KNX protocol integration and remote control system applied to smart home

Abstract

The invention discloses a KNX protocol integration and remote control system applied to a smart home, and relates to the technical field of smart home, Internet of Things, automatic control and computer vision, equipment is managed in a unified manner through a KNX bus main control module, and communication and task scheduling are guaranteed; the scene sensing module is fused with radar and image data to accurately recognize a scene; the energy consumption optimization module formulates an energy-saving strategy based on equipment energy consumption and use habits; the security protection module guarantees the security of the system by utilizing multiple authentication, encryption and anomaly detection; the remote control module realizes convenient remote operation by means of AR interaction; according to the system, centralized control, accurate sensing, high efficiency, energy saving, safety protection and convenient interaction of the smart home equipment are realized, and the intelligent level, user experience and energy utilization efficiency of the smart home are improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of smart home, Internet of Things, automation control, and computer vision, and specifically to a KNX protocol integration and remote control system applied to smart home. Background Art

[0002] With the continuous progress of information technology and Internet of Things technology, smart home has become an important part of people's lives. It aims to achieve automatic control and intelligent management of home appliances with the help of advanced technologies, creating a convenient, comfortable, safe, and energy-saving living environment for users. However, the current smart home systems face many challenges: the device communication protocols and interface standards are not unified, resulting in difficulties in integration and centralized management; scene perception mostly relies on a single sensor, with insufficient accuracy, coverage, and adaptability, and weak complex scene recognition ability; although there is energy consumption monitoring, there is a lack of energy-saving strategies and intelligent control means, and the cooperation between devices and overall energy efficiency are not fully considered; there are loopholes in security protection, with insufficient terminal authentication, data encryption, and anomaly detection; the remote control interaction experience and convenience are not good, and the operation is cumbersome.

[0003] To solve the above problems, the present invention proposes a KNX protocol integration and remote control system applied to smart home. Summary of the Invention

[0004] The purpose of the present invention is to provide a KNX protocol integration and remote control system applied to smart home to solve the problems raised in the prior art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: A KNX protocol integration and remote control system applied to smart home includes a KNX bus master control module, a scene perception module, an energy consumption optimization module, a security protection module, and a remote control module; the KNX bus master control module is responsible for centralized management and control of smart home devices through integrated protocol stack processing and multi-threaded scheduling; the scene perception module detects the state of the smart home space by fusing millimeter-wave radar and image sensors, providing environmental information for the system to automatically adjust the operation of devices; the energy consumption optimization module is responsible for optimizing the energy consumption of smart home devices according to dynamic electricity prices and device usage conditions, reducing electricity costs; the security protection module is responsible for ensuring system communication and data security, preventing intrusion and data leakage; the remote control module supports users to remotely control smart home devices through an AR interface, realizing instruction transmission and status feedback.

[0006] The KNX bus master control module includes a hardware interface unit, a protocol conversion unit, and a task scheduling unit; The hardware interface unit supports dual-mode adaptive access for TP-C / RF / IP / Zigbee / Wi-Fi6e; this unit automatically selects a communication method for data transmission by continuously monitoring the signal strength and stability of the communication environment; when a high and stable wireless signal is detected, it preferentially uses the wireless communication method; if the wireless signal is poor, it switches to the wired communication method to ensure device connection and data transmission; The protocol conversion unit performs a two-way mapping with the Internet of Things message queue through the constructed device object model; for different types of smart home devices, first convert their communication protocols into a unified device object model, and then map this model into the MQTT message format to achieve data interaction between the device and the system; in addition, the instructions sent from the system to the device are also converted through this two-way mapping mechanism to ensure interoperability between devices with different protocols; The task scheduling unit manages KNX communication and device control tasks using a priority cascading mechanism; assigns priorities to each task according to the urgency of the task, and when multiple tasks request processing simultaneously, preferentially processes the tasks with higher priorities; at the same time, through the cascading mechanism, ensures that after the high-priority tasks are processed, the low-priority tasks are processed in sequence; among them, the distinction of task urgency is based on the user's needs and is freely set by the user in the remote control module.

[0007] The scene perception module includes a millimeter-wave radar unit, an image sensor unit, and a data fusion unit; The millimeter-wave radar unit emits millimeter-wave signals and receives the reflected signals, and calculates the distance, speed, and angle of the object using the time difference and frequency change information of the signals to achieve the detection of objects in the space; The image sensor unit is used to collect image information in the space; captures light through a lens, converts it into an electrical signal, and then performs noise reduction and enhancement processing through an image processing algorithm to obtain clear image data, providing a visual basis for scene recognition; The data fusion unit is responsible for weighted fusion of the collected radar data and visual data and outputs the information for scene recognition.

[0008] The energy consumption optimization module includes a device energy consumption monitoring unit, a device usage habit analysis unit, and a device intelligent control unit; The device energy consumption monitoring unit continuously collects the current and voltage of the device by installing current sensors and voltage sensors in the device circuit; then, based on the collected current and voltage parameters, calculates the real-time energy consumption of the device using the average power calculation formula, and at the same time, uploads the calculated real-time energy consumption data to the system database for storage; The device usage habit analysis unit receives the energy consumption data uploaded by the device energy consumption monitoring unit, applies data analysis and machine learning technologies, and uses the clustering analysis method to obtain the rules and patterns of device usage. According to the data analysis results, a user's device usage habit model is established. This model can predict the user's future device usage behavior and provide a reference for the customization of energy-saving strategies; The device intelligent control unit first combines the device energy consumption characteristic data provided by the device energy consumption monitoring unit and the user usage habit model established by the device usage habit analysis unit to generate an energy-saving strategy; then, according to the determined energy-saving strategy, it generates corresponding device control instructions and sends the generated control instructions to the corresponding device through the KNX bus master module; at the same time, it real-time monitors the operation status and energy consumption of the device to ensure that the device operates according to the requirements of the energy-saving strategy.

[0009] The security protection module includes a terminal authentication unit, a data encryption unit, and an anomaly detection unit; The terminal authentication unit implements the EAP-TLS + national cryptography certificate two-factor authentication method; when the user terminal accesses the system, it first authenticates the user's username and password through the EAP-TLS protocol; then, it uses the national cryptography certificate for secondary authentication to ensure the legality and security of the terminal device; The data encryption unit establishes an AES-GCM end-to-end encryption tunnel to encrypt and protect the data transmitted within the system; at the data sending end, it uses the AES-GCM algorithm to encrypt the data to generate ciphertext and an authentication tag; at the data receiving end, it uses the same key and algorithm to decrypt the ciphertext and verify the validity of the authentication tag to ensure that the data is not stolen during transmission; at the same time, it sets the update period of the key to enhance the security of encryption; The anomaly detection unit detects abnormal activities by real-time monitoring and analyzing the user behavior and device status within the system; first, it establishes a normal behavior model and a device status baseline, recording various parameters and behavior patterns of the system under normal operation; then, it real-time collects the operation data of the system and compares and analyzes it with the normal model; when it finds that the data deviation exceeds the set threshold, it determines it as an abnormal activity and issues an alarm to the user.

[0010] The remote control module includes an AR interaction unit, an instruction transmission unit, an instruction parsing and forwarding unit, and a status feedback unit; The AR interaction unit triggers control instructions through the AR interface; the AR interface combines the virtual control interface with the real home scene using augmented reality technology, enabling users to interact with home devices intuitively; users operate on the AR interface through gestures and voices to trigger corresponding control instructions; The instruction transmission unit transmits the user's control instructions to the cloud server via WebSocket; WebSocket is a two-way communication protocol that can establish a real-time connection between the browser and the server, and WebSocket allows real-time two-way data transmission between the server and the client; in the smart home system, the user's control instructions can be directly sent to the cloud server, and at the same time, the server will feedback the device status information to the client; The instruction parsing and forwarding unit converts the user's natural language control instructions into a standard KNX command set that can be recognized and executed by the smart home system, and accurately forwards the parsed instructions to the local master control unit to achieve remote control of smart home devices; among them, the conversion of natural language is realized through the NLP engine of the cloud server. This engine can analyze the user's instruction intention through learning and training of natural language data, and then, according to the protocol specifications of the smart home system, convert this information into a standard KNX command set to ensure that the device can correctly identify and run; The status feedback unit is responsible for feeding back the operating status information of the smart home device to the user, enabling the user to understand the working conditions of the device in real time and dynamically adjust the device control strategy according to the feedback information; the implementation method is as follows: after the home device executes the instruction, it will send its own operating status information back to the system through the communication network; after the system receives this information, it will display it on the visualization terminal; then the system will dynamically adjust the device control strategy according to the feedback status information of the device based on the user's demand preferences.

[0011] The data fusion unit is responsible for performing weighted fusion on the collected radar data and visual data, and the output scene recognition information includes the following: After the data fusion unit receives the data transmitted by the millimeter-wave radar unit and the image sensor unit, it first preprocesses the data; for the millimeter-wave radar data, it checks the integrity and accuracy of the data, removes the existing outliers and noise interference; by setting the threshold range, it filters out the distance and speed data beyond the threshold range; for the image data, it will perform further feature extraction and normalization processing; extract the features in the image, and at the same time normalize the image data so that it is within a unified scale and range for subsequent fusion processing; The data fusion unit assigns different weights to radar data and visual data according to different scene characteristics and data characteristics; the specific weight assignment will be dynamically adjusted according to pre-set rules and real-time monitored scene changes; the data fusion unit can monitor the speed information in the millimeter-wave radar data in real time; when the radar detects that the speed of an object exceeds the preset stationary threshold, it is determined that the object starts to move, and at this time, the weight of the radar data starts to increase; when the object is in a stationary state, the weight of the visual data will be increased according to the preset rules; among them, the sum of the weights of the radar data and the visual data is always 100%, the adjustment range of the radar data is 30%-90%, and the adjustment range of the visual data weight is 10%-70%; according to the assigned weights, the pre-processed radar data and visual data are weighted and summed; the radar data and the visual data are weighted and combined in the same dimension to obtain the fused comprehensive data; Then, the fused comprehensive data is analyzed and processed, and the scene information in the space is recognized using the pattern recognition algorithm; the implementation method is as follows: First, feature extraction and enhancement are performed. For the fused comprehensive data, multi-dimensional feature extraction is first performed; in the spatial dimension, using the gradient algorithm of the image, the gradient value of the object edge in the image is calculated to highlight the contour features of the object; at the same time, based on the radar data, the distance distribution features of the object are extracted, and a distance histogram is constructed to describe the distribution of the object in space; in the time dimension, in a sliding window manner, the fused data over a period of time is processed, and the change rate of the data in adjacent time windows is calculated to obtain the motion trend features of the object. Finally, the principal component analysis method is used to reduce the dimension of the extracted high-dimensional features; First, the covariance matrix of the feature data is calculated, and this matrix reflects the correlation between different features; then, the covariance matrix is eigen-decomposed to obtain the eigenvalues and eigenvectors; the eigenvectors are sorted according to the size of the eigenvalues, and the first k eigenvectors corresponding to the largest k eigenvalues are selected as the principal components; through these k principal components, the original high-dimensional feature data is projected into a low-dimensional space to achieve the purpose of dimension reduction while retaining the data information. For model recognition and classification, the support vector machine algorithm is used for scene classification; in the training stage, the fused data samples after dimension reduction processing are used as inputs, and the corresponding scene categories of the samples are used as labels; the SVM algorithm separates data samples of different categories by finding an optimal classification hyperplane, and this classification hyperplane can maximize the interval between the two types of data; for linearly separable data, a linear kernel function is used to construct a classification model; for linearly inseparable data, the kernel trick is introduced, and a suitable non-linear kernel function is selected to map the data into a high-dimensional space to make it linearly separable, and then a classification model is constructed. In the classification stage, after the new fusion data undergoes the same feature extraction and dimensionality reduction processes, it is input into the trained SVM model, and the model determines the scene category to which the data belongs according to the learned classification rules; For the analysis of the scene change trend, a Hidden Markov Model is used to analyze the scene change trend; an HMM is a double stochastic process, which includes a hidden state sequence and an observable output sequence; the scene categories at different times are used as the observable outputs, while the hidden states represent the internal trends of the scene changes; First, initialize the parameters of the HMM, including the state transition probability matrix, the observation probability matrix, and the initial state probability distribution; the state transition probability matrix describes the probability of transitioning from one hidden state to another hidden state; the observation probability matrix represents the probability of generating the scene category under the hidden state; the initial state probability distribution defines the probabilities of each hidden state at the initial moment; Then, according to the historical scene category data, use the Baum - Welch algorithm to train and optimize the parameters of the HMM, so that the model can fit the actual scene change process; in the prediction stage, based on the current scene category and the trained HMM model, calculate the predicted scene category distribution through the forward algorithm, so as to predict the scene change trend; By comparing and analyzing the fusion data at different times, monitor the scene change trend; finally, output the recognized scene information to provide an environmental basis for other modules of the smart home system.

[0012] According to the data analysis results, establish a user's device usage habit model, which can predict the user's future device usage behavior and provide a reference for the formulation of energy - saving strategies, including the following content: First, select the long short-term memory network as the core model, which has an input layer, an LSTM layer, a fully connected layer, and an output layer. Among them, the input layer receives the feature vectors obtained in the feature extraction stage, and its dimension is determined by the number of extracted features. These features include time features, frequency features, and correlation features. Then, set multiple LSTM units to form a hidden layer. Each LSTM unit contains an input gate, a forget gate, and an output gate, which are used to control the inflow, retention, and output of information. By stacking multiple LSTM layers, the complexity and expressive ability of the model are increased. Connect the output of the LSTM layer to a fully connected layer to perform a linear transformation on the output of the LSTM layer to map it to the final output space. Then, initialize the parameters of the model, including the weights and biases of the LSTM layer, and the weights and biases of the fully connected layer. Then, input the data in the training set into the model in batches, perform forward propagation to calculate the predicted values of the model. Calculate the value of the loss function according to the predicted values and the true values, and use the backpropagation algorithm to calculate the gradient of the loss function with respect to the model parameters. The optimizer updates the parameters of the model according to the gradient to gradually reduce the value of the loss function. Finally, repeat the above steps until the loss function reaches the preset number of training epochs. After the model training is completed, input the current device usage data into the trained model. The model performs forward propagation calculation according to the input data and outputs the prediction results of the user's future device usage behavior. Finally, formulate an energy-saving strategy based on the energy consumption characteristics of the device, the predicted device usage behavior, and the energy-saving goal. For the formulation of the energy-saving strategy, first, clarify the energy consumption characteristics of the device, the predicted device usage behavior, and the energy-saving goal. Then, classify the devices into deferrable devices, non-deferrable devices, and adjustable devices, and set priorities. Based on the device usage prediction, formulate preliminary strategies for various types of devices. Deferrable devices perform off-peak operation and batch processing, non-deferrable devices optimize the operation mode and provide maintenance upgrade suggestions, and adjustable devices perform parameter optimization and intelligent control. Match the preliminary strategy with the energy-saving goal. If the goal is not achieved, adjust the strategy, which can increase energy-saving measures and optimize the device combination. Then, establish a strategy evaluation and feedback mechanism, set evaluation indicators to regularly evaluate the strategy effect, and collect user feedback to improve the strategy. Finally, implement the energy-saving strategy, convert it into control instructions to achieve automatic device control, and establish a real-time monitoring system to monitor the device operation status and energy consumption data, which is convenient for users to adjust the strategy through the system.

[0013] Compared with the prior art, the beneficial effects of the present invention are: 1. Precise Scenario Perception: The scenario perception module integrates a millimeter-wave radar and an image sensor. The millimeter-wave radar unit can accurately detect the distance, speed, and angle of an object, and the image sensor unit can collect clear image data. The data fusion unit performs weighted fusion and in-depth analysis on the data of both, and uses pattern recognition algorithms and principal component analysis methods to achieve high-precision recognition and classification of the spatial scenario, accurately perceive the status and changes of the home environment, and provide a reliable basis for the system to automatically adjust the operation of devices.

[0014] 2. Efficient Task Scheduling: The task scheduling unit uses a priority cascading mechanism to manage KNX communication and device control tasks, freely sets the urgency and priority of tasks according to user needs, gives priority to high-priority tasks, and ensures that low-priority tasks are processed in sequence after high-priority tasks are completed, avoiding problems such as untimely task processing and slow system response, and improving the overall operation efficiency of the system. 3. Intelligent Energy-saving Strategy Customization: The device energy consumption monitoring unit of the energy consumption optimization module collects device energy consumption data in real time. The device usage habit analysis unit uses data analysis and machine learning technologies to establish a user device usage habit model to predict the user's future device usage behavior. The device intelligent control unit generates energy-saving strategies based on the device energy consumption characteristics and prediction results, performs peak-shifting operation and batch processing on delayable devices, optimizes the operation mode and provides maintenance suggestions for non-delayable devices, and performs parameter optimization and intelligent control on adjustable devices, effectively reducing the energy consumption of the smart home system and improving the overall energy utilization efficiency. Description of the Drawings

[0015] Figure 1 It is a schematic diagram of the system architecture of the KNX protocol integration and remote control system applied to smart homes of the present invention. Detailed Embodiments

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment: As Figure 1 shown, the present invention provides a technical solution, KNX Protocol Integration and Remote Control System for Smart Home, including KNX Bus Master Module, Scene Sensing Module, Energy Consumption Optimization Module, Security Protection Module and Remote Control Module; the KNX Bus Master Module is responsible for centralized management and control of smart home devices through integrated protocol stack processing and multi-threaded scheduling; the Scene Sensing Module detects the state of the smart home space by integrating millimeter-wave radar and image sensors, providing environmental information for the system to automatically adjust device operation; the Energy Consumption Optimization Module is responsible for optimizing the energy consumption of smart home devices based on dynamic electricity prices and device usage, reducing electricity costs; the Security Protection Module is responsible for ensuring system communication and data security, preventing intrusion and data leakage; the Remote Control Module supports users to remotely control smart home devices through an AR interface, realizing instruction transmission and status feedback.

[0018] The KNX Bus Master Module includes a hardware interface unit, a protocol conversion unit and a task scheduling unit; The hardware interface unit supports TP-C / RF / IP / Zigbee / Wi-Fi6e dual-mode adaptive access; this unit automatically selects a communication method for data transmission by real-time monitoring the signal strength and stability of the communication environment; when detecting high and stable wireless signal strength, it preferentially uses wireless communication; if the wireless signal is poor, it switches to wired communication to ensure device connection and data transmission; The protocol conversion unit performs bidirectional mapping through the constructed device object model and the Internet of Things message queue; for different types of smart home devices, first convert their communication protocols into a unified device object model, and then map this model into the MQTT message format to realize data interaction between the device and the system; in addition, the instructions sent from the system to the device are also converted through this bidirectional mapping mechanism to ensure interoperability between devices with different protocols; The task scheduling unit manages KNX communication and device control tasks using a priority cascading mechanism; assigns priorities to each task according to the urgency of the task, and when multiple tasks request processing simultaneously, it preferentially processes tasks with higher priorities; at the same time, through the cascading mechanism, it ensures that after high-priority tasks are processed, low-priority tasks are processed in sequence; among them, the distinction of task urgency is based on user requirements and is freely set by the user in the remote control module.

[0019] The Scene Sensing Module includes a millimeter-wave radar unit, an image sensor unit and a data fusion unit; The millimeter-wave radar unit emits millimeter-wave signals and receives reflected signals, and calculates the distance, speed and angle of objects using the time difference and frequency change information of the signals to realize the detection of objects in the space; The image sensor unit is used to collect image information in space; it captures light through a lens, converts it into an electrical signal, and then performs noise reduction and enhancement processing through an image processing algorithm to obtain clear image data, providing a visual basis for scene recognition; The data fusion unit is responsible for weighted fusion of the collected radar data and visual data, and outputs information for scene recognition.

[0020] The energy consumption optimization module includes a device energy consumption monitoring unit, a device usage habit analysis unit, and a device intelligent control unit; The device energy consumption monitoring unit continuously collects the current and voltage of the device by installing current sensors and voltage sensors in the device circuit; then, based on the collected current and voltage parameters, it calculates the real-time energy consumption of the device using the average power calculation formula. At the same time, it uploads the calculated real-time energy consumption data to the system database for storage; The device usage habit analysis unit receives the energy consumption data uploaded by the device energy consumption monitoring unit, and uses data analysis and machine learning techniques. Through cluster analysis methods, it obtains the rules and patterns of device usage. According to the data analysis results, it establishes a device usage habit model for users. This model can predict the future device usage behavior of users and provide a reference for formulating energy-saving strategies; The device intelligent control unit first combines the device energy consumption characteristic data provided by the device energy consumption monitoring unit and the user usage habit model established by the device usage habit analysis unit to generate an energy-saving strategy; then, according to the determined energy-saving strategy, it generates corresponding device control instructions, and sends the generated control instructions to the corresponding devices through the KNX bus main control module; at the same time, it continuously monitors the operating status and energy consumption of the devices to ensure that the devices operate in accordance with the requirements of the energy-saving strategy.

[0021] The security protection module includes a terminal authentication unit, a data encryption unit, and an anomaly detection unit; The terminal authentication unit implements the EAP-TLS + national cryptography certificate two-factor authentication method; when a user terminal accesses the system, it first performs identity verification through the EAP-TLS protocol to verify the user's username and password; then, it uses the national cryptography certificate for secondary authentication to ensure the legality and security of the terminal device; The data encryption unit establishes an AES-GCM end-to-end encryption tunnel to encrypt and protect the data transmitted within the system; at the data sending end, it uses the AES-GCM algorithm to encrypt the data to generate ciphertext and an authentication tag; at the data receiving end, it uses the same key and algorithm to decrypt the ciphertext and verify the validity of the authentication tag to ensure that the data is not stolen during transmission; at the same time, it sets the update period of the key to enhance the security of encryption; The anomaly detection unit detects abnormal activities by monitoring and analyzing the user behaviors and device states within the system in real time. First, a normal behavior model and a device state baseline are established to record various parameters and behavior patterns of the system under normal operation. Then, the operation data of the system is collected in real time and compared with the normal model for analysis. When it is found that the data deviation exceeds the set threshold, it is determined as an abnormal activity, and an alarm is sent to the user.

[0022] The remote control module includes an AR interaction unit, an instruction transmission unit, an instruction parsing and forwarding unit, and a status feedback unit. The AR interaction unit triggers control instructions through the AR interface. The AR interface combines the virtual control interface with the real home scene using augmented reality technology, enabling users to interact with home devices intuitively. Users operate on the AR interface by means of gestures and voices to trigger corresponding control instructions. The instruction transmission unit transmits the user's control instructions to the cloud server through WebSocket. WebSocket is a two-way communication protocol that can establish a real-time connection between the browser and the server, and WebSocket allows real-time two-way data transmission between the server and the client. In the smart home system, the user's control instructions can be directly sent to the cloud server, and at the same time, the server will feedback the device status information to the client. The instruction parsing and forwarding unit converts the user's natural language control instructions into a standard KNX command set that can be recognized and executed by the smart home system, and accurately forwards the parsed instructions to the local master control unit to achieve remote control of smart home devices. Among them, the conversion of natural language is realized through the NLP engine of the cloud server. This engine can analyze the user's instruction intent through learning and training of natural language data, and then, according to the protocol specifications of the smart home system, convert this information into a standard KNX command set to ensure that the device can correctly identify and operate. The status feedback unit is responsible for feeding back the operation status information of smart home devices to the user, enabling the user to understand the working conditions of the devices in real time and dynamically adjust the device control strategy according to the feedback information. The implementation method is as follows: after the home device executes the instruction, it will send its own operation status information back to the system through the communication network. After the system receives this information, it will display it on the visualization terminal. Then the system will dynamically adjust the device control strategy according to the feedback status information of the device based on the user's demand preferences.

[0023] The data fusion unit is responsible for weighted fusion of the collected radar data and visual data, and the output scene recognition information includes the following: After the data fusion unit receives the data transmitted by the millimeter-wave radar unit and the image sensor unit, it first preprocesses the data. For the millimeter-wave radar data, it checks the integrity and accuracy of the data, removes existing outliers and noise interference, and filters out the distance and speed data beyond the threshold range by setting the threshold range. For the image data, further feature extraction and normalization processing are performed. The features in the image are extracted, and at the same time, the image data is normalized to be within a unified scale and range for subsequent fusion processing. The data fusion unit assigns different weights to the radar data and the visual data according to different scene characteristics and data characteristics. The specific weight assignment will be dynamically adjusted according to the preset rules and the real-time monitored scene changes. The data fusion unit can monitor the speed information in the millimeter-wave radar data in real time. When the radar detects that the speed of an object exceeds the preset stationary threshold, it determines that the object starts to move, and at this time, the weight of the radar data begins to increase. When the object is in a stationary state, the weight of the visual data will be increased according to the preset rules. Among them, the sum of the weights of the radar data and the visual data is always 100%. The adjustment range of the radar data is 30% - 90%, and the adjustment range of the visual data weight is 10% - 70%. According to the assigned weights, the preprocessed radar data and visual data are weighted and summed. The radar data and the visual data are weighted and combined in the same dimension to obtain the fused comprehensive data. Then, the fused comprehensive data is analyzed and processed, and the scene information in the space is recognized using the pattern recognition algorithm. The implementation method is as follows: First, feature extraction and enhancement are performed. For the fused comprehensive data, multi-dimensional feature extraction is first performed. In the spatial dimension, using the gradient algorithm of the image, the gradient value of the object edge in the image is calculated to highlight the contour features of the object. At the same time, based on the radar data, the distance distribution features of the object are extracted, and a distance histogram is constructed to describe the distribution of the object in the space. In the time dimension, in a sliding window manner, the fused data over a period of time is processed, and the change rate of the data in adjacent time windows is calculated to obtain the motion trend features of the object. Finally, the principal component analysis method is used to reduce the dimension of the extracted high-dimensional features. First, the covariance matrix of the feature data is calculated, which reflects the correlation between different features. Then, the covariance matrix is eigen-decomposed to obtain the eigenvalues and eigenvectors. The eigenvectors are sorted according to the size of the eigenvalues, and the first k eigenvectors corresponding to the largest k eigenvalues are selected as the principal components. Through these k principal components, the original high-dimensional feature data is projected into a low-dimensional space to achieve the purpose of dimension reduction while retaining the data information. For model recognition and classification, the support vector machine (SVM) algorithm is used for scene classification. During the training phase, the fused data samples that have undergone dimensionality reduction are used as inputs, and the corresponding scene categories of the samples are used as labels. The SVM algorithm separates data samples of different categories by finding an optimal classification hyperplane, which can maximize the margin between the two types of data. For linearly separable data, a linear kernel function is used to construct the classification model. For linearly inseparable data, the kernel trick is introduced, and an appropriate non - linear kernel function is selected to map the data into a high - dimensional space to make it linearly separable, and then a classification model is constructed. During the classification phase, the new fused data is subjected to the same feature extraction and dimensionality reduction processes and then input into the trained SVM model. The model determines the scene category to which the data belongs according to the learned classification rules. For the analysis of scene change trends, the hidden Markov model (HMM) is used to analyze the scene change trends. HMM is a double - stochastic process that includes a hidden state sequence and an observable output sequence. The scene categories at different times are used as the observable outputs, while the hidden states represent the internal trends of scene changes. First, the parameters of the HMM are initialized, including the state transition probability matrix, the observation probability matrix, and the initial state probability distribution. The state transition probability matrix describes the probability of transitioning from one hidden state to another. The observation probability matrix represents the probability of generating scene categories under the hidden states. The initial state probability distribution defines the probabilities of each hidden state at the initial time. Then, based on the historical scene category data, the Baum - Welch algorithm is used to train and optimize the parameters of the HMM so that the model can fit the actual scene change process. During the prediction phase, based on the current scene category and the trained HMM model, the forward algorithm is used to calculate the predicted scene category distribution, thereby predicting the scene change trend. By comparing and analyzing the fused data at different times, the scene change trend is monitored. Finally, the recognized scene information is output, providing an environmental basis for other modules of the smart home system.

[0024] According to the data analysis results, a user device usage habit model is established. This model can predict the user's future device usage behavior and provide a reference for the formulation of energy - saving strategies, including the following content: First, select the long short-term memory network as the core model. This model has an input layer, an LSTM layer, a fully connected layer, and an output layer. Among them, the input layer receives the feature vectors obtained in the feature extraction stage, and its dimension is determined by the number of extracted features. These features include time features, frequency features, and correlation features. Then, set multiple LSTM units to form a hidden layer. Each LSTM unit contains an input gate, a forget gate, and an output gate, which are used to control the inflow, retention, and output of information. By stacking multiple LSTM layers, the complexity and expressive ability of the model are increased. Connect the output of the LSTM layer to a fully connected layer to perform a linear transformation on the output of the LSTM layer to map it to the final output space. Then, initialize the parameters of the model, including the weights and biases of the LSTM layer, and the weights and biases of the fully connected layer. Next, input the data in the training set into the model in batches and perform forward propagation to calculate the predicted values of the model. Calculate the value of the loss function according to the predicted values and the true values, and use the backpropagation algorithm to calculate the gradient of the loss function with respect to the model parameters. The optimizer updates the parameters of the model according to the gradient to gradually reduce the value of the loss function. Finally, repeat the above steps until the loss function reaches the preset number of training epochs. After the model training is completed, input the current device usage data into the trained model. The model performs forward propagation calculation according to the input data and outputs the prediction results of the user's future device usage behavior. Finally, formulate an energy-saving strategy based on the energy consumption characteristics of the device, the predicted device usage behavior, and the energy-saving goal. For the formulation of the energy-saving strategy, first, clarify the energy consumption characteristics of the device, the predicted device usage behavior, and the energy-saving goal. Then, classify the devices into deferrable devices, non-deferrable devices, and adjustable devices, and set priorities. Based on the device usage prediction, formulate preliminary strategies for various types of devices. Deferrable devices perform peak-shifting operation and batch processing, non-deferrable devices optimize the operation mode and provide maintenance and upgrade suggestions, and adjustable devices perform parameter optimization and intelligent control. Match the preliminary strategy with the energy-saving goal. If the goal is not achieved, adjust the strategy, which can increase energy-saving measures and optimize the device combination. Then, establish a strategy evaluation and feedback mechanism, set evaluation indicators to regularly evaluate the strategy effect, and collect user feedback to improve the strategy. Finally, implement the energy-saving strategy, convert it into control instructions to achieve automatic device control, and establish a real-time monitoring system to monitor the device operation status and energy consumption data, facilitating users to adjust the strategy through the system. Embodiment

[0025] Suppose this smart home system is deployed in a set of residential houses; current sensors and voltage sensors are installed in the distribution box and connected to the circuits of each major electrical appliance (such as air conditioners, refrigerators, washing machines, lighting fixtures, etc.) so that the device energy consumption monitoring unit can collect data. Millimeter-wave radar units and image sensor units are installed in the main activity areas such as the living room and bedroom to ensure comprehensive monitoring of the space state. At the same time, a KNX bus master control module is integrated into the home network device, and a hardware interface unit supporting TP-C / RF / IP / Zigbee / Wi-Fi6e dual-mode adaptive access is configured to ensure stable communication of the device.

[0026] After the system starts, the hardware interface unit monitors the communication environment in real time and automatically selects Wi-Fi6e for data transmission because its signal strength is high and stable in the initial stage. The protocol conversion unit begins to construct a two-way mapping between the device object model and the Internet of Things message queue, converts the different protocols of devices such as smart lamps and smart sockets into a unified model, and maps them into the MQTT message format to realize the data interaction between the device and the system.

[0027] The device energy consumption monitoring unit starts to collect the current and voltage data of each device, calculates the real-time energy consumption every 5 minutes, and uploads it to the system database. The millimeter-wave radar unit and the image sensor unit also start to work. The millimeter-wave radar continuously emits and receives millimeter-wave signals to obtain the distance, speed, and angle information of the object; the image sensor collects image data every 10 seconds and transmits it to the data fusion unit after noise reduction and enhancement processing.

[0028] After receiving the radar and image data, the data fusion unit performs preprocessing. For the millimeter-wave radar data, a distance threshold of 10 meters and a speed threshold of 0.1 m / s are set to filter abnormal data; feature extraction and normalization processing are performed on the image data. In the living room scenario, when someone enters, the millimeter-wave radar detects that the object speed exceeds the stationary threshold, and the data fusion unit increases the weight of the radar data to 70%, and the weight of the visual data is adjusted to 30%. After the fused data is subjected to multi-dimensional feature extraction and principal component analysis for dimensionality reduction, it is input into the support vector machine model for scene recognition, and it is judged as a "human activity" scene. The system automatically turns on the living room lights according to this result and adjusts the light brightness to an appropriate level.

[0029] The device usage habit analysis unit receives the data uploaded by the device energy consumption monitoring unit and analyzes the data of one week using the clustering analysis method. It is found that the user usually uses the washing machine after 10 pm, and the air conditioner is used more frequently during the day on weekends. Based on this, a user device usage habit model is established, and the device usage behavior in the next week is predicted.

[0030] The device intelligent control unit formulates energy-saving strategies by combining the energy consumption characteristics of the device and the prediction results. For delayable devices such as washing machines, they are set to run automatically during the low electricity price period at night (from 12:00 am to 6:00 am); for adjustable devices such as air conditioners, the temperature setting value and wind speed are automatically adjusted according to the indoor and outdoor temperatures and the predicted usage duration. Suppose it is predicted that the user will use the living room air conditioner in 30 minutes and the outdoor temperature is 30°C, then the air conditioner temperature is set to 26°C 5 minutes in advance and the wind speed is adjusted to the low gear. At the same time, the system monitors the operation status and energy consumption of the device in real time to ensure the effective implementation of the energy-saving strategy.

[0031] The user accesses the smart home system through the mobile terminal. The terminal authentication unit first verifies the username and password through the EAP-TLS protocol, and then conducts secondary authentication using the national cryptography certificate. After successful authentication, the data encryption unit establishes an AES-GCM end-to-end encryption tunnel to ensure the security of data transmission. The anomaly detection unit continuously monitors the user behavior and device status. If it is found that someone attempts to log in multiple times during non-daily usage periods and the logged-in IP address is abnormal, an alarm message will be immediately sent to the user's mobile phone.

[0032] When the user is out, they open the AR interaction application on the mobile phone. Through the AR interface, the user can see the picture where the virtual control interface at home is superimposed on the real scene. The user issues the voice command "Turn on the living room air conditioner", and the command transmission unit sends this command to the cloud server via WebSocket. The NLP engine of the cloud server parses the command intention and converts it into a standard KNX command set. The command parsing and forwarding unit forwards the command to the local main control unit to control the living room air conditioner to turn on. After the air conditioner executes the command, the operation status information is fed back to the system via the communication network, and the status feedback unit displays the information on the AR interface. The user can view the operation status of the air conditioner in real time and adjust parameters such as temperature and wind speed according to needs.

[0033] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, in any regard, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claimed elements.

Claims

1. A KNX protocol integration and remote control system applied to smart home, characterized in that: It includes a KNX bus master control module, a scene perception module, an energy consumption optimization module, a security protection module, and a remote control module. The KNX bus master control module processes and multi-threadedly schedules through an integrated protocol stack, and is responsible for the centralized management and control of smart home devices. The scene perception module detects the state of the smart home space by integrating a millimeter-wave radar and an image sensor, providing environmental information for the system to automatically adjust the operation of the devices. The energy consumption optimization module is responsible for optimizing the energy consumption of smart home devices based on dynamic electricity prices and device usage conditions, reducing electricity costs. The security protection module is responsible for ensuring system communication and data security, preventing intrusion and data leakage. The remote control module supports users to remotely control smart home devices through an AR interface, realizing instruction transmission and status feedback.

2. The KNX protocol integration and remote control system applied to smart home according to claim 1, characterized in that: The KNX bus master control module includes a hardware interface unit, a protocol conversion unit, and a task scheduling unit. The hardware interface unit supports TP-C / RF / IP / Zigbee / Wi-Fi6e dual-mode adaptive access. This unit automatically selects a communication method for data transmission by real-time monitoring the signal strength and stability of the communication environment. When detecting high and stable wireless signal strength, it preferentially uses wireless communication. If the wireless signal is poor, it switches to wired communication to ensure device connection and data transmission. The protocol conversion unit performs a two-way mapping through the constructed device object model and the Internet of Things message queue. For different types of smart home devices, first convert their communication protocols into a unified device object model, and then map this model into the MQTT message format to realize data interaction between the devices and the system. In addition, the instructions sent from the system to the devices are also converted through this two-way mapping mechanism to ensure interoperability between devices with different protocols. The task scheduling unit manages KNX communication and device control tasks using a priority cascading mechanism. It assigns priorities to each task according to the urgency of the task. When multiple tasks request processing simultaneously, it preferentially processes tasks with higher priorities. At the same time, through the cascading mechanism, it ensures that after high-priority tasks are processed, low-priority tasks are processed in sequence. Among them, the distinction of task urgency is based on the user's needs and is freely set by the user in the remote control module.

3. The KNX protocol integration and remote control system applied to smart home according to claim 1, characterized in that: The scene perception module includes a millimeter-wave radar unit, an image sensor unit, and a data fusion unit. The millimeter-wave radar unit emits millimeter-wave signals and receives reflected signals, and calculates the distance, speed, and angle of objects by using the time difference and frequency change information of the signals, realizing the detection of objects in the space. The image sensor unit is used to collect image information in the space. It captures light through a lens, converts it into an electrical signal, and then performs noise reduction and enhancement processing through image processing algorithms to obtain clear image data, providing a visual basis for scene recognition. The data fusion unit is responsible for weighted fusion of the collected radar data and visual data, and outputs information for scene recognition.

4. The KNX protocol integration and remote control system applied to smart home according to claim 1, characterized in that: The energy consumption optimization module includes a device energy consumption monitoring unit, a device usage habit analysis unit, and a device intelligent control unit. The device energy consumption monitoring unit continuously collects the current and voltage of the device by installing current sensors and voltage sensors in the device circuit; then, based on the collected current and voltage parameters, it calculates the real-time energy consumption of the device using the average power calculation formula. At the same time, it uploads the calculated real-time energy consumption data to the system database for storage; The device usage habit analysis unit receives the energy consumption data uploaded by the device energy consumption monitoring unit, and uses data analysis and machine learning technologies. Through the clustering analysis method, it obtains the rules and patterns of device usage. According to the data analysis results, it establishes a user's device usage habit model, which can predict the user's future device usage behavior and provide a reference for the customization of energy-saving strategies; The device intelligent control unit first combines the device energy consumption characteristic data provided by the device energy consumption monitoring unit and the user usage habit model established by the device usage habit analysis unit to generate an energy-saving strategy; then, according to the determined energy-saving strategy, it generates corresponding device control instructions and sends the generated control instructions to the corresponding device through the KNX bus main control module; at the same time, it monitors the operating status and energy consumption of the device in real time to ensure that the device operates according to the requirements of the energy-saving strategy.

5. The KNX protocol integration and remote control system applied to smart home according to claim 1, characterized in that: The security protection module includes a terminal authentication unit, a data encryption unit, and an anomaly detection unit; The terminal authentication unit implements the EAP-TLS + national cryptographic certificate two-factor authentication method; when the user terminal accesses the system, it first authenticates the user's username and password through the EAP-TLS protocol; then, it uses the national cryptographic certificate for secondary authentication to ensure the legitimacy and security of the terminal device; The data encryption unit establishes an AES-GCM end-to-end encryption tunnel to encrypt and protect the data transmitted within the system; at the data sending end, it uses the AES-GCM algorithm to encrypt the data to generate ciphertext and an authentication tag; at the data receiving end, it uses the same key and algorithm to decrypt the ciphertext and verify the validity of the authentication tag to ensure that the data is not stolen during transmission; at the same time, it sets the update period of the key to enhance the security of encryption; The anomaly detection unit detects abnormal activities by continuously monitoring and analyzing the user behavior and device status within the system; first, it establishes a normal behavior model and a device status baseline, recording various parameters and behavior patterns of the system under normal operating conditions; then, it continuously collects the operating data of the system and compares and analyzes it with the normal model; when it finds that the data deviation exceeds the set threshold, it determines it as an abnormal activity and issues an alarm to the user.

6. The KNX protocol integration and remote control system applied to smart home according to claim 1, characterized in that: The remote control module includes an AR interaction unit, an instruction transmission unit, an instruction parsing and forwarding unit, and a status feedback unit; The AR interaction unit triggers control instructions through the AR interface; the AR interface combines the virtual control interface with the real home scene using augmented reality technology, enabling users to interact with home devices intuitively; users operate on the AR interface through gestures and voices to trigger corresponding control instructions; The instruction transmission unit transmits the user's control instructions to the cloud server via WebSocket; WebSocket is a two-way communication protocol that can establish a real-time connection between the browser and the server, and WebSocket allows real-time two-way data transmission between the server and the client; in the smart home system, the user's control instructions can be directly sent to the cloud server, and at the same time, the server will feedback the device status information to the client; The instruction parsing and forwarding unit converts the user's natural language control instructions into a standard KNX command set that can be recognized and executed by the smart home system, and accurately forwards the parsed instructions to the local main control unit to achieve remote control of smart home devices; among them, the conversion of natural language is achieved through the NLP engine of the cloud server. This engine can analyze the user's instruction intention through learning and training of natural language data, and then, according to the protocol specifications of the smart home system, convert this information into a standard KNX command set to ensure that the device can correctly identify and operate; The status feedback unit is responsible for feeding back the operating status information of the smart home device to the user, enabling the user to understand the working conditions of the device in real time and dynamically adjust the device control strategy according to the feedback information; the implementation method is as follows: after the home device finishes executing the instruction, it will send its own operating status information back to the system through the communication network; after the system receives this information, it will display it on the visualization terminal; then the system will dynamically adjust the device control strategy according to the feedback status information of the device and the user's demand preferences.

7. The KNX protocol integration and remote control system applied to smart home according to claim 3, characterized in that: The data fusion unit is responsible for weighted fusion of the collected radar data and visual data, and the output scene recognition information includes the following: After the data fusion unit receives the data transmitted by the millimeter-wave radar unit and the image sensor unit, it first preprocesses the data; for the millimeter-wave radar data, it checks the integrity and accuracy of the data, removes the existing outliers and noise interference; by setting a threshold range, it filters out the distance and speed data beyond the threshold range; for the image data, it will perform further feature extraction and normalization processing; extract the features in the image, and at the same time normalize the image data to make it within a unified scale and range for subsequent fusion processing; The data fusion unit assigns different weights to radar data and visual data according to different scene characteristics and data characteristics; the specific weight assignment will be dynamically adjusted according to the preset rules and the scene changes monitored in real time; the data fusion unit can monitor the speed information in the millimeter-wave radar data in real time; when the speed of the object detected by the radar exceeds the preset stationary threshold, it is determined that the object starts to move, and at this time, the weight of the radar data begins to increase; when the object is in a stationary state, the weight of the visual data will be increased according to the preset rules; among them, the sum of the weights of the radar data and the visual data is always 100%, the adjustment range of the radar data is 30%-90%, and the adjustment range of the visual data weight is 10%-70%; according to the assigned weights, the preprocessed radar data and visual data are weighted and summed; the radar data and the visual data are weighted and combined in the same dimension to obtain the fused comprehensive data; Then, the fused comprehensive data is analyzed and processed, and the scene information in the space is identified using the pattern recognition algorithm; the implementation method is as follows: first, feature extraction and enhancement are performed. For the fused comprehensive data, multi-dimensional feature extraction is first carried out; in the spatial dimension, using the gradient algorithm of the image, the gradient value of the object edge in the image is calculated to highlight the contour features of the object; at the same time, based on the radar data, the distance distribution features of the object are extracted, and a distance histogram is constructed to describe the distribution of the object in space; in the time dimension, a sliding window method is adopted to process the fused data within a period of time, and the change rate of the data in adjacent time windows is calculated to obtain the motion trend features of the object; Finally, the principal component analysis method is used to reduce the dimension of the extracted high-dimensional features; first, the covariance matrix of the feature data is calculated, and this matrix reflects the correlation between different features; then, the covariance matrix is eigen-decomposed to obtain the eigenvalues and eigenvectors; the eigenvectors are sorted according to the size of the eigenvalues, and the first k eigenvectors corresponding to the largest k eigenvalues are selected as the principal components; through these k principal components, the original high-dimensional feature data is projected into a low-dimensional space to achieve the purpose of dimension reduction while retaining the data information; For model recognition and classification, the support vector machine algorithm is used for scene classification; in the training stage, the fused data samples after dimension reduction processing are used as inputs, and the corresponding scene categories of the samples are used as labels; the SVM algorithm separates the data samples of different categories by finding an optimal classification hyperplane, and this classification hyperplane can maximize the interval between the two types of data; for linearly separable data, a linear kernel function is used to construct a classification model; for linearly inseparable data, the kernel trick is introduced, and a suitable non-linear kernel function is selected to map the data into a high-dimensional space to make it linearly separable, and then a classification model is constructed; In the classification stage, the new fused data is input into the trained SVM model after the same feature extraction and dimension reduction processing, and the model judges the scene category to which the data belongs according to the learned classification rules; For the analysis of the scene change trend, the hidden Markov model is used to analyze the scene change trend; HMM is a double stochastic process, which contains a hidden state sequence and an observable output sequence; the scene categories at different times are used as the observable outputs, while the hidden states represent the internal trends of scene changes; First, initialize the parameters of the HMM, including the state transition probability matrix, the observation probability matrix, and the initial state probability distribution; the state transition probability matrix describes the probability of transitioning from one hidden state to another hidden state; the observation probability matrix represents the probability of generating a scene category under a hidden state; the initial state probability distribution defines the probabilities of each hidden state at the initial moment; Then, according to the historical scene category data, use the Baum - Welch algorithm to train and optimize the parameters of the HMM, so that the model can fit the actual scene change process; in the prediction stage, based on the current scene category and the trained HMM model, calculate the predicted scene category distribution through the forward algorithm, so as to predict the scene change trend; By comparing and analyzing the fusion data at different times, monitor the scene change trend; finally, output the recognized scene information to provide an environmental basis for other modules of the smart home system.

8. The KNX protocol integration and remote control system applied to smart home according to claim 4, characterized in that: According to the data analysis results, establish a user's device usage habit model, which can predict the user's future device usage behavior and provide a reference for the formulation of energy - saving strategies, including the following content: First, select the long short - term memory network as the core model. This model has an input layer, an LSTM layer, a fully - connected layer, and an output layer; among them, the input layer receives the feature vectors obtained in the feature extraction stage, and its dimension is determined by the number of extracted features; these features include time features, frequency features, and correlation features; then set multiple LSTM units to form a hidden layer. Each LSTM unit contains an input gate, a forget gate, and an output gate, which are used to control the inflow, retention, and output of information; by stacking multiple LSTM layers, increase the complexity and expressive ability of the model; connect the output of the LSTM layer to a fully - connected layer to perform a linear transformation on the output of the LSTM layer to map it to the final output space; then initialize the parameters of the model, including the weights and biases of the LSTM layer, the weights and biases of the fully - connected layer; then input the data in the training set into the model in batches, and perform forward propagation to calculate the predicted values of the model; calculate the value of the loss function according to the predicted values and the true values, and use the backpropagation algorithm to calculate the gradient of the loss function with respect to the model parameters. The optimizer updates the model parameters according to the gradient to gradually reduce the value of the loss function; finally, repeat the above steps until the loss function reaches the preset number of training epochs; After the model training is completed, input the current device usage data into the trained model. The model performs forward propagation calculation according to the input data and outputs the prediction results of the user's future device usage behavior; finally, formulate an energy - saving strategy according to the energy consumption characteristics of the device, the predicted device usage behavior, and the energy - saving target; For the formulation of energy-saving strategies, it is necessary to first clarify the energy consumption characteristics of equipment, the predicted equipment usage behavior, and the energy-saving goals; then classify the equipment into deferrable equipment, non-deferrable equipment, and adjustable equipment, and set priorities; based on the equipment usage prediction, formulate preliminary strategies for various types of equipment. For deferrable equipment, off-peak operation and batch processing are carried out. For non-deferrable equipment, operation mode optimization and maintenance upgrade suggestions are provided. For adjustable equipment, parameter optimization and intelligent control are carried out; match the preliminary strategy with the energy-saving goal. If the goal is not achieved, adjust the strategy, which can increase energy-saving measures and optimize the equipment combination; then establish a strategy evaluation and feedback mechanism, set evaluation indicators to regularly evaluate the strategy effect, and collect user feedback to improve the strategy; finally, implement the energy-saving strategy, convert it into control instructions to achieve automatic control of equipment, and establish a real-time monitoring system to monitor the equipment operation status and energy consumption data, facilitating users to adjust the strategy through the system.

Images