Intelligent clothing wearing method and system based on wireless firmware upgrade

Abstract

The application relates to the technical field of intelligent wearing equipment, and discloses an intelligent clothing wearing method and system based on wireless firmware upgrading, the system comprises a multi-modal control data sensing and fusion module, a dynamic scene recognition and adaptation module, an intelligent strategy matching and decision module, a strategy accurate execution and feedback module and a remote firmware upgrading and self-learning optimization module; a multi-modal sensing unit integrated in a detachable functional component of clothing synchronously collects voice instruction data and entity key trigger signals of a wearer, and performs instruction type recognition and semantic analysis processing on an original control data set, thereby generating standardized control intention data flow, realizing synchronous collection and fusion analysis of voice and key multi-modal control data, and avoiding instruction misrecognition and response delay caused by a single control mode.

Description

Technical Field

[0001] This invention relates to the field of smart wearable device technology, specifically to a smart clothing wearing method and system based on wireless firmware upgrades. Background Technology

[0002] Wearable devices refer to devices that use wearable technology to intelligently configure people's daily clothing, embedding various sensors, identification, connection and cloud services into items such as glasses, watches, and clothing. They are characterized by small size, light weight and high level of informatization, covering multiple fields such as industrial manufacturing, medical health, and information entertainment.

[0003] Currently, the control methods in smart clothing systems are relatively simple, typically supporting only a single voice command or button trigger. This makes it impossible for multimodal sensing units to simultaneously collect and fuse voice and button commands. If the wearer is in a noisy environment or in motion, the single control method is prone to misrecognition of commands or response delays, which may result in a low accuracy rate of control intent recognition and fail to guarantee the stability of smart clothing control and user experience in different scenarios.

[0004] Therefore, a smart clothing wearable method and system based on wireless firmware upgrades is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a smart clothing wearing method and system based on wireless firmware upgrades, which solves the problems mentioned in the background technology, such as low accuracy of control intention recognition and inability to guarantee the control stability and user experience of smart clothing in different scenarios.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method and system for wearing smart clothing based on wireless firmware upgrades, the method comprising the following steps: S1. Through the multimodal sensing unit integrated into the detachable functional components of the clothing, the wearer's voice command data and physical button trigger signals are collected simultaneously, and a raw control data set is generated. S2. Perform instruction type identification and semantic parsing processing on the original control data set to generate a standardized control intention data stream; S3. Based on the standardized control intention data stream, the current wearable scene and environmental noise level are identified in real time through the dynamic scene recognition engine, and dynamic scene adaptation parameters are generated. S4. Based on the dynamic scene adaptation parameters, call the algorithm matching engine to intelligently match the optimal control response strategy algorithm model from the pre-built dual-mode control strategy algorithm library and generate the target control strategy algorithm identifier. S5. Based on the target control strategy algorithm identifier, call the corresponding strategy execution logic to dynamically control the functional execution unit of the clothing, and generate strategy execution logs and control effect feedback data. S6. Based on the strategy execution log, control effect feedback data and historical data, remote firmware iteration optimization and self-learning update of the dual-mode control strategy algorithm library and dynamic scene recognition engine are performed through wireless communication link.

[0007] Preferably, generating the raw control data set in S1 includes the following steps: S11. By deploying a miniature microphone array and intelligent noise reduction unit within the detachable functional components, audio data of the wearer's voice wake-up words and function commands are collected, and a subset of voice control data is generated. S12. By deploying waterproof physical button units on the surface of detachable functional components, the wearer's touch signals such as pressing, long pressing, and double-clicking are collected, and a subset of button operation data is generated. S13. The subset of voice control data and the subset of button control data are combined to form the original control data set.

[0008] Preferably, the instruction type identification and semantic parsing processing in S2 includes the following steps: S21. Receive the original control data set, perform endpoint detection and feature extraction on the voice control data subset, and convert the electrical signal of the key control data subset into a standardized digital pulse sequence; S22. Based on the preset voice command recognition model, decode and semantically map the extracted voice features to generate the corresponding textual control intention. At the same time, according to the preset key function mapping table, the digital pulse sequence is parsed into the corresponding function request; S23. The textual control intentions and function requests are integrated into a standardized control intention data stream according to the time sequence of event triggering.

[0009] Preferably, the dynamic scene recognition and adaptation parameter generation in step S3 includes the following steps: S31. Obtain the standardized control intention data stream and input it into the dynamic scene recognition engine. The engine has a built-in noise classifier and activity recognizer based on deep neural networks, which is suitable for extracting environmental noise features and wearer motion state features from the data stream. S32. Compare the extracted environmental noise features with the preset noise threshold database to determine whether the current environment is quiet, normal, or noisy. At the same time, the motion state characteristics are matched with the preset activity model to determine whether the wearer is in a stationary state, a walking state, or a state of vigorous exercise. S33. Based on the determined environmental noise level and the wearer's movement state, generate dynamic scene adaptation parameters including the preferred control mode, voice wake-up sensitivity, and button response threshold.

[0010] Preferably, generating the target manipulation strategy algorithm identifier in step S4 includes the following steps: S41. Construct a dual-mode control strategy algorithm library, which stores various preset control response strategy algorithm models in matrix form, including pure voice priority response model, pure button priority response model, hybrid intelligent arbitration response model and emergency mode response model. S42. Extract the key feature vector from the dynamic scene adaptation parameters. The key feature vector includes environmental noise level, motion state type, remaining battery level and user historical preferences. S43. Perform multi-dimensional similarity calculation between the key feature vector and the standard applicable feature vector of each algorithm model in the dual-mode control strategy algorithm library; S44. Based on the similarity calculation results, the optimal control response strategy algorithm model is matched through the algorithm matching engine, and its unique encoding is output as the target control strategy algorithm identifier.

[0011] Preferably, the generation of strategy execution logs and control effect feedback data in step S5 includes the following steps: S51. Based on the target manipulation strategy algorithm identifier, call the predefined strategy execution logic corresponding to the identifier; S52. The predefined strategy execution logic accesses the standardized control intent data stream and the dynamic scene adaptation parameters to generate a detailed instruction set containing specific response actions, execution objects, action parameters and execution timing. S53. The detailed instruction set is sent to the execution units of the clothing, such as the sound and light prompt module, voice broadcast module, vibration feedback module, and communication module. S54. Synchronously monitor the execution process of instructions, record the execution status, execution results and system feedback, and form a policy execution log; S55. Within a preset period after the strategy is executed, collect the user's subsequent operations and feedback signals, evaluate the satisfaction with the control response, compare it with the expected effect, and generate control effect feedback data.

[0012] Preferably, the remote firmware iterative optimization and self-learning update in S6 includes the following steps: S61. Perform correlation analysis between the strategy execution log and the control effect feedback data to calculate the effectiveness score and efficiency score of this strategy execution; S62. Store the complete data package of this response, including the dynamic scene adaptation parameters, the selected target control strategy algorithm identifier, the execution result and the effectiveness score, as a new case in the historical case library. S63. Based on the historical case library, the differential upgrade package sent from the cloud is received periodically through the wireless communication link to incrementally train and optimize the parameters of the noise classifier and activity recognizer in the dynamic scene recognition engine. S64. Based on the historical case library, the differential upgrade package sent from the cloud is received periodically through the wireless communication link to optimize and update the standard applicable feature vectors and matching weights of each algorithm model in the dual-mode control strategy algorithm library.

[0013] Preferably, the system includes: The multimodal control data perception and fusion module gathers the wearer's control signals through the voice acquisition unit and the button signal acquisition unit, performs standardized processing using the command recognition unit and the semantic parsing unit, and outputs a standardized control intention data stream. The dynamic scene recognition and adaptation module receives the standardized control intention data stream, performs scene analysis through the environmental noise identification unit and the motion state recognition unit, calculates the dynamic scene adaptation parameters through the adaptation parameter generation unit, and outputs structured dynamic scene adaptation parameter data. The intelligent strategy matching and decision-making module receives the dynamic scene adaptation parameter data, parses key features through the feature vectorization unit, uses the algorithm matching engine to search and make decisions in the preset dual-mode control strategy algorithm library, and outputs the target control strategy algorithm identifier.

[0014] Preferably, the system further includes: The strategy precision execution and feedback module receives the target control strategy algorithm identifier, generates detailed control instructions through the strategy instruction compilation unit, controls the function execution unit of the clothing through the execution mechanism drive interface unit, and outputs strategy execution logs and control effect feedback data through the execution monitoring and effect evaluation unit. The remote firmware upgrade and self-learning optimization module establishes a data link with the cloud server through the wireless communication unit, receives the strategy execution log and operation effect feedback data, builds a historical case library through the case management and analysis unit, updates the system core model and strategy using the differential upgrade package receiving and burning unit, and outputs system firmware upgrade instructions.

[0015] Preferably, the remote firmware upgrade and self-learning optimization module also includes a firmware verification and security protection unit. The firmware verification and security protection unit is used to perform integrity verification and signature verification on the downloaded differential upgrade package before performing the firmware upgrade, and to trigger an automatic rollback mechanism to restore to the previous stable firmware version when the upgrade fails.

[0016] Compared with existing technologies, the present invention provides a method and system for wearing smart clothing based on wireless firmware upgrades, which has the following beneficial effects: 1. In this invention, a multimodal sensing unit integrated into the detachable functional components of clothing synchronously collects the wearer's voice command data and physical button trigger signals, and performs command type recognition and semantic parsing processing on the original control data set to generate a standardized control intention data stream. This realizes the synchronous collection and fusion parsing of voice and button multimodal control data, avoiding command misidentification and response delay caused by a single control method.

[0017] 2. In this invention, based on the standardized control intention data stream, the current wearable scene and environmental noise level are identified in real time through the dynamic scene recognition engine, dynamic scene adaptation parameters are generated, and the algorithm matching engine is called according to the parameters to intelligently match the optimal control response strategy algorithm model from the pre-built dual-mode control strategy algorithm library, so as to ensure the control stability and adaptability of smart clothing under different environmental noise levels and movement states.

[0018] 3. In this invention, based on strategy execution logs, control effect feedback data, and historical data, the dual-mode control strategy algorithm library and dynamic scene recognition engine are remotely iterated and self-learned through a wireless communication link to achieve continuous optimization of the system's core model and strategy, thereby improving the control response effect during long-term use. Attached Figure Description

[0019] Figure 1 This is a flowchart of a smart clothing wearing method based on wireless firmware upgrade according to the present invention; Figure 2 This is a schematic diagram of an intelligent clothing wearable system based on wireless firmware upgrades according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] For specific implementation examples, please refer to: Figures 1-2A method and system for wearing smart clothing based on wireless firmware upgrades, the method comprising the following steps: S1. Through the multimodal sensing unit integrated into the detachable functional components of the clothing, the wearer's voice command data and physical button trigger signals are collected simultaneously, and a raw control data set is generated. S2. Perform instruction type identification and semantic parsing on the original control data set to generate a standardized control intent data stream; S3. Based on standardized control intent data stream, the dynamic scene recognition engine identifies the current wearable scene and environmental noise level in real time and generates dynamic scene adaptation parameters. S4. Based on the dynamic scene adaptation parameters, call the algorithm matching engine to intelligently match the optimal control response strategy algorithm model from the pre-built dual-mode control strategy algorithm library and generate the target control strategy algorithm identifier. S5. Based on the target control strategy algorithm identifier, call the corresponding strategy execution logic to dynamically control the functional execution unit of the clothing, and generate strategy execution logs and control effect feedback data. S6. Based on strategy execution logs, control effect feedback data, and historical data, remote firmware iteration optimization and self-learning updates are performed on the dual-mode control strategy algorithm library and dynamic scene recognition engine via wireless communication links.

[0022] The initial control data set is generated in S1, including the following steps: S11. By deploying a miniature microphone array and intelligent noise reduction unit within the detachable functional components, audio data of the wearer's voice wake-up words and function commands are collected, and a subset of voice control data is generated. S12. By deploying waterproof physical button units on the surface of detachable functional components, the wearer's touch signals such as pressing, long pressing, and double-clicking are collected, and a subset of button operation data is generated. S13. Combine the voice control data subset and the button control data subset to form the original control data set.

[0023] Instruction type identification and semantic parsing in S2 includes the following steps: S21. Receive the original control data set, perform endpoint detection and feature extraction on the voice control data subset, and convert the electrical signal of the key control data subset into a standardized digital pulse sequence. After acquiring the mixed sensor signals, the speech signal is first subjected to endpoint detection using a dual-threshold method to determine the start and end times of the speech activity. Simultaneously, the time-domain waveform features of the key press operation data subset are extracted. Subsequently, a quantization coding algorithm is used to convert the analog electrical signal into a discrete digital pulse sequence; the specific conversion formula is as follows: ; in express The key input at any time simulates electrical signals. express Level quantization function, express The standardized digital pulse sequence value output at each moment; S22. Based on the preset voice command recognition model, decode and semantically map the extracted voice features to generate the corresponding textual control intention. At the same time, according to the preset key function mapping table, the digital pulse sequence is parsed into the corresponding function request; The extracted speech feature vectors are input into a pre-trained deep learning model. The decoder calculates the probability distribution of each preset speech command, selects the command with the highest probability as the recognition result, and performs semantic mapping. For digital pulse sequences, the system traverses a preset key function mapping table to find matching signal feature combinations, thereby parsing the function request that the user expects to execute. The determination formula is as follows: ; in This represents the maximum predicted probability. For instruction mapping tables and rule bases, The extracted feature sequences and data sequences, The preset confidence threshold, Given the input digital pulse sequence, A null value indicates that no valid control request was detected. Functions representing mapping relationships The final parsed function request; S23. Integrate the textual control intentions and function requests into a standardized control intention data stream according to the time sequence of event triggering.

[0024] The dynamic scene recognition and adaptation parameter generation in S3 includes the following steps: S31. Obtain standardized control intention data stream and input it into the dynamic scene recognition engine. The engine has a built-in noise classifier and activity recognizer based on deep neural networks, which is suitable for extracting environmental noise features and wearer motion state features from the data stream. The dynamic scene recognition engine receives raw data streams from multimodal sensors; the noise classifier inside the engine analyzes the spectral entropy characteristics of audio signals and classifies them into different environmental noise levels; at the same time, the activity recognizer extracts the frequency domain characteristics of accelerometer data to identify the wearer's current body motion state, and the outputs of both together constitute the current scene context information. S32. Compare the extracted environmental noise features with the preset noise threshold database to determine whether the current environment is quiet, normal, or noisy. Noise level determination logic: ; Value To determine the environmental noise level, For the extracted environmental noise feature vector, and The preset threshold boundaries between quiet and noisy environments. Quiet level, It is a general level. Noise level; At the same time, the motion state characteristics are matched with the preset activity model to determine whether the wearer is in a stationary state, a walking state, or a state of vigorous exercise. Motion state matching logic: ; in To determine the wearer's movement status, To select the parameter that maximizes the conditional probability , Given the characteristics of the motion state, to belong to the th The posterior probability of a motion state; S33. Based on the determined environmental noise level and the wearer's movement state, generate dynamic scene adaptation parameters including the preferred control mode, voice wake-up sensitivity, and button response threshold. For high-noise scenarios, the system automatically improves voice wake-up sensitivity and lowers the key response threshold; for scenarios involving vigorous movement, it optimizes key debounce parameters and adjusts the voice recognition window; and finally outputs a set of dynamic configuration parameters.

[0025] The generation of the target manipulation strategy algorithm identifier in S4 includes the following steps: S41. Construct a dual-mode control strategy algorithm library, which stores various preset control response strategy algorithm models in matrix form, including pure voice priority response model, pure button priority response model, hybrid intelligent arbitration response model and emergency mode response model. S42. Extract the key feature vectors from the dynamic scene adaptation parameters. The key feature vectors include the environmental noise level, motion state type, remaining battery level, and user historical preferences. To facilitate subsequent strategy matching, the system transforms the discrete parameters in the dynamic configuration parameter set into points in a high-dimensional feature vector space; it then eliminates the influence of dimensions through normalization and extracts key feature vectors for similarity calculation. ; in For feature mapping function, To adapt the parameter set for dynamic scenarios, For normalization function, The extracted key feature vectors; S43. Perform multi-dimensional similarity calculation between the key feature vector and the standard applicable feature vector of each algorithm model in the dual-mode control strategy algorithm library; The weighted Euclidean distance is used to measure the difference between the two; the smaller the distance, the higher the similarity. ; in The function for calculating the weighted Euclidean distance is... For the first Feature vectors of candidate strategies For the first Weights of each dimension The total number of feature dimensions. For the key feature vector in the th Values ​​in each dimension For the first The value of the feature vector of each candidate strategy in the j-th dimension; S44. Based on the similarity calculation results, the optimal control response strategy algorithm model is matched through the algorithm matching engine, and its unique encoding is output as the target control strategy algorithm identifier. The algorithm matching engine sorts the candidates based on the calculated similarity scores and selects the candidate strategy with the smallest distance as the optimal matching result; then it loads the algorithm model and its hyperparameter configuration corresponding to the strategy, ready to execute the subsequent control response logic; ; in This represents the optimal control response strategy algorithm model. This represents taking the parameter that minimizes the function value. ; The process of generating strategy execution logs and control effect feedback data in S5 includes the following steps: S51. Based on the target manipulation strategy algorithm identifier, call the predefined strategy execution logic corresponding to the identifier; After the strategy model is loaded, the predefined control response logic is executed according to the control parameters output by the model; this includes driving the vibration motor on the clothing to provide tactile feedback and sending control signals to external devices via Bluetooth module, thereby completing the closed loop of the user's requested function. S52, Predefined strategy execution logic accesses standardized control intent data stream and dynamic scenario adaptation parameters to generate a detailed instruction set containing specific response actions, execution objects, action parameters and execution timing; S53, Send detailed instruction sets to the execution units of the clothing, such as the sound and light prompt module, voice broadcast module, vibration feedback module, and communication module; S54. Synchronously monitor the execution process of instructions, record the execution status, execution results and system feedback, and form a policy execution log; S55. Within a preset period after the strategy is executed, collect the user's subsequent operations and feedback signals, evaluate the satisfaction with the control response, compare it with the expected effect, and generate control effect feedback data. ; in To control the satisfaction rating of the response, For activation function, This is the weight matrix. For bias terms, This is the input feature vector.

[0026] Remote firmware iteration optimization and self-learning update in S6 includes the following steps: S61. Correlate the strategy execution log with the control effect feedback data to calculate the effectiveness score and efficiency score of this strategy execution; Validity score: ; in The validity score is set and the value range is [0,1]. The execution time was [time]. This is the shortest processing time in history. This is the longest time ever recorded. Efficiency rating: ; The normalized calculations for each sub-indicator are as follows: ; ; in Rate efficiency This represents the weighting coefficient of the satisfaction index in the overall evaluation. This represents the weighting coefficient of energy consumption indicators in the comprehensive evaluation. The normalized satisfaction score. The normalized energy consumption score, The actual satisfaction rating for the implementation of this strategy. and These are the minimum and maximum historical satisfaction ratings. This represents the actual energy consumption for executing this strategy. and These are the historical minimum and maximum energy consumption values; S62. Store the complete data package for this response, including dynamic scenario adaptation parameters, selected target manipulation strategy algorithm identifier, execution results and effectiveness score, as a new case in the historical case library. S63. Based on the historical case library, the differential upgrade package sent from the cloud is received periodically through the wireless communication link to perform incremental training and optimization on the parameters of the noise classifier and activity recognizer in the dynamic scene recognition engine. Using the new feature data and feedback labels generated from this interaction, the original deep learning model is incrementally trained; the model weights are updated and the loss function is optimized through the backpropagation algorithm, thereby improving the model's recognition accuracy in similar scenarios in the future; ; in For the updated model parameters, These are the model parameters before the update. For learning rate, The gradient of the loss function. For loss function, For new data batches; S64. Based on the historical case library, the differential upgrade package sent from the cloud is received periodically through the wireless communication link to optimize and update the standard applicable feature vectors and matching weights of each algorithm model in the dual-mode control strategy algorithm library. After completing incremental training, the relevant parameters in the policy library are fine-tuned; the parameter update process is smoothed by the moving average method to avoid model oscillation, and the optimized model parameters and policy configuration are sent to the terminal device to complete the entire adaptive optimization closed loop. ; in These are the smoothed parameters. The smoothing coefficient and , These are the parameters from the previous iteration.

[0027] The system includes: The multimodal control data perception and fusion module gathers the wearer's control signals through the voice acquisition unit and the button signal acquisition unit, performs standardized processing using the command recognition unit and the semantic parsing unit, and outputs a standardized control intention data stream. The dynamic scene recognition and adaptation module receives a standardized control intention data stream, performs scene analysis through the environmental noise identification unit and the motion state recognition unit, calculates dynamic scene adaptation parameters through the adaptation parameter generation unit, and outputs structured dynamic scene adaptation parameter data. The intelligent strategy matching and decision-making module receives dynamic scene adaptation parameter data, parses key features through feature vectorization units, uses the algorithm matching engine to search and make decisions in the preset dual-mode control strategy algorithm library, and outputs the target control strategy algorithm identifier.

[0028] The system also includes: The strategy precision execution and feedback module receives the target control strategy algorithm identifier, generates detailed control instructions through the strategy instruction compilation unit, controls the function execution unit of the clothing through the execution mechanism drive interface unit, and outputs strategy execution logs and control effect feedback data through the execution monitoring and effect evaluation unit. The remote firmware upgrade and self-learning optimization module establishes a data link with the cloud server through the wireless communication unit, receives policy execution logs and control effect feedback data, builds a historical case library through the case management and analysis unit, updates the system core model and policy using the differential upgrade package receiving and burning unit, and outputs system firmware upgrade instructions.

[0029] The remote firmware upgrade and self-learning optimization module also has a firmware verification and security protection unit. The firmware verification and security protection unit is used to perform integrity verification and signature verification on the downloaded differential upgrade package before performing firmware upgrade, and to trigger an automatic rollback mechanism to restore to the previous stable firmware version when the upgrade fails.

[0030] The operation steps of a smart clothing wearable method and system based on wireless firmware upgrade are as follows: Step 1: Multimodal manipulation data perception and fusion By integrating a multimodal sensing unit into the detachable functional components of the garment, voice command data and physical button trigger signals from the wearer are collected simultaneously to generate a raw control data set. Subsequently, the raw control data set undergoes command type recognition and semantic parsing. The voice control data subset is decoded into textual control intentions after endpoint detection and feature extraction, and the electrical signals of the button control data subset are converted into digital pulse sequences and parsed into function requests. Finally, the data is integrated into a standardized control intention data stream according to the event trigger time sequence.

[0031] Step 2: Dynamic Scene Recognition and Adaptation Parameter Generation Based on a standardized control intent data stream, it is input into a dynamic scene recognition engine with a built-in deep neural network-based noise classifier and activity recognizer to extract environmental noise features and wearer motion state features. The environmental noise features are compared with a preset noise threshold database to determine whether the environment is quiet, normal, or noisy; the motion state features are matched with a preset activity model to determine whether the wearer is stationary, walking, or in vigorous motion. Based on the determined environmental noise level and motion state, dynamic scene adaptation parameters, including preferred control mode, voice wake-up sensitivity, and key response threshold, are generated.

[0032] Step 3: Intelligent Strategy Matching and Decision Making Key feature vectors are extracted from the dynamic scene adaptation parameters, including ambient noise level, motion state type, remaining battery level, and user historical preferences. These key feature vectors are then compared with the standard applicable feature vectors of various algorithm models in a pre-built dual-mode control strategy algorithm library for multi-dimensional similarity calculation. The algorithm library stores pure voice-first response models, pure button-first response models, hybrid intelligent arbitration response models, and emergency mode response models in matrix form. Based on the similarity calculation results, the optimal control response strategy algorithm model is matched using an algorithm matching engine, and its unique encoding is output as the target control strategy algorithm identifier.

[0033] Step 4: Precise Implementation and Feedback of the Strategy Based on the target control strategy algorithm identifier, the corresponding predefined strategy execution logic is invoked. This logic accesses standardized control intent data streams and dynamic scene adaptation parameters to generate a detailed instruction set containing specific response actions, execution objects, action parameters, and execution timing. This detailed instruction set is then distributed to the apparel's audio-visual prompt module, voice broadcast module, vibration feedback module, and communication module, among other execution units. The execution process is simultaneously monitored, and execution status, results, and system feedback are recorded, forming a strategy execution log. Within a preset period after strategy execution, subsequent user actions and feedback signals are collected to assess control response satisfaction and compare it with expected results, generating control effect feedback data.

[0034] Step 5: Remote Firmware Upgrade and Self-Learning Optimization The strategy execution logs and control effect feedback data are correlated and analyzed to calculate the effectiveness and efficiency scores of the current strategy execution. A complete data package containing dynamic scene adaptation parameters, target control strategy algorithm identifiers, execution results, and effectiveness scores is stored as a new case in the historical case library. Based on the historical case library, differential upgrade packages sent from the cloud are periodically received via wireless communication links to incrementally train and optimize the parameters of the noise classifier and activity recognizer in the dynamic scene recognition engine. Simultaneously, the standard applicable feature vectors and matching weights of each algorithm model in the dual-mode control strategy algorithm library are tuned and updated, enabling remote firmware iterative optimization and self-learning updates of the system's core models and strategies.

[0035] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0036] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for wearing smart clothing based on wireless firmware upgrades, characterized in that: The method includes the following steps: S1. Through the multimodal sensing unit integrated into the detachable functional components of the clothing, the wearer's voice command data and physical button trigger signals are collected simultaneously, and a raw control data set is generated. S2. Perform instruction type identification and semantic parsing processing on the original control data set to generate a standardized control intention data stream; S3. Based on the standardized control intention data stream, the current wearable scene and environmental noise level are identified in real time through the dynamic scene recognition engine, and dynamic scene adaptation parameters are generated. S4. Based on the dynamic scene adaptation parameters, call the algorithm matching engine to intelligently match the optimal control response strategy algorithm model from the pre-built dual-mode control strategy algorithm library and generate the target control strategy algorithm identifier. S5. Based on the target control strategy algorithm identifier, call the corresponding strategy execution logic to dynamically control the functional execution unit of the clothing, and generate strategy execution logs and control effect feedback data. S6. Based on the strategy execution log, control effect feedback data and historical data, remote firmware iteration optimization and self-learning update of the dual-mode control strategy algorithm library and dynamic scene recognition engine are performed through wireless communication link.

2. The method for wearing smart clothing based on wireless firmware upgrade according to claim 1, characterized in that: The generation of the raw control data set in S1 includes the following steps: S11. By deploying a miniature microphone array and intelligent noise reduction unit within the detachable functional components, audio data of the wearer's voice wake-up words and function commands are collected, and a subset of voice control data is generated. S12. By deploying waterproof physical button units on the surface of detachable functional components, the wearer's touch signals such as pressing, long pressing, and double-clicking are collected, and a subset of button operation data is generated. S13. The subset of voice control data and the subset of button control data are combined to form the original control data set.

3. The method for wearing smart clothing based on wireless firmware upgrade according to claim 2, characterized in that: The instruction type identification and semantic parsing processing in S2 includes the following steps: S21. Receive the original control data set, perform endpoint detection and feature extraction on the voice control data subset, and convert the electrical signal of the key control data subset into a standardized digital pulse sequence; S22. Based on the preset voice command recognition model, decode and semantically map the extracted voice features to generate the corresponding textual control intention. At the same time, according to the preset key function mapping table, the digital pulse sequence is parsed into the corresponding function request; S23. The textual control intentions and function requests are integrated into a standardized control intention data stream according to the time sequence of event triggering.

4. The method for wearing smart clothing based on wireless firmware upgrade according to claim 3, characterized in that: The dynamic scene recognition and adaptation parameter generation in S3 includes the following steps: S31. Obtain the standardized control intention data stream and input it into the dynamic scene recognition engine. The engine has a built-in noise classifier and activity recognizer based on deep neural networks, which is suitable for extracting environmental noise features and wearer motion state features from the data stream. S32. Compare the extracted environmental noise features with the preset noise threshold database to determine whether the current environment is quiet, normal, or noisy. At the same time, the motion state characteristics are matched with the preset activity model to determine whether the wearer is in a stationary state, a walking state, or a state of vigorous exercise. S33. Based on the determined environmental noise level and the wearer's movement state, generate dynamic scene adaptation parameters including the preferred control mode, voice wake-up sensitivity, and button response threshold.

5. The method for wearing smart clothing based on wireless firmware upgrade according to claim 4, characterized in that: The generation of the target manipulation strategy algorithm identifier in S4 includes the following steps: S41. Construct a dual-mode control strategy algorithm library, which stores various preset control response strategy algorithm models in matrix form, including pure voice priority response model, pure button priority response model, hybrid intelligent arbitration response model and emergency mode response model. S42. Extract the key feature vector from the dynamic scene adaptation parameters. The key feature vector includes environmental noise level, motion state type, remaining battery level and user historical preferences. S43. Perform multi-dimensional similarity calculation between the key feature vector and the standard applicable feature vector of each algorithm model in the dual-mode control strategy algorithm library; S44. Based on the similarity calculation results, the optimal control response strategy algorithm model is matched through the algorithm matching engine, and its unique encoding is output as the target control strategy algorithm identifier.

6. The method for wearing smart clothing based on wireless firmware upgrade according to claim 5, characterized in that: The generation of strategy execution logs and control effect feedback data in S5 includes the following steps: S51. Based on the target manipulation strategy algorithm identifier, call the predefined strategy execution logic corresponding to the identifier; S52. The predefined strategy execution logic accesses the standardized control intent data stream and the dynamic scene adaptation parameters to generate a detailed instruction set containing specific response actions, execution objects, action parameters and execution timing. S53. The detailed instruction set is sent to the execution units of the clothing, such as the sound and light prompt module, voice broadcast module, vibration feedback module, and communication module. S54. Synchronously monitor the execution process of instructions, record the execution status, execution results and system feedback, and form a policy execution log; S55. Within a preset period after the strategy is executed, collect the user's subsequent operations and feedback signals, evaluate the satisfaction with the control response, compare it with the expected effect, and generate control effect feedback data.

7. The method for wearing smart clothing based on wireless firmware upgrade according to claim 6, characterized in that: The remote firmware iteration optimization and self-learning update in S6 includes the following steps: S61. Perform correlation analysis between the strategy execution log and the control effect feedback data to calculate the effectiveness score and efficiency score of this strategy execution; S62. Store the complete data package of this response, including the dynamic scene adaptation parameters, the selected target control strategy algorithm identifier, the execution result and the effectiveness score, as a new case in the historical case library. S63. Based on the historical case library, the differential upgrade package sent from the cloud is received periodically through the wireless communication link to incrementally train and optimize the parameters of the noise classifier and activity recognizer in the dynamic scene recognition engine. S64. Based on the historical case library, the differential upgrade package sent from the cloud is received periodically through the wireless communication link to optimize and update the standard applicable feature vectors and matching weights of each algorithm model in the dual-mode control strategy algorithm library.

8. A smart clothing wearable system based on wireless firmware upgrade, used to implement the smart clothing wearable method based on wireless firmware upgrade as described in any one of claims 1-7, characterized in that: The system includes: The multimodal control data perception and fusion module gathers the wearer's control signals through the voice acquisition unit and the button signal acquisition unit, performs standardized processing using the command recognition unit and the semantic parsing unit, and outputs a standardized control intention data stream. The dynamic scene recognition and adaptation module receives the standardized control intention data stream, performs scene analysis through the environmental noise identification unit and the motion state recognition unit, calculates the dynamic scene adaptation parameters through the adaptation parameter generation unit, and outputs structured dynamic scene adaptation parameter data. The intelligent strategy matching and decision-making module receives the dynamic scene adaptation parameter data, parses key features through the feature vectorization unit, uses the algorithm matching engine to search and make decisions in the preset dual-mode control strategy algorithm library, and outputs the target control strategy algorithm identifier.

9. A smart wearable clothing system based on wireless firmware upgrades according to claim 8, characterized in that: The system also includes: The strategy precision execution and feedback module receives the target control strategy algorithm identifier, generates detailed control instructions through the strategy instruction compilation unit, controls the function execution unit of the clothing through the execution mechanism drive interface unit, and outputs strategy execution logs and control effect feedback data through the execution monitoring and effect evaluation unit. The remote firmware upgrade and self-learning optimization module establishes a data link with the cloud server through the wireless communication unit, receives the strategy execution log and operation effect feedback data, builds a historical case library through the case management and analysis unit, updates the system core model and strategy using the differential upgrade package receiving and burning unit, and outputs system firmware upgrade instructions.

10. A smart clothing wearable system based on wireless firmware upgrade according to claim 9, characterized in that: The remote firmware upgrade and self-learning optimization module also has a firmware verification and security protection unit. The firmware verification and security protection unit is used to perform integrity verification and signature verification on the downloaded differential upgrade package before performing firmware upgrade, and to trigger an automatic rollback mechanism to restore to the previous stable firmware version when the upgrade fails.

Images