.

3. The artificial intelligence-based all-weather infant unmanned monitoring system according to claim 1, characterized in that: The second type of sensor network in S2 includes heart rate monitoring, body temperature monitoring and sound monitoring.

4. The artificial intelligence-based all-weather infant unmanned monitoring system according to claim 1, characterized in that: The data storage database constructed in S2 adopts MongoDB database.

5. The artificial intelligence-based all-weather infant unmanned monitoring system according to claim 1, characterized in that: In the S3, a chained infant behavior dataset is constructed, in which the infant's behavior is marked in the dataset and the key value matching of the guardian's response measures is performed. The physiological behaviors of infants in different growth months are collected and stored in the form of a linked list, and the main nodes of the chained dataset are defined as heart rate, body temperature and sound respectively; the data storage of each node is in the form of a key-value pair, where the key represents the attribute description and the value represents the monitoring data.

6. According to claim 1, the artificial intelligence-based all-weather infant unmanned monitoring system is characterized in that: In the S5, information feedback is provided according to the baby's behavior, and the resources required for baby monitoring are adjusted. On the premise that the current environmental data is consistent with the baby's life and the baby's body functions are normal, the information is pushed to the guardian; on the premise that the current environmental data is not consistent with the baby's life or the baby's body functions are not normal, the information is pushed to the guardian, and when the guardian cannot be contacted, other guardians are contacted step by step.

7. A method for performing unmanned infant monitoring using the artificial intelligence-based all-weather unmanned infant monitoring system as described in any one of claims 1 to 6.

Images