Self-adaptable biological wave resonance AI Bluetooth smart glasses

Abstract

This invention provides an adaptive wearable bio-wave resonance AI Bluetooth smart glasses, belonging to the field of smart glasses technology. It includes: a frame, an adaptive adjustment component, and an intelligent microcirculation control system. The adaptive adjustment component is used to adaptively adjust the microcirculation of the eyes, head, and ears. The Bluetooth intelligent delivery control system is symmetrically mounted on the two temples of the frame and is used to control the adaptive adjustment component. This invention, by combining the adaptive adjustment component and the intelligent microcirculation control system, fundamentally solves the problem of traditional smart Bluetooth glasses only having Bluetooth intelligent delivery functions. It adds a bio-wave resonance chip for synchronous resonance reception and conduction, addressing and improving sub-health issues related to microcirculation disorders in the head, eyes, and ears. The entire adjustment process is automated without manual user intervention. It also features dual-system modes with both passive and active motherboards, and multiple mechanisms such as low-battery safety degradation, combining ease of use, reliability, and security.

Description

Technical Field

[0001] This invention relates to the field of smart glasses technology, and more particularly to a bio-wave resonance AI Bluetooth smart glasses that can be worn adaptively. Background Technology

[0002] The technological evolution of eyeglasses stems from humanity's long-term exploration of vision correction, its development encompassing three core dimensions: the understanding of optical principles, material innovation, and structural optimization. As early as the Roman era, people discovered that water-filled glass spheres could magnify text. In the 11th century, the Arab scholar al-Hashm al-Islam, in his book *The Book of Optics*, first scientifically demonstrated the imaging principle of convex lenses, laying the theoretical foundation for subsequent technologies. In the late 13th century, Venetian craftsmen connected two convex lenses with a frame, inventing the world's first wearable riveted eyeglasses, solving the problem of reading for presbyopia. In the 16th century, German craftsmen invented concave lenses, achieving myopia correction for the first time. In the early 18th century, the British optician Edward Scarlett invented rigid temples, completely freeing the user's hands and establishing the basic form of modern eyeglasses. In terms of materials, there have been three leaps from natural crystal to glass, and then to the resin lenses invented by Essilor in France in 1954, solving the problems of thickness and fragility, and promoting the widespread adoption of eyeglasses. The advent of mechanical grinding equipment in the 19th century enabled lens design to shift from flat to biconcave / biconvex or even crescent curvature, achieving more precise vision correction and forming the modern optical system of eyeglasses.

[0003] Smart Bluetooth glasses are a product of the deep integration of traditional glasses with technologies such as wireless communication, microelectronics, and artificial intelligence. Their development is built upon the maturity of Bluetooth technology, breakthroughs in low-power chips, and the widespread adoption of wearable devices. In the 1990s, technology companies began exploring the feasibility of combining audio functions with glasses. In 1997, ODG launched the first Bluetooth smart glasses, the ODGR-8, marking the official birth of the industry. Core technological evolution has focused on four main directions: First, Bluetooth communication has been upgraded from early versions to BLE 5.x+, achieving lower power consumption, longer range, and multi-device collaborative connectivity; second, breakthroughs in audio technology, integrating miniature microphone arrays, open acoustic units, and bidirectional noise reduction algorithms to improve call clarity; third, improved hardware integration capabilities, with dedicated chips such as the Hengxuan BES2700 / 2800 and Qualcomm AR1 using a 6nm process, achieving a balance between high computing power and low power consumption in a very small size, supporting functions such as voice assistant wake-up and ambient sound recognition; and fourth, functional expansion, gradually integrating AI real-time translation, health monitoring, AR navigation, and other features from basic calls and music playback, becoming lightweight smart wearable terminals. The current product features a lightweight design and uses a core combination of Bluetooth main control chip, microphone, speaker and battery to connect to smartphones to achieve seamless scene interaction, completing a paradigm leap from "functional wearable device" to "human-machine symbiotic super terminal".

[0004] Most smart Bluetooth glasses are worn with a fixed frame, lacking the ability to be adjusted and adapted to different people and scenarios. This can cause the glasses to fall off during special use, resulting in some damage. Summary of the Invention

[0005] This invention provides an adaptive wearable bio-wave resonance AI Bluetooth smart glasses, which solves the shortcomings of existing technologies that are mostly worn with fixed frames and lack the ability to adjust and adapt to different people and scenarios.

[0006] On one hand, this invention provides an adaptively wearable bio-wave resonance AI Bluetooth smart glasses, comprising: a frame, an adaptive adjustment component, an intelligent microcirculation control system, a wearing data acquisition module, a wearing feature extraction module, a wearing state recognition module, a wearing intelligent decision-making module, and a control module. The adaptive adjustment component is used to adaptively adjust the tightness of the fit; the intelligent microcirculation control system includes an adjustment chip symmetrically mounted on the two temples of the frame for controlling the adaptive adjustment component; the wearing data acquisition module collects multi-source raw data related to the wearing state of the glasses; the wearing feature extraction module extracts key features from the multi-source raw data to generate a state recognition feature vector; the wearing state recognition module identifies the current wearing state based on the state recognition feature vector and a pre-trained lightweight recognition model, outputting a current wearing state label; the wearing intelligent decision-making module generates adaptive coordination control commands based on the current wearing state label, associates device functions with the wearing scenario, and outputs different control commands for different wearing states; the control module drives the adaptive adjustment component to perform adjustment operations according to the control commands, synchronously collects feedback data after adjustment and feeds it back to the intelligent control module, and monitors the command execution status in real time.

[0007] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses is provided, wherein the adaptive adjustment component includes a soft headrest and a control rod; the soft headrest and the control rod are fixedly connected and used to adjust the tightness between the smart Bluetooth glasses and the head when worn; the intelligent microcirculation control system also includes an optical wave gyroscope, which is used to detect the angular velocity of head rotation.

[0008] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses, the steps of the wear data acquisition module collecting multi-source raw data related to the wearing state of the glasses include: The data source was determined, and flexible capacitive sensors, pressure sensors, IMU inertial measurement units, hinge position sensors, miniature thermopile sensors, PCB resonant chip modules, and voice control systems were selected and deployed on the inner side of the temples, nose pads, frame hinges, and lens periphery, respectively. Dynamically adapt the sampling frequency, setting the sampling frequency when wearing and when not wearing; Simultaneously collect multi-dimensional data to obtain contact sensing data, IMU data, and auxiliary data. Contact sensing data includes the capacitance change rate between the frame and the skin and the contact pressure value; IMU data includes the three-dimensional attitude angle of the glasses and acceleration changes; auxiliary data includes hinge status, skin surface temperature, and resonance data.

[0009] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses includes a key feature extraction module that performs the following steps: Preprocess the raw data from multiple sources to output standardized multi-source data; Key features were extracted from standardized multi-source data, including temporal features from contact sensing data; instantaneous values, rates of change, and moving averages of three-dimensional attitude angles were extracted from IMU data; and features related to wearing position and attitude deviation were extracted in combination with auxiliary data. A weighted fusion strategy is adopted to integrate multi-dimensional features. Principal component analysis is used to reduce the dimensionality of the initial feature vector, retaining principal components with a cumulative contribution rate exceeding a preset size, and eliminating redundant features. Finally, a key feature vector that can be used for state recognition is generated.

[0010] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses, the wearing status recognition module identifies the current wearing status through the following steps: Five core wearing states and corresponding feature thresholds are preset. The core wearing states are normal wearing, wearing too loosely, wearing too tightly, abnormal posture, and not wearing. The state recognition feature vector generated by the wearing feature extraction module is input into the pre-trained lightweight recognition model to initially identify the current wearing state. The system combines a rule engine to calculate a confidence score for state recognition, corrects recognition biases caused by head movements, and outputs a label for the current wearing status.

[0011] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses, the steps for creating a pre-trained lightweight recognition model include: A multi-scenario wearing dataset was constructed, collecting multi-source raw data with different head shapes, different wearing postures, and different environments. After preprocessing and feature extraction, the corresponding wearing status labels were labeled to form training set, validation set and test set. A lightweight CNN model was selected as a lightweight model for local computing power of the glasses, simplifying the model network structure and reducing the number of parameters; The model is trained using a training set, with the cross-entropy loss function as the optimization objective, and the model parameters are iteratively optimized using gradient descent. The trained model is validated using a test set to ensure that the model's recognition accuracy is not lower than the preset accuracy range and the recognition delay does not exceed the preset delay time, thus obtaining a validated model. The validated model is subjected to lightweight compression to remove redundant parameters, and then deployed to the local edge computing module of the intelligent microcirculation control system to complete the creation of the pre-trained lightweight recognition model.

[0012] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses, the steps of calculating a state recognition confidence score using a rule engine include: The preset rule engine's scoring metrics include three core metrics: the matching degree between feature vectors and thresholds for various wearing states, the stability of data collection, and the consistency of multi-source data. The scores for the three core indicators are calculated as follows: feature matching degree is calculated based on the deviation between the feature vector and the corresponding state threshold, and the smaller the deviation, the higher the score; data stability is calculated based on the fluctuation range of data from five consecutive sampling points, and the smaller the fluctuation, the higher the score; multi-source data consistency is judged based on the synergy of contact sensing data, IMU data, and auxiliary data, and the higher the synergy, the higher the score. The overall confidence score is calculated using a weighted summation method. A preset score range and confidence threshold are set. If the overall confidence score is greater than or equal to the confidence threshold, the recognition result is deemed valid, and the corresponding wearing status label is directly output. If the overall confidence score is less than the confidence threshold, the recognition result is deemed questionable, and a second recognition is triggered.

[0013] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses method includes the following steps for correcting recognition errors caused by head movements: Extract the acceleration change and attitude angle change rate from the IMU data to determine whether there is active head movement. Set thresholds for acceleration change and attitude angle change rate. If the thresholds are exceeded, it is determined to be active head movement. If it is determined to be active head movement, retrieve the head movement trajectory data and adjust the state recognition weights in combination with the rule engine; The recognition result is recalculated based on the adjusted state recognition weights. The wearing state labels before and after the adjustment are compared. If the labels are consistent, the output is confirmed. If the labels are inconsistent, the feature data of three consecutive sampling points are combined for comprehensive judgment. Misidentification caused by head movement is eliminated, and the corrected wearing state label is output. If the movement is determined to be non-active head movement, the original recognition weights are maintained, and the initial recognition of the wearing status label is directly output.

[0014] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses is provided, wherein the step of generating adaptive coordination control commands by the wearable intelligent decision-making module includes: By combining the current wearing status tag with the device function and the corresponding wearing scenario, a mapping relationship between status and command is established; Based on the mapping relationship between status and command, differentiated control commands are generated for different wearing statuses; Perform a rationality check on the differentiated control commands to ensure that the differentiated control commands match the execution capabilities of the adaptive adjustment components.

[0015] According to the present invention, an adaptive wearable bio-wave resonance AI Bluetooth smart glasses are provided, wherein the differentiated control commands include: When wearing the glasses normally, maintain the frame position, adjust the sensor sampling frequency to 10Hz, enable core functions and optimize adaptation; When worn too loosely, the adaptive adjustment component is given a fine-tuning command, triggering a vibration alert, which adapts to the sampling frequency and switches to the corresponding function. When worn too tightly, the device outputs a relaxation command to the adaptive adjustment component, adjusting the force distribution and disabling unnecessary functions to reduce power consumption. When the attitude is abnormal, output a vibration prompt command, correct the attitude calibration parameters, and pause non-core functions if necessary; When not worn, it outputs a low-power command to disable core functions, retaining only intermittent status detection.

[0016] This invention provides an adaptive wearable bio-wave resonance AI Bluetooth smart glasses. By combining an adaptive adjustment component and an intelligent microcirculation control system, it fundamentally solves the problems of discomfort caused by head shape differences in traditional smart Bluetooth glasses, such as pressure marks from prolonged wear, and easy slippage during exercise. The combination of the adaptive adjustment component and intelligent control system effectively addresses these issues. The use of a medical-grade soft nose pad that conforms to the human body's physiological curve significantly improves comfort during extended wear. Furthermore, the dynamic sampling frequency adjustment strategy ensures both fast adjustment response and high recognition accuracy while improving battery life. The fully automated adjustment process requires no manual intervention from the user and includes multiple mechanisms such as head movement misjudgment correction and low-battery safety degradation, combining ease of use, reliability, and safety to significantly enhance the user experience of smart Bluetooth glasses. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the structure of an adaptive wearable bio-wave resonance AI Bluetooth smart glasses provided in an embodiment of the present invention; Figure 2 This is a front view of an adaptive wearable bio-wave resonance AI Bluetooth smart glasses provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a module of an adaptive wearable bio-wave resonance AI Bluetooth smart glasses provided in an embodiment of the present invention; Figure 4 This is a flowchart illustrating how the wearing status recognition module, combined with a rule engine, calculates the status recognition confidence score in an embodiment of the present invention. Figure 5 This is a flowchart illustrating how the wearing status recognition module corrects recognition deviations caused by head movements in an embodiment of the present invention.

[0019] Reference numerals: 1. Frame; 2. Adaptive adjustment assembly; 21. Soft headrest; 22. Control lever; 3. Intelligent microcirculation control system. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] Example: The following is combined Figures 1-5 This invention describes an adaptive wearable bio-wave resonance AI Bluetooth smart glasses.

[0022] like Figures 1-5As shown in the illustration, this invention provides an adaptive wearable bio-wave resonance AI Bluetooth smart glasses, comprising: a frame 1, an adaptive adjustment component 2, and an intelligent microcirculation control system 3. The frame 1 is typically a one-piece support structure made of carbon fiber reinforced polyamide composite material with a carbon fiber content of 20%~30%, including a frame, nose pads, and two foldable temples. The overall weight is controlled within 28g, the frame width ranges from 135~145mm, and the temple extension length is 140~150mm, fitting over 95% of adult users' head sizes. The adaptive adjustment component 2 is used to adaptively adjust the tightness of the fit. The adaptive adjustment component 2 is integrated into an electromechanical actuator at the end of the temple, which can steplessly adjust the clamping force between the temple and the head within a ±5mm stroke range, with a clamping force adjustment range of 0.5N~3N, covering the entire force range required for comfortable wear. The intelligent microcirculation control system 3 is symmetrically mounted on the two temples of the frame 1 and is used to control the adaptive adjustment component 2. The intelligent microcirculation control system 3 is typically an embedded edge computing system based on the ARM Cortex-M4F core, with a main frequency of 168MHz, equipped with 256KBSRAM and 2MB Flash memory, and has local real-time data processing and control capabilities. The standby power consumption of the whole machine is ≤50μA, the working power consumption is ≤15mA@3.7V, and it is equipped with a 110mAh lithium polymer battery, which can achieve 7 days of standby and 12 hours of continuous use on a single charge.

[0023] The intelligent microcirculation control system 3 includes an adjustment chip, which is symmetrically mounted on the two temples of the frame 1 to control the adaptive adjustment component 2. This improves microcirculation between the head, eyes, and ears when worn.

[0024] The modulation chip is a bio-wave resonance chip based on physical therapy. Its working principle may involve using specific physical energies (such as electricity, light, magnetism, sound waves, or heat) to produce biological effects on human tissues, thereby promoting treatment or rehabilitation. The following is a detailed explanation of its working principle, broken down by core modules: Depending on the application scenario, the chip may select one or more of the following physical factors for its physical energy type and mechanism of action: Electrical stimulation works by stimulating nerves or muscles with microcurrents to interfere with pain signal transmission (pain relief) or induce muscle contraction (rehabilitation). The core technology generates low-frequency pulses (e.g., 1-100Hz), adjusting the amplitude and frequency to suit different tissues (e.g., nerves, muscles). Phototherapy (red / near-infrared light) works by penetrating the skin at a specific wavelength (630-850nm), being absorbed by mitochondria, promoting ATP production, and accelerating cell repair.

[0025] The main functions of the regulating chip include: integrating a micro-LED array to control light intensity and pulse period; using pulsed electromagnetic fields (PEMF) at low frequencies to affect cell membrane potential, promote ion channel activity, reduce inflammation, or accelerate bone healing; generating alternating magnetic fields with adjustable frequencies (e.g., 1-50Hz) to target specific tissues (e.g., the eyeball, joints, or bones); and using ultra-high frequency light waves to generate micro-vibrations and thermal effects, promoting blood circulation in deep tissues.

[0026] The regulating chip system architecture includes an energy generation module and a microcontroller. The energy generation module designs an oscillation circuit based on the target physical factors. The microcontroller (MCU) regulates parameters (frequency, duty cycle, intensity) and integrates a biosensor, preamplifier (oscillator wafer), filter, and a special antenna module for wireless radio frequency transmission and reception to achieve closed-loop control, ensuring efficient energy transfer to the target tissue.

[0027] The wearing data acquisition module collects multi-source raw data related to the wearing status of the glasses. The wearing feature extraction module extracts key features from the multi-source raw data to generate a state recognition feature vector. The wearing status recognition module identifies the current wearing status based on the state recognition feature vector, combined with a pre-trained lightweight recognition model, and outputs a current wearing status label. The wearing intelligent decision-making module generates adaptive coordinated control commands based on the current wearing status label, associates device functions with the wearing scenario, and outputs different control commands for different wearing states. The control module drives the adaptive adjustment component 2 to perform adjustment operations according to the control commands, synchronously collects the feedback data after adjustment and feeds it back to the intelligent control module, monitors the command execution status in real time, and uses an incremental PID control algorithm to achieve precise closed-loop adjustment of the clamping force, with a steady-state adjustment error ≤0.05N.

[0028] The adaptive adjustment component 2 includes a soft headrest 21 and a control lever 22. The soft headrest 21 is typically made of medical-grade silicone coated with memory foam in an arc-shaped contact shape, with a Shore hardness of A20±5 and a contact area with the head of ≥120mm². 2This design evenly distributes clamping force, preventing localized indentations. The soft headrest 21 has a radius of curvature of 60-70mm, conforming to the physiological curve of the human temporal region, with rounded edges of R2mm. The control lever 22 is typically driven by a precision lead screw mechanism with a miniature DC geared motor, a lead of 0.5mm, a stepping accuracy of 0.01mm, a maximum output thrust of 5N, and a no-load operating speed of 2mm / s. It features a built-in travel limit switch to prevent overdrive. The soft headrest 21 and control lever 22 are fixedly connected and used to adjust the fit between the smart Bluetooth glasses and the head when worn. The control lever 22 converts the rotational motion of the motor into linear motion through a threaded joint, pushing the soft headrest 21 to extend and retract along the temple axis, thereby changing the pressure between the two soft headrests 21 and the head, and realizing continuous adjustment of the clamping force. The soft headrest 21 integrates a pressure sensor array, which can provide real-time feedback on the clamping force distribution. The pressure sensor array consists of 4 distributed piezoresistive sensors arranged in a 2×2 matrix, with a range of 0~5N, a sensitivity of 1mV / V, and a sampling frequency synchronized with the main sampling frequency of the system.

[0029] The steps involved in collecting multi-source raw data related to glasses wearing status using the wear data acquisition module include: The data sources were determined, and flexible capacitive sensors, pressure sensors, IMU inertial measurement units, hinge position sensors, and miniature thermopile sensors were selected and deployed on the inner side of the temples, nose pads, frame hinges, and lens periphery, respectively. This ensures comprehensive capture of state changes at all key locations during wear, eliminating blind spots.

[0030] The sampling frequency is dynamically adapted, setting different sampling frequencies for worn and unworn states. The main sampling frequency is 10Hz when worn, decreasing to 0.5Hz when unworn, and temporarily increasing to 20Hz when a suspected wearing event is triggered. While ensuring data acquisition accuracy, the system power consumption is dynamically adjusted based on device status to extend battery life.

[0031] Simultaneous acquisition of multi-dimensional data yields contact sensing data, IMU data, and auxiliary data. Contact sensing data includes the capacitance change rate between the frame and skin, and contact pressure values. IMU data includes the three-dimensional attitude angles of the glasses and acceleration changes. Auxiliary data includes hinge status and skin surface temperature. Cross-validation of multi-source data effectively improves data reliability and reduces data errors caused by single sensor failures or interference.

[0032] The steps for extracting key features using a feature extraction module include: The multi-source raw data is preprocessed to output standardized multi-source data. The preprocessing includes steps such as data filtering using a 5th-order Butterworth low-pass filter, outlier removal using the 3σ criterion, and normalization using the min-max normalization method to eliminate the impact of environmental noise, sensor drift, and sudden interference on data quality.

[0033] Key features are extracted from standardized multi-source data, including temporal features from contact sensing data, such as the mean, variance, peak value, and trend of contact pressure, which can effectively reflect the contact state between the frame and the head. Instantaneous values, rates of change, and moving averages of three-dimensional posture angles are extracted from IMU data, and combined with auxiliary data to extract features related to wearing position and posture deviation, which can accurately determine whether the glasses have shifted, slipped, or tilted.

[0034] A weighted fusion strategy is employed to integrate multi-dimensional features. Principal component analysis is used to reduce the dimensionality of the initial feature vector. Appropriate weights are assigned based on the contribution of different features to wearing status recognition, highlighting the role of core features and improving the accuracy of the recognition model. Principal components with a cumulative contribution exceeding a preset size are retained, and redundant features are eliminated, ultimately generating key feature vectors that can be used for status recognition.

[0035] The steps for the wearing status recognition module to identify the current wearing status include: Five core wearing states and corresponding characteristic thresholds are preset: normal wearing, too loose, too tight, abnormal posture, and not wearing. These cover all major wearing states that may occur during daily use of smart Bluetooth glasses, ensuring the system can handle various practical usage scenarios. The normal wearing clamping force threshold is 1N~2N, too loose is <1N, too tight is >2N, abnormal posture is a tilt angle >15°, and not wearing is when all contact sensor signals are below 0.1N and the capacitance change rate is <5%.

[0036] The state recognition feature vector generated by the wearing feature extraction module is input into a pre-trained lightweight recognition model to initially identify the current wearing state. The model can quickly complete feature matching and state classification, achieving a rapid initial recognition response.

[0037] By combining a rule engine to calculate the confidence score for state recognition, the system corrects for recognition biases caused by head movements and outputs a label indicating the current wearing status. Through a dual verification mechanism of model recognition and rule-based judgment, the false recognition rate is significantly reduced, improving the system's robustness.

[0038] In the wear status recognition module, the steps for creating the pre-trained lightweight recognition model include: A multi-scenario wearing dataset was constructed, collecting raw data from multiple sources under different head shapes, wearing postures, and environments. This data covers various usage scenarios, including indoor, outdoor, static, and active scenarios, as well as user samples from different genders and age groups, ensuring the diversity and representativeness of the dataset. After preprocessing and feature extraction, corresponding wearing status labels were added to form training, validation, and test sets. The total dataset size is ≥10,000 samples, with the training set accounting for 70%, the validation set for 15%, and the test set for 15%.

[0039] A lightweight CNN model was chosen as the lightweight model for the glasses' local computing power. The model's network structure was simplified, and the number of parameters was reduced. The model contains two convolutional layers, one pooling layer, and two fully connected layers. The convolutional kernel size is 3×3, and the number of channels is 8 and 16, respectively. While ensuring recognition accuracy, the computational complexity and memory usage of the model were minimized to enable it to run efficiently on embedded edge computing platforms.

[0040] The model was trained using a training set, with cross-entropy loss as the optimization objective. Gradient descent was used to iteratively optimize the model parameters. The initial learning rate was 0.001, decreasing to 0.9 times the original rate every 10 epochs, for a total of 50 training epochs. Data augmentation techniques were employed during training to further enhance the model's generalization ability, enabling it to adapt to differences in user experience and usage environments.

[0041] The trained model is validated using a test set to ensure that its recognition accuracy is not lower than a preset accuracy range (usually 98%) and its recognition latency does not exceed a preset latency time (usually 10ms). A validated model is obtained. Simultaneously, the model's recognition performance in extreme scenarios is tested to ensure that the system can still work stably in complex environments.

[0042] The validated model undergoes lightweight compression, significantly reducing its size while maintaining its original recognition performance, enabling rapid loading and execution. Redundant parameters are removed, and the model is deployed to the local edge computing module of the intelligent microcirculation control system 3 to complete the creation of the pre-trained lightweight recognition model.

[0043] The steps involved in calculating the confidence score for wear status recognition using the wear status recognition module combined with the rule engine include: The pre-defined scoring metrics of the rule engine include three core indicators: the matching degree between feature vectors and thresholds for various wearing states, the stability of data collection, and the consistency of multi-source data. The recognition results are comprehensively evaluated from three dimensions: data quality, feature matching, and multi-source collaboration, ensuring that the confidence score accurately reflects the reliability of the recognition results.

[0044] The system calculates scores for three core indicators. Feature matching degree is calculated based on the deviation between the feature vector and the corresponding state threshold; the smaller the deviation, the higher the score, and it effectively distinguishes between normal states with obvious features and critical states with ambiguous features. Data stability is calculated based on the fluctuation range of data from five consecutive sampling points; the smaller the fluctuation, the higher the score, effectively eliminating the impact of instantaneous data anomalies caused by sudden noise interference on the recognition results. Multi-source data consistency is judged based on the synergy of contact sensing data, IMU data, and auxiliary data; the higher the synergy, the higher the score. Through mutual verification of data from different types of sensors, it effectively identifies single sensor failures or false alarms.

[0045] A weighted summation method is used to calculate the overall confidence score. A preset scoring range and confidence threshold are established, with feature matching weighting at 0.6, data stability weighting at 0.2, and multi-source data consistency weighting at 0.2. The scoring range is 0-100 points, and the confidence threshold is set at 70 points. If the overall confidence score is greater than or equal to the confidence threshold, the recognition result is considered valid, and the corresponding wearing status tag is directly output, ensuring the system can quickly respond to clear status changes and improve the timeliness of adjustments. If the overall confidence score is less than the confidence threshold, the recognition result is considered questionable, triggering a second recognition. During the second recognition process, the sampling frequency is increased to 20Hz, adding 10 sampling points to further improve the accuracy of the recognition result.

[0046] The steps for the wear status recognition module to correct recognition errors caused by head movements include: Extract the acceleration change and attitude angle change rate from the IMU data to determine if there is active head movement. Set thresholds for acceleration change and attitude angle change rate, with the acceleration change threshold set at 2 m / s². 2 The threshold for the rate of change of attitude angle is 10° / s. If the rate exceeds the threshold, it is determined to be an active head movement.

[0047] If the head movement is determined to be active, the head movement trajectory data is retrieved, and the state recognition weight is adjusted in conjunction with the rule engine. The weight of the contact perception data is increased from 0.7 to 0.9, and the weight of the IMU posture data is decreased from 0.3 to 0.1. This reduces the impact of device posture changes caused by head movement on the wearing status recognition and avoids misjudgment.

[0048] The recognition result is recalculated based on the adjusted state recognition weights. The wearing state labels before and after the adjustment are compared. If the labels match, the result is confirmed and output. If the labels do not match, a comprehensive judgment is made by combining the feature data of three consecutive sampling points to eliminate misidentifications caused by head movements, and the corrected wearing state label is output.

[0049] If the movement is determined to be non-active head movement, the original recognition weights are maintained, and the initial recognition of the wearing status label is directly output.

[0050] The steps involved in generating adaptive coordinated control commands using a smart decision-making module include: By combining the current wearing status tag with the device function and the corresponding wearing scenario, a mapping relationship between status and command is established.

[0051] Based on the mapping relationship between status and command, differentiated control commands are generated for different wearing statuses.

[0052] The rationality of the differentiated control commands is verified. The verification includes whether the adjustment stroke is within ±5mm, whether the target clamping force is within 0.5N~3N, and whether the motor drive current exceeds the rated value, so as to ensure that the differentiated control commands match the execution capability of the adaptive adjustment component 2.

[0053] The differentiated control commands in the wearable intelligent decision-making module include: When worn normally, maintain the frame position, adjust the sensor sampling frequency to 10Hz, enable core functions and optimize adaptation. Core functions include Bluetooth audio calls, music playback, and voice assistant wake-up.

[0054] When the fit is too loose, the adaptive adjustment component 2 is given a fine-tuning command. The adjustment speed is 1mm / s, and the adjustment step is 0.5mm each time, until the clamping force reaches 1.5N. The vibration prompt is triggered. The vibration parameters are frequency 200Hz, duration 100ms, and intensity medium. The vibration frequency is adapted to the sampling frequency and the corresponding function is switched.

[0055] When worn too tightly, the adaptive adjustment component 2 is output to loosen the tension. The adjustment speed is 1mm / s, and the adjustment step is 0.5mm each time, until the clamping force reaches 1.5N. The force distribution is adjusted, and unnecessary functions are turned off to reduce power consumption. Unnecessary functions include voice wake-up and ambient sound enhancement.

[0056] When the attitude is abnormal, a vibration prompt command is output. The vibration parameters are a frequency of 200Hz, a duration of 200ms, a strong intensity, and two consecutive vibrations. The attitude calibration parameters are corrected, and non-core functions are paused if necessary.

[0057] When not worn, it outputs a low-power command, shuts down core functions, retains only intermittent status detection, puts the Bluetooth module into deep sleep mode, and switches the system clock to a 32kHz low-speed crystal oscillator.

[0058] The following is a detailed explanation of the working principle of this type of adaptive wearable bio-wave resonance AI Bluetooth smart glasses: Once the device is fully charged, it automatically enters standby mode. The soft headrest 21 on the inside of the temple remains in the initial retracted position, all functions are turned off, and only the basic wake-up detection capability is retained.

[0059] When the user picks up the glasses and unfolds the temples, the device automatically wakes up, the Bluetooth indicator light flashes, and it enters pairing mode, while preparing to perform wear detection.

[0060] After the user puts the glasses on the bridge of their nose, the device automatically detects the wearing action, and the soft headrests 21 on the inside of the left and right temples will slowly extend and gently fit the sides of the user's head.

[0061] When a comfortable clamping force is reached, the soft headrest 21 automatically stops moving, and the device emits a short vibration to indicate that the adjustment is complete.

[0062] After adjustment, all core functions such as Bluetooth audio and voice assistant will be automatically enabled, and users can use them normally.

[0063] After adjustment, all core functions, including Bluetooth audio and voice assistant, will automatically activate, allowing for normal user operation. If the glasses become loose or slip, the soft headrest 21 will automatically and slowly extend to increase the clamping force until a comfortable position is restored. If the user feels pressure on their head, the soft headrest 21 will automatically and slowly retract to decrease the clamping force until a comfortable position is restored. If the glasses are worn crookedly, the device will emit two short vibrations to remind the user to adjust the wearing position. Normal head movements such as turning, nodding, and shaking the head will not trigger false adjustments.

[0064] When the user removes the glasses, the soft headrest 21 automatically retracts to its initial position. All core functions are automatically turned off, Bluetooth is disconnected, and the device enters a deep sleep state, retaining only basic wake-up detection capabilities.

[0065] If any obstacles are encountered during adjustment, the device will immediately stop to prevent damage. When the battery is low, the device will automatically disable the adaptive adjustment function, retaining only basic Bluetooth calling functionality and reminding the user to charge it in time. Users can customize the comfortable clamping force via the accompanying mobile app, and the device will automatically remember and apply the preference settings.

[0066] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A type of adaptive wearable bio-wave resonance AI Bluetooth smart glasses, characterized in that, include: Eyeglass frame (1); Adaptive adjustment component (2) is used to adaptively adjust the tightness of the fit; The intelligent microcirculation control system (3) includes an adjustment chip, which is symmetrically installed on the two temples of the frame (1) for controlling the adaptive adjustment component (2). The wearable data acquisition module is used to collect multi-source raw data related to the wearing status of glasses; The wearable feature extraction module is used to extract key features from the multi-source raw data and generate a state recognition feature vector; The wearing status recognition module is used to identify the current wearing status based on the status recognition feature vector and a pre-trained lightweight recognition model, and output the current wearing status label. The wearable intelligent decision-making module is used to generate adaptive coordination control commands based on the current wearing status tag, associate device functions with wearing scenarios, and output different control commands for different wearing states; The control module is used to drive the adaptive adjustment component (2) to perform adjustment operations according to the control command, synchronously collect the feedback data after adjustment and feed it back to the intelligent control module, and monitor the command execution status in real time.

2. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 1, characterized in that, The adaptive adjustment component (2) includes a soft headrest (21) and a control rod (22); the soft headrest (21) and the control rod (22) are fixedly connected and used to adjust the tightness between the smart Bluetooth glasses and the head when worn; the smart microcirculation control system (3) also includes an optical gyroscope, which is used to detect the angular velocity of head rotation.

3. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 1, characterized in that, The steps of the wear data acquisition module to collect multi-source raw data related to the wearing status of glasses include: The data source was determined, and flexible capacitive sensors, pressure sensors, IMU inertial measurement units, hinge position sensors, miniature thermopile sensors, PCB resonant chip modules, and voice control systems were selected and deployed on the inner side of the temples, nose pads, frame hinges, and lens periphery, respectively. Dynamically adapt the sampling frequency, setting the sampling frequency when wearing and when not wearing; Multi-dimensional data is collected simultaneously to obtain contact sensing data, IMU data, and auxiliary data. The contact sensing data includes the capacitance change rate between the frame and the skin and the contact pressure value; the IMU data includes the three-dimensional attitude angle of the glasses and the acceleration change; the auxiliary data includes the hinge state, skin surface temperature, and resonance data.

4. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 3, characterized in that, The steps for key feature extraction by the wearable feature extraction module include: The multi-source raw data is preprocessed to output standardized multi-source data; Key features are extracted from the standardized multi-source data, including extracting temporal features from the contact sensing data; extracting instantaneous values, rates of change, and moving averages of three-dimensional attitude angles from the IMU data; and combining auxiliary data to extract features related to wearing position and attitude deviation. A weighted fusion strategy is adopted to integrate multi-dimensional features. The initial feature vector is reduced in dimensionality by principal component analysis. Principal components with a cumulative contribution rate exceeding a preset size are retained, and redundant features are eliminated. Finally, the key feature vector that can be used for state recognition is generated.

5. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 1, characterized in that, The steps for the wearing status recognition module to identify the current wearing status include: Five core wearing states and corresponding feature thresholds are preset. The core wearing states are normal wearing, wearing too loosely, wearing too tightly, abnormal posture, and not wearing. The state recognition feature vector generated by the wearing feature extraction module is input into the pre-trained lightweight recognition model to initially identify the current wearing state. The system combines a rule engine to calculate a confidence score for state recognition, corrects recognition biases caused by head movements, and outputs a label for the current wearing status.

6. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 5, characterized in that, The steps for creating the lightweight recognition model pre-trained by the wearing status recognition module include: A multi-scenario wearing dataset was constructed, collecting multi-source raw data with different head shapes, different wearing postures, and different environments. After preprocessing and feature extraction, the corresponding wearing status labels were labeled to form training set, validation set and test set. A lightweight CNN model was selected as a lightweight model for local computing power of the glasses, simplifying the model network structure and reducing the number of parameters; The model is trained using a training set, with the cross-entropy loss function as the optimization objective, and the model parameters are iteratively optimized using gradient descent. The trained model is validated using a test set to ensure that the model's recognition accuracy is not lower than the preset accuracy range and the recognition delay does not exceed the preset delay time, thus obtaining a validated model. The verified model is subjected to lightweight compression processing to remove redundant parameters and deployed to the local edge computing module of the intelligent microcirculation control system (3) to complete the creation of the pre-trained lightweight recognition model.

7. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 5, characterized in that, The steps of the wearing status recognition module in conjunction with the rule engine to calculate the status recognition confidence score include: The preset rule engine's scoring metrics include three core metrics: the matching degree between feature vectors and thresholds for various wearing states, the stability of data collection, and the consistency of multi-source data. The scores of the three core indicators are calculated. The feature matching degree is calculated based on the deviation between the feature vector and the corresponding state threshold. The smaller the deviation, the higher the score. The data stability is calculated based on the data fluctuation amplitude of five consecutive sampling points. The smaller the fluctuation, the higher the score. The multi-source data consistency is judged based on the synergy of contact sensing data, IMU data, and auxiliary data. The higher the synergy, the higher the score. A weighted summation method is used to calculate the overall confidence score. A preset score range and confidence threshold are set. If the overall confidence score is greater than or equal to the confidence threshold, the recognition result is deemed valid, and the corresponding wearing status label is directly output. If the overall confidence score is less than the confidence threshold, the recognition result is deemed questionable, and a second recognition is triggered.

8. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 5, characterized in that, The steps for the wearing status recognition module to correct recognition deviations caused by head movements include: Extract the acceleration change and attitude angle change rate from the IMU data to determine whether there is active head movement. Set thresholds for acceleration change and attitude angle change rate. If the thresholds are exceeded, it is determined to be active head movement. If it is determined to be active head movement, retrieve the head movement trajectory data and adjust the state recognition weights in combination with the rule engine; The recognition result is recalculated based on the adjusted state recognition weights. The wearing state labels before and after the adjustment are compared. If the labels are consistent, the output is confirmed. If the labels are inconsistent, the feature data of three consecutive sampling points are combined for comprehensive judgment. Misidentification caused by head movement is eliminated, and the corrected wearing state label is output. If the movement is determined to be non-active head movement, the original recognition weights are maintained, and the initial recognition of the wearing status label is directly output.

9. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 1, characterized in that, The steps for generating adaptive coordinated control commands using the wearable intelligent decision-making module include: By combining the current wearing status tag, the device functions are associated with the corresponding wearing scenarios, and a mapping relationship between status and commands is established; Based on the mapping relationship between the states and the commands, differentiated control commands are generated for different wearing states; The rationality of the differentiated control command is verified to ensure that the differentiated control command matches the execution capability of the adaptive adjustment component (2).

10. The adaptive wearable bio-wave resonance AI Bluetooth smart glasses according to claim 9, characterized in that, The differentiated control commands in the wearable smart decision-making module include: When wearing the glasses normally, maintain the frame position, adjust the sensor sampling frequency to 10Hz, enable core functions and optimize adaptation; When the fit is too loose, the adaptive adjustment component (2) outputs a fine-tuning command to trigger a vibration alert, adapts to the sampling frequency, and switches the corresponding function. When worn too tightly, the adaptive adjustment component (2) is output to relax the force distribution and turn off unnecessary functions to reduce power consumption. When the attitude is abnormal, output a vibration prompt command, correct the attitude calibration parameters, and pause non-core functions if necessary; When not worn, it outputs a low-power command to disable core functions, retaining only intermittent status detection.

Images