An intelligent home regulation method based on internet of things

Abstract

The application relates to the technical field of smart home, and particularly discloses a smart home regulation and control method based on the Internet of Things, which comprises the following steps: S1, an environment sensing module collects environment parameters through a distributed sensor network; S2, a user behavior analysis module constructs a user behavior probability model based on historical operation data and real-time action recognition, and outputs a behavior preference weight; S3, a cloud decision engine generates a device control instruction set by fusing the environment parameters and the behavior preference weight, and calculates a regulation and control priority; and S4, a device execution network drives home devices according to the instruction set, and optimizes regulation and control parameters through a feedback mechanism. The smart home regulation and control method based on the Internet of Things adopts the advanced Internet of Things technology, fuses multiple modules such as environment sensing, user behavior analysis, cloud decision and device execution, realizes the intelligent and automatic regulation and control of home devices, and creates a comfortable, convenient and safe home environment for users.

Description

Technical Field

[0001] This invention relates to the field of smart home technology, and in particular to a smart home control method based on the Internet of Things. Background Technology

[0002] With the rapid advancement of technology and the continuous improvement of people's quality of life, the smart home market is experiencing rapid development. Its vision of a comfortable and convenient home life attracts many consumers. However, behind this booming development, existing problems are gradually becoming apparent, seriously hindering further market expansion.

[0003] In traditional home systems, the independent operation of various devices has become a major obstacle to convenient living for users. Air conditioners, lighting, curtains, and other devices are like isolated "islands," forcing users to operate them separately and make tedious settings on different devices. This not only wastes time and energy, but also lacks synergy between devices, making it difficult to form an organic whole and thus failing to create the ideal comfortable and convenient home environment for users.

[0004] Even though some smart home systems have achieved network connectivity, their level of intelligence remains unsatisfactory. In terms of data collection, single sensors have limited functionality and cannot comprehensively perceive multiple environmental factors. For example, temperature sensors can only acquire temperature information but cannot take into account other important indicators such as humidity and air quality; furthermore, the data collection frequency is too low to capture dynamic changes in the environment in a timely manner. In smart home systems, this lag leads to device adjustments failing to meet actual needs, significantly reducing the user experience.

[0005] The same problems exist in the control decision-making stage. Due to a lack of efficient data processing and in-depth analysis methods, the system struggles to make scientific and intelligent decisions based on the collected data. Furthermore, traditional home control systems rely excessively on network stability, which introduces numerous hidden dangers in practical use. In remote areas with weak network signals or in congested urban environments, device control is prone to delays or even malfunctions. This not only affects normal user experience but may also threaten home security.

[0006] The rise of IoT, embedded systems, and machine learning technologies has offered new possibilities for overcoming these bottlenecks. This invention aims to leverage advanced technologies to solve existing problems, propel the smart home market to new heights, and bring users a truly intelligent, convenient, and secure home experience. Summary of the Invention

[0007] The purpose of this invention is to provide a smart home control method based on the Internet of Things, which solves the problems of traditional smart home devices being independent and difficult to coordinate, incomplete and lagging data collection, lack of intelligence in control decisions, and over-reliance on the network, thereby improving the level of home intelligence and creating a comfortable, convenient, and safe home environment.

[0008] To achieve the above objectives, the present invention provides a smart home control method based on the Internet of Things, comprising the following steps:

[0009] S1. The environmental perception module collects environmental parameters through a distributed sensor network.

[0010] S2. The user behavior analysis module constructs a user behavior probability model based on historical operation data and real-time action recognition, and outputs the behavior preference weights W(u).

[0011] S3, the cloud-based decision engine generates a device control instruction set C by integrating environmental parameters and behavioral preference weights, and calculates the control priority;

[0012] S4. The device execution network drives home appliances according to the instruction set and optimizes control parameters through a feedback mechanism.

[0013] Preferably, in S1, the environmental parameters include temperature T, humidity H, light intensity L, air quality index A, and sound decibels D. The environmental sensing module uses multi-sensor fusion technology to fuse the data, and the fusion formula is:

[0014]

[0015] Where S is the fused environmental state value, α i S represents the weights of the i-th type of sensor. i Let be the measurement value of the i-th type of sensor, β be the noise correction coefficient, and σ be the environmental disturbance factor.

[0016] Preferably, in S1, the sampling frequency range of the environmental sensing module is 1-60Hz.

[0017] Preferably, in S1, the sensor weight α i Updated using an adaptive strategy based on the covariance matrix:

[0018]

[0019] Among them, S avg S is the average value of similar sensors. max S min These represent the maximum and minimum values ​​of the sensor's measurement range.

[0020] Preferably, in S1, the collected environmental parameter data is transmitted using the AES-256 encryption protocol, and the encryption strength E satisfies:

[0021]

[0022] Where B represents the data packet size, K represents the dynamic key length, and M represents the security redundancy coefficient.

[0023] Preferably, in S2, the user behavior analysis module uses a Hidden Markov Model (HMM) to calculate the user behavior probability P(u):

[0024]

[0025] Among them, O t For the observed behavior at time t, s t Let be the hidden state at time t. Let ψ be the observation probability, and ψ be the state transition function.

[0026] Preferably, in S3, the equipment control instruction set C = f(T,H,L,A,D,W), and the control priority Q is calculated using the following optimization equation:

[0027]

[0028] Where k1, k2, and k3 are the normalization coefficients of environmental parameters, and k1+k2+k3=1.

[0029] Preferably, in S3, the normalization coefficients k1, k2, and k3 are dynamically adjusted using a user physiological comfort model:

[0030]

[0031] Where, θ i C represents the user's personalized preference parameters. p This indicates real-time physiological indicators.

[0032] Preferably, in S4, the device execution network uses a fuzzy PID controller to optimize the control parameters, and the corresponding time R satisfies:

[0033]

[0034] in, This represents the device's basic response threshold, and N represents the number of currently online devices.

[0035] Therefore, the present invention employs the above-mentioned smart home control method based on the Internet of Things, and the beneficial effects are as follows:

[0036] (1) Through comprehensive analysis of environmental parameters and user behavior, the system can automatically and intelligently control home appliances. Users do not need to operate manually. The system can adjust the air conditioner temperature, turn the lights on and off in advance according to environmental changes and user habits, which greatly improves the convenience of life and allows users to enjoy the comfort brought by smart home.

[0037] (2) This invention uses multi-source sensors to collect rich environmental parameters and processes them through advanced fusion technology. Combined with scientific control priority calculation, the system can accurately control home appliances and maintain the indoor environment in an optimal state. For example, it can precisely adjust the working mode of the air purifier according to the indoor air quality index to provide users with a healthy living environment.

[0038] (3) This invention employs the AES-256 encrypted transmission protocol to ensure secure data transmission and reduce the risk of data theft or tampering. Simultaneously, the device execution network optimizes control parameters through a fuzzy PID controller and adjusts the response time according to the number of online devices, enhancing the stability and reliability of the system under different network environments and device load conditions, thus ensuring the stable operation of home appliances.

[0039] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the overall process of an embodiment of a smart home control method based on the Internet of Things according to the present invention;

[0041] Figure 2 This is a schematic diagram of an improved CNN model structure according to an embodiment of the smart home control method based on the Internet of Things of the present invention;

[0042] Figure 3 This is a schematic diagram of the device execution network implementation of an embodiment of the smart home control method based on the Internet of Things according to the present invention. Detailed Implementation

[0043] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0044] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0045] A smart home control method based on the Internet of Things includes the following steps:

[0046] S1. The environmental perception module collects environmental parameters through a distributed sensor network.

[0047] Environmental parameters include temperature (T), humidity (H), light intensity (L), air quality index (A), and sound level (D). The environmental sensing module uses multi-sensor fusion technology to fuse data, and the fusion formula is as follows:

[0048]

[0049] Where S is the fused environmental state value, α i S represents the weights of the i-th type of sensor. i Let be the measurement value of the i-th type of sensor, β be the noise correction coefficient, and σ be the environmental disturbance factor.

[0050] The noise correction factor β is used to eliminate random noise interference that may occur during sensor measurement, including thermal noise from electronic components and external electromagnetic interference. The environmental disturbance factor σ takes into account sudden and transient interference factors in the environment, such as airflow changes caused by suddenly opening doors and windows, or brief periods of strong light.

[0051] The environmental sensing module has a sampling frequency range of 1-60Hz. Different environmental parameters can be set with different sampling frequencies based on their rate of change. Temperature and humidity, which change relatively slowly, can be set with lower sampling frequencies, such as 1-5Hz; while light intensity and sound decibels may change rapidly, requiring higher sampling frequencies, such as 10-60Hz. This ensures data accuracy while avoiding data redundancy and resource waste caused by oversampling.

[0052] Sensor weight α i The covariance matrix is ​​updated using an adaptive strategy, and the formula is as follows:

[0053]

[0054] Among them, S avg S is the average value of similar sensors. max S min These represent the maximum and minimum values ​​of the sensor's measurement range. As time progresses and the environment changes, sensor performance may drift. An adaptive covariance matrix strategy can dynamically adjust the sensor weights, making the fused data more accurate and reliable. When the measurement value of a certain temperature sensor deviates significantly from the mean of similar sensors, its weight will be reduced accordingly to minimize its impact on the fusion result.

[0055] Meanwhile, the collected environmental parameter data is transmitted using the AES-256 encryption protocol, with encryption strength E satisfying:

[0056]

[0057] Where B represents the data packet size, K represents the dynamic key length, and M represents the security redundancy factor. AES-256 is an advanced encryption standard with high security. The dynamic key length K is dynamically adjusted according to the security requirements of data transmission, while the security redundancy factor M further enhances the reliability of encryption, preventing data from being stolen or tampered with during transmission.

[0058] S2. The user behavior analysis module constructs a user behavior probability model based on historical operation data and real-time action recognition, and outputs the behavior preference weight W(u).

[0059] The system collects user data on their interaction with various home appliances over the past year, including operation time and operation type (such as turning lights on / off, adjusting air conditioning temperature, and controlling curtains). Operation time reflects the user's daily routine, while operation type reflects their usage habits for different devices. Simultaneously, motion recognition cameras are installed in key activity areas such as the living room and bedroom to capture user movements in real time. These cameras utilize convolutional neural networks (CNNs) improved with deep learning algorithms to accurately identify user actions, such as gestures and postures, thereby gaining a more comprehensive understanding of the user's intentions.

[0060] The overall structure of the improved CNN model is as follows:

[0061] Input layer:

[0062] In smart homes, motion recognition cameras capture video frame image data containing user actions. Considering the large amount of video data in real-world scenarios, the input layer preprocesses the original video frames to reduce the computational burden. Downsampling technology is used to reduce the size of high-resolution video frames by a certain ratio; for example, downsampling a 1920×1080 resolution image to 224×224 resolution. This preserves key action features while significantly reducing the amount of data.

[0063] Normalizing image data maps pixel values ​​to the [0,1] interval, making it easier for the model to converge during training and improving training efficiency.

[0064] Convolutional layer group:

[0065] Parallel convolutional layers with different kernel sizes are introduced to capture multi-scale features, using 3×3, 5×5, and 7×7 kernels. 3×3 kernels capture local details, 5×5 kernels capture slightly larger-scale features, and 7×7 kernels capture even broader contextual information. The outputs of these parallel convolutional layers are concatenated along the channel dimension to enrich the feature representation.

[0066] To alleviate the vanishing gradient problem and enhance the network's ability to learn deep features, residual connections are added between convolutional layers. Taking two consecutive convolutional layers as an example, the input is processed by the first convolutional layer and the ReLU activation function, then added to the output of the second convolutional layer, and then input into the next convolutional layer. This structure enables the network to learn complex action features more effectively and makes the training process more stable.

[0067] By employing dilated convolution, the receptive field of the convolution kernel is expanded without increasing parameters or computational cost. For a specific convolutional layer, an appropriate dilation rate is set. For example, when the dilation rate is 2, the actual effective range of the convolution kernel is equivalent to that of a 5×5 convolution kernel, and when the dilation rate is 3, the actual effective range of the convolution kernel is equivalent to that of a 7×7 convolution kernel. This improves the model's ability to capture features at long distances and provides a more comprehensive understanding of user actions.

[0068] Pooling layer:

[0069] Following the convolutional layers, an improved deformable pooling layer replaces traditional max pooling or average pooling. The deformable pooling layer adaptively adjusts the pooling region based on the distribution of input features, better preserving key features related to user actions. For example, when recognizing a waving gesture, deformable pooling can more accurately locate the hand region and perform pooling operations, avoiding the loss of important information due to a fixed pooling region.

[0070] The output dimension of the pooling layer is adjusted reasonably according to the model complexity and computing resources. For example, the feature map size can be reduced by half while keeping the number of channels unchanged, so as to balance the amount of computation and the preservation of feature information.

[0071] Attention mechanism module:

[0072] Following the pooling layer, this invention introduces channel attention mechanisms (such as the Squeeze-and-Excitation module) and spatial attention mechanisms (such as the SpatialAttentionModule). The channel attention mechanism obtains global features for each channel through global average pooling, then calculates channel weights using fully connected layers and activation functions, reweighting the features of different channels to highlight important channels relevant to action recognition. The spatial attention mechanism generates a spatial attention map through convolutional operations, weighting different spatial locations of the feature map to focus on key regions where user actions occur. This invention combines both mechanisms, enabling the model to more effectively focus on key action features and improve recognition accuracy.

[0073] Fully connected layer:

[0074] The feature maps processed by the attention mechanism are unfolded into one-dimensional vectors and input into the fully connected layer. To prevent overfitting, a Dropout layer is added between the fully connected layers to randomly drop a certain proportion (e.g., 0.5) of the neuron connections, thereby enhancing the model's generalization ability.

[0075] Adjust the number of neurons in the fully connected layer and optimize it according to actual needs and dataset size. For smart home action recognition tasks, the number of neurons can be appropriately reduced to reduce computational complexity while ensuring model performance, so as to adapt to the limited computing resources of smart home devices.

[0076] Output layer:

[0077] The output layer uses a Softmax classifier to output the probability distribution of various user actions. Assuming the model needs to identify 10 common user actions, the output layer will have 10 neurons, each corresponding to an action category, outputting the predicted probability of that category of action. The category with the highest probability value is the user action identified by the model.

[0078] The user behavior analysis module uses a Hidden Markov Model (HMM) to calculate the user behavior probability P(u):

[0079]

[0080] Among them, O t For the observed behavior at time t, s t Let be the hidden state at time t. Let ψ be the observation probability, and ψ be the state transition function.

[0081] Hidden Markov Models (HMMs) are suitable for processing data with sequential characteristics. By using historical operation data and real-time action recognition data as input, HMMs are trained to learn patterns and regularities in user behavior. For example, a model can learn that users typically turn off living room lights and turn on bedroom lights at specific times in the evening, thus predicting the probability of the user's behavior at the same time in the future.

[0082] The trained Hidden Markov Model can accurately output the weights of user behavior preferences for operating various devices in different scenarios, reflecting the likelihood of users operating different devices and providing an important basis for subsequent cloud-based decision-making.

[0083] S3, the cloud-based decision engine generates a device control instruction set C by integrating environmental parameters and behavioral preference weights, and calculates the control priority;

[0084] The cloud-based decision engine receives environmental parameters uploaded by the environmental perception module and behavioral preference weights output by the user behavior analysis module in real time. It then fuses these two parameters using a preset algorithm and generates a set of device control commands based on the fusion result. For example, when the environmental perception module detects an increase in indoor temperature and the user behavior preference weights indicate that the user frequently lowers the air conditioning temperature under similar conditions, it generates a corresponding air conditioning temperature adjustment command. Simultaneously, if the light intensity is too strong and the user has frequently drawn the curtains under such conditions in the past, it generates a command to close the curtains.

[0085] The equipment control instruction set C = f(T,H,L,A,D,W), and the control priority Q are calculated using the following optimization equation:

[0086]

[0087] Where k1, k2, and k3 are the normalization coefficients of environmental parameters, and k1+k2+k3=1.

[0088] The normalization coefficients k1, k2, and k3 are dynamically adjusted using a user physiological comfort model.

[0089]

[0090] Where, θ i C represents the user's personalized preference parameters. p This system displays real-time physiological indicators. User-specific preferences can be set by the user within the system, such as preferred temperature and humidity ranges. Real-time physiological indicators can be acquired through wearable devices (such as smart bracelets and smartwatches), including heart rate and body temperature. By comprehensively considering environmental parameters and user behavioral preferences, a control priority is calculated, prioritizing device control commands that have a greater impact on user comfort, ensuring the system can respond quickly and meet user needs.

[0091] S4. The device execution network drives home appliances according to the instruction set and optimizes control parameters through a feedback mechanism.

[0092] The device execution network receives instruction sets from the cloud-based decision engine and drives corresponding home appliances to perform operations. Examples include controlling air conditioning temperature, controlling light switches or brightness, and controlling curtain opening and closing. The device execution network can communicate with home appliances via wireless communication protocols (such as Wi-Fi, Bluetooth, ZigBee, etc.) to ensure that instructions are transmitted accurately to the devices.

[0093] After the equipment performs an operation, feedback sensors (such as temperature feedback sensors for air conditioners, brightness feedback sensors for lights, and position feedback sensors for curtains) collect real-time equipment operating status data and feed it back to the equipment execution network.

[0094] The equipment execution network uses a fuzzy PID controller to optimize control parameters, adjusting the equipment response time based on the number of currently online devices to ensure stable and efficient equipment operation, with the response time R satisfying the following:

[0095]

[0096] in, This represents the device's basic response threshold, and N represents the number of currently online devices. When the number of online devices is large, the device response time should be appropriately extended to avoid network congestion leading to command execution failures. The fuzzy PID controller combines the advantages of fuzzy control and PID control, and can dynamically adjust control parameters according to the actual operating conditions of the device, improving the system's control accuracy and stability.

[0097] Example

[0098] Environmental perception module deployment and data acquisition:

[0099] In the living room, install the temperature sensor on a wall at least 1.5 meters away from the air conditioner vent and windows. Since the temperature of the air conditioner vent is affected by the air conditioner's cooling or heating function, and the temperature near the windows is affected by the outdoor temperature, installing the temperature sensor in this location ensures that the measurement data is not affected by these factors and more accurately reflects the actual temperature of the living room.

[0100] Humidity sensors should be installed in corners away from water sources, such as bathrooms, where moisture can directly affect the sensor's measurement results. Keeping them away from water sources can avoid this effect.

[0101] The light sensor is installed on the edge of the window glass, facing the main outdoor light source, so that it can accurately capture changes in the intensity of outdoor light.

[0102] Air quality sensors are installed at a height of 1.5-2 meters above the ground in a location with good air circulation, such as the central area of ​​the living room. This height and location can better reflect the overall indoor air quality.

[0103] The sound decibel sensor is installed away from noise sources such as electrical appliances, doors, and windows, such as in a corner of the living room, to ensure accurate collection of ambient noise.

[0104] After installation, all sensors are calibrated using professional calibration equipment. The calibration process involves comparing the sensors with standard measuring equipment and adjusting the sensor readings to match the standard values. Temperature and humidity sensors are set to collect data every 5 minutes, while light, air quality, and sound decibel sensors collect data every 1 minute. The sensors transmit data to the environmental sensing module's receiver via wireless communication technologies such as Wi-Fi, Bluetooth, or ZigBee. During transmission, the data is encrypted using the AES-256 encryption protocol to ensure data security.

[0105] User behavior analysis module setup:

[0106] Collect user data on their operation of various home devices over the past year. This data can be obtained from the smart home system's log files. Clean and preprocess the data to remove outliers and errors.

[0107] Meanwhile, motion recognition cameras are installed in key activity areas such as the living room and bedroom to capture user movements in real time. High-definition cameras can be used to improve the accuracy of motion recognition. A Hidden Markov Model (HMM) is built using Python and machine learning libraries (such as Scikit-learn and TensorFlow). Historical operation data and real-time motion recognition data are used as input to train the model and construct a probabilistic model of user behavior.

[0108] During training, cross-validation is used to evaluate and optimize the model, and the model parameters are continuously adjusted so that it can accurately output the weights of user behavior preferences for operating various devices in different scenarios.

[0109] Cloud-based decision engine operation:

[0110] The cloud-based decision engine receives environmental parameters uploaded by the environmental perception module and behavioral preference weights output by the user behavior analysis module in real time. It then uses a preset algorithm to fuse the two and generates a set of device control instructions based on the fusion result.

[0111] When the environmental sensing module detects an increase in indoor temperature and user behavior preference weights indicate that the user frequently lowers the air conditioning temperature under similar conditions, it generates a corresponding air conditioning temperature adjustment command. By optimizing equations to calculate control priorities, it prioritizes device control commands corresponding to changes in environmental parameters such as temperature and air quality that have a significant impact on user comfort. The cloud-based decision engine employs a distributed computing architecture to improve data processing speed and efficiency, ensuring timely responses to environmental changes and user needs.

[0112] Device performs network implementation:

[0113] The device execution network receives instruction sets from the cloud-based decision engine, driving corresponding home devices to perform operations, including controlling air conditioners to adjust temperature, controlling light switches, or adjusting brightness. After the device performs an operation, feedback sensors, including temperature feedback sensors for air conditioners and brightness feedback sensors for lights, collect real-time device operating status data and feed it back to the device execution network.

[0114] The device execution network utilizes a fuzzy PID controller to optimize control parameters and adjust device response times based on the number of online devices, ensuring stable and efficient operation. For example, when there are many online devices, the response time can be appropriately extended to prevent network congestion from causing command execution failures. The device execution network can also employ multi-threading technology to process commands from multiple devices simultaneously, improving the system's concurrent processing capabilities.

[0115] Therefore, the present invention adopts the above-mentioned smart home control method based on the Internet of Things. Through comprehensive and detailed environmental perception, accurate user behavior analysis, intelligent cloud decision-making and efficient device execution, it can realize the intelligent and automated control of home devices, and provide users with a more comfortable, convenient and safe home life experience.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. 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 still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smart home control method based on the Internet of Things, characterized in that, Includes the following steps: S1. The environmental perception module collects environmental parameters through a distributed sensor network. S2, the user behavior analysis module, constructs a user behavior probability model based on historical operation data and real-time action recognition, and outputs behavior preference weights. ; The S3 cloud-based decision engine generates a set of device control instructions by integrating environmental parameters and behavioral preference weights. Calculate the priority of regulation; S4. The device execution network drives home appliances according to the instruction set and optimizes control parameters through a feedback mechanism. In S3, the device control instruction set Regulation priority Calculated using the following optimization equation: ； in, For environmental parameters, the normalization coefficients are given, and ; For temperature, For humidity, Light intensity, Air Quality Index. Sound in decibels; Environmental parameter normalization coefficient , , Dynamically adjusted based on the user's physiological comfort model: ； in, Indicates user personalized preference parameters, This indicates real-time physiological indicators.

2. The smart home control method based on the Internet of Things according to claim 1, characterized in that, In S1, environmental parameters include temperature. ,humidity Light intensity Air Quality Index and sound decibels The environmental perception module uses multi-sensor fusion technology to fuse data, and the fusion formula is as follows: ； in, The merged environmental state value. For the first Sensor-like weights, For the first Sensor-like measurement values This is the noise correction factor. These are environmental disturbance factors.

3. The smart home control method based on the Internet of Things according to claim 2, characterized in that, In S1, the sampling frequency range of the environmental perception module is 1-60Hz.

4. The smart home control method based on the Internet of Things according to claim 3, characterized in that, In S1, sensor weights Updated using an adaptive strategy based on the covariance matrix: ； in, This is the average value for similar sensors. , These represent the maximum and minimum values ​​of the sensor's measurement range.

5. A smart home control method based on the Internet of Things according to claim 4, characterized in that, In S1, the collected environmental parameter data is transmitted using the AES-256 encryption protocol, with a encryption strength of [missing information]. satisfy: ； in, Indicates the data packet size. Indicates the dynamic key length. This represents the safety redundancy coefficient.

6. The smart home control method based on the Internet of Things according to claim 5, characterized in that, In S2, the user behavior analysis module uses a Hidden Markov Model (HMM) to calculate the probability of user behavior. : ； in, For time The observation behavior, For time The hidden state, For the probability of observation, This is the state transition function.

7. A smart home control method based on the Internet of Things according to claim 6, characterized in that, In S4, the device execution network uses a fuzzy PID controller to optimize control parameters, and the response time is... satisfy: ； in, Indicates the device's basic response threshold. This indicates the number of devices currently online.

Images