An internet of things smart home control method and system based on personalized service

Abstract

The application relates to the technical field of smart home, and discloses an Internet of Things smart home control method and system based on personalized services, which comprises the following steps: collecting bioelectric signals, environmental reflection signals, environmental point cloud data and light intensity distribution data of a user on a smart home; performing space-time coding processing on the bioelectric signals to obtain a biological feature vector, coding the environmental point cloud data and the light intensity distribution data into a behavior mode vector, and extracting a space thermal field corresponding to the environmental reflection signals and the environmental point cloud data; establishing a topological relationship graph among the smart homes, constructing a cross-device collaborative strategy among the smart homes based on the biological feature vector, the behavior mode vector and the topological relationship graph, and determining a real-time control instruction of the smart home. The application can deeply integrate multidimensional biological and environmental signals, establish intelligent topological correlations among devices, and effectively utilize space energy information for collaborative decision-making and accurate control.

Description

Technical Field

[0001] This invention relates to an IoT smart home control method and system based on personalized services, belonging to the field of smart home technology. Background Technology

[0002] While significant progress has been made in IoT smart home control technology, there are still a series of interconnected and significant shortcomings in achieving truly personalized, collaborative, and intelligent services.

[0003] First, user state perception is limited in scope and superficial. Existing solutions generally rely on single sensors or limited data types to infer user intentions or states. This makes it difficult to accurately capture complex physiological states and subtle behavioral patterns, hindering the formation of a deep, personalized understanding. Second, due to the limitations of perception dimensions and the lack of effective inter-device connections, smart home devices often operate in silos. Control strategies between devices are typically independent or simply preset, lacking dynamic collaborative decision-making mechanisms based on global topology relationships and real-time multimodal user / environment information. This not only makes device interaction rigid and unintelligent but also directly leads to low resource utilization efficiency. For example, it cannot determine the user's actual activity area and... Third, the neglect of dynamic energy distribution in the environment is one of the key factors causing resource waste and decreased comfort. Existing technologies rarely effectively model the heat distribution generated by user activities and its diffusion effect in the environment. This makes it impossible for the system to identify high-energy-consuming areas or to carry out precise and predictive energy-saving control based on the user's actual thermal comfort needs. Ultimately, the combined effect of the above-mentioned defects such as insufficient perception, fragmented equipment, and extensive energy management seriously restricts the system's ability to provide deeply personalized services. The system is unable to generate highly adapted and cross-device collaborative fine-grained control commands based on the user's real-time and comprehensive physiological-behavioral-environmental state, and the improvement of user experience encounters a bottleneck.

[0004] In summary, existing IoT smart home control technologies are primarily limited by insufficient depth and breadth of multimodal information perception of users and the environment. This leads to a lack of dynamic intelligent collaboration between devices based on global topology and neglects the modeling and utilization of the environmental spatial thermal field. Ultimately, these factors hinder the system from providing truly personalized, energy-efficient, and seamlessly collaborative smart home services. There is an urgent need for a new method that can deeply integrate multidimensional biological and environmental signals, establish intelligent topological connections between devices, and effectively utilize spatial energy information for collaborative decision-making and precise control of smart homes. Summary of the Invention

[0005] This invention provides an IoT smart home control method and system based on personalized services. Its main purpose is to deeply integrate multi-dimensional biological and environmental signals, establish intelligent topological connections between devices, and effectively utilize spatial energy information for collaborative decision-making and precise control.

[0006] To achieve the above objectives, the present invention provides an IoT smart home control method based on personalized services, comprising:

[0007] Collect users' bioelectrical signals, environmental reflection signals, environmental point cloud data, and light intensity distribution data in smart homes;

[0008] The bioelectric signal is spatiotemporally encoded to obtain a biological feature vector. The environmental point cloud data and the light intensity distribution data are encoded into a behavior pattern vector. The spatial thermodynamic field corresponding to the environmental reflection signal and the environmental point cloud data is extracted.

[0009] Establish a topology graph between various smart home devices, and construct a cross-device collaboration strategy between various smart home devices based on the biometric vector, the behavior pattern vector, and the topology graph.

[0010] Based on the biometric vector, the behavioral pattern vector, and the spatial thermal field, the real-time control commands for the smart home are determined.

[0011] The smart home is controlled through the cross-device collaboration strategy and the real-time control commands to obtain home control results.

[0012] Optionally, the collection of user bioelectrical signals, environmental reflection signals, environmental point cloud data, and light intensity distribution data in the smart home includes:

[0013] Identify frequently touched surfaces in smart homes;

[0014] A flexible AMOLED skin is deployed on the high-frequency contact surface;

[0015] The flexible AMOLED skin includes a base layer and a light-transmitting channel. The base layer includes an interdigitated electrode array, silver nanowire electrodes, and a varistor array.

[0016] When the user's skin conductivity signal penetrates the light-transmitting channel, the skin conductivity corresponding to the skin conductivity signal is captured by the interdigital electrode array;

[0017] When the user's muscle contraction electrical signal penetrates the light-transmitting channel, the final electromyographic signal corresponding to the muscle contraction electrical signal is captured by the silver nanowire electrode.

[0018] When the user's pressure signal penetrates the light-transmitting channel, the pressure distribution corresponding to the pressure signal is collected by the pressure-sensitive resistor array;

[0019] The skin conductivity, the final electromyographic signal, and the pressure distribution are used as bioelectrical signals.

[0020] A millimeter-wave radar system and a photodiode are embedded in the smart home device.

[0021] The millimeter-wave radar system includes a TDM-MIMO antenna array and an adaptive beamforming algorithm.

[0022] Based on the adaptive beamforming algorithm, frequency-modulated continuous waves are transmitted into the environment where the smart home is located through the TDM-MIMO antenna array;

[0023] The millimeter-wave radar system receives environmental reflection signals related to the frequency-modulated continuous wave.

[0024] The environmental reflection signal is transmitted to the central controller corresponding to the smart home.

[0025] In the central controller, the environmental reflection signal is converted into environmental point cloud data;

[0026] The photodiode is used to collect the ambient light intensity distribution in the environment where the smart home is located;

[0027] The ambient light intensity distribution is reconstructed into light intensity distribution data using spherical harmonic expansion.

[0028] Optionally, the spatiotemporal coding processing of the bioelectric signal to obtain a biological feature vector includes:

[0029] The skin conductivity, final electromyographic signal, and pressure distribution in the bioelectrical signals were obtained.

[0030] The skin conductivity is piecewise integrated to obtain a sequence of area values;

[0031] Extract the fluctuation period from the area value sequence;

[0032] The fluctuation period is processed by Poisson encoding to obtain a pulse sequence;

[0033] The pulse sequence is input into a preset LIF neuron model to obtain the emotion label output by the LIF neuron model;

[0034] The final electromyography signal is filtered by the SNN encoder in the central controller to obtain a filtered electromyography signal.

[0035] Calculate the spatiotemporal gradient of the pressure distribution;

[0036] The filtered electromyographic signal and the spatiotemporal gradient are fused through cross-modal product to obtain the intent intensity;

[0037] The emotion label, the intent intensity, and the spatiotemporal gradient are compressed into a biological feature vector through the fully connected SNN layer in the central controller.

[0038] Optionally, encoding the environmental point cloud data and the light intensity distribution data into a behavior pattern vector includes:

[0039] The alpha-shape algorithm in the central controller is used to extract the joints from the environmental point cloud data;

[0040] The joint motion vectors corresponding to the joint points are analyzed by combining the graph convolutional network in the central controller.

[0041] Calculate the gradient optical flow characteristics of the light intensity distribution data;

[0042] Based on the joint motion vector and the gradient optical flow features, identify the motion features corresponding to the environmental point cloud data and the light intensity distribution data;

[0043] The spatiotemporal hash value of the action feature is generated by the SHA-256 algorithm in the central controller.

[0044] Optionally, extracting the spatial thermal field corresponding to the environmental reflection signal and the environmental point cloud data includes:

[0045] The environmental reflection signal was subjected to Wigner-Ville distribution analysis to determine its time-frequency distribution.

[0046] Extract the user's breathing signal from the time-frequency distribution;

[0047] The user's respiratory rate is calculated using the respiratory signal;

[0048] Based on the breathing frequency, the spatial thermal field of the environment in which the smart home is located is determined.

[0049] Optionally, determining the spatial thermal field of the smart home environment based on the breathing frequency includes:

[0050] Take environmental point cloud data and thermal diffusivity;

[0051] The spatial coordinates of the user and the coordinates of objects in the environment where the smart home is located are determined using the environmental point cloud data.

[0052] Calculate the point cloud density parameters corresponding to the environmental point cloud data;

[0053] The user's activity power is calculated based on the thermal diffusion coefficient, the spatial coordinates, the object coordinates, the breathing frequency, and the point cloud density parameter.

[0054] The spatial thermal field of the smart home environment is generated by the user activity power.

[0055] Optionally, establishing the topology diagram between the various smart home devices includes:

[0056] An initial adjacency matrix between smart home devices is generated by pulse signalers embedded in each smart home device.

[0057] Obtain the intensity of intent;

[0058] The initial adjacency matrix is ​​updated with weights based on the intent strength to obtain the updated adjacency matrix;

[0059] The updated adjacency matrix is ​​converted into a topological graph of the relationships between the various smart home devices.

[0060] Optionally, the step of constructing a cross-device collaboration strategy among various smart home devices based on the biometric vector, the behavioral pattern vector, and the topological relationship graph includes:

[0061] The biometric vector, the behavioral pattern vector, and the topological relationship graph are input into a preset adversarial reinforcement learning model;

[0062] Obtain the cross-device collaborative strategy output by the adversarial reinforcement learning model regarding the biometric vector, the behavioral pattern vector, and the topological relationship graph.

[0063] Optionally, determining the real-time control commands for the smart home based on the biometric vector, the behavioral pattern vector, and the spatial thermal field includes:

[0064] Obtain the emotion-related, intention-related, and stress-related home furnishings corresponding to the biometric vectors;

[0065] Based on the biometric vectors, emotion control instructions, intention control instructions, and stress control instructions are generated for the emotion-related home furnishings, the intention-related home furnishings, and the stress-related home furnishings, respectively.

[0066] The spatial thermal field is used to identify high-energy-density home appliances within the smart home system.

[0067] Based on the high energy density value corresponding to the high energy density home furnishing in the spatial thermal field, an energy-saving instruction for the high energy density home furnishing is generated;

[0068] The smart home's special control commands are initiated through the aforementioned behavior pattern vector;

[0069] The emotion control command, the intention control command, the stress control command, the energy-saving command, and the special control command are used as the real-time control commands for the smart home.

[0070] To address the above problems, the present invention also provides an IoT smart home control system based on personalized services, the system comprising:

[0071] The data acquisition module is used to collect users' bioelectrical signals, environmental reflection signals, environmental point cloud data, and light intensity distribution data in smart homes;

[0072] The thermal field extraction module is used to perform spatiotemporal encoding processing on the bioelectric signal to obtain a biological feature vector, encode the environmental point cloud data and the light intensity distribution data into a behavior pattern vector, and extract the spatial thermal field corresponding to the environmental reflection signal and the environmental point cloud data.

[0073] The strategy construction module is used to establish a topology graph between various smart home devices, and to construct cross-device collaboration strategies between various smart home devices based on the biometric vectors, the behavioral pattern vectors, and the topology graph.

[0074] The instruction determination module is used to determine the real-time control instructions of the smart home based on the biometric vector, the behavioral pattern vector, and the spatial thermal field.

[0075] The home control module is used to control the smart home through the cross-device collaboration strategy and the real-time control commands to obtain home control results.

[0076] Compared to the problems described in the background technology, the embodiments of the present invention address the issue of a single and superficial dimension of user state perception by collecting multi-dimensional data. The embodiments of the present invention perform spatiotemporal encoding processing on the bioelectrical signals to preprocess and convert the collected biosignals into computer-recognizable data. Furthermore, the embodiments of the present invention encode the environmental point cloud data and the light intensity distribution data into behavioral pattern vectors to identify user behavior categories. Furthermore, the embodiments of the present invention extract the spatial thermal field corresponding to the environmental reflection signal and the environmental point cloud data to identify which areas are frequently used by the user. Furthermore, the embodiments of the present invention establish a topological relationship graph between various smart home devices to establish intelligent topological associations between devices. Furthermore, the embodiments of the present invention construct cross-device collaborative strategies between various smart home devices based on the biofeature vectors, the behavioral pattern vectors, and the topological relationship graph to perform collaborative decision-making for smart homes. The embodiments of the present invention determine the real-time control commands of the smart home devices based on the biofeature vectors, the behavioral pattern vectors, and the spatial thermal field to effectively utilize spatial energy information for precise control of smart homes. Therefore, the IoT smart home control method and system based on personalized services provided in this embodiment of the invention can deeply integrate multi-dimensional biological and environmental signals, establish intelligent topological associations between devices, and effectively utilize spatial energy information for collaborative decision-making and precise control. Attached Figure Description

[0077] Figure 1 A flowchart illustrating an embodiment of the IoT smart home control method based on personalized services provided by the present invention;

[0078] Figure 2 A schematic diagram illustrating user activity power in an embodiment of the IoT smart home control method based on personalized services provided by the present invention;

[0079] Figure 3 This is a schematic diagram illustrating the generation of a cross-device collaboration strategy in an IoT smart home control method based on personalized services, according to an embodiment of the present invention.

[0080] Figure 4 This is a schematic diagram of a module for implementing the IoT smart home control system based on personalized services, according to an embodiment of the present invention.

[0081] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0082] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0083] This application provides an IoT smart home control method based on personalized services. The executing entity of the IoT smart home control method based on personalized services includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application embodiment: a server, a terminal, etc. In other words, the IoT smart home control method based on personalized services can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0084] Example 1:

[0085] Reference Figure 1 The diagram shown is a flowchart illustrating an IoT smart home control method based on personalized services according to an embodiment of the present invention. In this embodiment, the IoT smart home control method based on personalized services includes:

[0086] S1. Collect bioelectric signals, environmental reflection signals, environmental point cloud data, and light intensity distribution data of users in smart homes.

[0087] This invention addresses the problem of a single and superficial dimension of user status perception by collecting multi-dimensional data.

[0088] In one embodiment of the present invention, the acquisition of bioelectrical signals, environmental reflection signals, environmental point cloud data, and light intensity distribution data of a user on a smart home device includes: identifying high-frequency contact surfaces of the user on the smart home device; deploying a flexible AMOLED skin on the high-frequency contact surface; wherein the flexible AMOLED skin includes a substrate layer and a light-transmitting channel, the substrate layer including an interdigital electrode array, silver nanowire electrodes, and a pressure-sensitive resistor array; when the user's skin conductivity signal penetrates the light-transmitting channel, the skin conductivity corresponding to the skin conductivity signal is captured by the interdigital electrode array; when the user's muscle contraction electrical signal penetrates the light-transmitting channel, the final electromyographic signal corresponding to the muscle contraction electrical signal is captured by the silver nanowire electrodes; when the user's pressure signal penetrates the light-transmitting channel, the pressure signal is acquired by the pressure-sensitive resistor array. The corresponding pressure distribution; the skin conductivity, the final electromyographic signal, and the pressure distribution are used as bioelectrical signals; a millimeter-wave radar system and a photodiode are embedded in the smart home; wherein, the millimeter-wave radar system includes a TDM-MIMO antenna array and an adaptive beamforming algorithm; based on the adaptive beamforming algorithm, a frequency-modulated continuous wave is transmitted to the environment where the smart home is located through the TDM-MIMO antenna array; the environmental reflection signal of the frequency-modulated continuous wave is received in the millimeter-wave radar system; the environmental reflection signal is transmitted to the central controller corresponding to the smart home; in the central controller, the environmental reflection signal is converted into environmental point cloud data; the ambient light intensity distribution in the environment where the smart home is located is collected through the photodiode; the ambient light intensity distribution is reconstructed into light intensity distribution data through spherical harmonic expansion.

[0089] The high-frequency contact surface refers to the surface of home furnishing components that users frequently touch, such as handrails, tabletops, and handles. For example, a flexible AMOLED skin is deployed on the surface of a smart toilet handrail to monitor the user's physiological state when holding it in real time. The flexible AMOLED skin is a flexible display and sensing integrated thin-film structure used to deploy on high-frequency contact surfaces. It mainly includes a substrate layer and a light-transmitting channel. The substrate layer uses a double-layer transparent polyimide (CPI) film, with AMOLED pixels integrated on the upper layer for displaying information and a three-dimensional sensing array embedded on the lower layer for sensing. The light-transmitting channel has small holes (5 micrometers in diameter) between the RGB pixels to allow biosignals to penetrate to the substrate layer. The interdigital electrode array is an electrode structure used to detect skin conductivity. It consists of multiple slender electrodes with a spacing of 20 micrometers. Applying 0.1 mA of alternating current to the interdigital electrode array allows the detection of sweat gland activity on the user's fingers through the impedance phase shift of the interdigital electrode array, thereby capturing skin conductivity. The nano-silver wire electrode refers to a sensitive electrode used to capture the action potential of muscle fibers. It can capture the electrical signals generated by the user's muscle contraction to obtain the final electromyographic signal. The pressure-sensitive resistor array refers to a resistor array used to detect pressure distribution. It can collect the pressure signal applied by the user to a high-frequency contact surface and generate a pressure distribution map with a resolution of 0.1 Newtons. The TDM-MIMO antenna array refers to a time-division multiplexed multiple-input multiple-output antenna used to transmit frequency-modulated continuous wave (FMCW) and receive the reflected signal. The adaptive beamforming algorithm refers to an algorithm that optimizes the signal reception and transmission direction to improve signal quality and resolution. The central controller refers to the core control unit of the smart home system, used to receive and process data from various sensors and issue control commands to the smart home. The photodiode refers to a semiconductor device used to detect light intensity and convert light intensity into a current signal. For example, a built-in diode in a curtain motor monitors the light intensity in the window area.

[0090] Optionally, the process of capturing the skin conductivity corresponding to the user's skin conductivity signal through the interdigital electrode array when the user's skin conductivity signal penetrates the light-transmitting channel is as follows: After applying 0.1 mA of alternating current to the interdigital electrode array, an alternating electric field is formed between the electrodes. When the skin conductivity signal penetrates the light-transmitting channel, it affects the impedance of this electric field. By detecting the impedance change, the interdigital electrode array can capture the skin conductivity. Skin conductivity is related to sweat gland activity and reflects the user's emotional state. Furthermore, the process of capturing the final electromyographic signal corresponding to the user's muscle contraction electrical signal through the nano-silver wire electrode when the user's muscle contraction electrical signal penetrates the light-transmitting channel is as follows: The nano-silver wire electrode has high conductivity and good biocompatibility, and can accurately detect the weak electrical signals generated by muscle contraction (i.e., the final electromyographic signal). These signals internally reflect the intensity and frequency of muscle contraction and can be used for subsequent intention judgment. The process of acquiring the pressure distribution corresponding to the user's pressure signal through the light-transmitting channel by the pressure-sensitive resistor array is as follows: When the user's finger pressure is applied to the pressure-sensitive resistor array, the resistance value of the pressure-sensitive resistor array changes, and the degree of change is related to the pressure magnitude. By measuring the resistance change, the pressure-sensitive resistor array acquires pressure distribution information, thereby generating a pressure distribution map with a resolution of 0.1 Newtons, accurately presenting the pressure distribution on the skin. Furthermore, in the central controller, the conversion of the environmental reflection signal into environmental point cloud data is as follows: The received reflection signal is amplified and filtered to remove noise and interference and improve signal quality. The pre-processed analog signal is converted into a digital signal by an analog-to-digital converter (ADC). The digital signal is then analyzed in the frequency domain using FFT. The frequency offset between the frequency-modulated continuous wave and the environmental reflection signal is calculated by FFT, and the distance information of the target can be obtained. The specific formula is as follows:

[0091]

[0092] in, Indicates frequency offset. Indicates the distance to the target. Represents the speed of light. The frequency is indicated. After obtaining the distance between each target (indoor person, object) and the millimeter-wave radar system, the three-dimensional point cloud coordinates of each target can be analyzed, which is the environmental point cloud data.

[0093] Optionally, the process of reconstructing the ambient light intensity distribution into light intensity distribution data using the spherical harmonic expansion is as follows: Since the photodiodes are not located in a large continuous space indoors, but rather dispersed across different pieces of furniture, the collected ambient light intensity distribution is not necessarily regularly distributed. Therefore, spherical Dirichlet interpolation is first performed on the discrete sampling points (ambient light intensity distribution) to fill in the angular areas not directly covered. The spherical harmonic expansion of the ambient light intensity distribution is then constructed using the following formula:

[0094]

[0095] in, Represents the spherical harmonic basis functions, used to describe spatial angles. Ambient light intensity distribution in the direction, The polar angle represents the angle from the positive z-axis to that direction. The azimuth angle represents the angle from the x-axis to that direction on the xy-plane. These represent the spherical harmonic coefficients, which are the weights corresponding to the spherical harmonic function. These coefficients are calculated by fitting the collected ambient light intensity distribution and are used to describe the characteristics of the ambient light intensity distribution. Represents the angle in spherical coordinates. (Polar angle) and Illumination intensity in the (azimuth) direction (this illumination intensity is not the ambient light intensity distribution, but is obtained through modeling and calculation). This represents a double summation of the spherical harmonic basis functions, where l is the order of the spherical harmonics and m is the degree of the spherical harmonics. The spherical harmonic basis functions are used to expand the illumination intensity into components of different frequencies and directions. The purpose of constructing the spherical harmonic expansion of the ambient light intensity distribution is to perform the following steps. The specific determination of the value means that the above-mentioned In the formula, Since it is unknown, the expansion coefficients can be solved using the least squares method. The required formula is as follows:

[0096]

[0097] in, This represents the measured value of the i-th photodiode, which is the ambient light intensity distribution. This represents the spatial azimuth angle of the i-th photodiode. express and The error between them can be solved using the least squares method. In simple terms, the specific value of S is the value of S with respect to each coefficient. Taking the partial derivatives and setting them to zero yields a system of linear equations. Solving this system of equations gives the spherical harmonic coefficients. and will As light intensity distribution data.

[0098] S2. Perform spatiotemporal coding processing on the bioelectric signal to obtain a biological feature vector, encode the environmental point cloud data and the light intensity distribution data into a behavior pattern vector, and extract the spatial thermodynamic field corresponding to the environmental reflection signal and the environmental point cloud data.

[0099] In this embodiment of the invention, the bioelectric signals are spatiotemporally encoded to preprocess the collected biosignals and convert them into computer-recognizable data.

[0100] In one embodiment of the present invention, the step of performing spatiotemporal encoding processing on the bioelectric signal to obtain a biofeature vector includes: acquiring skin conductivity, final electromyography (EMG) signal, and pressure distribution from the bioelectric signal; performing piecewise integration on the skin conductivity to obtain an area value sequence; extracting fluctuation periods from the area value sequence; performing Poisson encoding processing on the fluctuation periods to obtain a pulse sequence; inputting the pulse sequence into a preset LIF neuron model to obtain an emotion label output by the LIF neuron model; performing signal filtering processing on the final EMG signal through an SNN encoder in the central controller to obtain a filtered EMG signal; calculating the spatiotemporal gradient of the pressure distribution; performing cross-modal multiplication and fusion of the filtered EMG signal and the spatiotemporal gradient to obtain an intent intensity; and compressing the emotion label, the intent intensity, and the spatiotemporal gradient into a biofeature vector through a fully connected SNN layer in the central controller.

[0101] The fluctuation period refers to the periodic change of the area value sequence over time, that is, the time interval between the repetition of each area value in the area value sequence. For example, when a user is in an anxious state, the fluctuation period of the area value sequence is usually less than 0.5Hz, indicating that the area values ​​change more frequently. When the user is in a calm state, the fluctuation period of the area value sequence is greater than 2Hz, and the changes are relatively sparse. This periodic change reflects the activity changes of the autonomic nervous system, thereby revealing the user's emotional state. The LIF neuron model refers to a model that simulates the behavior of biological neurons. It describes how neurons accumulate input electrical signals (integration process) and fire action potentials (pulses) when a certain threshold is reached. Afterward, the membrane potential of the neuron is rapidly reset (leakage process). The membrane potential formula in the LIF neuron model is: , As the input current, in a spiking neural network, when a pulse sequence reaches a neuron, each pulse carries a certain amount of charge; this charge is... , The membrane potential at the current moment. This represents the membrane potential at the previous moment. This represents the natural decay of the membrane potential. The term "membrane capacitance" represents the neuron's membrane's ability to store charge. The emotion labels include anxiety labels, calm labels, etc. For example, when the pulse sequence frequency is high (e.g., period less than 0.5 Hz), the LIF neuron model rapidly accumulates potentials and fires action potentials, outputting a binary code "01" to indicate the user is in an anxious state. When the pulse sequence frequency is low (e.g., period greater than 2 Hz), the LIF neuron model slowly accumulates potentials, outputting a binary code "10" to indicate the user is in a calm state. The SNN encoder refers to a signal processing module based on a spiking neural network, used to load band-stop filter weights to suppress 50 Hz power frequency interference. The band-stop filter in the SNN encoder has a center frequency of 50 Hz and a bandwidth of... The fully connected layer of the SNN is a layer structure in a Spiking Neural Network (SNN) where neurons are fully connected to neurons in the preceding and following layers. The function of the fully connected SNN is to comprehensively process the input data, fusing the input data through weights. For example, using an SNN fully connected layer for feature compression (256 nodes in the input layer, 64 nodes in the hidden layer, and 32 nodes in the output layer), a 128-dimensional biometric feature vector [emotional tag] can be output. Intent strength [62-bit spatiotemporal gradient].

[0102] Optionally, the process of performing piecewise integration on the skin conductivity to obtain the area value sequence is as follows: the continuous signal is divided into segments according to a time window, and integration is performed on each segment to obtain the area value. For example, the skin conductivity is integrated piecewise per second to obtain the signal area value per second. The area value sequence refers to a sequence composed of a series of area values ​​obtained by piecewise integration. Further, the process of performing Poisson coding on the fluctuation period to obtain the pulse sequence is as follows: based on the fluctuation period of the area value sequence, a pulse sequence with intervals corresponding to the period is generated. For example, if the fluctuation period extracted from the area value sequence is 0.3Hz (anxiety state), then according to Poisson coding... The rules generate a corresponding pulse sequence, where the pulse interval corresponds to the fluctuation period. This pulse sequence retains the key features of the original signal while representing the signal in a form more suitable for subsequent digital signal processing. Poisson coding is a coding method that converts analog signals into pulse sequences. Specifically, it is a non-random coding method that, based on the periodic characteristics of the signal and the probability distribution of the Poisson process, converts the signal into a series of discrete pulse events. The pulse sequence is a series of discrete pulse signals, each pulse having a short duration and a certain amplitude. Furthermore, the formula for calculating the spatiotemporal gradient of the pressure distribution is as follows:

[0103]

[0104] in, Represents the spatiotemporal gradient, characterizing the rate of change of pressure. Indicates pressure distribution. Represents a time variable. Represents the XY axis variables in space.

[0105] Furthermore, the formula for obtaining the intent intensity by performing cross-modal product fusion of the filtered electromyographic signal and the spatiotemporal gradient is as follows: Intent , The input represents the normalized filtered electromyographic signal, where Intent represents the intensity of intent, and a and b are trainable parameters fitted using methods such as least squares. This represents the normalized spatiotemporal gradient.

[0106] Furthermore, in this embodiment of the invention, the environmental point cloud data and the light intensity distribution data are encoded into a behavior pattern vector to identify the user's behavior category.

[0107] In one embodiment of the present invention, encoding the environmental point cloud data and the light intensity distribution data into a behavior pattern vector includes: extracting joints from the environmental point cloud data using the alpha-shape algorithm in the central controller; analyzing the joint motion vectors corresponding to the joints using the graph convolutional network in the central controller; calculating the gradient optical flow features of the light intensity distribution data; identifying the action features corresponding to the environmental point cloud data and the light intensity distribution data based on the joint motion vectors and the gradient optical flow features; and generating the spatiotemporal hash value of the action features using the SHA-256 algorithm in the central controller.

[0108] The Alpha-shape algorithm is used to extract geometric shapes and structures from point cloud data. It can identify topological features in point cloud data, such as boundaries, holes, and connected regions. In a smart home environment, the Alpha-shape algorithm can be used to process environmental point cloud data generated by millimeter-wave radar systems and extract the positional information of human joints, such as elbow and knee joints. The graph convolutional network (GCN) is a deep learning model used to process graph-structured data. It can learn the relationships and features between nodes in the graph and is suitable for processing data with complex relationships. In a smart home system, GCN can be used to analyze the relationships between joints extracted by the Alpha-shape algorithm and identify joint motion vectors, such as the swing direction and speed of the arm. The calculation formula for the gradient optical flow feature is as follows:

[0109]

[0110] in, Current frame -Previous frame , Indicates the rate of change of light intensity. Indicates gradient optical flow characteristics,

[0111] The motion features refer to the features that can be extracted from a user, such as walking speed, gesture trajectory, and fall actions, by analyzing the joint motion vectors and the gradient optical flow features. For example, the joint motion vector is the elbow joint angle. Knee curvature ,when and The fall event is time-marked as the action feature. Further, the spatiotemporal hash value of the action feature is generated by the SHA-256 algorithm in the central controller: after the user's fall action is identified, the action feature data of the action is input into the SHA-256 algorithm. The SHA-256 algorithm will encrypt the data and generate a 64-character hash value. This hash value can be used to record and identify the user's fall event, while protecting the user's privacy, because the hash value cannot directly restore the original action feature.

[0112] Furthermore, in this embodiment of the invention, the spatial thermal field corresponding to the environmental reflection signal and the environmental point cloud data is extracted to identify which areas are frequently used by users.

[0113] In one embodiment of the present invention, the step of extracting the spatial thermal field corresponding to the environmental reflection signal and the environmental point cloud data includes: performing Wigner-Ville distribution analysis on the environmental reflection signal to determine the time-frequency distribution of the environmental reflection signal; extracting the user's breathing signal from the time-frequency distribution; calculating the user's breathing frequency using the breathing signal; and determining the spatial thermal field of the environment in which the smart home is located based on the breathing frequency.

[0114] It should be noted that Wigner-Ville distribution analysis (WVD) unfolds the signal in the time and frequency domain through high-resolution time-frequency analysis, clearly presenting the distribution of different frequency components. Respiratory signals (0.1-0.5Hz) and heartbeat signals (about 1Hz) can be distinguished on the time-frequency graph because of their different frequencies. Then, by using power spectral density calculation or frequency domain analysis, the respiratory frequency corresponding to the respiratory signal can be determined.

[0115] In another embodiment of the present invention, determining the spatial thermal field of the environment in which the smart home is located based on the breathing frequency includes: acquiring environmental point cloud data and a thermal diffusion coefficient; using the environmental point cloud data to determine the user's spatial coordinates and the coordinates of objects in the environment in which the smart home is located; calculating the point cloud density parameter corresponding to the environmental point cloud data; calculating the user's activity power based on the thermal diffusion coefficient, the spatial coordinates, the object coordinates, the breathing frequency, and the point cloud density parameter; and generating the spatial thermal field of the environment in which the smart home is located based on the user's activity power.

[0116] For example, the formula for calculating the user's activity power is as follows:

[0117]

[0118] in, For user activity power, These are the object coordinates, that is, the spatial coordinates of each object in the millimeter-wave point cloud data. , These are spatial coordinates, specifically the user coordinates used for millimeter-wave radar positioning. It is the control parameter for spatial decay in the Gaussian function, with values ​​ranging from 0.3 to 1 meter, used to characterize the radius of thermal diffusion. Respiratory rate, This represents the metabolic energy within each time period, for example... =5J / time 20 times / minute = 100J / minute. The space thermal field is composed of the power of user activities at different coordinates. This represents the total number of point clouds globally.

[0119] See Figure 2 The diagram shown illustrates the user activity power of an IoT smart home control method based on personalized services, according to an embodiment of the present invention.

[0120] S3. Establish a topology diagram between various smart home devices. Based on the biometric vector, the behavior pattern vector, and the topology diagram, construct a cross-device collaboration strategy between various smart home devices.

[0121] This invention establishes a topological relationship diagram between various smart home devices to create a smart topological association between them.

[0122] In one embodiment of the present invention, establishing a topology graph between various smart home devices includes: generating an initial adjacency matrix between various smart home devices using a pulse signaler embedded in each smart home device; obtaining an intent strength; updating the weights of the initial adjacency matrix based on the intent strength to obtain an updated adjacency matrix; and converting the updated adjacency matrix into a topology graph between various smart home devices.

[0123] Optionally, the process of generating the initial adjacency matrix between smart home devices by using pulse signalers embedded in each smart home device is as follows: the devices are directly connected via millimeter-wave pulses (non-IP protocol) to form a neural synapse-like structure. During the initial network setup, the central controller broadcasts topology-generated pulses, and each device generates initial weights based on the signal strength. (Attenuation model) Equipment spacing, Maximum communication distance Let be the value in the i-th row and j-th column of the initial adjacency matrix, representing the edge weight between the i-th and j-th homes. Further, the formula for updating the weights of the initial adjacency matrix based on the intent strength is as follows: , It is 0.7. The value is 0.3. Both are parameters that have been fitted in advance using optimization algorithms such as least squares. The topological graph includes edges, nodes, and weights on the edges.

[0124] Furthermore, in this embodiment of the invention, a cross-device collaboration strategy is constructed between various smart home devices based on the biometric vector, the behavioral pattern vector, and the topological relationship graph, so as to carry out collaborative decision-making for smart homes.

[0125] In one embodiment of the present invention, the step of constructing a cross-device collaboration strategy among various smart home devices based on the biometric vector, the behavior pattern vector, and the topology graph includes: inputting the biometric vector, the behavior pattern vector, and the topology graph into a preset adversarial reinforcement learning model; and obtaining the cross-device collaboration strategy output by the adversarial reinforcement learning model regarding the biometric vector, the behavior pattern vector, and the topology graph.

[0126] The adversarial reinforcement learning model is a model architecture that integrates generative adversarial networks (GANs) and reinforcement learning (RL). It includes an adversarial mechanism and a reinforcement learning framework. The adversarial mechanism learns the potential distribution of data through a game between a generator and a discriminator. The reinforcement learning framework uses a policy network as the decision core and optimizes the collaborative policy by maximizing the cumulative reward. The generated cross-device collaborative policy includes the parameter adjustment amount of each home device. For example, the air conditioner lowers the temperature by 3 degrees Celsius and the refrigerator raises the temperature by 5 degrees Celsius.

[0127] See Figure 3 The diagram shown is a schematic representation of the generation of a cross-device collaboration strategy in an IoT smart home control method based on personalized services, according to an embodiment of the present invention.

[0128] S4. Determine the real-time control commands for the smart home based on the biometric vector, the behavioral pattern vector, and the spatial thermal field.

[0129] This invention determines the real-time control commands for the smart home based on the biometric vector, the behavioral pattern vector, and the spatial thermal field, thereby effectively utilizing spatial energy information for precise control of the smart home.

[0130] In one embodiment of the present invention, determining the real-time control command of the smart home based on the biometric vector, the behavioral pattern vector, and the spatial thermal field includes: acquiring emotion-related, intention-related, and stress-related home appliances corresponding to the biometric vector; generating emotion control commands, intention control commands, and stress control commands for the emotion-related, intention-related, and stress-related home appliances, respectively, based on the biometric vector; identifying high-energy-density home appliances in the smart home using the spatial thermal field; generating energy-saving commands for the high-energy-density home appliances based on the high-energy-density values ​​corresponding to the high-energy-density home appliances in the spatial thermal field; activating special control commands for the smart home using the behavioral pattern vector; and using the emotion control commands, intention control commands, stress control commands, energy-saving commands, and special control commands as the real-time control commands for the smart home.

[0131] It should be noted that the "emotion-related home furnishings" refer to home furnishings that affect the user's emotions, such as smart lighting, aromatherapy diffusers, integrated massage chairs, and sleep aid speakers. The "intention-related home furnishings" refer to home furnishings that are adjusted according to the user's intentions, such as the ability to adjust the air conditioner temperature at any time. The "stress-related home furnishings" refer to home furnishings that can relieve the user's stress. When a person is under high stress, it may manifest as subconscious physical actions, such as pressing buttons on devices with excessive force or at a rapid pace. In such cases, the detected stress distribution information can be used to determine whether to activate stress-relieving devices such as massage chairs. The emotion control commands, intention control commands, and stress control commands all involve turning on the device and adjusting the intensity level. Commands, such as when a user needs to relieve stress, if the stress-relieving device is not turned on, can be turned on and the user's usual setting adjusted to achieve a stress control command. The high-energy-density home refers to areas where the thermal field exceeds a threshold value, such as the refrigerator area where E is greater than 0.8, indicating high-frequency activity by the user in the refrigerator area. In this case, energy saving is required for the refrigerator. Therefore, the refrigerator temperature is increased by 5 degrees Celsius at fixed energy-saving intervals, such as 5 degrees Celsius. Increasing the refrigerator temperature by 5 degrees Celsius is an energy-saving command. Special control commands refer to the activation of an alarm system when abnormal user behavior is detected, such as a fall. Activating the alarm system is considered a special control command.

[0132] S5. Control the smart home through the cross-device collaboration strategy and the real-time control command to obtain the home control result.

[0133] Compared to the problems described in the background technology, the embodiments of the present invention address the issue of a single and superficial dimension of user state perception by collecting multi-dimensional data. The embodiments of the present invention perform spatiotemporal encoding processing on the bioelectrical signals to preprocess and convert the collected biosignals into computer-recognizable data. Furthermore, the embodiments of the present invention encode the environmental point cloud data and the light intensity distribution data into behavioral pattern vectors to identify user behavior categories. Furthermore, the embodiments of the present invention extract the spatial thermal field corresponding to the environmental reflection signal and the environmental point cloud data to identify which areas are frequently used by the user. Furthermore, the embodiments of the present invention establish a topological relationship graph between various smart home devices to establish intelligent topological associations between devices. Furthermore, the embodiments of the present invention construct cross-device collaborative strategies between various smart home devices based on the biofeature vectors, the behavioral pattern vectors, and the topological relationship graph to perform collaborative decision-making for smart homes. The embodiments of the present invention determine the real-time control commands of the smart home devices based on the biofeature vectors, the behavioral pattern vectors, and the spatial thermal field to effectively utilize spatial energy information for precise control of smart homes. Therefore, the IoT smart home control method and system based on personalized services provided in this embodiment of the invention can deeply integrate multi-dimensional biological and environmental signals, establish intelligent topological associations between devices, and effectively utilize spatial energy information for collaborative decision-making and precise control.

[0134] Example 2:

[0135] like Figure 4 The diagram shown is a functional block diagram of an IoT smart home control system based on personalized services according to the present invention.

[0136] The IoT smart home control system 400 based on personalized services described in this invention can be installed in an electronic device. Depending on the functions implemented, the IoT smart home control system based on personalized services may include a data acquisition module 401, a thermal field extraction module 402, a strategy construction module 403, an instruction determination module 404, and a home control module 405. The module described in this invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0137] In this embodiment of the invention, the functions of each module / unit are as follows:

[0138] The data acquisition module 401 is used to collect the user's bioelectric signals, environmental reflection signals, environmental point cloud data and light intensity distribution data in the smart home.

[0139] The thermal field extraction module 402 is used to perform spatiotemporal encoding processing on the bioelectric signal to obtain a biological feature vector, encode the environmental point cloud data and the light intensity distribution data into a behavior pattern vector, and extract the spatial thermal field corresponding to the environmental reflection signal and the environmental point cloud data.

[0140] The strategy construction module 403 is used to establish a topology graph between various smart home devices, and to construct a cross-device collaboration strategy between various smart home devices based on the biometric vector, the behavior pattern vector and the topology graph.

[0141] The instruction determination module 404 is used to determine the real-time control instructions of the smart home based on the biometric vector, the behavior pattern vector and the spatial thermal field.

[0142] The home control module 405 is used to control the smart home through the cross-device collaboration strategy and the real-time control commands to obtain home control results.

[0143] In detail, the modules in the IoT smart home control system 400 based on personalized services described in this embodiment of the invention employ the same methods as described above. Figure 1 The method uses the same technical means as the IoT smart home control method based on personalized services described above, and can produce the same technical effect, so it will not be elaborated here.

[0144] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for controlling an Internet of Things (IoT) smart home based on personalized services, the method comprising: The method comprises: collecting bioelectric signals, environmental reflection signals, environmental point cloud data and light intensity distribution data of a user on a smart home; temporally and spatially encoding the bioelectric signals to obtain a biological feature vector, encoding the environmental point cloud data and the light intensity distribution data into a behavior pattern vector, and extracting a spatial thermal field corresponding to the environmental reflection signals and the environmental point cloud data, comprising: performing Wigner-Ville distribution analysis on the environmental reflection signals to determine the time-frequency distribution of the environmental reflection signals; extracting a user's breathing signal from the time-frequency distribution; calculating the user's breathing frequency using the breathing signal; determining the spatial thermal field in the environment where the smart home is located according to the breathing frequency, comprising: obtaining environmental point cloud data; determining the spatial coordinates of the user and the object coordinates in the environment where the smart home is located using the environmental point cloud data; calculating the global point cloud number corresponding to the environmental point cloud data; calculating the user activity power of the user according to the spatial coordinates, the object coordinates, the breathing frequency and the global point cloud number; generating the spatial thermal field in the environment where the smart home is located through the user activity power; establishing a topological relationship diagram between each smart home, and constructing a cross-device collaboration strategy between each smart home based on the biological feature vector, the behavior pattern vector and the topological relationship diagram; determining the real-time control instruction of the smart home according to the biological feature vector, the behavior pattern vector and the spatial thermal field; controlling the smart home through the cross-device collaboration strategy and the real-time control instruction to obtain a home control result. 2.The IoT smart home control method based on personalized service of claim 1, wherein, The method comprises: identifying a high-frequency contact surface of a user on a smart home; deploying a flexible AMOLED skin on the high-frequency contact surface; wherein the flexible AMOLED skin comprises a substrate layer and a light transmission channel, the substrate layer comprising an interdigital electrode array, a nano-silver wire electrode and a piezoresistor array; when the skin conductivity signal of the user penetrates the light transmission channel, capturing the skin conductivity corresponding to the skin conductivity signal through the interdigital electrode array; when the muscle contraction electrical signal of the user penetrates the light transmission channel, capturing the final electromyographic signal corresponding to the muscle contraction electrical signal through the nano-silver wire electrode; when the pressure signal of the user penetrates the light transmission channel, collecting the pressure distribution corresponding to the pressure signal through the piezoresistor array; taking the skin conductivity, the final electromyographic signal and the pressure distribution as bioelectric signals; embedding a millimeter wave radar system and a photodiode on the smart home; wherein the millimeter wave radar system comprises a TDM-MIMO antenna array and an adaptive beam algorithm; based on the adaptive beam algorithm, transmitting a frequency-modulated continuous wave into the environment where the smart home is located through the TDM-MIMO antenna array; receiving the environmental reflection signal of the frequency-modulated continuous wave in the millimeter wave radar system; The ambient reflection signal is transmitted to the corresponding central controller of the smart home; In the central controller, the ambient reflection signal is converted into ambient point cloud data; The ambient light intensity distribution in the environment where the smart home is located is collected through the photosensitive diode; The ambient light intensity distribution is reconstructed into light intensity distribution data through spherical harmonic expansion. 3.The IoT smart home control method based on personalized service of claim 1, wherein, The biological characteristic vector is obtained by spatiotemporal coding processing of the bioelectric signal, including: Obtaining skin conductivity, final electromyographic signal and pressure distribution in the bioelectric signal; Segmented integration is performed on the skin conductivity to obtain an area value sequence; The fluctuation period is extracted from the area value sequence; Poisson coding processing is performed on the fluctuation period to obtain a pulse sequence; The pulse sequence is input into a preset LIF neuron model to obtain an emotional label output by the LIF neuron model; The final electromyographic signal is filtered through an SNN encoder in the central controller to obtain a filtered electromyographic signal; The spatiotemporal gradient of the pressure distribution is calculated; The filtered electromyographic signal and the spatiotemporal gradient are cross-modally multiplied and fused to obtain an intention intensity; The emotional label, the intention intensity and the spatiotemporal gradient are compressed into a biological characteristic vector through an SNN fully connected layer in the central controller. 4.The IoT smart home control method based on personalized service of claim 1, wherein, The ambient point cloud data and the light intensity distribution data are encoded into a behavior mode vector, including: The alpha-shape algorithm in the central controller is used to extract the key nodes in the ambient point cloud data; The joint motion vector corresponding to the key nodes is analyzed by combining the graph convolution network in the central controller; The gradient optical flow feature of the light intensity distribution data is calculated; According to the joint motion vector and the gradient optical flow feature, the motion feature corresponding to the ambient point cloud data and the light intensity distribution data is identified; The spatiotemporal hash value of the motion feature is generated through the SHA-256 algorithm in the central controller. 5.The IoT smart home control method based on personalized service of claim 1, wherein, The topological relationship graph between each smart home is established, including: An initial adjacency matrix between each smart home is generated by a pulse signal generator embedded on each smart home; An intention intensity is obtained; The initial adjacency matrix is updated based on the intention intensity to obtain an updated adjacency matrix; The updated adjacency matrix is converted into a topological relationship graph between each smart home. 6.The IoT smart home control method based on personalized service of claim 1, wherein, Based on the biological characteristic vector, the behavior mode vector and the topological relationship graph, a cross-device collaboration strategy between each smart home is constructed, including: The biological characteristic vector, the behavior mode vector and the topological relationship graph are input into a preset adversarial reinforcement learning model; The cross-device collaboration strategy output by the adversarial reinforcement learning model about the biological characteristic vector, the behavior mode vector and the topological relationship graph is obtained. 7.The IoT smart home control method based on personalized service of claim 1, wherein, According to the biological characteristic vector, the behavior mode vector and the spatial thermal field, a real-time control instruction of the smart home is determined, including: The emotional related home, the intention related home and the pressure related home corresponding to the biological characteristic vector are obtained; Generate emotion control instructions, intention control instructions and stress control instructions of the emotion-related home, the intention-related home and the stress-related home respectively according to the biological feature vector; Identify high energy density homes in the intelligent home by using the space thermal field; Generate energy saving instructions of the high energy density home based on the corresponding high energy density value of the high energy density home in the space thermal field; Start special control instructions of the intelligent home by using the behavior pattern vector; Take the emotion control instructions, the intention control instructions, the stress control instructions, the energy saving instructions and the special control instructions as real-time control instructions of the intelligent home.

8. An Internet of Things smart home control system based on personalized service, characterized by, The system comprises: A data acquisition module for acquiring bioelectric signals, environmental reflection signals, environmental point cloud data and light intensity distribution data of a user on an intelligent home; A thermal field extraction module for performing space-time coding processing on the bioelectric signals to obtain a biological feature vector, coding the environmental point cloud data and the light intensity distribution data into a behavior pattern vector, and extracting a space thermal field corresponding to the environmental reflection signals and the environmental point cloud data, comprising: Performing Wigner-Ville distribution analysis on the environmental reflection signals to determine the time-frequency distribution of the environmental reflection signals; Extracting the breathing signal of the user from the time-frequency distribution; Calculating the breathing frequency of the user by using the breathing signal; Determining the space thermal field in the environment where the intelligent home is located according to the breathing frequency, comprising: Obtaining environmental point cloud data; Determining the spatial coordinates of the user and the object coordinates in the environment where the intelligent home is located by using the environmental point cloud data; Calculating the global point cloud number corresponding to the environmental point cloud data; Calculating the user activity power of the user according to the spatial coordinates, the object coordinates, the breathing frequency and the global point cloud number; Generating the space thermal field in the environment where the intelligent home is located by using the user activity power; A strategy construction module for establishing a topological relationship graph between each intelligent home, and constructing a cross-device collaboration strategy between each intelligent home based on the biological feature vector, the behavior pattern vector and the topological relationship graph; An instruction determination module for determining real-time control instructions of the intelligent home according to the biological feature vector, the behavior pattern vector and the space thermal field; A home control module for controlling the intelligent home by using the cross-device collaboration strategy and the real-time control instructions to obtain a home control result.

Images