EEG closed-loop feedback driven eye massager adaptive control method and device

Abstract

The application relates to the technical field of closed-loop control based on biometric recognition, and discloses an electroencephalogram closed-loop feedback driven eye massager adaptive control method and equipment, wherein the method comprises the following steps: collecting electroencephalogram signals of a target user in a resting closed-eye state, and performing power spectrum density average value and power spectrum density standard deviation calculation on each frequency band of the electroencephalogram signals, and constructing a resting state baseline; determining a mental state category according to the power spectrum density average value of each frequency band, matching audio from a pre-set soothing audio library based on the mental state category, and associating the audio identifier corresponding to the matched audio, a pre-set audio volume and a pre-set audio weight to serve as audio configuration; and controlling the eye massager to perform a massage operation in an electroencephalogram closed-loop feedback driven adaptive control mode according to the resting state baseline and the audio configuration. Through the synergistic effect of the establishment of a personalized reference system and the electroencephalogram closed-loop feedback driven adaptive control, the massage strategy always operates around the optimal comfort interval of the target user.

Description

Technical Field

[0001] This application relates to the field of closed-loop control technology based on biometric recognition, and in particular to an adaptive control method and device for an eye massager driven by EEG closed-loop feedback. Background Technology

[0002] In the field of personal health care electronic devices, eye massagers have gained widespread attention for their ability to relieve eye strain and improve sleep. Currently, consumer-grade eye massagers primarily utilize airbag pressure technology combined with heat therapy and vibration. They rely on built-in micro-pumps to drive multiple airbag zones distributed around the eye sockets, inflating and deflating them according to a pre-set sequence. This physical pressure promotes blood circulation around the eyes, while constant-temperature heat therapy and micro-vibration provide multi-dimensional sensory relaxation. However, current eye massagers rely solely on preset programs for operation, failing to recognize individual user differences and real-time physiological feedback, thus making it difficult to meet the personalized massage needs of different users and the same user in different states. Summary of the Invention

[0003] Based on this, it is necessary to address the technical problem that existing eye massagers use preset fixed program control, which cannot recognize individual differences and real-time physiological feedback of users, and cannot meet the personalized massage needs of different users and the same user in different states. Therefore, an adaptive control method and device for eye massagers driven by EEG closed-loop feedback is proposed.

[0004] In a first aspect, an adaptive control method for an eye massager driven by EEG closed-loop feedback is provided, the method comprising: In response to the initialization command, the eye massager is controlled to execute the guided mode permission according to the set of guided parameters; During the execution of the guided mode, the target user's EEG signal in a resting, closed-eye state is collected, and the average power spectral density and standard deviation of the power spectral density of each frequency band are calculated, and a resting-state baseline is constructed. The mental state category is determined based on the average power spectral density of each frequency band. Audio is matched from a preset soothing audio library based on the mental state category. The audio identifier, preset audio volume and preset audio weight corresponding to the matched audio are associated as the audio configuration. Based on the resting-state baseline and audio configuration, an adaptive control method driven by EEG closed-loop feedback is used to control the eye massager to perform massage operations.

[0005] In a second aspect, an eye massager is provided, comprising: a main control module, a massage execution module, an EEG acquisition module, and an audio output module, wherein the main control module is electrically connected to the massage execution module, the EEG acquisition module, and the audio output module respectively; The massage execution module includes a miniature air pump, a solenoid valve group, multiple independent airbag zones, and a heating element. The miniature air pump is connected to each airbag zone through an air guide tube and its inflation and deflation sequence is independently controlled by the solenoid valve group. The heating element, the miniature air pump, and the solenoid valve group are controlled by the PWM signal output by the main control module. The EEG acquisition module includes electrode pads and a signal conditioning circuit electrically connected to the electrode pads. The signal conditioning circuit converts the EEG signals acquired by the electrode pads into digital signals and sends them to the main control module. The audio output module is used to play audio; The main control module is configured to perform the steps of the adaptive control method for an eye massager driven by EEG closed-loop feedback as described in any of the first aspects.

[0006] This application acquires EEG signals from target users in a resting, closed-eye state and constructs a resting-state baseline. This provides a personalized reference system reflecting the target user's brain activity, allowing all subsequent EEG changes to be quantified relative to this reference system. This avoids misjudgments caused by using fixed thresholds to uniformly evaluate different target users. Based on this, the application determines the mental state category and matches corresponding audio configurations according to the average power spectral density of each frequency band of the EEG signal. This achieves personalized initial audio matching based on objective physiological data, avoiding the tediousness and inaccuracy of manual selection by the target user. Furthermore, an adaptive control method driven by EEG closed-loop feedback is used to execute the massage operation. This allows the eye massager to continuously sense the target user's real-time physiological feedback and dynamically adjust subsequent massage strategies based on the established resting-state baseline during operation. Therefore, this application, through the synergistic effect of establishing a personalized reference system and adaptive control driven by EEG closed-loop feedback, ensures from the source that the benchmark on which the massage strategy is based is individualized, and also achieves dynamic tracking of the target user's state changes throughout the massage process. This ensures that the massage strategy always operates around the target user's optimal comfort range, avoiding the problems of overstimulation or ineffective massage caused by a fixed "one-size-fits-all" procedure, thereby improving the personalized adaptation ability and massage effectiveness of eye massage. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of this application 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 only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] in: Figure 1This is an application environment diagram of an adaptive control method for an eye massager driven by EEG closed-loop feedback in one embodiment. Figure 2 This is a flowchart of an adaptive control method for an eye massager driven by EEG closed-loop feedback in one embodiment. Figure 3 This is a structural block diagram of an eye massager in one embodiment. Detailed Implementation

[0009] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0010] In related technologies, different users have significantly different tolerances to massage intensity, and the massage needs of the same user also vary depending on their level of fatigue. However, current eye massagers only use preset fixed programs for control during operation, failing to recognize individual user differences and real-time physiological feedback, and lacking the ability to dynamically adjust massage strategies. This easily leads to situations where "excessive force" causes user discomfort, or "insufficient force" results in poor massage effects. To address the above problems, this application provides an adaptive control method for eye massagers driven by EEG closed-loop feedback. By collecting and analyzing the user's EEG signals in real time, using a resting-state baseline as a personalized reference, the method dynamically adapts massage parameters during the massage process, achieving an intelligent upgrade from "blindly executing preset fixed programs" to "perception-understanding-adaptation." The following is combined with... Figure 1 The application environment shown will be used to provide a detailed description of the technical solution of this application.

[0011] The EEG closed-loop feedback-driven adaptive control method for eye massagers provided in this application embodiment can be applied to, for example... Figure 1 The application environment shown. Please refer to [link / reference]. Figure 1 As shown, the application environment includes an eye massager 01 and a control terminal 02. The control terminal 02 can be integrated into the eye massager 01 (for example, as the main control module of the eye massager 01), or it can be set independently of the eye massager 01 (for example, in the form of a smartphone, tablet, smart bracelet, smartwatch or dedicated remote control). The embodiments of this application do not impose specific limitations on this.

[0012] When the control terminal 02 is set independently of the eye massager 01, the control terminal 02 and the eye massager 01 are connected via wired or wireless means (such as Bluetooth, Wi-Fi, ZigBee, etc.) to send control commands to the main control module of the eye massager 01 and receive its feedback status data.

[0013] The eye massager 01 is a wearable device, its shape adapted to the contour of the human eye. The target user fixes the eye massager 01 to the head by wearing a strap, so that its inner side fits against the target user's eye area. The eye massager 01 includes a main control module, a massage execution module, an EEG acquisition module, and an audio output module. The main control module is electrically connected to the massage execution module, the EEG acquisition module, and the audio output module. The modules are connected to each other through an I2C (Inter-Integrated Circuit) bus and a PWM (Pulse Width Modulation) control line to form a local control network, realizing data interaction and command distribution. The main control module is an embedded processor with an integrated microcontroller unit (MCU). It internally stores an EEG analysis algorithm library, a comfort zone mapping database, and a multi-channel PWM driver program. It is used to receive and process the EEG signals transmitted by the EEG acquisition module, construct a resting-state baseline and determine the audio configuration based on the EEG signals, and monitor the target user based on the EEG signals.

[0014] The massage execution module includes a miniature air pump, a solenoid valve assembly, multiple independent airbag zones, and a heating element. The miniature air pump is connected to each airbag zone via an air guide tube, and the solenoid valve assembly independently controls the inflation and deflation sequence of each airbag zone. The heating element, the miniature air pump, and the solenoid valve assembly are controlled by a PWM signal output from the main control module to achieve precise adjustment of the pressure, timing, and temperature of each airbag zone. The airbag zones are distributed around multiple acupoints around the eye socket (e.g., temples, brow ridges, Chengqi acupoints), and the alternating inflation and deflation simulates the kneading motion of human hand massage, promoting blood circulation around the eyes.

[0015] Airbag compartments are independent, controllable, and expandable cavities in eye massagers. They are often made of TPU (Thermoplastic Polyurethane), which has high elasticity and durability.

[0016] The miniature air pump is a built-in miniature electric pump that provides positive pressure air to the airbag compartments. The miniature air pump is typically fixed inside the outer casing of the eye massager and connected to the solenoid valve assembly via an air tube.

[0017] Optionally, the number of airbag sections is 4 to 8.

[0018] A solenoid valve assembly is a valve body combination consisting of multiple independent solenoid valves. Under the control of the main control module, it is used to open or close the air passages of each airbag compartment individually, achieving independent control of the inflation and deflation sequence of each airbag compartment. The solenoid valve assembly is installed inside the housing, located between the miniature air pump and the airbag compartment.

[0019] The heating element is preferably a graphene heating film, which is a planar heating element. It is fitted into the internal structure of the side of the eye mask that contacts the body, or embedded in the back of the airbag section to provide a warm compress function.

[0020] Optionally, the massage execution module further includes a vibration element controlled by a PWM signal output from the main control module to precisely adjust the vibration intensity of each airbag zone. The vibration element is preferably an eccentric wheel vibration motor or a linear resonant motor, fitted to the side in contact with the human body or integrated into the back side of the airbag zone (i.e., the surface of the airbag zone facing away from the skin, or in the receiving space between the airbag zone and the device housing), to provide micro-vibration stimulation with adjustable frequency and amplitude to relax the muscles around the eyes and relieve nerve tension.

[0021] Taking the layered structure of an eye massager as an example, from the inside out (i.e., from the side that contacts the skin to the side that is away from the skin), the sequence is: skin contact layer (such as a flexible fabric or silicone layer) → airbag section body → back surface of the airbag section → device shell or internal support. Therefore, placing the vibration element (such as an eccentric wheel vibration motor or a linear resonant motor) on the "back side of the airbag section" means fixing the vibration element on the surface of the airbag section away from the skin, or embedding it in the receiving space between the airbag section and the device shell.

[0022] The EEG acquisition module includes electrode pads and a signal conditioning circuit electrically connected to the electrode pads. The electrode pads are preferably dry electrodes made of Ag / AgCl (silver / silver chloride) conductive rubber, positioned inside the eye massager 01 in contact with the target user's forehead area, for acquiring the target user's EEG signals. The signal conditioning circuit sequentially includes an instrumentation amplifier, a power frequency notch filter, and an analog-to-digital converter. The instrumentation amplifier differentially amplifies the microvolt-level EEG signals acquired by the electrode pads to suppress common-mode interference; the power frequency notch filter filters out 50Hz mains coupling noise; and the analog-to-digital converter converts the amplified and filtered analog EEG signals into digital signals, which are then sent to the main control module.

[0023] Instrumentation amplifier: Composed of a high-precision differential amplifier and external resistors, it is usually a three-op-amp or two-op-amp structure and is used to extract weak differential EEG signals.

[0024] Power frequency notch filter: An active band-stop filter consisting of resistors, capacitors and operational amplifiers, typically designed with a center frequency of 50Hz, used to suppress mains interference.

[0025] Analog-to-digital converters (ADCs): Composed of sample-and-hold circuits, quantizers, and encoders, they are typically integrated into successive approximation or delta-Sigma (Δ-Sigma) chips to convert analog voltages into digital quantities. A delta-Sigma chip is a high-precision ADC based on oversampling and noise shaping techniques, often used to convert weak analog electroencephalogram (EEG) signals into digital signals.

[0026] To address the impact of electromagnetic interference during the inflation and deflation of the airbag on EEG signal acquisition, the signal conditioning circuit of the EEG acquisition module includes a hardware filtering unit to pre-filter high-frequency electromagnetic interference before the EEG signal enters the instrumentation amplifier. Simultaneously, an optocoupler is used between the signal conditioning circuit and the main control module to cut off electromagnetic noise generated by the miniature air pump and solenoid valve assembly from coupling to the EEG signal acquisition channel via the electrical path. When acquiring EEG signals, the main control module uses the moment the solenoid valve assembly actuates as a synchronization reference point. Within a preset blanking time window (preferably 50ms to 100ms) after the solenoid valve actuates, the acquired EEG signal is not recorded or analyzed temporarily to eliminate the influence of mechanical vibration and electromagnetic transient noise on the EEG signal.

[0027] Optionally, the number of electrode pads is one, which is set in the forehead region for collecting global EEG signals.

[0028] Optionally, there are multiple air sac partitions and electrode pads, with each air sac partition and electrode pad corresponding to another. The electrode pad is positioned directly behind or above its corresponding air sac partition. The electrode pad is used to collect EEG signals from the local brain region corresponding to the air sac partition. When any air sac partition is activated, the EEG signal from the local brain region corresponding to that air sac partition is collected through the electrode pad corresponding to that air sac partition. This achieves spatial decoupling and precise attribution of the EEG signal source.

[0029] When acquiring EEG signals through any electrode pad, the main control module simultaneously monitors the contact impedance value of that electrode pad. When the contact impedance value exceeds a preset impedance threshold (optional, 5kΩ), the main control module marks the currently acquired EEG signal as low-confidence data and excludes it from the relaxation score calculation. When the contact impedance value is within the preset range, the acquired EEG signal is marked as high-confidence data and included in the relaxation score calculation. Through the above contact impedance monitoring and confidence marking, the influence of signal quality fluctuations caused by changes in electrode contact state during airbag inflation and deflation on the relaxation score is eliminated.

[0030] The audio output module includes an audio decoding chip and a speaker or Bluetooth communication unit. The audio decoding chip has a built-in digital signal processor (DSP) for decoding compressed audio files. The speaker or Bluetooth communication unit receives playback commands from the main control module and outputs sound through the built-in speaker, or plays audio from a preset soothing audio library (such as white noise, alpha wave induced music, and natural sounds) through a Bluetooth player (such as Bluetooth headphones or Bluetooth speakers) connected via the Bluetooth communication unit, to assist the target user in relaxing. The preset soothing audio library includes various audio genres.

[0031] In practical use, after the target user wears the eye massager 01 on their head and starts the device, the main control module first responds to the initialization command and controls the massage execution module to start running in gentle mode according to the set of guidance parameters. Simultaneously, it controls the EEG acquisition module to collect the target user's EEG signals in a resting, closed-eye state. The main control module calculates the average power spectral density and standard deviation of each frequency band of the EEG signal and constructs a resting-state baseline to obtain a personalized reference system reflecting the target user's baseline brain activity level. Subsequently, the main control module determines the target user's current mental state category (e.g., anxiety / tension, alertness / calmness, or relaxation / drowsiness) based on the average power spectral density of each frequency band, and matches corresponding audio from a preset soothing audio library based on this mental state category. The matching audio's corresponding audio identifier, preset audio volume, and preset audio weight are associated as the audio configuration. Based on this, the main control module uses an adaptive control method driven by EEG closed-loop feedback, according to the resting-state baseline and the audio configuration, to control the eye massager 01 to perform massage operations. During the massage process, the main control module continuously receives real-time EEG signals collected by the EEG acquisition module. Using the resting baseline as a reference, it dynamically adjusts the massage strategy to adjust the inflation pressure, inflation and deflation sequence, heating temperature, and audio playback parameters of each airbag zone, so that the massage strategy always operates around the optimal comfort range of the target user.

[0032] In other words, this application achieves closed-loop control of an eye massager based on biometric recognition. This application constructs a closed-loop control system based on multimodal biometric recognition, with electroencephalogram (EEG) signals as the primary basis and heart rate variability signals as an auxiliary reference.

[0033] The overall working principle of the eye massager provided in this application embodiment is as follows: After activation, the EEG acquisition module captures the target user's EEG signals in real time, and performs Fourier transform on the EEG signals to extract characteristic frequency bands. In the initial stage of the massage, an exploratory strategy is adopted, exploring different massage areas, pressure, and temperature combinations, recording the parameter configuration of the EEG signals that induce high relaxation scores. Subsequently, the subsequent massage strategy is dynamically adjusted based on the EEG signals, prioritizing the activation of areas with high partition weights and inhibiting areas with low partition weights, while simultaneously matching an audio therapy program of the same frequency, achieving a personalized adaptive closed-loop massage. The specific implementation of the above working principle will be described in detail in the following steps.

[0034] Please see Figure 2 As shown, Figure 2 A flowchart illustrating the adaptive control method for an eye massager driven by EEG closed-loop feedback provided in this application embodiment includes the following steps: S1: Respond to the initialization command and, based on the set of guidance parameters, control the eye massager to execute the guidance mode enabled; An initialization command is a command signal used to trigger the eye massager to start and enter the guided mode. The sources of initialization commands include button operations during the target user's first power-on, start signals actively triggered by the target user, or control commands automatically generated by the system after detecting that the target user is wearing the device. The target user inputs the initialization command through operations via device buttons, touch panel, or remote control, or a trigger signal is sent to the main control module by a wear detection sensor (such as a capacitive contact sensor or pressure sensor) after detecting that the device has been correctly worn by the target user.

[0035] The guidance parameter set refers to the default set of operating parameters used by the eye massager when it is first activated, or when historical data of the target user is lacking, or when a resting-state baseline needs to be rebuilt, to guide the target user into a resting, closed-eye state suitable for EEG signal acquisition. This initial configuration is based on the principles of gentleness and low interference, and its parameter settings are designed to allow the target user to relax naturally and maintain a resting, closed-eye state after wearing the eye massager, without overstimulating the target user or causing discomfort. The guidance parameter set includes gentle pressure applied to each airbag zone at low pressure, mild heat applied to the heating film at a moderate temperature, and initial audio playback at a low volume from the audio output module.

[0036] Specifically, upon receiving the initialization command, the system reads the set of guiding parameters from a preset storage space (e.g., the main control module's memory). Control signals with corresponding duty cycles are sent via the PWM control line to the drive port of the miniature air pump, the control port of the solenoid valve assembly, and the power adjustment port of the heating element in the massage execution module. This controls the eye massager to enable the guided mode, guiding the target user into a natural, relaxed, and restful state with their eyes closed. Simultaneously, a start command is sent to the EEG acquisition module, putting the signal conditioning circuit into a data acquisition preparation state; a load command is sent to the audio output module, causing the audio decoding chip to load the audio library index and establish a communication link.

[0037] S2: During the execution of the guided mode, the target user’s EEG signal in a resting and closed-eye state is collected, and the average power spectral density and standard deviation of the power spectral density of each frequency band are calculated, and the resting state baseline is constructed. Electroencephalogram (EEG) signals refer to the weak potential changes on the scalp surface caused by the synchronous electrical activity of neuronal groups in the cerebral cortex.

[0038] The resting, closed-eye state refers to the physiological state of the target user when they are not receiving any external active stimulation, keeping their body relaxed, and their eyes closed. In this state, the brain is mainly engaged in spontaneous rhythmic activity, and alpha waves usually exhibit relatively obvious power characteristics, which can accurately reflect the baseline EEG activity level of the target user when they are not subject to external interference.

[0039] The average power spectral density (PSD) is a statistical measure obtained by averaging the power spectral density across preset frequency bands after performing a Fast Fourier Transform (FFT) on an EEG signal. Power spectral density is the distribution of signal power per unit frequency within a specific frequency band after performing a FFT on an EEG signal; it is a key indicator for quantifying the intensity of EEG rhythms in that frequency band.

[0040] The resting-state baseline is a statistical reference system composed of the average power spectral density and standard deviation of each frequency band (delta wave, theta wave, alpha wave, and beta wave) calculated based on EEG signals collected from a target user in a resting, closed-eye state. It serves as a personalized reference system reflecting the baseline level of brain activity for that target user. The resting-state baseline includes the average power spectral density of each frequency band (called the average power spectral density) and the standard deviation of the power spectral density of each frequency band (called the standard deviation of the power spectral density). The average power spectral density of the alpha wave is labeled as the resting alpha baseline, and the ratio of the average power spectral density of the theta wave to that of the alpha wave is labeled as the relaxation index baseline. The standard deviation of the power spectral density of each frequency band is used to quantify the degree of fluctuation and dispersion of the EEG signals during resting-state acquisition, serving as a statistical basis for determining whether real-time power spectral density changes are significant during subsequent closed-loop control. The average power spectral density and standard deviation of each frequency band in the resting-state baseline are stored together in the "Baseline Spectrum Baseline" field of the comfort zone mapping database.

[0041] Delta waves: frequency 0.5-4 Hz, reflecting deep sleep or a state of brain inhibition. Theta waves: frequency 4-8 Hz, reflecting drowsiness, meditation, or a relaxed but not awake state. Alpha waves: frequency 8-13 Hz, reflecting a relaxed and calm state with eyes closed and at rest. Beta waves: frequency 13-30 Hz, reflecting an alert, anxious, or mentally active state.

[0042] Specifically, during the process of the guidance mode being operated in a gentle pressing manner and the target user being in a resting and closed-eye state, the EEG acquisition module continuously acquires and digitizes the target user's EEG signals in the resting and closed-eye state at a preset sampling rate.

[0043] First, the raw digitized EEG signals were preprocessed. Fast Fourier Transform (FFT) was performed on the preprocessed EEG signals within fixed time windows to calculate the power spectral density (PSD) for each frequency band (δ, θ, α, and β waves). After the acquisition phase corresponding to the resting, eye-closed state, the PSD calculation results for each time window were statistically analyzed. The average PSD and standard deviation for each frequency band were calculated to construct a resting-state baseline.

[0044] The preprocessing operations specifically include: first, performing power frequency notch filtering on the raw digitized EEG signal to eliminate periodic interference caused by mains coupling; performing high-pass filtering to remove baseline drift caused by electrode polarization or slow skin potential changes; and performing low-pass filtering to eliminate high-frequency electromyographic noise.

[0045] The method for determining whether the target user is in a resting, closed-eye state includes: after the target user wears the eye massager, they naturally close their eyes. During the guided mode operation, the main control module continuously acquires EEG signals within a first time window (e.g., 30 seconds), calculates the average power spectral density and standard deviation of alpha waves (8–13 Hz) within this time window, and calculates the average power spectral density of beta waves (13–30 Hz). When the average power spectral density of alpha waves is significantly higher than that of beta waves (e.g., the ratio of the average power spectral density of alpha waves to the average power spectral density of beta waves is greater than a preset proportional threshold), and the standard deviation of the power spectral density of alpha waves is lower than a preset fluctuation threshold, it is determined that the target user has entered a stable resting, closed-eye state, and new EEG signals are then acquired to construct a resting-state baseline.

[0046] S3: Determine the mental state category based on the average power spectral density of each frequency band, match audio from the preset soothing audio library based on the mental state category, and associate the audio identifier, preset audio volume and preset audio weight of the matched audio as the audio configuration; Mental state category refers to the type of current brain activity of a target user, characterized by the relative distribution of the average power spectral density of each frequency band in the electroencephalogram (EEG). Mental state categories include anxious or tense state, awake or calm state, and relaxed or drowsy state.

[0047] A preset soothing audio library refers to a collection of audio files pre-stored to assist the target user in relaxing. The preset soothing audio library includes various audio genres, such as white noise, alpha wave-induced music, natural sounds, and light music. The data source for the preset soothing audio library is the audio files pre-installed at the device's factory. Each audio genre corresponds to one or more audio files, and each audio file has a unique audio identifier.

[0048] The audio configuration is the control data for the audio played by the eye massager. The audio configuration includes the audio identifier corresponding to the matched audio, the audio volume, and the audio weight. It can be understood that the audio identifier in the audio configuration includes at least two elements.

[0049] Preset audio volume refers to the pre-set audio playback loudness value, which is used to determine the initial playback volume of the audio in the audio configuration.

[0050] Preset audio weight refers to the pre-set audio selection priority coefficient, which is used to control the probability of an audio being selected among multiple audio candidates.

[0051] Specifically, the average power spectral density of each frequency band calculated in S2 is read, and the ratio of the average power spectral density of the β wave to that of the α wave is calculated. Based on the comparison of this ratio with a preset threshold range, the target user's current mental state category is determined. The device executing this application internally stores a preset soothing audio library, which includes an audio recommendation mapping table and audio files. The audio recommendation mapping table defines the correspondence between different standard categories and audio genres. Based on the determined mental state category, a matching category is searched in the audio recommendation mapping table. Based on the audio genre corresponding to the matching standard category, an audio file under that genre is selected from the preset soothing audio library. The audio identifier, preset audio volume value, and audio weight corresponding to the genre of each selected audio file are associated and combined into an audio configuration data structure.

[0052] For example, when the target user's mental state is determined to be either awake or slightly stressed, the first type of soothing music (including white noise or alpha wave music) is selected from the preset soothing audio library as the matching audio file; when alpha or theta waves are dominant, other corresponding categories of audio are selected from the preset soothing audio library as the matching audio file according to the mental state category.

[0053] S4: Based on the resting baseline and audio configuration, an adaptive control method driven by EEG closed-loop feedback is used to control the eye massager to perform massage operations.

[0054] The adaptive control method driven by EEG closed-loop feedback refers to a control mode that uses real-time acquired EEG signals as feedback, adjusts the massage strategy of the eye massager continuously according to the audio configuration and the resting state baseline as a personalized reference, so that the massage strategy matches the real-time physiological state of the target user.

[0055] Massage operation refers to a series of physical stimulation actions performed by the massage execution module according to the drive instructions generated by the main control module when the eye massager is worn by the target user. Massage operation includes the inflation and deflation of each airbag zone according to a specific time sequence, the micro-vibration generated by the eccentric wheel vibration motor at a specific frequency, and the heat application of the heating element at a specific power.

[0056] Specifically, based on the resting-state baseline established in S2 and the audio configuration determined in S3, an initial massage strategy is generated. An adaptive control method driven by EEG closed-loop feedback is used to control the eye massager to perform massage operations according to the massage strategy. During the massage operation, the EEG acquisition module continuously acquires and digitizes the target user's real-time EEG signals at a preset sampling rate. The same preprocessing and fast Fourier transform as in S2 are performed on the real-time acquired EEG signals to calculate the power spectral density of each frequency band at the current moment. The average power spectral density and standard deviation of each frequency band in the resting-state baseline are used as reference benchmarks. For the power spectral density of each frequency band calculated at the current moment, the difference between the power spectral density and the average power spectral density of the corresponding frequency band in the resting-state baseline is calculated. When the absolute value of the difference exceeds a preset multiple (such as 1 or 2 times) of the standard deviation of the power spectral density (referring to the standard deviation of the power spectral density in the resting-state baseline), it is determined that the power spectral density of that frequency band has changed significantly. When the absolute value of the difference does not exceed a preset multiple of the standard deviation of the power spectral density, it is determined that the change belongs to normal physiological fluctuations. This quantifies the target user's current relaxation level and the trend of EEG activity changes. The massage strategy is dynamically adjusted based on the quantitative results: when the target user's relaxation level is detected to be continuously increasing or stable at a high relaxation level, the airbag pressure intensity and heating power are appropriately maintained or reduced to avoid overstimulation; when insufficient relaxation is detected and there are no obvious signs of stress, the stimulation intensity of the corresponding area is appropriately increased to promote relaxation. Simultaneously, the audio weight distribution in the audio configuration is dynamically adjusted according to the dominant frequency characteristics of the current EEG signal. The next audio is selected from the preset soothing audio library according to the updated weights, and the audio volume is adjusted according to the amplitude change trend of the EEG signal. Then, the audio is played according to the selected audio and the adjusted audio volume. The above adjustment process continues throughout the massage operation, forming a closed-loop control cycle of "acquisition-analysis-adjustment-reacquisition" until the massage duration reaches the preset value or the target user actively terminates the operation.

[0057] This embodiment acquires EEG signals from the target user in a resting, closed-eye state and constructs a resting-state baseline. This provides a personalized reference system reflecting the target user's brain activity, allowing all subsequent EEG changes to be quantified relative to this reference system. This avoids misjudgments caused by using fixed thresholds to uniformly evaluate different target users. Based on this, the mental state category is determined according to the average power spectral density of each frequency band of the EEG signal, and corresponding audio configurations are matched. This achieves personalized initial audio matching based on objective physiological data, avoiding the tediousness and inaccuracy of manual selection by the target user. Furthermore, an adaptive control method driven by EEG closed-loop feedback is used to execute the massage operation. This allows the eye massager to continuously sense the target user's real-time physiological feedback and dynamically adjust subsequent massage strategies accordingly, using the established resting-state baseline as a reference. Therefore, this application, through the synergistic effect of establishing a personalized reference system and adaptive control driven by EEG closed-loop feedback, ensures from the source that the benchmark on which the massage strategy is based is individualized, and also achieves dynamic tracking of the target user's state changes throughout the massage process. This ensures that the massage strategy always operates around the target user's optimal comfort range, avoiding the problems of overstimulation or ineffective massage caused by a fixed "one-size-fits-all" procedure, thereby improving the personalized adaptation ability and massage effectiveness of eye massage.

[0058] In one embodiment, the adaptive control method driven by EEG closed-loop feedback, based on resting-state baseline and audio configuration, to control the eye massager to perform massage operations includes: S41: Responding to the exploration command, based on the resting baseline and audio configuration, it performs trial sampling for the target user and constructs comfort zone mapping data; An exploration command is a control signal used to trigger the eye massager to perform a personalized parameter exploration process based on the target user's current physiological state. Exploration commands can originate from control signals automatically generated by the system after completing resting-state baseline construction and audio configuration, active confirmation signals issued by the target user via device buttons or touch panel after completing resting-state data acquisition, or active confirmation signals generated by the eye massager itself when the target user activates the device after establishing a resting-state baseline and audio configuration. Upon receiving an exploration command, the system begins a trial sampling process based on the target user's current state prior to the current massage operation.

[0059] The target user refers to a specific individual who is currently wearing an eye massager and receiving EEG signal collection and massage services.

[0060] Trial sampling refers to the process of activating the eye massager according to a preset trial strategy while simultaneously collecting the target user's electroencephalogram (EEG) signals to assess the impact of various massage parameters on the user's relaxation level, under the target user's current physiological state. Trial sampling includes sequentially activating each airbag zone individually to assess the massage effect at different spatial locations, performing gradient heating within a preset temperature range to assess the effect of different heat application temperatures, and playing audio according to the configured audio settings to assess the effect of different audio content.

[0061] Comfort zone mapping data refers to a dataset that characterizes the mapping relationship between different combinations of massage parameters and the target user's level of relaxation. Comfort zone mapping data includes the zone identifier, zone weight, optimal pressure, optimal temperature, temperature constraints, and adjusted audio configuration for each airbag zone. This data is stored in a comfort zone mapping database and used to generate personalized massage strategies for the current massage session.

[0062] Specifically, upon responding to the exploration command, exploratory sampling is performed under the target user's current physiological state, using the average power spectral density and standard deviation of each frequency band stored in the resting-state baseline established when the target user is at rest with their eyes closed as a reference. Each airbag zone is activated individually according to a preset duration, and EEG signals are continuously collected during the activation of each airbag zone. The relaxation score corresponding to that zone is evaluated based on the resting-state baseline, and the weighting coefficient and optimal pressure for that zone are determined. Subsequently, the heating element is controlled to perform gradient heating within a preset temperature range, maintaining the temperature at each point for a preset duration while collecting EEG signals. The relaxation score is evaluated based on the resting-state baseline, the optimal temperature is locked, and temperature constraints are determined. Simultaneously, audio is played according to the audio configuration, and EEG signals are collected during playback. The mental state category is determined based on the EEG signals, and the audio weights in the audio configuration are adjusted. The audio volume in the audio configuration is adjusted based on the amplitude change trend of the EEG signals. The partition identifier, partition weight, optimal pressure, optimal temperature, temperature constraint, and adjusted audio configuration corresponding to each airbag partition are associated and stored in the comfort zone mapping database as the comfort zone mapping data of the target user in the current physiological state.

[0063] By responding to exploration commands, trial sampling is performed under the target user's current physiological state to construct comfort zone mapping data. This ensures that the constructed comfort zone mapping data reflects the target user's actual relaxation sensitivity and massage preferences before the current massage, rather than using historical data or default parameters. Since the target user's physiological state changes with factors such as fatigue levels, emotional fluctuations, and daily routines, the optimal combination of massage parameters may differ for the same target user at different times. By re-performing trial sampling before each massage, the individualized needs of the target user in their current state can be captured, thus avoiding the problem of historically optimal parameters becoming inapplicable due to state changes. Therefore, constructing comfort zone mapping data for the target user's current state at each use provides an accurate data foundation for subsequently generating massage strategies adapted to the current usage state.

[0064] S42: Responding to massage commands, a massage strategy is generated based on comfort zone mapping data. An adaptive control method driven by EEG closed-loop feedback is adopted. The eye massager is controlled to perform massage operations based on the resting-state baseline and the massage strategy. During the massage operation, real-time EEG signals are continuously collected based on the resting-state baseline to monitor the target user. The massage strategy is dynamically adjusted based on the monitoring results to achieve adaptive control driven by EEG closed-loop feedback.

[0065] Massage commands are control signals used to trigger the eye massager to switch from the exploration phase to the formal massage execution phase. Massage commands can originate from control signals automatically generated by the system after completing trial sampling and comfort zone mapping data construction, confirmation commands issued by the target user via device buttons or touch panel after the exploration phase is completed, or confirmation commands issued by the target user via device buttons or touch panel after the eye massager is turned on.

[0066] A massage strategy refers to a set of instructions generated based on the target user's comfort zone mapping data, used to control the various actuators of the eye massager to work collaboratively according to a specific timing and parameters during massage operation. The massage strategy includes the activation sequence, inflation pressure, and holding time of each airbag zone; the target temperature and temperature maintenance range of the heating element; and the audio identifier, audio volume, and audio weight of the audio output module.

[0067] Specifically, upon responding to a massage command, the massage duration and activation frequency of each airbag zone are determined based on the zone weights in the comfort zone mapping data—zone weights are directly proportional to massage duration and activation frequency—and combined with zone identifiers, optimal pressure, optimal temperature, temperature constraints, and adjusted audio configurations from the comfort zone mapping data, a massage strategy is generated that includes zone timing, pressure, temperature, audio, and volume. Then, the eye massager is controlled to perform the massage operation based on the resting-state baseline and the massage strategy. During the massage operation, real-time EEG signals of the target user are continuously collected at a preset sampling rate and digitized. The target user is monitored based on the resting-state baseline and the collected EEG signals. The massage strategy is dynamically adjusted based on the monitoring results. After each adjustment, the eye massager is controlled to perform the massage operation based on the resting-state baseline and the adjusted massage strategy, thereby achieving adaptive control driven by EEG closed-loop feedback.

[0068] Optionally, cumulative fatigue protection can also be implemented during the massage procedure: (1) After the actual massage duration of a single session exceeds the first massage duration (optional, 20 minutes), for each additional second massage duration (optional, 5 minutes), the overall inflation pressure of all airbag zones will be automatically reduced by 5%; (2) If, during the third consecutive massage duration (optional, 15 minutes), a linear regression is performed with time as the independent variable and the power spectral density of the α wave as the dependent variable, and the regression slope is negative and the decrease in the power spectral density of the α wave exceeds a specified proportion (optional, 10%), then it is determined that the user is experiencing massage fatigue, and the system automatically switches to rest mode—pausing the inflation and deflation of all airbag zones for a rest duration (optional, 5 minutes), retaining only the lowest heat setting and extremely low volume audio; during rest mode, the timer corresponding to the third massage duration is paused.

[0069] This embodiment, by responding to exploration commands to perform trial sampling and construct comfort zone mapping data, can systematically assess the actual impact of different massage parameters on the target user's relaxation level before the formal massage begins. This results in personalized parameter configurations specific to the target user's current state, avoiding adaptation biases caused by relying on fixed programs or general templates. Based on this, a massage strategy is generated and executed according to the comfort zone mapping data in response to massage commands. This ensures that the parameters of each dimension of the massage strategy originate from the target user's own physiological feedback data, rather than factory-preset fixed patterns. During the massage operation, real-time EEG signals are continuously collected and monitored using a resting-state baseline as a personalized reference. The massage strategy is dynamically adjusted based on the monitoring results, allowing the eye massager to maintain real-time tracking of the target user's state changes throughout the massage. Thus, through the phased design of the exploration and execution phases and the continuous implementation of a closed-loop control cycle, it is ensured that the personalized basis of the massage strategy originates from the target user's own physiological response data, while also ensuring the timeliness and relevance of strategy adjustments during the massage process. This allows the massage operation to always revolve around the target user's actual comfort zone.

[0070] In one embodiment, the response to the exploration command, based on the resting-state baseline and audio configuration, performs trial sampling for the target user and constructs comfort zone mapping data, including: S411: In response to the exploration command, activate each airbag section individually and sequentially based on the preset maintenance duration, and execute steps S412 to S416 when activating each airbag section.

[0071] The preset duration of maintenance refers to the length of time each air sac segment remains inflated during individual activation. The data for the preset duration of maintenance is sourced from pre-stored system default parameters, which are used to ensure that each air sac segment has sufficient time to elicit a measurable EEG response during activation, while avoiding discomfort to the target user due to excessive stimulation time.

[0072] Specifically, upon responding to the exploration command, each airbag section is activated individually in a preset order. For the currently activated airbag section, a drive signal is sent to the drive port of the micro air pump via the PWM control line, and the corresponding solenoid valve of that airbag section is opened to inflate it independently, while the solenoid valves of the other airbag sections remain closed. The airbag section remains inflated until the preset duration is reached, after which the solenoid valve is opened to deflate the airbag section. An interval is set between the activation of each airbag section, during which all airbags are in a deflated state to allow the periocular tissue to return to its baseline state. Steps S412 to S415 are executed simultaneously each time an airbag section is activated, until all airbag sections have been explored, at which point step S416 is executed.

[0073] S412: Acquire EEG signals, assess relaxation scores based on resting baseline, determine zone weights and optimal stress; The relaxation score is a quantitative index calculated based on the degree of change in the target user's electroencephalogram (EEG) signal relative to the resting baseline under stimulation by specific massage parameters. It is used to characterize the promoting effect of the specific massage parameters on the target user's relaxation state.

[0074] Optionally, the relaxation score S_i = h1 + h2 × tanh(Δα × h3). h1 is the baseline score, used to determine the center position of the score, ensuring the score is distributed within a preset range. h2 is the scaling factor, used to control the magnitude of the score change as Δα is mapped, and together with h1, determines the upper and lower limits of the score. h3 is the adjustment factor, used to control the sensitivity of the tanh function input signal, i.e., amplifying or reducing the influence of Δα on the score. tanh is the hyperbolic tangent function, used to map Δα to the (-1, 1) interval, making the relaxation score change smoother and avoiding drastic jumps in the relaxation score due to extreme values ​​of Δα. Δα is the rate of change of the power spectral density of the α wave, used to quantify the influence of massage stimulation on the target user's relaxation state. Δα = (PSD_stimulation_interval) / (Δα + h2 × tanh(Δα × h3)). PSD_baseline) / PSD_baseline, where PSD_baseline is the average power spectral density of the alpha band in the resting baseline, and PSD_stimulation is the average power spectral density of the alpha band in the EEG signal collected during the period when the air sac area is in an activated state within a preset time window.

[0075] Optionally, h1 is set to 50 to determine the center position of the score; h2 is set to 50 to control the range of change of the score as it maps with Δα, so that the relaxation score is distributed in the range of 0 to 100; and h3 is set to 3 to control the sensitivity of the input signal of the tanh function.

[0076] Optionally, h3 = 1 / σ_α, where σ_α is the standard deviation of the power spectral density of the α wave in the resting baseline. This setting normalizes the variation of Δα to the standard deviation of the individual's own fluctuations, preventing users with naturally large fluctuations in α waves from entering the scoring saturation zone too early, and also preventing users with naturally stable α waves from being scored too slowly.

[0077] Zone weight is a quantitative value used to characterize the contribution of each airbag zone to the target user's relaxation level. Zone weight is calculated based on the relaxation level scores obtained by each airbag zone during the trial process.

[0078] Optimal pressure refers to the airbag inflation pressure that enables the target user to achieve the highest level of relaxation for each airbag section.

[0079] Specifically, during the active period of the current airbag zone, the target user's EEG signals are continuously acquired and digitized. The digitized EEG signals acquired during activation undergo the same preprocessing and Fast Fourier Transform (FFT) operations as in the resting-state baseline construction phase. The power spectral density of each frequency band is calculated, and the rate of change of the alpha wave power spectral density relative to the average power spectral density of the resting-state alpha baseline is calculated using the average power spectral density of the alpha wave as a reference. This rate of change is mapped to a relaxation score. Based on this relaxation score, a weighting algorithm is used to calculate the weight coefficient of the zone—the higher the relaxation score, the larger the weight coefficient of the zone. Subsequently, independent activation is performed on the airbag zone at multiple preset pressure gradients. EEG signals are acquired and the corresponding relaxation score is calculated under each pressure gradient. A quadratic polynomial fitting method is used to perform curve fitting on the data points (including pressure and relaxation score) under each pressure gradient. The pressure corresponding to the highest point in the fitted curve is determined as the optimal pressure for that zone.

[0080] W_i = W_i_old × k1 + (S_i / 100) × k2, where S_i is the relaxation score calculated in step S412, W_i_old is the historical partition weight of the airbag partition before this partition weight update, and W_i is the updated partition weight. The denominator 100 is a preset score normalization factor used to normalize the relaxation score (ranging from 0 to 100) to the 0-1 range, ensuring that the current response item (i.e., (S_i / 100) × k2) and the historical weight item (i.e., W_i_old × k1) are on the same order of magnitude.

[0081] The historical weighting factor is used to prevent drastic changes in weighting due to a single, accidental physiological fluctuation, ensuring the continuity and stability of the massage strategy. The current response factor allows the eye massager to capture the target user's current physiological state changes in real time (such as being particularly tired or stressed today), thereby allocating massage resources to the most sensitive and effective areas at that moment.

[0082] Understandably, when the partition weights are determined for the first time, the Wi_old of all airbag partitions takes the preset initial value; in subsequent updates, Wi_old takes the value of Wi after the last update.

[0083] Optionally, the default initial value is set to 0.5.

[0084] k1 is the historical weight retention coefficient, which controls the proportion of the historical weight before the airbag partition update in this update. The larger k1 is, the stronger the influence of the historical weight.

[0085] k2 is the current response coefficient, which controls the proportion of the relaxation score obtained in this calculation in this update. The larger the k2 is, the stronger the influence of the current measured feedback. The sum of k1 and k2 is 1.

[0086] S413: Perform gradient heating within a preset temperature range, collect EEG signals, assess relaxation score based on resting baseline, lock the optimal temperature, and calculate the left and right descent rates based on the optimal temperature within a preset temperature range. The optimal temperature refers to the heat therapy temperature within a preset temperature range that allows the target user to achieve the highest level of relaxation.

[0087] The left-side rate of decline refers to the average rate of change of the relaxation score as the temperature decreases within a preset temperature range (e.g., from the beginning to the end of the preset temperature range) to the left of the optimal temperature on the temperature comfort curve. The right-side rate of decline refers to the average rate of change of the relaxation score as the temperature increases within a preset temperature range to the right of the optimal temperature on the temperature comfort curve. Both the left-side and right-side rates of decline are average slopes within this preset temperature range, not instantaneous rates of change at a single temperature point.

[0088] The left-side rate of decline refers to the rate at which the relaxation score decreases as the temperature decreases to the left of the optimal temperature (i.e., the region below the optimal temperature) on the temperature comfort curve. The right-side rate of decline refers to the rate at which the relaxation score decreases as the temperature increases to the right of the optimal temperature (i.e., the region above the optimal temperature) on the temperature comfort curve. The data for both the left-side and right-side rates of decline are obtained by the main control module performing curve fitting based on the relaxation scores corresponding to each temperature point, and then calculating the absolute values ​​of the slopes of the curves on both sides of the optimal temperature.

[0089] Specifically, within a preset temperature range, the heating element is controlled to gradually increase the temperature in a gradient manner, maintaining each temperature point for a specific duration. During the maintenance of each temperature point, the target user's electroencephalogram (EEG) signals are continuously acquired and digitized. The relaxation score corresponding to each temperature point is calculated based on the resting-state baseline using the digitized EEG signals acquired during the maintenance of each temperature point. After testing all temperature points, a curve fitting is performed on each data point with temperature as the x-axis and relaxation score as the y-axis to generate a temperature comfort curve. The temperature corresponding to the highest point on the curve is locked as the optimal temperature for the target user. Subsequently, the average slope of the temperature comfort curve within the preset temperature range to the left and right of the optimal temperature is calculated, serving as the left-side descent rate and the right-side descent rate, respectively.

[0090] S414: If the descent rate on the left is greater than the descent rate on the right, then the optimal temperature is subtracted from the first temperature to obtain the lower limit of the temperature constraint; otherwise, the optimal temperature is added to the first temperature to obtain the upper limit of the temperature constraint. Temperature constraints refer to the boundary values ​​used to limit the adjustable range of heat application temperature during subsequent massage operations. Temperature constraints include a lower limit and an upper limit. The lower limit is used to prevent the temperature from being too low, resulting in insufficient heat application effect, while the upper limit is used to prevent the temperature from being too high, which may cause burns or discomfort to the target user.

[0091] The first temperature refers to the preset temperature offset used to calculate the boundary values ​​of temperature constraints. The data source for the first temperature is pre-stored system default parameters, which are used to generate a safe temperature regulation boundary based on the optimal temperature.

[0092] Specifically, the left-side descent rate calculated in S413 is compared with the right-side descent rate. If the left-side descent rate is greater than the right-side descent rate, it indicates that the target user is more sensitive to low temperatures—relaxation decreases faster when the temperature is below the optimal temperature. Therefore, the optimal temperature is subtracted from the first temperature to obtain the lower limit of the temperature constraint, prioritizing the avoidance of excessively low temperatures leading to a significant decrease in relaxation. If the left-side descent rate is not greater than the right-side descent rate, it indicates that the target user is more sensitive to high temperatures or that the sensitivity on both sides is equal. Therefore, the optimal temperature is added to the first temperature to obtain the upper limit of the temperature constraint, prioritizing the avoidance of excessively high temperatures causing discomfort. The determined lower or upper limit value is stored as the temperature constraint for the target user in the comfort zone mapping data.

[0093] Optionally, if the descent rate on the left is greater than the descent rate on the right, then the optimal temperature is subtracted from the first temperature to obtain the lower limit of the temperature constraint, and the optimal temperature is added to the fifth temperature to obtain the upper limit of the temperature constraint; otherwise, the optimal temperature is added to the first temperature to obtain the upper limit of the temperature constraint, and the optimal temperature is subtracted from the fifth temperature to obtain the lower limit of the temperature constraint, wherein the fifth temperature is less than the first temperature.

[0094] The fifth temperature refers to the preset temperature offset used to calculate the boundary values ​​of temperature constraints. The data source for the fifth temperature is pre-stored system default parameters, which are used to generate a safe temperature regulation boundary based on the optimal temperature.

[0095] Optionally, if the descent rate on the left is greater than the descent rate on the right, then the optimal temperature minus the first temperature is used as the lower limit of the temperature constraint, and the default upper limit temperature is used as the upper limit of the temperature constraint; otherwise, the optimal temperature plus the first temperature is used as the upper limit of the temperature constraint, and the default lower limit temperature is used as the lower limit of the temperature constraint.

[0096] S415: Play audio based on audio configuration, collect EEG signals, determine mental state category and adjust audio weights in audio configuration, determine amplitude change trend and adjust audio volume in audio configuration; Specifically, according to the audio configuration determined in S3, the audio file corresponding to the audio configuration is played at the preset audio volume. During audio playback, the target user's EEG signals are continuously acquired and digitized. The average power spectral density of each frequency band is calculated for the digitized EEG signals acquired during playback, and the target user's current mental state category is determined based on the ratio of the average power spectral density of beta waves to that of alpha waves. The audio weight distribution in the audio configuration is adjusted according to the determined mental state category—increasing the audio weight of audio genres with a high degree of matching with the current mental state, and decreasing the audio weight of genres with a low degree of matching. Simultaneously, the trend of alpha wave amplitude changes within a specified time window is monitored: if the alpha wave amplitude shows a continuous upward trend, the audio volume in the audio configuration is decreased; if the alpha wave amplitude shows a continuous downward trend or beta waves show a sudden increase, the audio volume in the audio configuration is increased. The adjusted audio weights and audio volumes are updated in the audio configuration.

[0097] Alpha wave amplitude refers to the peak value of the alpha wave band in the time domain of an electroencephalogram (EEG), reflecting the intensity of the alpha wave rhythm. A larger alpha wave amplitude indicates stronger alpha wave activity, typically corresponding to a deeper state of relaxation. The relationship between alpha wave amplitude and alpha wave power spectral density is as follows: alpha wave power spectral density is the distribution of the square of the alpha wave amplitude in the frequency domain; they are positively correlated but have different physical meanings—alpha wave amplitude reflects the instantaneous intensity of the time-domain waveform, while alpha wave power spectral density reflects the concentration of energy in the frequency domain.

[0098] Optionally, the adjustment of audio weights based on mental state categories specifically involves: when the dominant frequency of the target user's EEG signal is alpha wave dominant, audio in the alpha wave-induced music category of the preset soothing audio library is selected first; when the dominant frequency of the target user's EEG signal is theta wave dominant, audio in the theta wave-induced music category of the preset soothing audio library is selected first, thereby achieving frequency matching between audio therapy and the dominant frequency of the target user's current EEG signal.

[0099] Understandably, this priority selection is achieved by dynamically adjusting the audio weights in the audio configuration. The specific logic is as follows: After the main control module detects a dominant frequency characteristic (such as alpha wave dominance), it multiplies the preset audio weight of the corresponding audio genre (such as alpha wave-induced music) by a coefficient greater than 1 (e.g., increasing it to 1.5 times), while proportionally reducing the weights of other genres. Based on the updated weights, the audio output module performs a weighted random selection during the next audio switching, significantly increasing the probability of selecting audio with higher weights. For example, if the weight of alpha wave music is increased from 0.3 to 0.6, and others are reduced to 0.2, the probability of playing alpha wave music next time reaches 60%, thus achieving "priority."

[0100] S416: Associate the zone identifier, zone weight, optimal pressure, optimal temperature, temperature constraint, and adjusted audio configuration corresponding to the airbag zone, and use them as the mapping data corresponding to the airbag zone in the comfort zone mapping data.

[0101] A zone identifier is a number or mark used to uniquely identify each airbag zone. The data for the zone identifier comes from the preset numbers assigned to each airbag zone at the factory and is stored in the memory of the main control module.

[0102] Specifically, the partition identifier of the current airbag partition, the partition weight and optimal pressure determined in S412, the optimal temperature locked in S413, the temperature constraints determined in S414, and the adjusted audio configuration in S415 are associated and combined to form a mapping data entry corresponding to the airbag partition. After all airbag partitions have completed the processes from S411 to S415, the mapping data entries corresponding to each airbag partition are summarized as the complete comfort zone mapping data of the target user in the current physiological state and stored in the comfort zone mapping database.

[0103] It should be noted that the optimal pressure for each airbag zone, determined during the sequential and individual activation of each zone, is a baseline value obtained under the condition of eliminating the traction interference caused by the simultaneous inflation of adjacent zones. In actual massage, when multiple airbag zones are activated alternately or sequentially, adjacent airbag zones may experience mutual traction due to different inflation states, thus affecting the actual contact pressure of a single airbag zone and the subjective experience of the target user. To eliminate this effect, when generating a massage strategy that includes zone timing and pressure, the actual inflation pressure of the currently activated airbag zone is dynamically compensated based on the degree of overlap in the activation timing of adjacent airbag zones.

[0104] Dynamic compensation is as follows: when adjacent airbag zones are simultaneously inflated, the target pressure of the current airbag zone is reduced by a preset first compensation coefficient; when adjacent airbag zones are deflated, the target pressure of the current airbag zone is restored to the optimal pressure. The first compensation coefficient is stored in the comfort zone mapping database and is associated with the optimal pressure of each airbag zone.

[0105] The first compensation coefficient ranges from 0.05 to 0.15. When adjacent airbag sections are simultaneously inflated, the target pressure of the currently activated airbag section is multiplied by (1 - the first compensation coefficient) to obtain the actual inflation pressure; when adjacent airbag sections are deflated, the actual inflation pressure of the currently activated airbag section is restored to its optimal pressure.

[0106] This embodiment, by responding to exploration commands and sequentially activating each airbag zone individually while simultaneously acquiring EEG signals, can accurately assess the independent contribution of massage at each spatial location to the target user's relaxation level, while eliminating interactive interference caused by simultaneous stimulation of multiple airbag zones. This allows for precise attribution of massage effects. Furthermore, by probing multiple pressure gradients on each airbag zone and performing curve fitting based on relaxation scores to determine the optimal pressure, the optimal stimulation intensity for each airbag zone can be precisely locked with fewer trials, avoiding the time consumption of exhaustive searching across a wide range. Simultaneously, by performing gradient heating within a preset temperature range and determining the optimal temperature and temperature constraints based on a temperature comfort curve, the optimal heat therapy temperature for the target user is locked, and differentiated temperature adjustment boundaries are established based on the target user's sensitivity to low or high temperatures, ensuring the safety of temperature adjustment during subsequent massage. By synchronously playing audio and dynamically adjusting the audio configuration based on EEG signals, real-time matching of audio content with the target user's current mental state is achieved. Therefore, the optimal parameters of each dimension are correlated and stored in a comfort zone mapping database, providing a complete and accurate data foundation for generating personalized massage strategies.

[0107] In one embodiment, the response to a massage command, generating a massage strategy based on comfort zone mapping data, employing an EEG closed-loop feedback-driven adaptive control method, and controlling the eye massager to perform massage operations based on the resting-state baseline and the massage strategy, includes: S421: Responds to massage commands, generates a massage strategy that includes zone timing, pressure, temperature, audio and volume based on comfort zone mapping data, and adopts an adaptive control method driven by EEG closed-loop feedback to control the eye massager to perform massage operations based on the resting baseline and massage strategy. The zonal timing refers to the time arrangement in which each airbag zone is activated sequentially according to a specific order and time interval during the massage process, including the activation order of each zone, the inflation start time, the pressure holding duration, and the deflation start time.

[0108] Specifically, upon responding to a massage command, the system reads the comfort zone mapping data constructed in this cycle from the comfort zone mapping database. Based on the weights of each zone in the comfort zone mapping data, and following the principle that zone weights are proportional to massage duration and activation frequency, each airbag zone is assigned a number of activations and a single activation duration within a complete massage cycle. Zones with higher weights are assigned more activations and longer single activation durations. Based on the activation counts and single activation durations of each airbag zone, and combined with the zone identifier, optimal pressure, optimal temperature, temperature constraints, and adjusted audio configuration in the comfort zone mapping data, a massage strategy is generated sequentially along the timeline. This strategy includes parameters such as the activation sequence of each zone, the inflation pressure of each zone, the target temperature and temperature maintenance range of the heating element, the audio identifier, and the audio volume. The activation sequence and inflation pressure of each zone in the massage strategy are converted into a PWM control command sequence and loaded into the execution queue of the PWM driver. Simultaneously, the activation sequence and temperature parameters (target temperature and temperature maintenance range of the heating element) are sent as control commands to the power adjustment port of the heating element, and the audio parameters (audio identifier and audio volume) are sent as control commands to the audio output module. Using the average power spectral density and standard deviation of each frequency band stored in the resting-state baseline as a reference, an adaptive control method driven by EEG closed-loop feedback is adopted to control the eye massager to perform massage operations according to the massage strategy.

[0109] S422: In response to the end command, save the updated comfort zone mapping data during this massage operation into the target user's target user profile; An end command is a control signal used to trigger the eye massager to terminate the current massage operation and enter the end process. The source of the end command includes a signal that is automatically triggered after the internal timer of the program file implementing this application reaches the preset massage duration, or a stop signal actively issued by the target user through the device buttons or touch panel on the eye massager.

[0110] The target user profile refers to a structured storage space used to store the comfort zone mapping data of the target user. The comfort zone mapping data stored in the target user profile can serve as an important reference during the initialization phase in subsequent use: when the same target user uses the eye massager again, if the user has previously set to skip the acquisition of comfort zone mapping data and issues a command to skip the acquisition of comfort zone mapping data through the operation button, the main control module will first read the comfort zone mapping data in the target user profile as the comfort zone mapping data for this massage.

[0111] A target user profile refers to a structured storage space used to store the comfort zone mapping data of target users. The comfort zone mapping data stored in the target user profile can serve as an important reference in the initialization phase during subsequent use: when the same target user uses the eye massager again, the comfort zone mapping data in the target user profile is read first as the initial parameters for the current trial sampling. Based on this, a rapid verification sampling is performed within a shortened time window—only a small number of pressure gradient and temperature points are sampled for verification in the region of the partition weight in the initial parameters to quickly confirm or fine-tune the initial parameters. The confirmed or fine-tuned initial parameters are used as the comfort zone mapping data for the current time, thereby shortening the time required to build the comfort zone mapping data for the current time.

[0112] Optionally, the target user profile may also be used to store the target user's historical massage data.

[0113] Specifically, the end process is triggered when the massage duration reaches a preset value and an end command is actively triggered, or when the target user actively presses the stop button. First, a closing control sequence is executed: all airbag zones are completely deflated, the heating element temperature is reduced to a safe level, and the audio volume is gradually reduced to zero. While the closing control sequence is being executed, the background system collects the final comfort zone mapping data and various data from the entire massage process (i.e., historical massage data) and stores them in the target user's profile. This historical massage data includes the actual number of activations and average relaxation scores for each airbag zone, the actual fluctuation range of the heat therapy temperature and the corresponding average relaxation level, the audio playback list and the average relaxation level for each audio track, whether the soothing mode was triggered and the number of triggers, and whether the sleep protection mode was entered and the duration of the protection.

[0114] During the massage procedure, real-time electroencephalogram (EEG) signals are continuously collected for monitoring, and one or more of the following control operations are implemented based on the monitored data: Operation (1): When the relaxation level is detected to decrease transiently to the stress response condition based on the resting baseline, the eye massager is controlled to switch to the soothing mode and when the relaxation level is detected to recover to the recovery condition, the eye massager is controlled to resume the massage operation according to the resting baseline and massage strategy. Stress response conditions refer to the criteria used to determine whether a target user experiences a sudden unpleasant reaction due to external stimuli or internal factors.

[0115] The Soothing Mode is a low-intervention operating state that is temporarily switched to when a stress response is detected in the target user. In Soothing Mode, inflation commands for all airbag compartments are terminated and rapid deflation is performed, heating power is reduced, audio volume is lowered, and a preset soft audio tone is switched. The data source for Soothing Mode is a set of pre-configured operating parameters that are automatically triggered when stress response conditions are detected.

[0116] Recovery criteria refer to the standards used to determine whether a target user has recovered from stress and can exit the soothing mode. Recovery criteria are determined based on the degree of alpha wave power recovery continuously monitored during the execution of the soothing mode.

[0117] Operation (2): When the target user is detected to be in the early stage of light sleep, control the eye massager to switch to sleep protection mode. When the target user is detected to be in deep sleep in sleep protection mode, control the eye massager to switch to deep sleep care mode. When the target user is detected to be waking up in deep sleep care mode, control the eye massager to switch to low intensity wake-up mode. Pre-light sleep refers to the initial sleep stage in which the human body transitions from a waking state to a sleep state, also known as N1 stage sleep.

[0118] Sleep Guardian Mode refers to a low-intervention operating state that is automatically switched to when the target user is detected to have entered the early stage of light sleep. The data source for Sleep Guardian Mode is a set of pre-configured operating parameters that are automatically triggered after the target user is determined to have entered the early stage of light sleep.

[0119] Deep sleep refers to N3 stage sleep in the human sleep cycle.

[0120] Deep sleep care mode refers to an operational state where the intensity of intervention is further reduced after the target user is detected to have entered deep sleep. The data source for deep sleep care mode is a set of pre-configured operating parameters that are automatically triggered after the target user is determined to have entered deep sleep.

[0121] Awakening trend refers to the physiological signal changes that indicate a target user's transition from a sleep state to a wakeful state.

[0122] Low-intensity wake-up mode refers to a gentle wake-up operation state that is switched to when a target user is detected to be waking up. The data source for low-intensity wake-up mode is a set of pre-configured operating parameters that are automatically triggered after the target user is determined to be waking up.

[0123] Operation (3): Based on the resting baseline, evaluate the comprehensive relaxation value of each airbag zone in the first cycle, and dynamically adjust the zone weight in the comfort zone mapping data according to the comprehensive relaxation value, and update the massage strategy according to the updated comfort zone mapping data.

[0124] The first cycle refers to the time interval set for periodically evaluating the overall relaxation level of each airbag zone. The duration of the first cycle is consistent with the duration of a complete airbag activation sequence, ensuring that each airbag zone is activated at least once during each evaluation. The data for the first cycle is sourced from the cycle preset in the program documentation of this application.

[0125] The overall relaxation rating is a quantitative indicator used to characterize the overall contribution of each airbag zone to the target user's relaxation level in the current massage cycle.

[0126] This embodiment generates a massage strategy based on comfort zone mapping data in response to massage commands. This strategy includes zonal timing, pressure, temperature, audio, and volume, ensuring that all parameters of the massage strategy originate from the target user's own physiological feedback data rather than factory-preset fixed patterns. By continuously collecting real-time EEG signals during the massage operation and monitoring them using a resting-state baseline as a personalized reference, the massage strategy is dynamically adjusted based on the monitoring results. This allows the eye massager to maintain real-time tracking of the target user's state changes throughout the massage. Furthermore, during the massage operation, through the coordinated operation of (1), (2), and (3), this solution achieves differentiated adaptive control under different target user states. Operation (1) uses beta wave surge detection and soothing mode switching to identify and respond to the target user's discomfort (i.e., stress response) in a short time, thereby quickly reducing the intervention intensity to a minimum; Operation (2) uses sleep state recognition and graded sleep protection mode switching to ensure that the target user will not be awakened by continuous physical stimulation after falling asleep during the massage; Operation (3) uses periodic evaluation of the comprehensive relaxation evaluation value of each airbag zone and dynamic adjustment of zone weights to ensure that the allocation of massage resources can be continuously optimized as the target user's physiological state slowly drifts. Thus, through the phased design of the exploration phase and the execution phase and the full-process closed-loop control cycle, it is ensured that the personalized basis of the massage strategy comes from the target user's own physiological response data, and that the timeliness and pertinence of the strategy adjustment during the massage process are ensured, so that the massage operation can always run around the actual comfort range of the target user.

[0127] In one embodiment, the response to a massage command, generating a massage strategy including zone timing, pressure, temperature, audio, and volume based on comfort zone mapping data, includes: S4211: Determine the massage duration and activation frequency of each airbag zone based on the zone weights in the comfort zone mapping data. The zone weights are directly proportional to both the massage duration and the activation frequency. Massage duration refers to the total duration for which a single airbag zone is activated and maintained in an inflated, compressed state during a complete massage cycle. The data for massage duration is derived from time-series parameters calculated based on zone weights in a proportional relationship.

[0128] Activation frequency refers to the number of times a single airbag compartment is activated per unit of time. The data source for activation frequency is a time-series parameter calculated based on the compartment weights in a proportional relationship.

[0129] The positive correlation refers to the positive correlation between the partition weight and the massage duration and activation frequency. That is, the larger the partition weight value, the longer the corresponding massage duration and the higher the activation frequency.

[0130] Specifically, the weights of each airbag zone are read from the comfort zone mapping data. Based on the relative proportions of these weights, the total duration of a complete massage cycle is allocated to each airbag zone according to their respective weights, with higher weights resulting in longer massage durations. Simultaneously, the number of activations per unit time for each zone is determined based on their relative weights, with higher weights resulting in higher activation frequencies. The massage duration and activation frequency of each airbag zone are then used as the basis for generating the subsequent zone timing sequence and stored in a temporary strategy data area in memory.

[0131] S4212: Based on the massage duration and activation frequency of each airbag zone, as well as the zone identifier, optimal pressure, optimal temperature, temperature constraints, and adjusted audio configuration in the comfort zone mapping data, generate a massage strategy that includes zone timing, pressure, temperature, audio, and volume.

[0132] The partition identifier, optimal pressure, optimal temperature, temperature constraint, and adjusted audio configuration of each airbag partition are read from the comfort zone mapping data. Based on the massage duration and activation frequency of each airbag partition determined in step S4211, a correspondence between airbag partitions and parameters is established by combining the partition identifier of each airbag partition. The activation time, duration, and interval of each airbag partition are arranged into partition timing sequences according to the time axis. The optimal pressure of each airbag partition is associated with the corresponding partition activation period as the pressure control parameter for that period. The optimal temperature and temperature constraint are used as the temperature control parameters of the heating element throughout the entire massage cycle. The audio identifier, audio volume, and audio weight in the adjusted audio configuration are used as the audio control parameters of the audio output module. The above partition timing, pressure control parameters, temperature control parameters, and audio control parameters are integrated into a structured massage strategy and stored in memory for execution by the PWM driver and the audio output module.

[0133] This embodiment determines the massage duration and activation frequency of each airbag zone based on zone weights, thus preventing a uniform distribution of massage resources. Instead, it differentiates the allocation according to the quantitative results of the target user's physiological feedback, concentrating more massage energy on the most sensitive and responsive areas while reducing ineffective stimulation of less sensitive areas. By integrating the massage duration and activation frequency of each airbag zone with zone identifiers, optimal pressure, optimal temperature, temperature constraints, and adjusted audio configurations, a complete massage strategy is generated. This ensures that all parameters of the massage strategy originate from the target user's own physiological feedback data, rather than factory-preset fixed patterns. Consequently, the target user achieves a higher level of relaxation.

[0134] In one embodiment, operation (1) is: S4231: Using the second cycle, transient change data is calculated based on the collected EEG signals. The transient change data includes the β wave power, the ratio of α wave to β wave power, and the periodic change rate of the total power. The second cycle refers to the time interval set for monitoring transient changes in EEG signals during the massage process. The data for the second cycle comes from the preset trigger cycle of the timer interrupt service routine inside the main control module, which is used to repeatedly calculate transient change data at fixed time intervals.

[0135] Transient change data refers to a set of quantitative indicators used to characterize the degree of change in EEG signals between cycles corresponding to the second cycle. Transient change data includes the periodic rate of change of beta wave power, the periodic rate of change of the ratio of alpha wave to beta wave power, and the periodic rate of change of total power.

[0136] Specifically, with a second cycle as the time interval, the digitized EEG signal acquired within that cycle is acquired at the beginning of each cycle. A Fast Fourier Transform is performed on the EEG signal to calculate the power spectral density of each frequency band. The power spectral density stored from the previous cycle is read from memory, and the rate of change of the current cycle's β-wave power spectral density relative to the previous cycle's β-wave power spectral density is calculated (as the periodic rate of change of β-wave power), the rate of change of the ratio of the current cycle's α-wave power spectral density to the β-wave power spectral density relative to the previous cycle's ratio (as the periodic rate of change of the α-wave to β-wave power ratio), and the rate of change of the sum of the power spectral densities of each frequency band in the current cycle (called total power) relative to the total power of the previous cycle (as the periodic rate of change of the total power for each band).

[0137] S4232: When a transient decrease in relaxation level is detected based on transient change data and resting baseline to reach the stress response condition, the eye massager is controlled to immediately terminate all airbag inflation commands and perform rapid deflation, reduce heating power by a first ratio, reduce audio volume by a second ratio, and switch to preset soft audio to achieve soothing mode operation. Rapid deflation refers to the operation of expelling gas from the airbag compartment at a faster rate than normal deflation. The data for rapid deflation comes from the main control module, which controls the solenoid valve to achieve a greater opening to accelerate deflation.

[0138] The first percentage refers to a preset percentage value used to reduce heating power. The data for the first percentage comes from pre-configured system parameters and is used to reduce heating power to a safe level in soothing mode. The first percentage is a value greater than 0 and less than 1.

[0139] The second percentage refers to a preset percentage value used to reduce the audio volume. The data for the second percentage comes from pre-configured system parameters and is used to reduce the audio volume to a low-interference level in soothing mode. The second percentage is a value greater than 0 and less than 1.

[0140] Preset soothing audio refers to pre-stored audio files used in soothing mode, and its audio genre is natural sound, white noise, or other low-stimulation audio types. The data source for preset soothing audio is the device's factory preset.

[0141] Specifically, transient change data are compared with the average power spectral density and standard deviation of each frequency band stored in the resting baseline to monitor whether the transient decrease in relaxation reaches the stress response condition.

[0142] Optionally, when a transient decrease in relaxation level is detected based on transient change data and resting baseline to meet the stress response condition, specifically including: when the following two conditions are met for at least two consecutive second cycles, it is determined that the transient decrease in relaxation level has met the stress response condition: Condition 1: The periodic rate of change of the power spectral density of the β wave shows an upward trend, and its increase exceeds the product of the standard deviation of the power spectral density of the β wave in the resting baseline and the fifth threshold. Condition 2: The periodic rate of change of the ratio of the power spectral density of the α wave to that of the β wave shows a decreasing trend, and the magnitude of the decrease exceeds the product of the standard deviation of the ratio at the resting baseline and the sixth threshold.

[0143] Specifically, in the next control cycle, the current operating state is marked as soothing mode. The control of the eye massager in soothing mode is as follows: an opening command is immediately sent to all solenoid valves via the PWM control line, causing each airbag zone to deflate rapidly and simultaneously; a control signal is sent to the power adjustment port of the heating element via the PWM control line, reducing the current heating power by a first percentage; and a control command is sent to the audio output module via the I2C bus, reducing the current audio volume by a second percentage and switching to a preset soft audio for playback.

[0144] S4233: During the operation period of the soothing mode, continuously monitor the alpha wave power spectral density. If the current alpha wave power spectral density recovers to more than the third proportion of the alpha wave power spectral density before reaching the stress response condition, it is determined that the relaxation level has recovered to the recovery condition. Then, control the eye massager to exit the soothing mode and gradually return to performing massage operations according to the resting baseline and massage strategy.

[0145] The third percentage refers to a preset percentage value used to determine whether the alpha wave power has recovered to the pre-stress level. The data source for the third percentage is pre-configured system parameters used to determine the recovery conditions for exiting the soothing mode.

[0146] The progressive mode refers to a gradual return to normal massage intensity after exiting the soothing mode. This is achieved by progressively increasing the output parameters of each component of the eye massager (e.g., the components within the massage execution module and the audio output module) through multiple steps. The data source for the progressive mode is a preset step-recovery rule, including the duration of each step and the increment of each step.

[0147] During the soothing mode, the system continuously receives digitized EEG signals from the EEG acquisition module in a second cycle. It performs a Fast Fourier Transform on the current cycle's EEG signal and calculates the power spectral density of the alpha wave band. The system reads the alpha wave power spectral density recorded before reaching the stress response condition from memory and calculates the ratio of the current alpha wave power spectral density to the alpha wave power spectral density before reaching the stress response condition. When this ratio exceeds a third proportion, the relaxation level is determined to have recovered to the recovery condition. The eye massager exits the soothing mode and gradually restores the output parameters of each actuator according to a preset step-by-step recovery rule: in each step, airbag inflation, heating, and audio playback are performed at preset proportions corresponding to the parameters in the massage strategy. After the current step continues for a specified duration, the system moves to the next step, gradually increasing to the complete parameter values ​​in the massage strategy, ultimately restoring the massage operation to be performed entirely based on the resting baseline and the massage strategy.

[0148] This embodiment employs a second-cycle continuous calculation of transient change data and determines stress response conditions based on a resting-state baseline. This allows for rapid identification of the target user's emergency response when the combined indicators of a rapid increase in the power spectral density of the beta wave and a rapid decrease in the ratio of the power spectral density of the alpha wave to the beta wave appear, avoiding potential misjudgments that might arise from relying on a single indicator. By immediately terminating all airbag inflation commands and rapidly deflating, reducing heating power and audio volume, and switching to a gentle audio tone upon stress detection, all physical stimuli that could cause discomfort to the target user are eliminated as quickly as possible, helping them recover rapidly from the stress state. By continuously monitoring the alpha wave power spectral density during the soothing mode and exiting the soothing mode when it recovers to above the third ratio, a gradual return to normal massage is achieved. This avoids premature recovery that could cause re-stress in the target user and also prevents sudden increases in intensity from causing discomfort again. Thus, through the synergistic effect of rapid response, comprehensive intervention reduction, and gradual recovery, stress management can be safely and effectively completed when used for sudden emergency responses, and a smooth return to normal massage can be achieved after the target user recovers.

[0149] In one embodiment, operation (2) is: S4241: Using the third cycle, if the proportion of the theta wave power spectral density exceeds the first threshold and the proportion of the alpha wave power spectral density is lower than the second threshold in multiple consecutive third cycles, then the target user is determined to have entered the early stage of light sleep. The third cycle refers to the set time interval for monitoring EEG signals to determine sleep state during the massage process. The data source for the third cycle is the preset trigger cycle of the internal timer interrupt service routine of the device implementing this application, which is used to repeatedly perform sleep state determination at fixed time intervals.

[0150] The theta wave power spectral density ratio refers to the proportion of the power spectral density of the theta wave band in the current EEG signal to the sum of the power spectral densities of the entire frequency band.

[0151] The first threshold is a preset critical value used to determine whether the proportion of theta wave power spectral density reaches the characteristic level of early light sleep. The data source for the first threshold is pre-configured system parameters, which can be adjusted based on the proportion of theta wave power spectral density in the resting baseline of the target user.

[0152] Alpha wave power spectral density ratio refers to the proportion of the power spectral density of the alpha wave band in the current EEG signal to the sum of the power spectral densities of the entire frequency band.

[0153] The second threshold is a preset critical value used to determine whether the proportion of alpha wave power spectral density is lower than the characteristic level of early light sleep. The data source for the second threshold is pre-configured system parameters, which can be adjusted based on the proportion of alpha wave power spectral density in the resting baseline of the target user.

[0154] Specifically, using a three-cycle interval, digitized EEG signals within the current time window are acquired from the EEG acquisition module at the end of each cycle. A Fast Fourier Transform (FFT) is performed on the digitized EEG signals of the current cycle to calculate the power spectral density (PSD) of the theta wave band, the PAD band, and the total PSD across all frequency bands. The percentages of the theta wave PSD and the PAD wave PSD are then calculated. The percentage of the theta wave PSD in the current cycle is compared to a first threshold, and the percentage of the PAD wave PSD is compared to a second threshold. If, for several consecutive third cycles, the percentage of the theta wave PSD exceeds the first threshold and the percentage of the PAD wave PSD is below the second threshold, the target user is determined to have entered the pre-light sleep stage.

[0155] Optionally, during the massage operation, if the mental state category determined in S3 is relaxed or drowsy, and the sleep state determination conditions in steps S4241 to S4244 are met, then the eye massager is controlled to operate according to the corresponding sleep protection mode. If the mental state category determined in S3 is anxious or tense, then the stress response detection in operation (1) is performed first during the massage, and the sleep state determination process is temporarily suspended until the mental state category changes to relaxed or drowsy.

[0156] S4242: Within the first duration, control the eye massager to stop all airbag inflation and deflation operations, reduce the heat therapy temperature to the second temperature at a preset cooling rate, reduce the audio volume to the first volume and switch to pink noise to achieve sleep protection mode operation; The first duration refers to the time required from determining that the target user has entered the pre-light sleep stage to completing the switch to sleep protection mode. The data for the first duration comes from pre-configured system parameters to ensure the smoothness of the mode switching process.

[0157] The preset cooling rate refers to the temperature change per unit time as the heat application temperature drops from the current temperature to the target temperature. The data for the preset cooling rate comes from pre-configured system parameters and is used to control the smoothness of the cooling process.

[0158] The second temperature refers to the target value of the heat therapy temperature in sleep protection mode. The data for the second temperature comes from pre-configured system parameters and is set to a value lower than the normal massage temperature but higher than the ambient temperature.

[0159] The first volume refers to the target value of the audio volume in sleep protection mode. The data source for the first volume is a pre-configured system parameter, set to a value lower than the normal massage volume.

[0160] Pink noise refers to noise signals whose power spectral density is inversely proportional to frequency. It is gentler and more suitable for a sleep environment than white noise. The data source for pink noise is the audio file preset by the device at the factory.

[0161] Specifically, the current operating state is marked as sleep protection mode in the next control cycle. The control of the eye massager in sleep protection mode includes: after determining that the target user has entered the early stage of light sleep, switching to sleep protection mode with a first duration as a transition window; sending a stop command to the micro air pump via the PWM control line and opening all solenoid valves to completely deflate each airbag section and stop subsequent inflation operations; sending a control signal to the power adjustment port of the heating element to gradually reduce the heat therapy temperature to a second temperature at a preset cooling rate; and sending a control command to the audio output module via the I2C bus to reduce the current audio volume to a first volume and switch to a preset pink noise audio file for playback.

[0162] S4243: During the operation of the sleep protection mode, when the proportion of delta wave power spectral density continuously rises above the third threshold based on the collected EEG signals in the fourth cycle, it is determined that the target user has entered deep sleep. The eye massager is then controlled to reduce the heat temperature to the third temperature and the audio volume to the second volume to switch to the deep sleep care mode. The fourth cycle refers to the set time interval for monitoring EEG signals to determine deep sleep in sleep guardian mode. The data source for the fourth cycle is a preset trigger cycle, and its duration may differ from that of the third cycle.

[0163] The delta wave power spectral density ratio refers to the proportion of the power spectral density of the delta wave band in the current EEG signal to the sum of the power spectral densities of the entire frequency band.

[0164] The third threshold is a preset critical value used to determine whether the proportion of delta wave power spectral density reaches the characteristic level of deep sleep. The data source for the third threshold is pre-configured system parameters used to identify EEG characteristics of deep sleep.

[0165] The third temperature refers to the target value of the heat therapy temperature in the deep sleep care mode. The data for the third temperature comes from the pre-configured system parameters in the main control module's memory. The third temperature is set to a value lower than the second temperature.

[0166] The second volume refers to the target audio volume value in deep sleep protection mode. The data source for the second volume is pre-configured system parameters, which are set to a value lower than the first volume.

[0167] Specifically, during the sleep protection mode, EEG signals are continuously acquired at four-cycle intervals. A Fast Fourier Transform (FFT) is performed on the EEG signal of the current cycle to calculate the power spectral density of the delta wave band and the total power spectral density of the entire band, and the percentage of delta wave power spectral density (i.e., the power spectral density of the delta wave band divided by the total power spectral density of the entire band) is calculated. The trend of the percentage of delta wave power spectral density over time is monitored. When the percentage of delta wave power spectral density shows a continuous increase and its value exceeds a third threshold, it is determined that the target user has entered deep sleep. The eye massager is then controlled to lower the heat therapy temperature from the second temperature to the third temperature, and the audio volume is lowered from the first volume to the second volume to switch to deep sleep protection mode.

[0168] S4244: During the operation of the deep sleep care mode, when the proportion of β wave power spectral density is detected to rise above the fourth threshold and the duration exceeds the second duration, it is confirmed that the target user has a waking trend. The eye massager is controlled to play a wake-up prompt sound and the heat compress temperature is raised back to the fourth temperature to achieve a low-intensity wake-up mode.

[0169] The fourth threshold is a preset critical value used to determine whether the proportion of beta wave power spectral density reaches the level characteristic of awakening trend. The data source for the fourth threshold is pre-configured system parameters used to identify EEG characteristics of awakening trend.

[0170] The second duration refers to the shortest duration required for the proportion of beta wave power spectral density to continuously exceed the fourth threshold. The data for the second duration comes from pre-configured system parameters and is used to exclude instantaneous beta wave fluctuations to avoid false alarms of awakening.

[0171] The fourth temperature refers to the target value of the heat therapy temperature in the low-intensity wake-up mode. The data for the fourth temperature comes from pre-configured system parameters and is set to a value that is higher than the third temperature but lower than the normal massage temperature.

[0172] A wake-up tone is an audio signal used to gently alert a user as they transition from sleep to wakefulness during their natural awakening process. The data for wake-up tones comes from audio files pre-installed at the device's factory.

[0173] Specifically, during the deep sleep care mode, digitized EEG signals are continuously acquired and the percentage of beta-wave power spectral density is calculated in the fourth cycle. The percentage of beta-wave power spectral density is monitored to see if it exceeds a fourth threshold. When the percentage of beta-wave power spectral density rises above the fourth threshold and this state (i.e., exceeding the fourth threshold) lasts for more than the second duration, it is determined that the target user is showing signs of awakening. In the next control cycle, the current operating state is marked as low-intensity wake-up mode. The control of the eye massager in low-intensity wake-up mode specifically includes: sending control commands to the audio output module via the I2C bus to play a preset wake-up prompt tone at a volume lower than normal; and simultaneously sending control signals to the power adjustment port of the heating element to raise the heat therapy temperature from the third temperature to the fourth temperature.

[0174] This embodiment employs a third cycle to continuously monitor the proportions of theta wave power spectral density and alpha wave power spectral density, ensuring that conditions are met for multiple consecutive cycles. This provides a temporal continuity constraint for identifying the early stages of light sleep, avoiding misjudgments caused by occasional EEG fluctuations. Upon determining entry into the early stages of light sleep, the airbag compression is stopped, the temperature is gradually lowered, and pink noise is switched to a first-duration transition window, achieving a smooth transition from active intervention mode to passive protection mode and preventing the target user from being awakened by sudden changes in movement. A fourth cycle is used to continuously monitor the proportion of delta wave power spectral density in sleep protection mode. When the proportion continuously rises above the third threshold, the system switches to deep sleep care mode, further reducing the heat therapy temperature and audio volume to minimize interference with deep sleep. In deep sleep care mode, when the proportion of beta wave power spectral density rises above the fourth threshold and lasts for more than the second duration, a wake-up trend is confirmed, and the system switches to low-intensity wake-up mode, avoiding discomfort caused by sudden resumption of strong stimulation during the target user's natural awakening process. Therefore, by monitoring the characteristic EEG signals of different sleep stages and switching to the corresponding differentiated intervention intensity operating mode, the eye massager can dynamically adjust the protection strategy according to the change of sleep depth after the target user falls asleep, providing moderate warmth and acoustic companionship without disturbing sleep, and providing a gentle wake-up transition when the target user wakes up naturally.

[0175] In one embodiment, operation (3) is: S4251: In the first cycle, based on the resting baseline, determine the relaxation score of each air sac area during the current round of activation according to the collected EEG signals, and calculate the relaxation increment of each air sac area compared to the previous round. The relaxation increment refers to the difference between the relaxation score of each airbag zone during the current activation cycle and the relaxation score during the previous activation cycle.

[0176] Specifically, taking the first cycle as the evaluation cycle, after completing a full round of airbag activation sequence within each cycle, the EEG signals stored in each airbag region during the activation period of this round are read. A Fast Fourier Transform is performed on the EEG signals corresponding to each airbag region to calculate the relaxation score for each airbag region, which is used as the relaxation score for each region in the current round. The relaxation scores of each airbag region stored in the previous round are read, and the relaxation scores of the corresponding regions in the previous round are subtracted from the relaxation scores of the current round's regions to obtain the relaxation increment for each airbag region.

[0177] S4252: Calculate the overall relaxation score based on the relaxation score during the current activation period and the relaxation increment compared to the previous round; S4253: Calculate the mean and standard deviation of the overall relaxation evaluation value for all airbag zones, and determine the high threshold and low threshold based on the mean and standard deviation; The mean of the overall relaxation rating is the arithmetic mean of the overall relaxation ratings of all airbag zones.

[0178] The standard deviation of the overall relaxation rating is a statistical measure of the dispersion of the overall relaxation rating of all airbag zones relative to the mean.

[0179] The high threshold is a critical value used to determine whether the overall relaxation level evaluation value is significantly higher than the average level.

[0180] The low threshold is a critical value used to determine whether the overall relaxation level evaluation value is significantly lower than the average level.

[0181] S4254: When the comprehensive relaxation evaluation value of any airbag zone is higher than the high threshold for a consecutive preset number of cycles, the zone weight corresponding to that airbag zone in the comfort zone mapping data is increased according to the first function. S4255: When the overall relaxation evaluation value of any airbag zone is lower than the low threshold for a consecutive preset number of cycles, the zone weight corresponding to that airbag zone in the comfort zone mapping data is reduced according to the second function. S4256: Update the massage strategy based on the updated comfort zone mapping data; in, F(i)=β1×S_i(t)+β2×ΔS_i Th_high = F_mean + c × F_std Th_low = F_mean d×F_std P_i_new1 = min(1.0, P_i_old × m) P_i_new2 = max(g, P_i_old×q) F(i) is the overall relaxation evaluation value of the i-th airbag partition, S_i(t) is the relaxation score of the i-th airbag partition during the current round of activation, ΔS_i is the relaxation increment of the i-th airbag partition compared to the previous round, β1 and β2 are preset weight coefficients, β1>β2, and i is the partition identifier of the airbag partition. Th_high is the high threshold, Th_low is the low threshold, F_mean is the mean, F_std is the standard deviation, and c and d are preset weight coefficients. P_i_new1 is the first function, P_i_new2 is the second function, min calculates the minimum value, max calculates the maximum value, P_i_old is the current partition weight of the airbag partition, m is a constant greater than 1, q is a constant less than 1 and greater than 0, and g is a constant less than 0.5 and greater than 0.

[0182] β1 > β2, to reflect the weighting principle of "current rating as the main factor and trend change as the secondary factor".

[0183] Optionally, β1 can be 0.7 and β2 can be 0.3.

[0184] Optional, optional, c takes the value 0.5, d takes the value 0.5.

[0185] Optionally, m can be 1.2, q can be 0.8, and g can be 0.05.

[0186] The first function is a mathematical expression used to increase the weight of the corresponding airbag partition.

[0187] The second function is a mathematical expression used to reduce the partition weight corresponding to the airbag partition.

[0188] For step S4256, after updating the weights of all airbag zones, complete comfort zone mapping data containing the updated weights is read from the comfort zone mapping database. Based on the updated weights, the massage duration and activation frequency of each airbag zone are redefined. Combined with the zone identifier, optimal pressure, optimal temperature, temperature constraints, and adjusted audio configuration from the comfort zone mapping data, a new massage strategy containing zone timing, pressure, temperature, audio, and volume is generated to replace the currently executed massage strategy.

[0189] Optionally, during the generation of the massage strategy, the activation order and activation interval of each airbag zone are not completely fixed, but are generated in a random walk manner under the guidance of the probability distribution determined by the zone weight, so that the massage path has a certain degree of random change in each use, avoiding adaptive fatigue or anticipatory tension caused by completely repetitive time sequence.

[0190] The random walk approach refers to a massage sequence that is not fixed. The activation probability is set based on the weight of each area, and the order of the airbag areas is randomly selected according to the activation probability. For example, the temple area has the highest weight (activation probability set to 60%), and the brow bone area has a medium weight (activation probability set to 30%), but the next activation may not start with the temples; the brow bone area might be randomly selected instead. This constantly changing sequence avoids long-term repetition that can lead to adaptive fatigue, while also ensuring high-frequency stimulation of high-weight areas.

[0191] This embodiment continuously evaluates the relaxation score of each airbag zone in the first cycle and calculates the relaxation increment compared to the previous cycle, thereby sensing the real-time relaxation contribution and its changing trend of each airbag zone in the current cycle. By calculating a comprehensive relaxation evaluation value based on the current cycle's relaxation score and relaxation increment, the comprehensive evaluation value not only reflects the absolute value of the current stimulation effect of the airbag zone, but also reflects the trend information of whether the effect is increasing or decreasing, avoiding misjudgment of the contribution of airbag zones due to the slow drift of the target user's baseline state. By calculating the mean and standard deviation of the comprehensive relaxation evaluation values ​​of all airbag zones and determining the high and low thresholds accordingly, the judgment threshold is adaptively drifted with the overall state of the target user, rather than using a fixed threshold to uniformly judge different target users in different states. By increasing or decreasing the zone weight of airbag zones whose comprehensive relaxation evaluation values ​​are higher than the high threshold or lower than the low threshold for multiple consecutive cycles, the zone weight adjustment has a time continuity constraint, avoiding frequent misadjustments caused by single random fluctuations. Therefore, through periodic comprehensive evaluation and adaptive threshold determination, the eye massager can continuously optimize the partition weight allocation of each airbag zone according to the slow drift of the target user's physiological state, so as to put more massage energy into the most effective area at present, while reducing stimulation to inefficient areas, thereby achieving continuous optimization of massage resources while maintaining the stability of the massage strategy.

[0192] In one embodiment, the eye massager further includes a PPG (photoplethysmography) sensor, which is positioned on the inner side of the eye massager in contact with the user's forehead or earlobe, for collecting the photoplethysmography signal of the target user; the main control module is also used to extract heart rate variability signals from the photoplethysmography signals, calculate the ratio of low-frequency power to high-frequency power (LF / HF) as an assessment value of autonomic nervous activity, and adjust the massage strategy using the heart rate variability index as an auxiliary judgment basis when there is an inconsistency between the relaxation score of the EEG signal assessment and the heart rate variability index.

[0193] LF: Low-frequency power (0.04–0.15 Hz), reflecting the joint regulatory activity of the sympathetic and parasympathetic nervous systems. HF: High-frequency power (0.15–0.40 Hz), primarily reflecting parasympathetic activity; a high value indicates relaxation. LF / HF: The ratio of low-frequency power to high-frequency power, measuring autonomic nervous system balance; an increased value indicates tension, while a decreased value indicates relaxation.

[0194] Specifically, the main control module compares the trend of relaxation score changes with the trend of LF / HF changes in real time. When the relaxation score increases while the LF / HF increases, it is determined to be a Type I inconsistency. In this case, it is believed that the EEG signal may be affected by electromyography interference or poor electrode contact, resulting in a false relaxation representation. The tension state indicated by the increase in LF / HF is more reliable. The main control module preferentially performs at least one of the following operations: reducing the inflation pressure of the currently activated airbag zone (by 10% to 20% of the current pressure), shortening the pressure holding time of the airbag zone (by 15% to 25%), lowering the temperature of the heat compress (e.g., 1°C to 2°C), lowering the audio volume (e.g., 10% to 15%), and switching to soft audio. When the relaxation score decreases while the LF / HF decreases, it is determined to be a Type II inconsistency. In this case, it is believed that the EEG signal may be affected by environmental noise or a brief shift in the user's attention, thus underestimating the actual degree of relaxation. The relaxation state indicated by the decrease in LF / HF is more reliable. The main control module maintains the current massage intensity or slightly increases it by no more than 5% of the current pressure of the airbag zone. If the aforementioned inconsistency persists for more than a preset time (preferably 30 to 60 seconds) without converging, the main control module performs a weighted comprehensive evaluation of the normalized relaxation score and the normalized autonomic nervous activity assessment value. Specifically, the autonomic nervous activity assessment value and the relaxation score are first normalized to the 0-1 range, and then the comprehensive evaluation value is calculated as: 0.6 × normalized autonomic nervous activity assessment value + 0.4 × normalized relaxation score. The massage strategy is then adjusted based on this comprehensive evaluation value. This multimodal fusion mechanism effectively identifies artifacts in the EEG signals, avoids misjudgments, and improves the accuracy and safety of massage strategy adjustments.

[0195] Please see Figure 3 As shown, in one embodiment, an eye massager is provided, the eye massager including: a main control module 1, a massage execution module 2, an EEG acquisition module 3 and an audio output module 4, the main control module 1 being electrically connected to the massage execution module 2, the EEG acquisition module 3 and the audio output module 4 respectively; The massage execution module 2 includes a miniature air pump 21, a solenoid valve group 22, multiple independent airbag sections 23, and a heating element 24. The miniature air pump 21 is connected to each airbag section 23 through an air guide tube 25 and the inflation and deflation sequence is independently controlled by the solenoid valve group 22. The heating element 24, the miniature air pump 21, and the solenoid valve group 22 are controlled by the PWM signal output by the main control module 1. The EEG acquisition module 3 includes an electrode pad 31 and a signal conditioning circuit 32 electrically connected to the electrode pad 31. The signal conditioning circuit 32 converts the EEG signal acquired by the electrode pad 31 into a digital signal and sends it to the main control module 1. The audio output module 4 is used to play audio; The main control module 1 is configured to execute the steps of the adaptive control method for an eye massager driven by EEG closed-loop feedback as described above.

[0196] This embodiment acquires EEG signals from the target user in a resting, closed-eye state and constructs a resting-state baseline. This provides a personalized reference system reflecting the target user's brain activity, allowing all subsequent EEG changes to be quantified relative to this reference system. This avoids misjudgments caused by using fixed thresholds to uniformly evaluate different target users. Based on this, the mental state category is determined according to the average power spectral density of each frequency band of the EEG signal, and corresponding audio configurations are matched. This achieves personalized initial audio matching based on objective physiological data, avoiding the tediousness and inaccuracy of manual selection by the target user. Furthermore, an adaptive control method driven by EEG closed-loop feedback is used to execute the massage operation. This allows the eye massager to continuously sense the target user's real-time physiological feedback and dynamically adjust subsequent massage strategies accordingly, using the established resting-state baseline as a reference. Therefore, this application, through the synergistic effect of establishing a personalized reference system and adaptive control driven by EEG closed-loop feedback, ensures from the source that the benchmark on which the massage strategy is based is individualized, and also achieves dynamic tracking of the target user's state changes throughout the massage process. This ensures that the massage strategy always operates around the target user's optimal comfort range, avoiding the problems of overstimulation or ineffective massage caused by a fixed "one-size-fits-all" procedure, thereby improving the personalized adaptation ability and massage effectiveness of eye massage.

[0197] In one embodiment, there are multiple airbag partitions 23 and electrode pads 31. The airbag partitions 23 and electrode pads 31 are arranged in a one-to-one correspondence, and the electrode pads 31 are arranged directly behind or above their corresponding airbag partitions 23. The electrode pads 31 are used to collect the electroencephalogram (EEG) signals of the local brain regions corresponding to the airbag partitions 23. When any air sac region 23 is activated, electroencephalogram (EEG) signals are collected through the electrode pad 31 corresponding to that air sac region 23.

[0198] A local brain region refers to the scalp area and the cerebral cortex area located directly behind a specific air sac region in terms of spatial location. The data for local brain regions are derived from the mapping locations determined based on the spatial correspondence between acupoints around the eyes and forehead and the lobes of the brain in human anatomy. This mapping relationship is stored in the memory of the main control module.

[0199] Specifically, in the layered structure of the eye massager, electrode pad layers and airbag partition layers are arranged sequentially from the side in contact with the skin to the side away from the skin. The number of electrode pads is the same as the number of airbag partitions, and each electrode pad has a one-to-one spatial correspondence with its corresponding airbag partition, with the electrode pad located directly behind or above the corresponding airbag partition. Each electrode pad is located in the receiving space between the surface of the corresponding airbag partition facing away from the skin and the device casing, or integrated into the surface of the airbag partition facing the skin. When any airbag partition is activated via the PWM control line, an EEG signal is simultaneously acquired via the I2C bus from the signal conditioning circuit connected to the electrode pad corresponding to that airbag partition, rather than from other non-corresponding electrode pads. The acquired EEG signal is labeled as the EEG signal corresponding to the local brain region covered by the airbag partition.

[0200] This embodiment sets the number of airbag zones and electrode pads to be equal and in one-to-one correspondence, with each electrode pad positioned directly behind or above its corresponding airbag zone. This spatially binds each electrode pad to a specific airbag zone, and the acquired EEG signals originate from the local brain region corresponding to that airbag zone, rather than using a single frontal electrode to collect global signals. By establishing a fixed spatial correspondence between each airbag zone and each electrode pad, when an airbag zone is activated, signals can be obtained from the electrode pad corresponding to that airbag zone, while signals collected from electrodes corresponding to other airbag zones serve as background references. Therefore, EEG changes at a specific moment can be accurately attributed to massage stimulation of that specific airbag zone, rather than aliased responses or global interference from other airbag zones. This improves the signal-to-noise ratio and attribution accuracy of relaxation assessment for each airbag zone, providing a more reliable source data foundation for subsequent massage strategy optimization.

[0201] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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. Such 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 this application, and should all be included within the protection scope of this application.

Claims

1. An adaptive control method for an eye massager driven by EEG closed-loop feedback, characterized in that, The method includes: In response to the initialization command, the eye massager is controlled to execute the guided mode permission according to the set of guided parameters; During the execution of the guided mode, the target user's EEG signal in a resting, closed-eye state is collected, and the average power spectral density and standard deviation of the power spectral density of each frequency band are calculated, and a resting-state baseline is constructed. The mental state category is determined based on the average power spectral density of each frequency band. Audio is matched from a preset soothing audio library based on the mental state category. The audio identifier, preset audio volume and preset audio weight corresponding to the matched audio are associated as the audio configuration. Based on the resting-state baseline and audio configuration, an adaptive control method driven by EEG closed-loop feedback is used to control the eye massager to perform massage operations.

2. The adaptive control method for an eye massager driven by EEG closed-loop feedback according to claim 1, characterized in that, The adaptive control method, driven by EEG closed-loop feedback, based on the resting-state baseline and audio configuration, controls the eye massager to perform massage operations, including: In response to the exploration command, based on the resting baseline and audio configuration, perform trial sampling for the target user and construct comfort zone mapping data; Responding to massage commands, a massage strategy is generated based on comfort zone mapping data. An adaptive control method driven by EEG closed-loop feedback is adopted. The eye massager is controlled to perform massage operations based on the resting-state baseline and the massage strategy. During the massage operation, real-time EEG signals are continuously collected based on the resting-state baseline to monitor the target user. The massage strategy is dynamically adjusted based on the monitoring results to achieve adaptive control driven by EEG closed-loop feedback.

3. The adaptive control method for an eye massager driven by EEG closed-loop feedback according to claim 2, characterized in that, The response exploration command, based on the resting-state baseline and audio configuration, performs trial sampling for the target user and constructs comfort zone mapping data, including: In response to the exploration command, each airbag zone is activated individually and sequentially based on a preset duration, and the following steps are performed when activating each airbag zone: Collect EEG signals, assess relaxation scores based on resting-state baseline, determine zone weights and optimal stress; The system performs gradient heating within a preset temperature range, collects EEG signals, assesses relaxation scores based on resting baseline, locks in the optimal temperature, and calculates the left and right descent rates based on the optimal temperature within a preset temperature range. If the rate of descent on the left is greater than the rate of descent on the right, then the optimal temperature is subtracted from the first temperature to obtain the lower limit of the temperature constraint; otherwise, the optimal temperature is added to the first temperature to obtain the upper limit of the temperature constraint. Based on the audio configuration, the system plays audio, collects EEG signals, determines the mental state category, adjusts the audio weights in the audio configuration, determines the amplitude change trend, and adjusts the audio volume in the audio configuration. The partition identifier, partition weight, optimal pressure, optimal temperature, temperature constraint, and adjusted audio configuration corresponding to the airbag partition are associated and used as the mapping data corresponding to the airbag partition in the comfort zone mapping data.

4. The adaptive control method for an eye massager driven by EEG closed-loop feedback according to claim 2, characterized in that, The response to massage commands generates a massage strategy based on comfort zone mapping data, employs an EEG-driven closed-loop feedback adaptive control method, and controls the eye massager to perform massage operations based on the resting-state baseline and the massage strategy, including: Responding to massage commands, the system generates massage strategies based on comfort zone mapping data, including zone timing, pressure, temperature, audio, and volume. It also employs an adaptive control method driven by EEG closed-loop feedback to control the eye massager to perform massage operations based on the resting baseline and the massage strategy. In response to the end command, the updated comfort zone mapping data from this massage operation will be stored in the target user's target user profile. During the massage procedure, real-time electroencephalogram (EEG) signals are continuously collected for monitoring, and one or more of the following control operations are implemented based on the monitored data: Operation (1): When the relaxation level is detected to decrease transiently to the stress response condition based on the resting baseline, the eye massager is controlled to switch to the soothing mode and when the relaxation level is detected to recover to the recovery condition, the eye massager is controlled to resume the massage operation according to the resting baseline and massage strategy. Operation (2): When the target user is detected to be in the early stage of light sleep, control the eye massager to switch to sleep protection mode. When the target user is detected to be in deep sleep in sleep protection mode, control the eye massager to switch to deep sleep care mode. When the target user is detected to be waking up in deep sleep care mode, control the eye massager to switch to low intensity wake-up mode. Operation (3): Based on the resting baseline, evaluate the comprehensive relaxation value of each airbag zone in the first cycle, and dynamically adjust the zone weight in the comfort zone mapping data according to the comprehensive relaxation value, and update the massage strategy according to the updated comfort zone mapping data.

5. The adaptive control method for an eye massager driven by EEG closed-loop feedback according to claim 4, characterized in that, The response to the massage command generates a massage strategy based on comfort zone mapping data, including zone timing, pressure, temperature, audio, and volume, comprising: The massage duration and activation frequency of each airbag zone are determined based on the zone weights in the comfort zone mapping data. The zone weights are directly proportional to both the massage duration and the activation frequency. Based on the massage duration and activation frequency of each airbag zone, as well as the zone identifier, optimal pressure, optimal temperature, temperature constraints, and adjusted audio configuration in the comfort zone mapping data, a massage strategy is generated that includes zone timing, pressure, temperature, audio, and volume.

6. The adaptive control method for an eye massager driven by EEG closed-loop feedback according to claim 4, characterized in that, The operation (1) is as follows: The second cycle is used to calculate transient change data based on the collected EEG signals. The transient change data includes the power of beta waves, the ratio of alpha wave power to beta wave power, and the periodic change rate of total power. When the transient decrease in relaxation level is detected based on transient change data and resting baseline to reach the stress response condition, the eye massager immediately terminates all airbag inflation commands and performs rapid deflation, reduces heating power by a first ratio, reduces audio volume by a second ratio, and switches to preset soft audio to achieve soothing mode operation. During the operation period of the soothing mode, the alpha wave power spectral density is continuously monitored. If the current alpha wave power spectral density recovers to more than the third proportion of the alpha wave power spectral density before the stress response condition is reached, it is determined that the relaxation level has recovered to the recovery condition. Then, the eye massager is controlled to exit the soothing mode and gradually return to performing massage operations according to the resting baseline and massage strategy.

7. The adaptive control method for an eye massager driven by EEG closed-loop feedback according to claim 4, characterized in that, The operation (2) is as follows: Using the third cycle, if the proportion of the theta wave power spectral density exceeds the first threshold and the proportion of the alpha wave power spectral density is lower than the second threshold in multiple consecutive third cycles, then the target user is determined to have entered the early stage of light sleep. Within the first duration, control the eye massager to stop all airbag inflation and deflation operations, reduce the heat therapy temperature to the second temperature at a preset cooling rate, reduce the audio volume to the first volume and switch to pink noise to achieve sleep protection mode operation; During the operation of the sleep protection mode, when the fourth cycle is used to monitor the collected EEG signals and the proportion of delta wave power spectral density continuously rises above the third threshold, it is determined that the target user has entered deep sleep. The eye massager is then controlled to reduce the heat temperature to the third temperature and the audio volume to the second volume to switch to the deep sleep care mode. During the deep sleep care mode, when the proportion of beta wave power spectral density is detected to rise above the fourth threshold and the duration exceeds the second duration, it is confirmed that the target user has a tendency to wake up. The eye massager is then controlled to play a wake-up prompt sound and the heat therapy temperature is raised back to the fourth temperature to achieve a low-intensity wake-up mode.

8. The adaptive control method for an eye massager driven by EEG closed-loop feedback according to claim 4, characterized in that, The operation (3) is as follows: In the first cycle, based on the resting-state baseline, the relaxation score of each air sac region during the current round of activation is determined according to the collected EEG signals, and the relaxation increment of each air sac region compared with the previous round is calculated. Calculate the overall relaxation score based on the relaxation score during the current activation period and the relaxation increment compared to the previous round; Calculate the mean and standard deviation of the overall relaxation evaluation value for all airbag zones, and determine the high and low thresholds based on the mean and standard deviation; When the overall relaxation evaluation value of any airbag zone is higher than the high threshold for a consecutive preset number of cycles, the zone weight corresponding to that airbag zone in the comfort zone mapping data is increased according to the first function. When the overall relaxation evaluation value of any airbag zone is lower than the low threshold for a consecutive preset number of cycles, the zone weight corresponding to that airbag zone in the comfort zone mapping data is reduced according to the second function. The massage strategy is updated based on the updated comfort zone mapping data; in, F(i)=β1×S_i(t)+β2×ΔS_i Th_high = F_mean + c × F_std Th_low = F_mean d×F_std P_i_new1 = min(1.0, P_i_old × m) P_i_new2 = max(g, P_i_old×q) F(i) is the overall relaxation evaluation value of the i-th airbag partition, S_i(t) is the relaxation score of the i-th airbag partition during the current round of activation, ΔS_i is the relaxation increment of the i-th airbag partition compared to the previous round, β1 and β2 are preset weight coefficients, β1>β2, and i is the partition identifier of the airbag partition. Th_high is the high threshold, Th_low is the low threshold, F_mean is the mean, F_std is the standard deviation, and c and d are preset weight coefficients. P_i_new1 is the first function, P_i_new2 is the second function, min calculates the minimum value, max calculates the maximum value, P_i_old is the current partition weight of the airbag partition, m is a constant greater than 1, q is a constant less than 1 and greater than 0, and g is a constant less than 0.5 and greater than 0.

9. An eye massager, characterized in that, The eye massager includes: a main control module, a massage execution module, an EEG acquisition module, and an audio output module. The main control module is electrically connected to the massage execution module, the EEG acquisition module, and the audio output module, respectively. The massage execution module includes a miniature air pump, a solenoid valve group, multiple independent airbag zones, and a heating element. The miniature air pump is connected to each airbag zone through an air guide tube and its inflation and deflation sequence is independently controlled by the solenoid valve group. The heating element, the miniature air pump, and the solenoid valve group are controlled by the PWM signal output by the main control module. The EEG acquisition module includes electrode pads and a signal conditioning circuit electrically connected to the electrode pads. The signal conditioning circuit converts the EEG signals acquired by the electrode pads into digital signals and sends them to the main control module. The audio output module is used to play audio; The main control module is configured to perform the steps of the adaptive control method for an eye massager driven by EEG closed-loop feedback as described in any one of claims 1 to 8.

10. The eye massager according to claim 9, characterized in that, The number of airbag sections and electrode pads are both multiple, and the airbag sections and electrode pads are set one-to-one. The electrode pads are set directly behind or directly above their corresponding airbag sections. The electrode pads are used to collect the electroencephalogram (EEG) signals of the local brain regions corresponding to the airbag sections. When any air sac region is activated, electroencephalogram (EEG) signals are acquired through the electrode pads corresponding to that air sac region.

Images