Wireless control method and system for intelligent furniture and intelligent furniture

Abstract

The invention relates to the technical field of Internet of Things, in particular to a wireless control method and system for intelligent furniture and the intelligent furniture. The method comprises the following steps: capturing different types of radio signals through a user terminal, and extracting time-frequency characteristics of each signal by using fast Fourier transform to obtain multi-signal time-frequency characteristic data; carrying out weighted fusion on the multi-signal time-frequency characteristic data, and carrying out dynamic adjustment on the weight of each signal based on transfer learning to obtain a channel stability index; dynamically selecting a wireless communication protocol according to the channel stability index by using an SDN technology, and recording a protocol switching process on a private block chain to obtain protocol switching log data; and performing federal learning framework aggregation based on switching feedback according to the protocol switching log data so as to obtain protocol efficiency matrix data. According to the method, multiple radio signals are processed by dynamically adjusting weights through transfer learning, the adaptability of the system to a complex electromagnetic environment is improved, and the communication stability is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a wireless control method and system for smart furniture and the smart furniture. Background Art

[0002] With the development of the Internet of Things (IoT) and the widespread adoption of smart homes, users are increasingly demanding home automation and intelligence. They hope to wirelessly control their furniture for remote operation, automated management, and personalized services. Advances in wireless communication technology, data processing capabilities, and artificial intelligence algorithms have laid the foundation for the implementation of smart furniture. However, as the number of smart devices increases, traditional control methods face numerous challenges in data processing and real-time response.

[0003] Traditional wireless control methods for smart furniture often have the following problems: smart furniture relies on wireless communication for data transmission, but wireless channels are greatly affected by interference and environmental changes, and traditional methods find it difficult to ensure channel stability in real time; different devices may support different wireless communication protocols, and traditional methods find it difficult to dynamically select and switch according to real-time channel status, resulting in low communication efficiency; smart homes need to accurately execute user commands, but users' multimodal interactive behaviors (such as voice, gestures, etc.) are complex, and traditional methods find it difficult to accurately understand and execute user intentions. Summary of the Invention

[0004] Based on this, it is necessary for the present invention to provide a wireless control method and system for smart furniture to solve at least one of the above technical problems.

[0005] To achieve the above object, a wireless control method for smart furniture includes the following steps:

[0006] Step S1: Capture different types of radio signals through the user terminal and extract the time-frequency characteristics of each signal using fast Fourier transform to obtain multi-signal time-frequency feature data; perform weighted fusion on the multi-signal time-frequency feature data and dynamically adjust the weights of each signal based on transfer learning to obtain a channel stability index;

[0007] Step S2: Using SDN technology to dynamically select wireless communication protocols based on the channel stability index, and recording the protocol switching process on a private blockchain to obtain protocol switching log data; performing aggregation based on the switching feedback federated learning framework based on the protocol switching log data to obtain protocol effectiveness matrix data;

[0008] Step S3: Constructing a semantic context network based on pre-acquired environmental sensor data and protocol effectiveness matrix data; obtaining user instruction data, parsing the user instruction data, extracting instruction context semantics, and semantically updating the semantic context network to obtain semantic network update data; generating an instruction credibility score based on the semantic network update data;

[0009] Step S4: Analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; use GAN simulation to generate misoperation scenarios as interaction verification test sets, and verify the operation intention based on the multimodal behavior data and interaction verification test sets to obtain intention verification result data; generate intention verification results based on the instruction credibility score and intention verification result data. Figure 1 consistency indicator data;

[0010] Step S5: Figure 1 The consistency index data is used to conduct command conflict arbitration based on fog computing, and the multi-device collaborative logical relationship analysis is performed to obtain collaborative execution decision data.

[0011] The present invention captures a variety of radio signals to provide the system with a rich data source, which helps to more comprehensively understand the furniture environment and user behavior; the fast Fourier transform (FFT) extracts the time-frequency characteristics of the signal, providing a scientific basis for evaluating signal quality and stability; the weights are dynamically adjusted through weighted fusion and transfer learning to generate a channel stability index, providing stable and reliable channel selection for wireless communication of smart furniture. SDN technology allows the dynamic selection of the optimal communication protocol based on the channel stability index, improving communication efficiency and reliability; recording the protocol switching process on a private blockchain ensures the immutability and transparency of the data and increases the security of the system; the protocol switching log provides data support for subsequent federated learning framework aggregation, which helps to optimize the selection and adjustment of communication protocols. Constructing a semantic context network helps the system better understand the contextual relationship between the furniture environment and user instructions; by parsing user instructions and extracting contextual semantics, the accuracy of instruction execution and user satisfaction are improved; the generated instruction credibility score provides the system with a reliability assessment of user instructions, providing a basis for subsequent decision-making. Analyze the user's multimodal interactive behaviors, such as voice and gestures, so that the system can understand the user's intention more comprehensively; use GAN to simulate misoperation scenarios, improve the system's ability to identify and handle abnormal situations, and verify the intention generated by the result data. Figure 1Consistency indicators ensure consistency between user commands and system execution. The command conflict arbitration mechanism based on fog computing effectively solves the command conflict problem that may arise in a multi-device environment; the multi-device collaborative logical relationship analysis optimizes the collaborative work of devices, improves the execution efficiency and response speed of the overall system, and the collaborative execution decision data provides intelligent decision-making support for the system, making furniture control more intelligent and automated. In the wireless control method of smart furniture, these steps together construct a complete process from signal capture to command execution. Each step contributes to improving the intelligence level of the system, the user interaction experience and the accuracy of operation. Through this comprehensive approach, the smart furniture system can better adapt to user needs and provide more personalized and efficient services.

[0012] The present invention further provides a wireless control system for smart furniture, which is used to execute the above-mentioned wireless control method for smart furniture. The wireless control system for smart furniture includes:

[0013] The channel stability assessment module is used to capture different types of radio signals through user terminals and extract the time-frequency characteristics of each signal using fast Fourier transform to obtain multi-signal time-frequency feature data. The module also performs weighted fusion of the multi-signal time-frequency feature data and dynamically adjusts the weights of each signal based on transfer learning to obtain a channel stability index.

[0014] The dynamic protocol selection module is used to dynamically select wireless communication protocols based on the channel stability index using SDN technology, and record the protocol switching process on the private blockchain to obtain protocol switching log data. The protocol switching log data is aggregated using a federated learning framework based on switching feedback to obtain protocol effectiveness matrix data.

[0015] The semantic context processing module is used to construct a semantic context network based on pre-acquired environmental sensor data and protocol effectiveness matrix data; obtain user command data, parse the user command data, extract the command context semantics, and semantically update the semantic context network to obtain semantic network update data; and generate a command credibility score based on the semantic network update data;

[0016] The multimodal intention verification module is used to analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; use GAN simulation to generate misoperation scenarios as interaction verification test sets, and verify the operation intention based on the multimodal behavior data and interaction verification test sets to obtain intention verification result data; generate intention verification results based on the instruction credibility score and intention verification result data. Figure 1 consistency indicator data;

[0017] Intelligent collaborative decision-making module for Figure 1The consistency index data is used to conduct command conflict arbitration based on fog computing, and the multi-device collaborative logical relationship analysis is performed to obtain collaborative execution decision data.

[0018] The channel stability assessment module of the present invention provides an assessment of wireless communication channel quality. It captures different types of radio signals and extracts their time-frequency features to obtain multi-signal time-frequency feature data. This multi-signal time-frequency feature data is weighted and fused, and the signal weights are dynamically adjusted to obtain a channel stability index. This assessment module improves the stability and reliability of wireless communication, enabling the system to select the optimal communication method based on real-time channel conditions to ensure good communication quality. The dynamic protocol selection module utilizes software-defined networking (SDN) technology to dynamically select wireless communication protocols based on the channel stability index. By recording protocol switching log data and aggregating it using a federated learning framework, protocol performance matrix data is obtained. This module improves the accuracy and efficiency of communication protocol selection, enabling the system to select the most suitable communication protocol based on real-time channel conditions, thereby providing better communication performance and user experience. The semantic context processing module utilizes environmental sensor data and protocol performance matrix data to construct a semantic context network, parse user command data, extract command context semantics, and update the semantic context network. The updated data from the semantic network generates a command credibility score. The effect of this module is to improve the accuracy of understanding user instructions and the ability to perceive context. The system can better understand the user's intentions and make decisions and executions based on the semantic context to provide more accurate and personalized services. The multimodal intention verification module obtains multimodal behavior data by analyzing the user's multimodal interaction behavior. Generate misoperation scenarios using generative adversarial networks (GANs) as interaction verification test sets. Verify the operation intention based on the multimodal behavior data and the interaction verification test set to obtain the intention verification result data. Generate intentions based on the instruction credibility score and the intention verification result data. Figure 1 The module improves the accuracy and credibility of user intention verification. The system can better understand the user's intention and accurately perform corresponding operations, thus providing a more satisfactory user experience. Figure 1 This module uses consistency indicator data to conduct fog computing-based command conflict resolution and analyze the logical relationships of multi-device collaboration to generate collaborative execution decision data. This module effectively resolves conflicts that may arise when multiple devices simultaneously execute commands, ensuring consistent and coordinated command execution. Furthermore, multi-device collaborative logical relationship analysis helps determine the collaboration method and execution order between devices, improving the efficiency and effectiveness of multi-device collaborative work.

[0019] Preferably, the present invention further provides a smart furniture, comprising:

[0020] memory for storing computer programs;

[0021] A processor is used to implement the above-mentioned wireless control method of smart furniture when executing the computer program. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0023] Figure 1 Schematic diagram of the steps of the wireless control method of smart furniture of the present invention;

[0024] Figure 2 for Figure 1 Detailed step flow diagram of step S1;

[0025] Figure 3 for Figure 1 Detailed step flow chart of step S2 in FIG. DETAILED DESCRIPTION

[0026] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0027] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0028] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0029] To achieve this, please refer to Figures 1 to 3The present invention provides a wireless control method for smart furniture, the method comprising the following steps:

[0030] Step S1: Capture different types of radio signals through the user terminal and extract the time-frequency characteristics of each signal using fast Fourier transform to obtain multi-signal time-frequency feature data; perform weighted fusion on the multi-signal time-frequency feature data and dynamically adjust the weights of each signal based on transfer learning to obtain a channel stability index;

[0031] Step S2: Using SDN technology to dynamically select wireless communication protocols based on the channel stability index, and recording the protocol switching process on a private blockchain to obtain protocol switching log data; performing aggregation based on the switching feedback federated learning framework based on the protocol switching log data to obtain protocol effectiveness matrix data;

[0032] Step S3: Constructing a semantic context network based on pre-acquired environmental sensor data and protocol effectiveness matrix data; obtaining user instruction data, parsing the user instruction data, extracting instruction context semantics, and semantically updating the semantic context network to obtain semantic network update data; generating an instruction credibility score based on the semantic network update data;

[0033] Step S4: Analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; use GAN simulation to generate misoperation scenarios as interaction verification test sets, and verify the operation intention based on the multimodal behavior data and interaction verification test sets to obtain intention verification result data; generate intention verification results based on the instruction credibility score and intention verification result data. Figure 1 consistency indicator data;

[0034] Step S5: Figure 1 The consistency index data is used to conduct command conflict arbitration based on fog computing, and the multi-device collaborative logical relationship analysis is performed to obtain collaborative execution decision data.

[0035] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a wireless control method for smart furniture of the present invention. In this example, the wireless control method for smart furniture includes the following steps:

[0036] Step S1: Capture different types of radio signals through the user terminal and extract the time-frequency characteristics of each signal using fast Fourier transform to obtain multi-signal time-frequency feature data; perform weighted fusion on the multi-signal time-frequency feature data and dynamically adjust the weights of each signal based on transfer learning to obtain a channel stability index;

[0037] The embodiment of the present invention captures radio signals on a user terminal and uses a fast Fourier transform (FFT) to extract time-frequency features from the captured signals. This can be achieved by using appropriate hardware devices (such as a radio receiver) and software libraries (such as a digital signal processing library). The captured signals may include different types of radio signals, such as Wi-Fi, Bluetooth, LTE, etc. By applying the FFT algorithm to each signal, it can be converted into a spectrum diagram to obtain the energy distribution of the signal at different frequencies and times. These time-frequency features are weighted and fused, and the weights of each signal are dynamically adjusted using a transfer learning-based method to obtain a channel stability index.

[0038] Step S2: Using SDN technology to dynamically select wireless communication protocols based on the channel stability index, and recording the protocol switching process on a private blockchain to obtain protocol switching log data; performing aggregation based on the switching feedback federated learning framework based on the protocol switching log data to obtain protocol effectiveness matrix data;

[0039] The embodiment of the present invention utilizes software-defined networking (SDN) technology to achieve dynamic selection of wireless communication protocols. SDN allows the network control plane to be separated from the data forwarding plane, and network devices are managed and configured through a centralized controller. In this step, a wireless communication protocol suitable for the current channel conditions is selected based on the channel stability index. The selection process can be completed by implementing a protocol selection algorithm in the SDN controller. At the same time, the log data of the protocol switching process is recorded and stored on a private blockchain to ensure the security and non-tamperability of the data. The protocol switching log data of different user devices is aggregated through a federated learning framework based on switching feedback to obtain protocol performance matrix data.

[0040] Step S3: Constructing a semantic context network based on pre-acquired environmental sensor data and protocol effectiveness matrix data; obtaining user instruction data, parsing the user instruction data, extracting instruction context semantics, and semantically updating the semantic context network to obtain semantic network update data; generating an instruction credibility score based on the semantic network update data;

[0041] The embodiment of the present invention constructs a semantic context network using pre-acquired environmental sensor data and protocol performance matrix data. Environmental sensors may include sensors for temperature, humidity, light, etc., which are used to obtain status information of the current environment. The protocol performance matrix data is obtained in step S2 and is used to represent the performance of different protocols under different channel conditions. These data are combined to construct a semantic context network, in which nodes represent different semantic concepts and edges represent the relationships between them. Appropriate algorithms and data structures are used to represent and manage the semantic context network, and it is semantically updated according to user instruction data.

[0042] Step S4: Analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; use GAN simulation to generate misoperation scenarios as interaction verification test sets, and verify the operation intention based on the multimodal behavior data and interaction verification test sets to obtain intention verification result data; generate intention verification results based on the instruction credibility score and intention verification result data. Figure 1 consistency indicator data;

[0043] The embodiment of the present invention analyzes the multimodal interactive behavior of the user through human-computer interaction technology. Multimodal interactive behavior can include various forms of input such as voice instructions, gestures, touch, etc. Use appropriate sensors and algorithms to convert these interactive behaviors into processable data formats. At the same time, use the generative adversarial network (GAN) to simulate and generate misoperation scenarios as a test set for interaction verification. The multimodal behavior data and the interaction verification test set are used to verify the operation intention, determine the user's operation intention and generate corresponding intention verification result data. Generate intention according to the instruction credibility score and the intention verification result data. Figure 1 Consistency indicator data.

[0044] Step S5: Figure 1 The consistency index data is used to conduct command conflict arbitration based on fog computing, and the multi-device collaborative logical relationship analysis is performed to obtain collaborative execution decision data.

[0045] The embodiments of the present invention are intended to Figure 1 The consistency index data is used to conduct fog computing-based instruction conflict arbitration and multi-device collaboration logic relationship analysis. In this step, fog computing technology is used to Figure 1 Consistency indicator data is processed and analyzed. Fog computing is a distributed computing model that pushes computing tasks and data processing to the edge of the network to reduce transmission latency and network congestion. By performing computations and decision-making on edge devices, faster responses and more efficient collaboration can be achieved. Here, fog computing is used to resolve conflicting instructions, resolving conflicts that may arise when multiple instructions are issued simultaneously. Simultaneously, multi-device collaborative logical relationship analysis is performed to determine the collaborative methods and logical relationships between multiple devices. Ultimately, collaborative execution decision data is generated to guide device collaboration.

[0046] The present invention captures a variety of radio signals to provide the system with a rich data source, which helps to more comprehensively understand the furniture environment and user behavior; the fast Fourier transform (FFT) extracts the time-frequency characteristics of the signal, providing a scientific basis for evaluating signal quality and stability; the weights are dynamically adjusted through weighted fusion and transfer learning to generate a channel stability index, providing stable and reliable channel selection for wireless communication of smart furniture. SDN technology allows the dynamic selection of the optimal communication protocol based on the channel stability index, improving communication efficiency and reliability; recording the protocol switching process on a private blockchain ensures the immutability and transparency of the data and increases the security of the system; the protocol switching log provides data support for subsequent federated learning framework aggregation, which helps to optimize the selection and adjustment of communication protocols. Constructing a semantic context network helps the system better understand the contextual relationship between the furniture environment and user instructions; by parsing user instructions and extracting contextual semantics, the accuracy of instruction execution and user satisfaction are improved; the generated instruction credibility score provides the system with a reliability assessment of user instructions, providing a basis for subsequent decision-making. Analyze the user's multimodal interactive behaviors, such as voice and gestures, so that the system can understand the user's intention more comprehensively; use GAN to simulate misoperation scenarios, improve the system's ability to identify and handle abnormal situations, and verify the intention generated by the result data. Figure 1 Consistency indicators ensure consistency between user commands and system execution. The command conflict arbitration mechanism based on fog computing effectively solves the command conflict problem that may arise in a multi-device environment; the multi-device collaborative logical relationship analysis optimizes the collaborative work of devices, improves the execution efficiency and response speed of the overall system, and the collaborative execution decision data provides intelligent decision-making support for the system, making furniture control more intelligent and automated. In the wireless control method of smart furniture, these steps together construct a complete process from signal capture to command execution. Each step contributes to improving the intelligence level of the system, the user interaction experience and the accuracy of operation. Through this comprehensive approach, the smart furniture system can better adapt to user needs and provide more personalized and efficient services.

[0047] Preferably, step S1 includes the following steps:

[0048] Step S11: collecting multiple radio signals from the surrounding environment through the radio receiving module of the user terminal to obtain original radio signal data; denoising the original radio signal data and performing channel equalization to obtain multi-source radio signal data;

[0049] Step S12: Performing a fast Fourier transform on the multi-source radio signal data and extracting the time domain and frequency domain features of each signal to obtain multi-signal time-frequency feature data, wherein the time-frequency features include power spectral density, symbol delay spread, signal envelope, pulse width, and spectral moment;

[0050] Step S13: performing weighted fusion on the multi-signal time-frequency feature data to generate fused signal feature data;

[0051] Step S14: collecting environmental state vector data around the terminal, and using transfer learning technology to dynamically adjust the signal weights in the fused signal feature data according to the environmental state vector data, thereby obtaining signal feature weight coefficients;

[0052] Step S15: weight correction is performed on the fused signal feature data according to the signal feature weight coefficient, and multiple quality index analyses of the signal are performed to generate a channel stability index, where the multiple quality indicators include signal-to-interference-plus-noise ratio, multipath spread, phase error, and carrier-to-noise ratio.

[0053] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S1 in the embodiment of the present invention, step S1 includes the following steps:

[0054] Step S11: collecting multiple radio signals from the surrounding environment through the radio receiving module of the user terminal to obtain original radio signal data; denoising the original radio signal data and performing channel equalization to obtain multi-source radio signal data;

[0055] In this embodiment of the present invention, the radio receiving module of the user terminal collects radio signals from the surrounding environment and performs denoising and channel equalization processing on the raw data to obtain multi-source radio signal data. This can be achieved using a software-defined radio (SDR) device, such as RTL-SDR or USRP, and applying technologies such as wavelet denoising algorithms and adaptive equalizers.

[0056] Step S12: Performing a fast Fourier transform on the multi-source radio signal data and extracting the time domain and frequency domain features of each signal to obtain multi-signal time-frequency feature data, wherein the time-frequency features include power spectral density, symbol delay spread, signal envelope, pulse width, and spectral moment;

[0057] This embodiment of the present invention performs a Fast Fourier Transform (FFT) on multi-source radio signal data to extract the time and frequency domain features of each signal. By applying the FFT algorithm, the signal is converted from the time domain to the frequency domain. Time domain features are then extracted using methods such as peak detection and autocorrelation functions. Frequency domain features such as power spectral density and spectral moments are also calculated.

[0058] Step S13: performing weighted fusion on the multi-signal time-frequency feature data to generate fused signal feature data;

[0059] This embodiment of the present invention performs weighted fusion on the time-frequency feature data of multiple signals. A corresponding weight is assigned to the time-frequency features of each signal, and these are weighted summed to obtain fused signal feature data. This can be achieved using weighted fusion methods such as simple weighted summation or principal component analysis (PCA).

[0060] Step S14: collecting environmental state vector data around the terminal, and using transfer learning technology to dynamically adjust the signal weights in the fused signal feature data according to the environmental state vector data, thereby obtaining signal feature weight coefficients;

[0061] This embodiment of the present invention collects environmental state vector data around the terminal and uses transfer learning technology to dynamically adjust the signal weights in the fused signal feature data based on this data. By collecting data from environmental sensors such as temperature, humidity, and light intensity, combined with a transfer learning algorithm (such as a deep learning-based transfer learning method), the signal feature weights are adaptively adjusted.

[0062] Step S15: weight correction is performed on the fused signal feature data according to the signal feature weight coefficient, and multiple quality index analyses of the signal are performed to generate a channel stability index, where the multiple quality indicators include signal-to-interference-plus-noise ratio, multipath spread, phase error, and carrier-to-noise ratio.

[0063] In this embodiment of the present invention, the fused signal feature data is weighted and combined according to the signal feature weight coefficient to obtain the final environment recognition result. The signal feature weight coefficient is multiplied by the fused signal feature data, and an appropriate classification algorithm (such as a support vector machine, decision tree, neural network, etc.) is used to identify and classify the weighted combined feature data to obtain the final environment recognition result.

[0064] The present invention collects multiple radio signals from the surrounding environment to obtain comprehensive information about the wireless communication environment, including signals from different frequency bands and protocols. Denoising and channel equalization can improve signal quality and reduce interference, ensuring the accuracy of subsequent analysis and processing. Fast Fourier transform can convert signals from the time domain to the frequency domain, providing information about the signal's energy distribution at different frequencies. Time-frequency feature extraction can analyze signal characteristics from both the time and frequency domain perspectives. For example, power spectral density reflects the signal's energy distribution, and symbol delay spread reflects the signal's propagation delay. These features help understand the signal's characteristics and quality. Weighted fusion can comprehensively consider the characteristics and quality of different signals to provide a comprehensive signal feature representation. By fusing signal feature data, the data dimension can be reduced, subsequent processing can be simplified, and key information can be retained. Environmental state vector data provides information about the user terminal's surrounding environment, such as noise level and channel fading. Using transfer learning technology, the signal feature weights in the current environment can be dynamically adjusted based on previously learned knowledge and models to adapt to changes in different environments. Weight correction adjusts the importance of fused signal feature data based on signal feature weight coefficients, more accurately reflecting signal quality. Multiple quality metrics provide a comprehensive assessment of signal quality, such as signal-to-interference-and-noise ratio (SIN / INR), which reflects the ratio of signal to interference and noise; multipath spread, which reflects the multipath effects of signal propagation; phase error, which reflects the accuracy of signal phase; and carrier-to-noise ratio, which reflects the ratio of signal to noise. These metrics can help assess channel stability and signal quality. By collecting multiple radio signals and performing denoising and channel equalization, comprehensive information about the wireless communication environment can be obtained. Denoising and channel equalization improve signal quality and reduce the impact of interference and noise. Fast Fourier transforms and time-frequency feature extraction enable analysis of signal features from multiple perspectives in the time and frequency domains, providing a foundation for subsequent processing. Using environmental state vector data and transfer learning techniques, signal feature weights can be dynamically adjusted based on the current environment, adapting to changing conditions. By weight correction and analyzing multiple quality metrics on the fused signal feature data, a channel stability index can be generated, comprehensively assessing multiple aspects of signal quality.

[0065] Preferably, step S13 includes the following steps:

[0066] Step S131: normalizing the time-frequency feature data of multiple signals and removing outliers to obtain standard time-frequency feature data;

[0067] This embodiment of the present invention normalizes and removes outliers from multi-signal time-frequency feature data to obtain standardized time-frequency feature data. Using the Python programming language and related libraries, such as NumPy and Pandas, the data can be normalized using min-max or Z-score normalization to ensure that different features have the same scale range. Statistical analysis or outlier detection algorithms, such as boxplots or isolation forests, can also be used to remove outliers to maintain data accuracy and consistency.

[0068] Step S132: performing key feature screening on each feature in the time-frequency feature standard data based on the distinguishing capability of feature indicators in wireless control, thereby generating simplified time-frequency feature data;

[0069] This embodiment of the present invention screens standard time-frequency feature data for key features and generates simplified time-frequency feature data. By calculating feature discrimination metrics, such as information gain or variance, the importance of each feature for distinguishing different environments can be assessed. Using feature selection algorithms in Python, such as mutual information or recursive feature elimination, key features with high discriminative power can be selected based on the size of the feature metrics, thereby generating simplified data.

[0070] Step S133: Initial weights are assigned to the simplified time-frequency feature data through a gradient boosting tree, and weighted fusion is performed to obtain fused signal feature data.

[0071] This embodiment of the present invention uses a gradient boosting tree to assign initial weights to the simplified time-frequency feature data and performs weighted fusion to obtain fused signal feature data. Using a gradient boosting tree algorithm, such as XGBoost or LightGBM, a model can be trained to obtain feature importance weights. The initial weights are multiplied by the simplified time-frequency feature data and combined using a weighted fusion method, such as weighted summation or principal component analysis, to obtain fused signal feature data.

[0072] The feature normalization of the present invention can unify the value ranges of different features, making them comparable and avoiding preference problems caused by numerical differences; outlier removal can eliminate abnormal or erroneous feature values, thereby improving the accuracy and stability of the data. Key feature screening based on the distinguishing ability of feature indicators can identify and select the features with the most distinguishing ability for wireless control tasks; simplified data of time-frequency features can reduce the dimension of features, retain the most relevant and representative features, and reduce the complexity of data processing. Using a gradient boosting tree to assign initial weights to simplified time-frequency feature data can perform weighted processing based on the importance of the data, better reflecting the contribution of each feature to the signal quality; weighted fusion can comprehensively consider the weights of each feature, generate fused signal feature data, and provide a comprehensive signal feature representation. Normalizing and removing outliers from multi-signal time-frequency feature data can improve data accuracy and stability; key feature screening based on the distinguishing ability of feature indicators in wireless control can select the features that are most discriminative for the task, thereby improving the discriminative ability and effectiveness of the features; simplifying data by generating time-frequency features can reduce the dimension of the data, reduce computational complexity, and retain the most relevant feature information; assigning initial weights to the simplified data through a gradient boosting tree and performing weighted fusion can better reflect the contribution and importance of each feature and provide a comprehensive signal feature representation.

[0073] Preferably, step S2 includes the following steps:

[0074] Step S21: Initialize the SDN controller according to the channel stability index and load the available wireless communication protocol to obtain the wireless communication protocol library and SDN controller status data;

[0075] Step S22: Perform protocol matching on the wireless communication protocol library based on the channel stability index, dynamically adjust the terminal's MAC / PHY layer protocol stack using the SDN controller status data, and record the new and old protocol switching details and context on the private blockchain to obtain protocol switching log data;

[0076] Step S23: performing performance index analysis of throughput, delay, and power consumption of different protocols under various channel conditions based on the protocol switching log data, thereby obtaining protocol performance index data;

[0077] Step S24: Utilize the federated learning framework to aggregate the protocol performance indicator data of each node to obtain protocol effectiveness matrix data.

[0078] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 1 Detailed step flow diagram of step S2 in the embodiment of the present invention, step S2 includes the following steps:

[0079] Step S21: Initialize the SDN controller according to the channel stability index and load the available wireless communication protocol to obtain the wireless communication protocol library and SDN controller status data;

[0080] This embodiment of the present invention initializes the SDN controller based on the channel stability index and loads available wireless communication protocols, thereby generating a wireless communication protocol library and SDN controller status data. First, a suitable algorithm or model is used to calculate the channel stability index, taking into account factors such as channel quality, interference level, and signal strength. The SDN controller is then initialized based on the stability index and loaded with available wireless communication protocols to form a protocol library. Simultaneously, the SDN controller records and updates status data related to network devices, protocols, and channels.

[0081] Step S22: Perform protocol matching on the wireless communication protocol library based on the channel stability index, dynamically adjust the terminal's MAC / PHY layer protocol stack using the SDN controller status data, and record the new and old protocol switching details and context on the private blockchain to obtain protocol switching log data;

[0082] This embodiment of the present invention matches the wireless communication protocol library based on the channel stability index and dynamically adjusts the terminal's MAC / PHY layer protocol stack using SDN controller status data. First, an appropriate protocol is selected for matching based on the current channel stability index, employing a predefined protocol selection strategy. Then, using the SDN controller status data, the terminal device's protocol stack is dynamically adjusted, including modifying protocol parameters, switching protocols, and reconfiguring. Furthermore, the details and context of the protocol switching are recorded on a private blockchain to ensure data security and immutability.

[0083] Step S23: performing performance index analysis of throughput, delay, and power consumption of different protocols under various channel conditions based on the protocol switching log data, thereby obtaining protocol performance index data;

[0084] This embodiment of the present invention analyzes performance indicators such as throughput, latency, and power consumption of different protocols under various channel conditions based on protocol switching log data. By analyzing protocol switching log data, various performance indicators, including throughput, latency, and power consumption, can be extracted. Statistical analysis or machine learning methods can be used to analyze the data and obtain protocol performance indicator data. This data can be used to evaluate the performance of different protocols under different channel conditions.

[0085] Step S24: Utilize the federated learning framework to aggregate the protocol performance indicator data of each node to obtain protocol effectiveness matrix data.

[0086] This embodiment of the present invention utilizes a federated learning framework to aggregate protocol performance indicator data from each node, thereby generating protocol performance matrix data. Federated learning allows for model training and parameter aggregation while protecting data privacy. By aggregating protocol performance indicator data from each node, a global protocol performance matrix can be generated, which can be used to evaluate and compare the performance of different protocols across the entire system.

[0087] Initializing the SDN controller based on the channel stability index provides a foundation for subsequent protocol selection and adjustment, ensuring the SDN controller is in an appropriate state. Loading available wireless communication protocols and building a wireless communication protocol library provides multiple protocol options, enabling the SDN controller to perform protocol switching and adjustment. Protocol matching within the wireless communication protocol library based on the channel stability index allows the selection of protocols suitable for current channel conditions, improving communication performance and efficiency. Dynamically adjusting the terminal's MAC / PHY layer protocol stack using SDN controller status data optimizes the protocol based on real-time channel conditions, improving system adaptability and flexibility. Recording protocol switching details and context on a private blockchain provides reliable log data for subsequent analysis and verification. Protocol performance indicator analysis under different channel conditions evaluates the performance of various protocols in different environments, revealing their strengths and weaknesses. The resulting protocol performance indicator data provides a reference for subsequent protocol selection and optimization, helping select the best-performing protocol solution. Aggregating protocol performance indicator data from each node using a federated learning framework integrates data from each node to generate a more comprehensive and accurate protocol performance matrix. This protocol performance matrix provides a performance comparison of different protocols across multiple nodes, helping to understand the performance and adaptability of the protocols across the entire network. Based on the channel stability index, the SDN controller is initialized and the wireless communication protocol library is loaded, ensuring that the SDN controller is in an appropriate state and providing multiple protocol options. The protocol library is matched based on the channel stability index, and the terminal protocol stack is dynamically adjusted using SDN controller status data. The details and context of protocol switching are recorded, providing reliable log data for subsequent analysis. Based on the protocol switching log data, the performance indicators of different protocols under various channel conditions are analyzed to evaluate the performance and advantages of the protocols. A federated learning framework is used to aggregate the protocol performance indicator data of each node to obtain protocol effectiveness matrix data. This data is combined from each node to provide a comprehensive and accurate protocol performance evaluation. These steps help achieve dynamic wireless communication protocol selection and adjustment, improving the adaptability and performance of wireless communication systems. By selecting the appropriate protocol based on real-time channel conditions and optimizing it based on performance indicators, higher throughput, lower latency, and lower power consumption can be achieved. Furthermore, through federated learning aggregation, data from different nodes can be combined to obtain a global protocol performance evaluation, providing a reference for overall network optimization.

[0088] Preferably, step S3 includes the following steps:

[0089] Step S31: Initially construct a semantic context network based on a graph neural network based on the pre-acquired environmental sensor data and the protocol effectiveness matrix data, where nodes represent environmental factors and protocol effectiveness indicators, and edges represent relationships between nodes, thereby obtaining a semantic context network;

[0090] Step S32: Obtain user instruction data, and use natural language processing technology to perform keyword and intent recognition on the user instruction data, thereby obtaining user instruction parsed data;

[0091] Step S33: performing semantic analysis on the user instruction parsed data to obtain contextual semantics including time, location, and object information, thereby obtaining instruction contextual semantic data;

[0092] Step S34: Mapping the instruction context semantic data to relevant nodes and edges in the semantic context network, thereby obtaining semantic mapping result data;

[0093] Step S35: dynamically adjusting the node weights and edge strengths of the semantic context network according to the semantic mapping result data, thereby obtaining updated semantic network data;

[0094] Step S36: Generate an instruction credibility score based on the semantic network update data.

[0095] This embodiment of the present invention collects environmental sensor data and protocol performance matrix data and uses graph neural network technology to construct a semantic context network. Nodes represent environmental factors and protocol performance indicators, while edges represent relationships between nodes. A graph convolutional network (GCN) is used to encode the graph structure, and an association rule learning algorithm is used to determine the relationships between nodes. Subsequently, user commands are captured through a speech recognition system or text input interface and preprocessed using natural language processing techniques, including word segmentation, stop word removal, and part-of-speech tagging. Next, machine learning or deep learning models are used to identify intent and generate user command parsed data. Furthermore, in-depth semantic analysis is performed on the parsed user command data, using syntactic analysis tools to identify sentence structure and semantic roles and extract contextual information. A semantic framework is constructed based on this contextual information, and knowledge graph technology is used to store and query this information. The command context semantic data is mapped to the semantic context network. Mapping rules are defined, and named entity recognition and entity linking techniques are used to map entities. Semantic relationships within the commands are identified and mapped to corresponding edges in the network. The semantic context network is dynamically adjusted based on the resulting semantic mapping data. A weight adjustment algorithm is designed to update node weights and edge strengths. Using the graph neural network's update mechanism, the network structure is dynamically adjusted based on the contextual semantics of user commands. Finally, a command credibility score is generated based on the updated semantic network data. A scoring model is constructed, extracting features from the updated semantic network, such as node centrality and clustering coefficient, as input. The model outputs a credibility score for the command, which is then thresholded or classified.

[0096] The semantic context network based on the graph neural network of the present invention can integrate environmental sensor data and protocol performance matrix data, establish relationships between nodes and edges, and form a comprehensive semantic network. The semantic context network provides a structured representation that can better express the relationship between environmental factors and protocol performance, providing a foundation for subsequent instruction parsing and optimization. Using natural language processing technology to perform keyword and intent recognition on user instruction data can extract key information about the user instruction and understand the user's intentions and needs. The user instruction parsed data provides a foundation for subsequent semantic analysis and contextual semantic construction, enabling the system to better understand the user's instructions and perform corresponding operations. By performing semantic analysis on the user instruction parsed data, contextual semantic information such as time, place, and object in the instruction can be further understood. The instruction context semantic data can more accurately describe the context and constraints of the user instruction, providing more comprehensive contextual information for subsequent semantic mapping and instruction scoring. Mapping the instruction context semantic data into the semantic context network can establish a connection between the instruction and the network, associating the instruction's semantic information with environmental factors and protocol performance. The semantic mapping result data provides the position and association information of the instruction in the semantic context network, providing a foundation for subsequent network adjustment and instruction scoring. Dynamically adjusting the semantic context network based on the semantic mapping results optimizes the network's structure and association strength based on the command's semantic context information. The semantic network update data reflects the impact of the command on the semantic context network, enabling the network to better adapt to the current command and contextual semantics, improving the system's intelligence and adaptability. Generating a command credibility score based on the semantic network update data can assess the rationality and reliability of the command and determine whether the command matches the current semantic context network. This credibility score allows the system to better determine the accuracy and feasibility of user commands, improving its understanding and response capabilities to user commands.

[0097] Preferably, step S36 includes the following steps:

[0098] Step S361: performing semantic conflict or inconsistency detection on the semantic network update data, thereby obtaining a semantic consistency report;

[0099] The embodiment of the present invention performs semantic conflict or inconsistency detection on the semantic network update data, thereby obtaining a semantic consistency report. This can be achieved using natural language processing technology and semantic similarity calculation methods. First, the semantic similarity of the nodes and edges in the semantic network update data is calculated to determine whether there is a conflict or inconsistency between them. For example, the cosine similarity between the node feature vectors can be calculated or a text similarity algorithm (such as TF-IDF, cosine similarity, edit distance, etc.) can be used to compare the semantic similarity between the nodes. Based on the similarity threshold, it is determined whether there is a semantic conflict or inconsistency, and a corresponding semantic consistency report is generated.

[0100] Step S362: updating the semantic context network according to the semantic network update data, and handling semantic conflicts according to the semantic consistency report, thereby obtaining a semantic update network;

[0101] The embodiment of the present invention performs a network update on the semantic context network based on the semantic network update data, and performs semantic conflict processing based on the semantic consistency report, thereby obtaining a semantic update network. A graph operation library (such as NetworkX) can be used to implement network updates and processing. According to the changes in the nodes and edges in the semantic network update data, the semantic context network is subjected to corresponding node and edge addition, deletion or modification operations. At the same time, according to the conflict information in the semantic consistency report, conflict processing is performed, such as node attribute adjustment, edge weight update, etc., to ensure the consistency of the semantic update network.

[0102] Step S363: performing context matching evaluation on the semantic update network to obtain a context matching score;

[0103] Embodiments of the present invention perform a contextual match evaluation on the semantic update network, thereby obtaining a contextual match score. A similarity calculation method can be used to evaluate the match between the semantic update network and the instruction context. For example, the similarity between nodes and edges in the semantic update network and keywords and intents in the instruction can be calculated. Based on the similarity calculation results, a contextual match score is generated, reflecting the degree of match between the semantic update network and the instruction context.

[0104] Step S364: Evaluate the degree of network change of the semantic update network and the semantic context network to obtain a network change degree score, where the network change includes changes in network node weights, changes in network connection strengths, and newly added or deleted nodes and links;

[0105] Embodiments of the present invention assess the degree of network change in the semantic update network and the semantic context network, thereby obtaining a network change score. The degree of network change can be assessed by considering factors such as changes in node weights, changes in network connection strength, and the number of newly added or deleted nodes and links. For example, the magnitude of change in node weights, the magnitude of change in edge strength, and the number of newly added or deleted nodes and links can be calculated. Based on the results of the change assessment, a network change score is generated to reflect the overall changes in the semantic update network and the semantic context network.

[0106] Step S365: Use the sigmoid function to assign weights to the context matching score and the network change degree score to generate an instruction credibility score.

[0107] In an embodiment of the present invention, a sigmoid function is used to weight the context match score and the network variability score to generate an instruction credibility score. The sigmoid function can be used to map the context match score and the network variability score to a range of [0, 1], and the weighting factors can be adjusted based on specific needs to obtain the final instruction credibility score. For example, the importance of the context match and network variability in the score can be adjusted based on actual application scenarios and user needs, thereby generating an instruction credibility score that takes into account trade-offs.

[0108] The present invention can timely discover conflicts or inconsistencies between the updated data and the existing semantic context network by performing semantic conflict or inconsistency detection on the semantic network update data; the semantic consistency report provides information about the differences and conflicts between the updated data and the semantic context network, providing guidance for subsequent network updates and conflict handling. Updating the semantic context network based on the semantic network update data can integrate new semantic information into the network, enabling the network to better reflect the current semantic state; handling semantic conflicts based on the semantic consistency report can resolve conflicts between the updated data and the existing network, ensuring the consistency and accuracy of the network. Context matching evaluation of the semantic update network can measure the degree of match between the updated network and the current context, and evaluate the network's adaptability and accuracy to the current context; the context matching score provides a quantitative indicator for judging the degree of adaptability of the updated network to the current context, thereby affecting the credibility score of the instruction. Network change evaluation of the semantic update network and the semantic context network can quantify the degree of change in the network, including changes in node weights, changes in connection strength, and changes in structure; the network change score provides an indicator for evaluating network changes, thereby affecting the credibility score of the instruction. Using a sigmoid function to weight the context match score and the network change score comprehensively considers the network's adaptability and variability, generating a comprehensive instruction credibility score. This score reflects the impact of the network's match and variability on instruction parsing, enabling the system to more accurately assess the credibility and feasibility of instructions. In summary, the above steps include detecting semantic conflicts or inconsistencies, updating the semantic context network, resolving semantic conflicts, and evaluating context match and network change. These effects help maintain the consistency and accuracy of the semantic network, assess the network's adaptability to the current context, and generate instruction credibility scores based on both network match and variability, thereby improving the system's intelligence and instruction understanding capabilities. These steps work together to provide an effective mechanism and evaluation metrics for updating the semantic network and parsing instructions.

[0109] Preferably, step S4 includes the following steps:

[0110] Step S41: Analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data;

[0111] The embodiments of the present invention use sensors or devices (such as cameras, microphones, touch screens, etc.) to capture multimodal interactive behavior data of users. For example, images and video data are acquired through a camera, audio data is acquired through a microphone, touch input data is acquired through a touch screen, and the like. Computer vision and audio processing technologies are used to pre-process and analyze the captured data. For example, image features are extracted using image processing algorithms, and audio is converted into text using speech recognition algorithms. Combined with analysis algorithms and machine learning technologies, feature extraction, behavior recognition, and classification are performed on multimodal data. For example, deep learning models (such as convolutional neural networks, recurrent neural networks, etc.) are used to process and analyze image and audio data to identify user interactive behaviors.

[0112] Step S42: Obtain historical misoperation data and normal operation data, and perform preset GAN model training based on the historical misoperation data and normal operation data to obtain a misoperation adversarial model; use the misoperation adversarial model to simulate and generate misoperation scenarios as an interactive verification test set;

[0113] The embodiment of the present invention collects historical misoperation data and normal operation data. The historical misoperation data can come from the user's misoperation records or misoperation data in the test scenario, while the normal operation data can come from the user's normal operation records or standard operation procedures. The historical misoperation data and normal operation data are trained using a preset GAN model. The GAN model consists of a generator and a discriminator, wherein the generator is used to generate misoperation scenarios, and the discriminator is used to distinguish between the generated misoperation scenarios and the real misoperation scenarios. By training the GAN model, the generator is able to generate false misoperation scenarios that are similar to the real misoperation scenarios, while the discriminator learns to distinguish between false misoperation scenarios and real misoperation scenarios; the misoperation adversarial model is used to generate misoperation scenarios as an interactive verification test set for subsequent verification of operation intentions.

[0114] Step S43: performing operation intention verification based on the multimodal behavior data and the interactive verification test set to obtain intention verification result data;

[0115] The embodiment of the present invention uses natural language processing technology to perform semantic analysis and operation intention extraction on multimodal behavior data. For example, voice recognition and natural language understanding technology are used to convert audio data into text, and the user's operation intention or instruction is extracted. The misoperation scenarios in the interactive verification test set are compared and analyzed with the multimodal behavior data to identify the characteristics and patterns of the misoperation. Machine learning or deep learning models are used to classify and verify the operation intention. Supervised learning algorithms (such as support vector machines, random forests, etc.) or sequence models (such as recurrent neural networks, Transformers, etc.) can be used to train the intention verification model; based on the results of the verification model, intention verification result data is generated to indicate the verification accuracy and credibility of each operation intention.

[0116] Step S44: Generate intent based on instruction credibility score and intent verification result data. Figure 1 Consistency indicator data.

[0117] The embodiment of the present invention comprehensively considers the instruction credibility score and the intention verification result data to define the intention Figure 1 The calculation method of consistency index can be based on specific needs and weight settings, by weighting the instruction credibility score and the intention verification result data, or by using other appropriate methods. Figure 1 The consistency index calculation method evaluates each operation intention and generates the corresponding intention Figure 1 Consistency indicator data. Figure 1 The consistency indicator data can be visualized or further analyzed so that users or systems can make decisions or optimize based on these indicators.

[0118] The present invention can obtain the user's multimodal behavior data such as voice, gestures, and expressions during the interaction process by analyzing the user's multimodal interaction behavior; the multimodal behavior data provides information about the different methods and signals used by the user in the interaction, which is helpful for the subsequent operation intention verification and intention Figure 1Provide basic data for consistency evaluation. Obtaining historical misoperation data and normal operation data can provide sample data of users' misoperation situations and normal operation behaviors in past interactions. By using the preset GAN model to train historical misoperation data and normal operation data, an misoperation adversarial model can be generated; using the misoperation adversarial model to simulate and generate misoperation scenarios as interactive verification test sets, the system's ability to identify and handle misoperations can be evaluated, thereby improving the system's robustness and user experience. Verifying operation intentions based on multimodal behavioral data can analyze user behaviors and interaction patterns, and infer user operation intentions. By using interactive verification test sets, the system's ability to judge and correct misoperations in different scenarios can be verified, thereby improving the system's reliability and user experience; the intention verification result data provides an evaluation of the accuracy of the system's operation intentions, providing information for subsequent intention verification. Figure 1 According to the instruction credibility score, the credibility of the semantic network and the user's operation intention verification results can be comprehensively considered to generate a comprehensive instruction credibility score; Figure 1 The consistency index data combines the instruction credibility score and the operation intention verification results to evaluate the consistency and accuracy of the system operation intention; Figure 1 The consistency index data provides a quantitative indicator for measuring the system's performance in command parsing and operation intention recognition, further improving the system's intelligence and user satisfaction. In summary, the effects of the above steps include obtaining multimodal behavior data, training the misoperation adversarial model, evaluating the system's ability to recognize and handle misoperations, verifying operation intentions, and generating intentions. Figure 1 Consistency indicator data; these effects help improve the robustness, reliability and user experience of the system, ensure that the system can accurately understand the user's operating intentions, and generate consistent command parsing results; by comprehensively considering the command credibility score and intent verification results, it can provide more accurate evaluation indicators and decision-making basis for the system's intelligent interaction and command understanding.

[0119] Preferably, step S41 includes the following steps:

[0120] Step S411: collecting information on voice, gesture, and touch interaction modes based on the user's interaction behavior with the smart furniture, thereby obtaining original multimodal interaction data;

[0121] The embodiments of the present invention use appropriate sensors and devices (such as microphones, cameras, touch screens, etc.) to capture multimodal interaction behavior information between users and smart furniture; for voice interaction, a microphone is used to record the user's voice input and obtain voice signals; for gesture interaction, a camera or depth sensor is used to capture the user's gesture movements and record gesture information such as joint angles and motion trajectories; for touch interaction, a touch screen is used to record the user's touch behavior information, including contact point coordinates, pressure values, and timestamps.

[0122] Step S412: Filter the voice modal data in the original multimodal interaction data, perform endpoint detection, remove the silent part, and enhance the high-frequency part to obtain a voice signal; calculate the Mel-frequency cepstral coefficients of the voice signal, and extract the fundamental frequency contour and energy envelope to obtain a voice feature vector; perform semantic recognition on the voice feature vector, and map the recognition result to a predefined command set to obtain voice command data;

[0123] This embodiment of the present invention filters the speech signal to remove noise and irrelevant frequency components, and performs endpoint detection on the signal to eliminate silent portions. Feature extraction is performed on the filtered speech signal. Common feature extraction methods include Mel-Frequency Cepstral Coefficient (MFCC) calculation, fundamental frequency contour extraction, and energy envelope extraction. The extracted speech feature vector is input into a semantic recognition model, which performs speech recognition and converts the speech into text. The recognition results are mapped to a predefined command set to obtain voice command data, representing the user's voice interaction intent.

[0124] Step S413: extracting the joint angle, motion trajectory, and posture description based on the directional gradient histogram from the gesture modal data in the original multimodal interaction data to obtain a gesture feature vector; performing temporal change analysis on the gesture feature vector and performing , thereby obtaining a gesture command sequence;

[0125] The embodiments of the present invention calculate the joint angles of gesture modal data and extract the joint angle information in the gesture action; use the motion trajectory analysis method to calculate the motion trajectory characteristics of the gesture action, such as speed and acceleration; use the posture description method based on the histogram of oriented gradients (HOG) to extract the posture characteristics of the gesture action; perform time series change analysis on the extracted gesture feature vectors, identify the gesture command sequence, and indicate the user's gesture interaction intention.

[0126] Step S414: performing touch point coordinates, pressure values, and timestamp analysis on the touch modal data in the original multimodal interaction data to obtain preliminary touch feature data; performing touch trajectory, pressure change, and multi-touch feature analysis based on the timestamp on the preliminary touch feature data to obtain a touch feature vector; performing touch gesture type recognition on the touch feature vector to obtain touch command data;

[0127] An embodiment of the present invention extracts preliminary feature data such as contact coordinates, pressure values, and timestamps from touch modal data; performs touch trajectory analysis based on timestamps to calculate trajectory features of touch actions, such as moving direction and speed; analyzes pressure change features in touch actions, such as duration and peak pressure; performs multi-touch feature analysis, such as multi-touch and number of fingers, and uses appropriate algorithms or models to identify touch gesture types for the extracted touch feature vectors, mapping touch behaviors to corresponding touch command data.

[0128] Step S415: Merge the touch command data, gesture command sequence, and voice command data into multimodal behavior data.

[0129] The embodiment of the present invention integrates the touch command data, gesture command sequence and voice command data obtained in the previous steps, and assigns weights to the data of different modalities as needed to determine the influence of different modalities in the multimodal behavior data; and arranges the data of different modalities in a certain order or time series to form a sequence of multimodal behavior data.

[0130] By collecting the user's interaction behavior with smart furniture, the present invention can obtain the original data of the user in various interaction modes such as voice, gesture and touch. The original multimodal interaction data provides the user's behavior pattern and signal information in different interaction modes, providing basic data for subsequent feature extraction and command recognition. Preprocessing operations such as filtering, endpoint detection and high-frequency enhancement on the voice signal help to extract effective voice information and remove noise; by calculating voice features such as Mel-frequency cepstral coefficients, fundamental frequency contour and energy envelope, the voice signal can be converted into a voice feature vector with semantic information; by performing semantic recognition and command mapping on the voice feature vector, the voice interaction can be converted into specific voice command data for subsequent intent verification and command parsing. Extracting features such as joint angles, motion trajectories and posture descriptions based on directional gradient histograms from gesture modal data helps to capture the spatial and dynamic information of gestures; by performing temporal change analysis on the gesture feature vector, the temporal evolution characteristics of the gesture can be obtained, thereby obtaining a gesture command sequence; the gesture command sequence provides a description of the user's gesture interaction behavior for subsequent intent verification and command parsing. Extracting preliminary features such as touch point coordinates, pressure values, and timestamps from touch modal data helps describe the user's basic actions during touch interaction. Timestamp-based touch trajectory, pressure changes, and multi-touch feature analysis can extract richer touch feature information. By identifying the touch gesture type of the touch feature vector, touch interactions can be converted into specific touch command data. This touch command data provides a description of the user's touch interaction behavior, which is used for subsequent intent verification and command parsing. Combining touch command data, gesture command sequences, and voice command data into multimodal behavior data comprehensively considers the user's behavioral instructions under different interaction modes. This multimodal behavior data provides a more comprehensive and accurate expression of user intent and can be used for the integrated control and response of smart furniture.

[0131] The present invention further provides a wireless control system for smart furniture, which is used to execute the above-mentioned wireless control method for smart furniture. The wireless control system for smart furniture includes:

[0132] The channel stability assessment module is used to capture different types of radio signals through user terminals and extract the time-frequency characteristics of each signal using fast Fourier transform to obtain multi-signal time-frequency feature data. The module also performs weighted fusion of the multi-signal time-frequency feature data and dynamically adjusts the weights of each signal based on transfer learning to obtain a channel stability index.

[0133] The dynamic protocol selection module is used to dynamically select wireless communication protocols based on the channel stability index using SDN technology, and record the protocol switching process on the private blockchain to obtain protocol switching log data. The protocol switching log data is aggregated using a federated learning framework based on switching feedback to obtain protocol effectiveness matrix data.

[0134] The semantic context processing module is used to construct a semantic context network based on pre-acquired environmental sensor data and protocol effectiveness matrix data; obtain user command data, parse the user command data, extract the command context semantics, and semantically update the semantic context network to obtain semantic network update data; and generate a command credibility score based on the semantic network update data;

[0135] The multimodal intention verification module is used to analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; use GAN simulation to generate misoperation scenarios as interaction verification test sets, and verify the operation intention based on the multimodal behavior data and interaction verification test sets to obtain intention verification result data; generate intention verification results based on the instruction credibility score and intention verification result data. Figure 1 consistency indicator data;

[0136] Intelligent collaborative decision-making module for Figure 1 The consistency index data is used to conduct command conflict arbitration based on fog computing, and the multi-device collaborative logical relationship analysis is performed to obtain collaborative execution decision data.

[0137] The channel stability assessment module of the present invention provides an assessment of wireless communication channel quality. It captures different types of radio signals and extracts their time-frequency features to obtain multi-signal time-frequency feature data. This multi-signal time-frequency feature data is weighted and fused, and the signal weights are dynamically adjusted to obtain a channel stability index. This assessment module improves the stability and reliability of wireless communication, enabling the system to select the optimal communication method based on real-time channel conditions to ensure good communication quality. The dynamic protocol selection module utilizes software-defined networking (SDN) technology to dynamically select wireless communication protocols based on the channel stability index. By recording protocol switching log data and aggregating it using a federated learning framework, protocol performance matrix data is obtained. This module improves the accuracy and efficiency of communication protocol selection, enabling the system to select the most suitable communication protocol based on real-time channel conditions, thereby providing better communication performance and user experience. The semantic context processing module utilizes environmental sensor data and protocol performance matrix data to construct a semantic context network, parse user command data, extract command context semantics, and update the semantic context network. The updated data from the semantic network generates a command credibility score. The effect of this module is to improve the accuracy of understanding user instructions and the ability to perceive context. The system can better understand the user's intentions and make decisions and executions based on the semantic context to provide more accurate and personalized services. The multimodal intention verification module obtains multimodal behavior data by analyzing the user's multimodal interaction behavior. Generate misoperation scenarios using generative adversarial networks (GANs) as interaction verification test sets. Verify the operation intention based on the multimodal behavior data and the interaction verification test set to obtain the intention verification result data. Generate intentions based on the instruction credibility score and the intention verification result data. Figure 1 The module improves the accuracy and credibility of user intention verification. The system can better understand the user's intention and accurately perform corresponding operations, thus providing a more satisfactory user experience. Figure 1 This module uses consistency indicator data to conduct fog computing-based command conflict resolution and analyze the logical relationships of multi-device collaboration to generate collaborative execution decision data. This module effectively resolves conflicts that may arise when multiple devices simultaneously execute commands, ensuring consistent and coordinated command execution. Furthermore, multi-device collaborative logical relationship analysis helps determine the collaboration method and execution order between devices, improving the efficiency and effectiveness of multi-device collaborative work.

[0138] Preferably, the present invention further provides a smart furniture, comprising:

[0139] memory for storing computer programs;

[0140] A processor is used to implement the steps of any of the above-mentioned wireless control methods for smart furniture when executing the computer program.

[0141] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0142] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A wireless control method for smart furniture, characterized in that: The following steps are involved: Step S1: Capturing different types of radio signals through a user terminal, and extracting the time-frequency characteristics of each signal using a fast Fourier transform to obtain multi-signal time-frequency characteristic data; The multi-signal time-frequency feature data is weightedly fused, and the weights of each signal are dynamically adjusted based on transfer learning to obtain the channel stability index; Step S2: Using SDN technology to dynamically select wireless communication protocols based on the channel stability index, and recording the protocol switching process on a private blockchain to obtain protocol switching log data; performing aggregation based on the switching feedback federated learning framework based on the protocol switching log data to obtain protocol effectiveness matrix data; Step S3: Constructing a semantic context network based on pre-acquired environmental sensor data and protocol effectiveness matrix data; obtaining user instruction data, parsing the user instruction data, extracting instruction context semantics, and semantically updating the semantic context network to obtain semantic network update data; generating an instruction credibility score based on the semantic network update data; Step S4: Analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; use GAN simulation to generate misoperation scenarios as interaction verification test sets, and verify the operation intention based on the multimodal behavior data and the interaction verification test sets to obtain intention verification result data; generate intention consistency index data based on the instruction credibility score and the intention verification result data; Step S5: Perform fog computing-based instruction conflict arbitration on the intent consistency index data, and perform multi-device collaborative logical relationship analysis to obtain collaborative execution decision data.

2. The wireless control method of smart furniture according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: collecting multiple radio signals from the surrounding environment through the radio receiving module of the user terminal to obtain original radio signal data; denoising the original radio signal data and performing channel equalization to obtain multi-source radio signal data; Step S12: Performing a fast Fourier transform on the multi-source radio signal data and extracting the time domain and frequency domain features of each signal to obtain multi-signal time-frequency feature data, wherein the time-frequency features include power spectral density, symbol delay spread, signal envelope, pulse width, and spectral moment; Step S13: performing weighted fusion on the multi-signal time-frequency feature data to generate fused signal feature data; Step S14: collecting environmental state vector data around the terminal, and using transfer learning technology to dynamically adjust the signal weights in the fused signal feature data according to the environmental state vector data, thereby obtaining signal feature weight coefficients; Step S15: weight correction is performed on the fused signal feature data according to the signal feature weight coefficient, and multiple quality index analyses of the signal are performed to generate a channel stability index, where the multiple quality indicators include signal-to-interference-plus-noise ratio, multipath spread, phase error, and carrier-to-noise ratio.

3. The wireless control method of smart furniture according to claim 2, characterized in that: Step S13 includes the following steps: Step S131: normalizing the time-frequency feature data of multiple signals and removing outliers to obtain standard time-frequency feature data; Step S132: performing key feature screening on each feature in the time-frequency feature standard data based on the distinguishing capability of feature indicators in wireless control, thereby generating simplified time-frequency feature data; Step S133: Initial weights are assigned to the simplified time-frequency feature data through a gradient boosting tree, and weighted fusion is performed to obtain fused signal feature data.

4. The wireless control method of smart furniture according to claim 3, characterized in that: Step S2 includes the following steps: Step S21: Initialize the SDN controller according to the channel stability index and load the available wireless communication protocol to obtain the wireless communication protocol library and SDN controller status data; Step S22: Perform protocol matching on the wireless communication protocol library based on the channel stability index, dynamically adjust the terminal's MAC / PHY layer protocol stack using the SDN controller status data, and record the new and old protocol switching details and context on the private blockchain to obtain protocol switching log data; Step S23: performing performance index analysis of throughput, delay, and power consumption of different protocols under various channel conditions based on the protocol switching log data, thereby obtaining protocol performance index data; Step S24: Utilize the federated learning framework to aggregate the protocol performance indicator data of each node to obtain protocol effectiveness matrix data.

5. The wireless control method of smart furniture according to claim 4, characterized in that: Step S3 includes the following steps: Step S31: Initially constructing a semantic context network based on a graph neural network based on the pre-acquired environmental sensor data and the protocol effectiveness matrix data, where nodes represent environmental factors and protocol effectiveness indicators, and edges represent relationships between nodes, thereby obtaining a semantic context network; Step S32: Obtain user instruction data, and use natural language processing technology to perform keyword and intent recognition on the user instruction data, thereby obtaining user instruction parsed data; Step S33: performing semantic analysis on the user instruction parsed data to obtain contextual semantics including time, location, and object information, thereby obtaining instruction contextual semantic data; Step S34: Mapping the instruction context semantic data to relevant nodes and edges in the semantic context network, thereby obtaining semantic mapping result data; Step S35: dynamically adjusting the node weights and edge strengths of the semantic context network according to the semantic mapping result data, thereby obtaining updated semantic network data; Step S36: Generate an instruction credibility score based on the semantic network update data.

6. The wireless control method of smart furniture according to claim 5, characterized in that: Step S36 includes the following steps: Step S361: performing semantic conflict or inconsistency detection on the semantic network update data, thereby obtaining a semantic consistency report; Step S362: updating the semantic context network according to the semantic network update data, and handling semantic conflicts according to the semantic consistency report, thereby obtaining a semantic update network; Step S363: performing context matching evaluation on the semantic update network to obtain a context matching score; Step S364: Evaluate the degree of network change of the semantic update network and the semantic context network to obtain a network change degree score, where the network change includes changes in network node weights, changes in network connection strengths, and newly added or deleted nodes and links; Step S365: Use the sigmoid function to assign weights to the context matching score and the network change degree score to generate an instruction credibility score.

7. The wireless control method of smart furniture according to claim 6, characterized in that: Step S4 includes the following steps: Step S41: Analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; Step S42: Obtain historical misoperation data and normal operation data, and perform preset GAN model training based on the historical misoperation data and normal operation data to obtain a misoperation adversarial model; use the misoperation adversarial model to simulate and generate misoperation scenarios as an interactive verification test set; Step S43: performing operation intention verification based on the multimodal behavior data and the interactive verification test set to obtain intention verification result data; Step S44: Generate intention consistency index data based on the instruction credibility score and intention verification result data.

8. The wireless control method of smart furniture according to claim 7, characterized in that: Step S41 includes the following steps: Step S411: collecting information on voice, gesture, and touch interaction modes based on the user's interaction behavior with the smart furniture, thereby obtaining original multimodal interaction data; Step S412: Filter the voice modal data in the original multimodal interaction data, perform endpoint detection, remove the silent part, and enhance the high-frequency part to obtain a voice signal; calculate the Mel-frequency cepstral coefficients of the voice signal, and extract the fundamental frequency contour and energy envelope to obtain a voice feature vector; perform semantic recognition on the voice feature vector, and map the recognition result to a predefined command set to obtain voice command data; Step S413: extracting the joint angle, motion trajectory, and posture description based on the directional gradient histogram from the gesture modal data in the original multimodal interaction data to obtain a gesture feature vector; performing temporal change analysis on the gesture feature vector and performing , thereby obtaining a gesture command sequence; Step S414: performing touch point coordinates, pressure values, and timestamp analysis on the touch modal data in the original multimodal interaction data to obtain preliminary touch feature data; performing touch trajectory, pressure change, and multi-touch feature analysis based on the timestamp on the preliminary touch feature data to obtain a touch feature vector; performing touch gesture type recognition on the touch feature vector to obtain touch command data; Step S415: Merge the touch command data, gesture command sequence, and voice command data into multimodal behavior data.

9. A wireless control system for smart furniture, characterized in that: The method for wirelessly controlling smart furniture according to claim 1 is configured to: The channel stability assessment module is used to capture different types of radio signals through user terminals and extract the time-frequency characteristics of each signal using fast Fourier transform to obtain multi-signal time-frequency feature data. The module also performs weighted fusion of the multi-signal time-frequency feature data and dynamically adjusts the weights of each signal based on transfer learning to obtain a channel stability index. The dynamic protocol selection module is used to dynamically select wireless communication protocols based on the channel stability index using SDN technology, and record the protocol switching process on the private blockchain to obtain protocol switching log data. The protocol switching log data is aggregated using a federated learning framework based on switching feedback to obtain protocol effectiveness matrix data. The semantic context processing module is used to construct a semantic context network based on pre-acquired environmental sensor data and protocol effectiveness matrix data; obtain user command data, parse the user command data, extract the command context semantics, and semantically update the semantic context network to obtain semantic network update data; and generate a command credibility score based on the semantic network update data; The multimodal intention verification module is used to analyze the user's multimodal interaction behavior through human-computer interaction technology to obtain multimodal behavior data; use GAN simulation to generate misoperation scenarios as interaction verification test sets, and verify the operation intention based on the multimodal behavior data and interaction verification test sets to obtain intention verification result data; generate intention consistency index data based on the instruction credibility score and intention verification result data; The intelligent collaborative decision-making module is used to perform fog computing-based instruction conflict arbitration on the intent consistency index data, and conduct multi-device collaborative logical relationship analysis to obtain collaborative execution decision data.

10. A smart furniture, characterized in that, include: memory for storing computer programs; A processor, configured to implement the wireless control method for smart furniture as described in any one of claims 1 to 8 when executing the computer program.

Images