A smart control method and system for a multifunctional eye mask

Abstract

This invention relates to the field of intelligent control technology for eye masks, and discloses an intelligent control method and system for multifunctional eye masks. The method includes acquiring the skin temperature around the eyes and the current ambient temperature to obtain an individual thermal inertia coefficient; calculating a second output power based on the thermal inertia coefficient and the current ambient temperature; executing the second output power, acquiring a skin temperature sequence, and querying a comfort temperature reference value in a preset comfort temperature reference table to calculate a heat loss power; performing power compensation calculation based on the heat loss power to obtain a third output power; executing the third output power, acquiring an instantaneous skin temperature matrix, and adjusting it to obtain an optimized power command sequence; adjusting and determining the closed-loop control gain based on the optimized power command sequence to form a final temperature control scheme. This method can accurately sense and control temperature to adapt to individual differences, thereby improving user comfort and personalized experience.

Description

Technical Field

[0001] This invention relates to the field of intelligent control of eye masks, and more particularly to an intelligent control method and system for multifunctional eye masks. Background Technology

[0002] In the fields of health and personal care, smart wearable devices based on SCADA platforms need to dynamically adjust according to the user's actual feelings and physical differences to improve user comfort and personalized experience. Especially for eye care products, such as multifunctional eye masks, how to improve user comfort and personalized experience through technological means has become a focus of industry attention.

[0003] In existing technology, eye mask products typically offer a uniform multi-level temperature control function. For example, users can select a low temperature of 39℃ (green light), a medium temperature of 42℃ (blue light), or a high temperature of 45℃ (red light). However, due to significant differences in how human skin perceives temperature and the rate of heat transfer, this individualized thermal inertia makes it difficult for a uniform temperature setting to meet the needs of all users. Furthermore, because of these differences, the device struggles to capture real-time changes in the user's skin condition during operation, making it unable to adjust the output power promptly, resulting in temperature fluctuations or prolonged deviations from the comfortable range.

[0004] In summary, existing multifunctional eye masks have shortcomings in data collection and control, making it difficult to effectively collect data and judge the user's actual feelings and physical differences, resulting in the inability to dynamically adjust the heat therapy temperature, thus affecting the user's experience. Summary of the Invention

[0005] This invention provides an intelligent control method and system for multifunctional eye masks, which can accurately sense and regulate temperature to adapt to individual differences, thereby improving user comfort and personalized experience.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides an intelligent control method for a multifunctional eye mask, comprising: The sensor in the eye mask continuously acquires the skin temperature around the eyes and the current ambient temperature, and inputs them into a pre-trained thermal inertia assessment model for analysis to obtain the heat exchange efficiency and the individual thermal inertia coefficient. The first output power and power amplification factor are obtained by querying the preset coefficient-power mapping table based on the individual thermal inertia coefficient. The first output power is then adjusted based on the current ambient temperature and the power amplification factor to obtain the second output power. The second output power is executed to control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain a skin temperature sequence. The skin temperature sequence is then queried from a preset comfort temperature reference table for matching to obtain a comfort temperature reference value, and the heat loss power is calculated. The heat loss power is converted to obtain a power compensation value, which is then superimposed with the second output power to obtain the third output power. The third output power is executed to control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix. The power of the heating element is continuously adjusted according to the skin temperature matrix, and the power matrix is ​​recorded over time to obtain an optimized power command sequence. Based on the optimized power command sequence, the closed-loop control gain of each heating element is adjusted and locked to obtain the final temperature control scheme.

[0007] Secondly, the present invention provides an intelligent control system for a multifunctional eye mask, comprising: The thermal inertia assessment module is used to continuously acquire the skin temperature around the eyes and the current ambient temperature from the eye mask sensor, and input them into the pre-trained thermal inertia assessment model for analysis to obtain the heat exchange efficiency and the individual thermal inertia coefficient. The two-factor correction module is used to query a preset coefficient-power mapping table based on the individual thermal inertia coefficient to obtain a first output power and a power amplification factor, and adjust the first output power based on the current ambient temperature and the power amplification factor to obtain a second output power; The heat loss calculation module is used to execute the second output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain a skin temperature sequence, query a preset comfort temperature reference table according to the skin temperature sequence to obtain a comfort temperature reference value, and calculate the heat loss power. A heat conduction compensation module is used to convert the heat loss power into a power compensation value, and then superimpose it with the second output power to obtain a third output power. The differential allocation module is used to execute the third output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix, continuously adjust the power of the heating element according to the skin temperature matrix, record the change process of the power matrix over time, and obtain an optimized power command sequence. The simulation verification module is used to adjust and lock the closed-loop control gain of each heating element according to the optimized power command sequence to obtain the final temperature control scheme.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention continuously acquires the skin temperature around the eyes and the current ambient temperature from the eye mask sensor, analyzes and evaluates the user's individual thermal inertia coefficient using a pre-trained thermal inertia assessment model, and adjusts and determines the actual output power based on the individual thermal inertia coefficient by querying a preset coefficient-power mapping table. Since human skin's temperature perception and heat transfer speed vary significantly, compared to existing technologies, this technology can assess the user's individual thermal inertia coefficient, thereby dynamically adjusting the eye mask's heating power in advance based on the user's individual thermal inertia coefficient. This solves the problems of existing technologies having a single temperature setting and being unable to dynamically adjust the heat therapy temperature.

[0009] (2) This invention obtains a skin temperature sequence by controlling an eye mask sensor to measure the skin temperature around the eyes at a preset frequency. The skin temperature sequence is then matched against a preset comfort temperature reference table to obtain a comfort temperature reference value, and the heat loss power is calculated. The heat loss power is then converted into power, and the output power is adaptively adjusted based on weighting parameters calculated in real time. This technology quantifies fluctuation characteristics into specific control commands, enabling the heating system to actively counteract unwanted body temperature fluctuations and maintain a stable skin temperature around the eyes within a narrow comfort zone. This significantly improves the constant temperature sensation and comfort during use, thus solving the problem of drastic temperature fluctuations during power adjustment in existing technologies, which affect user experience.

[0010] (3) The present invention obtains the final temperature control scheme by adjusting and locking the closed-loop control gain of each heating element according to the optimized power command sequence. This technology can dynamically adapt to environmental changes through closed-loop control, ensuring long-term stability, while optimizing energy use. Furthermore, it performs time-sharing and zone-based differentiated processing of power commands through simulation and verification, ensuring continuous improvement of user experience. This avoids the problems of existing technologies, such as inability to dynamically adapt to environmental changes, poor long-term stability, low energy efficiency, and fluctuations in user experience. Attached Figure Description

[0011] Figure 1 This is a schematic flowchart of an intelligent control method for a multifunctional eye mask provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of an intelligent control system for a multifunctional eye mask provided in the second embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] Reference Figure 1 The first embodiment of the present invention provides an intelligent control method for a multifunctional eye mask, comprising the following steps: S11: The initial skin temperature and current ambient temperature are continuously acquired from the eye mask sensor and input into the pre-trained thermal inertia assessment model for analysis to obtain the heat exchange efficiency and the individual thermal inertia coefficient. S12, based on the individual thermal inertia coefficient, query the preset coefficient-power mapping table to obtain the first output power and power amplification factor, and adjust the first output power according to the current ambient temperature and the power amplification factor to obtain the second output power; S13, execute the second output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain a skin temperature sequence, query a preset comfort temperature reference table according to the skin temperature sequence to obtain a comfort temperature reference value, and calculate the heat loss power. S14, the heat loss power is converted into a power compensation value, and then superimposed with the second output power to obtain the third output power; S15, execute the third output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix, and continuously adjust the power of the heating element according to the skin temperature matrix, record the change process of the power matrix over time, and obtain an optimized power command sequence; S16. Based on the optimized power command sequence, adjust and lock the closed-loop control gain of each heating element to obtain the final temperature control scheme.

[0014] In step S11, the initial skin temperature and current ambient temperature are continuously acquired from the eye mask sensor and input into a pre-trained thermal inertia assessment model for analysis to obtain the heat exchange efficiency and the individual thermal inertia coefficient, including: Within a preset time period, the temperature of the skin around the eyes and the current ambient temperature are continuously acquired from the eye mask sensor; Record the fluctuation range of skin temperature around the eyes over time as thermal sensitivity data, and calculate the dynamic change value of the current ambient temperature. The thermosensitive response data is input into the pre-trained thermal inertia assessment model to analyze the heat conduction rate and obtain the heat exchange efficiency. Based on the heat exchange efficiency and the dynamic change value, the thermal hysteresis effect is calculated to obtain the individual thermal inertia coefficient.

[0015] It should be noted that the preset time period is generally 30 seconds to avoid excessively long time intervals where large changes in ambient temperature would interfere with the assessment of individual thermal inertia coefficients. This value is determined based on the thermophysiological characteristics of the skin around the eyes, ensuring that sufficient data on the skin temperature around the eyes and the ambient temperature can be collected within this period. This comprehensively captures the thermal response process of the skin around the eyes to temperature changes, accurately reflecting individual differences in thermal sensitivity, while avoiding the problem of delayed temperature control response caused by excessively long periods. This achieves a balance between the completeness of data acquisition and the timeliness of temperature control. The sensor's acquisition frequency in this step is consistent with that in the subsequent step S13, set to 5Hz.

[0016] It should be noted that the thermosensitive response data is defined as the fluctuation range of the skin temperature around the eyes over time. Its calculation must be based on the initial skin temperature within the preset time period to ensure accurate reflection of the skin's sensitivity to temperature changes. The specific calculation method is as follows: First, extract the initial skin temperature around the eyes collected at the beginning of the preset time period. Then extract all validly collected periorbital skin temperature data within the period. ( (Sampling was performed 150 times at a frequency of 5Hz within 30 seconds). The absolute difference between the skin temperature at each moment and the initial temperature was calculated, and the maximum value of this absolute difference was taken as the thermosensitive response data. The calculation formula is as follows: ,in Data for thermosensitive reactions (unit: °C). This represents the total number of data collections within the preset period. The dynamic change value of ambient temperature is used to quantify the real-time fluctuation range of ambient temperature within the preset period. A moving average method is used to smooth out instantaneous changes in ambient temperature and ensure data stability. The calculation formula is as follows: ,in This represents the dynamic change value of ambient temperature (unit: °C). For the first The ambient temperature of the second sample collection For the first The ambient temperature of the second sample was taken, and the denominator was taken as... The aim is to match the correspondence between the number of data collections and the number of changes in ambient temperature, ensuring accurate calculation results. When calculating the thermal inertia coefficient, absolute values ​​are used in the dynamic change values. The core purpose is to eliminate interference from the direction of ambient temperature rise or fall, considering only the amplitude of environmental fluctuations, and avoiding the impact of ambient temperature changes on the sensor's temperature measurement accuracy, which could lead to misjudgments of the thermal inertia coefficient.

[0017] It should be noted that the pre-trained thermal inertia assessment model is a backpropagation (BP) neural network. The input layer of this network receives normalized thermosensitive response data derived from experiments on the thermal properties of periocular skin, covering temperature fluctuations and heat conduction information of the periocular skin under different ambient temperatures and initial temperature differences in different individuals. Two fully connected hidden layers are then connected, with ReLU activation functions used to extract and fuse deep features related to the heat conduction rate layer by layer. The final layer is the output layer, using a linear activation function to output the predicted heat conduction rate, which, when divided by the power of the heating element, represents the heat exchange efficiency. (Dimensionless parameter), actual measured values ​​are between 0.1 and 0.5.

[0018] The model is trained using supervised learning. The training data comes from a large amount of experimental data on the thermal properties of the periocular skin of different individuals. The dataset comprehensively covers samples of different ages, skin types, ambient temperatures, and initial temperature differences to ensure that the model can adapt to the thermophysiological differences of a wide range of people. Normalized thermosensitive response data is used as input, along with the corresponding heat exchange efficiency. The target output is used. Mean squared error (MSE) is used as the loss function, and the Adam optimizer is iteratively trained on a large number of samples, continuously adjusting the network weights until the model converges. The trained model can accurately predict the heat exchange efficiency between the periocular skin and the eye mask interface from new, unknown individuals' thermosensitive response data. .

[0019] It should be noted that the calculation of the individual thermal inertia coefficient needs to exclude the interference of ambient temperature changes on the sensor temperature measurement, and a safety margin should be reserved based on the range of environmental fluctuations to avoid misjudgment of thermal inertia due to environmental interference, which could lead to excessive heating power. This embodiment uses the following formula to evaluate the user's individual thermal inertia coefficient: in, The individual thermal inertia coefficient (dimensionless) is generally greater than 1.0. It characterizes the resistance of the user's skin to heat absorption and temperature rise. It is related to the physiological and thermodynamic properties of the user's skin. The measurement results are also related to factors such as the position and degree of fit of the eye mask and environmental interference. It is an environmental impact correction item. Use in environments with drastic temperature changes may cause [problems]. > In this case, the denominator is negative, therefore the environmental impact correction term uses absolute values; where This is a fitting coefficient used to accurately assess the impact of environmental factors. In this embodiment, it is obtained through linear fitting based on historical statistical data. In the formula, heat exchange efficiency is... With individual thermal inertia coefficient Negative correlation; and The impact of changes in ambient temperature on skin temperature was assessed. When this impact is significant, to avoid overestimating the user's thermal inertia and resulting in excessively high heating power, the thermal inertia coefficient needs to be appropriately increased. They are positively correlated. From the physical meaning of thermal hysteresis, the individual thermal inertia coefficient... The larger the value, the stronger the heat storage capacity of the skin tissue around the eyes, the slower the heat conduction speed, and the more obvious the thermal lag effect. Subsequent heating power needs to be adjusted reasonably based on a safety margin to compensate for the thermal lag effect while avoiding discomfort caused by excessive power; conversely, a smaller value indicates a weaker heat storage capacity. The smaller the size, the weaker the skin's heat storage capacity, the faster the heat conduction speed, and the weaker the thermal hysteresis effect. This allows for a suitable reduction in heating power, balancing comfort and energy saving.

[0020] For example, the preset time period is 30 seconds, and the sensor collects data at a frequency of 5 Hz. A total of [number] data are collected within the preset period. A set of valid data on the skin temperature around the eyes and the current ambient temperature; initial skin temperature around the eyes collected at the initial moment. Initial ambient temperature The maximum difference between skin temperature and initial temperature at any given time point within the cycle was 0.5℃, which corresponds to the thermal sensitivity data. The dynamic changes in ambient temperature were calculated using the moving average method, and the results were obtained. ; the thermosensitive reaction data The heat exchange efficiency is obtained after inputting into a pre-trained thermal inertia evaluation model and analyzing the heat conduction rate. Substituting all parameters into the formula for calculating the individual thermal inertia coefficient, we obtain... In this example, the ambient temperature rises. The corresponding thermal inertia coefficient is 1.71, which can effectively avoid interference with temperature measurement and misjudgment of thermal inertia due to ambient temperature rise, thereby preventing the problem of excessive heating power.

[0021] In step S12, a preset coefficient-power mapping table is consulted based on the individual thermal inertia coefficient to obtain the first output power and power amplification factor. The first output power is then adjusted based on the current ambient temperature and the power amplification factor to obtain the second output power, including: The corresponding first output power is obtained by querying the preset coefficient-power mapping table based on the individual thermal inertia coefficient. When the individual thermal inertia coefficient is greater than the preset coefficient threshold, the corresponding power amplification factor is obtained by querying the coefficient-power mapping table. Multiplying the power amplification factor by the first output power yields the enhanced power; The current ambient temperature is converted into an additional power value, and then smoothed to obtain the environmental compensation power. The enhanced power and the environmental compensation power are superimposed to obtain the second output power.

[0022] The preset coefficient threshold is set to 2.0. This preset value is based on the average individual thermal inertia coefficient of users during internal product testing and is used to determine whether additional power amplification compensation is needed. Users can also adjust this threshold themselves during use. If the individual thermal inertia coefficient is less than or equal to this threshold, there is no need to search for the power amplification factor; the first output power is directly used as the enhanced power. If it is greater than this threshold, the first output power is amplified according to the amplification factor in the preset coefficient-power mapping table. Enhanced Power The calculation formula is ,in The first output power (unit: W). This is the power amplification factor.

[0023] For example, when the thermal inertia coefficient obtained in step S11 is 1.71, the corresponding first output power is 3.52W, and the corresponding power amplification factor is 1.07. Since the thermal inertia coefficient is actually less than the preset threshold, there is no additional amplification. The actual amplification factor is 1.00, and the amplified power is... When the thermal inertia coefficient is 2.35, the corresponding first output power is 3.65W, the power amplification factor is 1.12, and the amplified power is... .

[0024] It should be noted that the preset coefficient-power mapping table contains a one-to-one correspondence between individual thermal inertia coefficients, first output power, and power amplification factor. Individual thermal inertia coefficients are positively correlated with both first output power and power amplification factor. A larger thermal inertia coefficient indicates stronger skin heat storage capacity, slower heat conduction, and a higher required base heating power. The corresponding power amplification factor is also larger, used to compensate for heat conduction lag and ensure heating efficiency adapts to individual differences. This mapping table is constructed based on a large amount of experimental data on different individual thermal inertia coefficients and corresponding comfortable heating powers. Specifically, it adopted and statistically analyzed usage data from 500 test subjects of different physical conditions and ages. The 3σ rule was used to eliminate extreme values. After evaluating the thermal inertia coefficient, the relationship between the thermal inertia coefficient and comfortable power was determined using a quadratic polynomial fitting method with least squares. Finally, a table of applicable engineering intervals was summarized using the K-means clustering algorithm. The values ​​can be determined according to the actual level of precision required. Then, the amplification factor corresponding to each hot spot is calculated through preset coefficient thresholds. Finally, this table is obtained by verifying and summarizing through A / B testing.

[0025] It should be noted that the power conversion based on ambient temperature follows a linear relationship between ambient temperature and heat loss. The core principle is to quantify the ambient temperature as a corresponding additional power value to compensate for heat loss under different environments. The specific conversion formula is as follows: ,in Additional power value (unit: W). The ambient power conversion factor (preset to 0.1W / ℃, based on the performance of the heating element). The reference ambient temperature is 25℃ (preset to match the human comfort environment). The current ambient temperature is displayed in °C. The smoothing process uses a moving average filter with a sliding window length of 5 sampling periods to ensure stable adjustment of the environmental compensation power and avoid discomfort caused by sudden power output changes.

[0026] It should be noted that the core calculation formula for the second output power is the superposition of the enhanced power and the environmental compensation power, and the formula is as follows: ,in This is the second output power (unit: W). Environmental compensation power after smoothing (unit: W).

[0027] For example, Current ambient temperature Substituting into the additional power conversion formula, take , The additional power value is obtained. After moving average filtering, the environmental compensation power The enhanced power is superimposed with the environmental compensation power to obtain the second output power. .

[0028] In step S13, the second output power is executed, controlling the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain a skin temperature sequence. The skin temperature sequence is then matched against a preset comfort temperature reference table to obtain a comfort temperature reference value, and the heat loss power is calculated, including: The instantaneous temperature rise rate calculated based on the skin temperature sequence is matched with a preset comfort temperature reference value table to obtain a comfort temperature reference value. When the absolute value of the difference between the skin temperature sequence and the comfort temperature reference value exceeds a preset fluctuation threshold, the extreme value distribution characteristics of the absolute value of the difference on the time axis are extracted, and the temperature fluctuation amplitude is calculated. When the temperature fluctuation exceeds the preset temperature fluctuation range, the heat loss power is calculated based on the temperature fluctuation. The preset comfort temperature reference value table stores the steady-state temperature of the skin around the eyes that is perceived as comfortable by the user, corresponding to different heating rate ranges.

[0029] It should be noted that the preset sampling frequency of the sensor array in this step is consistent with that in step S11, which is 5Hz, to ensure the continuity and comprehensiveness of temperature data and provide consistent data support for the calculation of instantaneous heating rate and the analysis of temperature fluctuation amplitude. Skin temperature sequence measured by a sensor During the data collection process, the 3σ criterion was used to remove outlier data, i.e., data that exceeded the mean of the skin temperature sequence by ±3 times the standard deviation was removed. This avoids interference from outlier data in the calculation of instantaneous heating rate, matching of comfort temperature benchmark value, and calculation of heat loss power, ensuring the accuracy of calculations in each step.

[0030] It should be noted that the instantaneous heating rate is used to quantify how quickly skin temperature changes over time, and its calculation formula is as follows: ,in The instantaneous heating rate (unit: ℃ / s, generally greater than zero due to the continuous heating of the goggles). The first in the skin temperature sequence Temperature values ​​at each sampling point For the first Temperature values ​​at each sampling point The time interval between two adjacent sampling points is determined by the sampling frequency, such as when sampling at 5Hz. .

[0031] It should be noted that the preset comfort temperature baseline table stores values ​​based on the instantaneous heating rate. Segmented matching corresponds to the corresponding comfort temperature benchmark value The construction of a preset comfort temperature baseline table is based on the physiological characteristics of the skin around the eyes and a large amount of user comfort experience experimental data to ensure that the baseline values ​​are consistent with the temperature change trend of the skin around the eyes. Specifically, product usage data and questionnaire feedback from 1000 testers of different physical conditions and ages were adopted and statistically analyzed. 968 valid questionnaires were adopted. Hampel filters were used to remove extreme values, and the relationship between the instantaneous heating rate and the comfort temperature baseline value was determined using the least squares method for quadratic polynomial fitting. The continuous fitting curve was discretized using the equal-width binning method to form an engineering-practical step lookup table. Based on the subject data, the actual step length was set at 0.02℃ / s. Finally, a preliminary push was conducted, and the results were validated based on subject feedback. For example: 0.05℃ / s < When the temperature rises at a rate ≤0.07℃ / s (slow skin temperature increase), 0.07℃ / s When the rate of skin temperature rise is less than 0.09℃ / s, To ensure that the baseline value matches the trend of skin temperature changes, a feedforward control strategy based on the rate of temperature rise is adopted to suppress temperature overshoot and ensure a comfortable fit for the human body.

[0032] It should be noted that the preset fluctuation threshold is set to 0.8℃ to distinguish between normal physiological temperature fluctuations and abnormal fluctuations requiring intervention; when each sampling point in the skin temperature sequence deviates from the comfort temperature baseline... absolute value of the difference When the temperature exceeds 0.8℃, the temperature fluctuation amplitude calculation is triggered. Extreme value distribution feature extraction requires obtaining the maximum, minimum, and average absolute values ​​of the differences within a preset time window (e.g., 10 seconds). The formula for calculating the temperature fluctuation amplitude is: ,in This refers to the temperature fluctuation range (unit: °C). The preset temperature fluctuation range is set to 0-0.5 °C. When the temperature fluctuation exceeds the comfort range, it indicates that the heat loss power needs to be calculated; the formula for calculating heat loss power is: ,in Heat loss power (unit: J / s). The equivalent convective heat transfer coefficient is preset according to the product operating conditions. 10 preferred ; The contact area between the eye mask and the skin around the eyes is preset to 120cm according to product specifications. 2 This is equivalent to 0.012m. 2 This formula, based on the law of thermal conduction, quantifies the heat loss corresponding to temperature fluctuations, providing a precise basis for subsequent power compensation.

[0033] For example, the second output power is executed. The sensor collects skin temperature at a frequency of 5Hz. (Sampling frequency consistent with steps S11 and S12) to obtain the skin temperature sequence; calculate the instantaneous temperature rise rate at each sampling point, where the average gradient value within a certain time period is 0.07℃ / s, and consult the comfort temperature reference value table to obtain... ; Some sampling points in the skin temperature sequence and If the absolute value of the difference exceeds 0.8℃, the maximum absolute value of the difference within a 10-second time window is extracted, which is 0.7℃ and the minimum is 0.3℃. The temperature fluctuation amplitude is then calculated. This value does not exceed the preset temperature fluctuation range (0-0.5℃), so there is no need to calculate the heat loss power. In another example, the average instantaneous heating rate is 0.08℃ / s, matching... Within a 10-second window, the maximum absolute value of the temperature difference was 1.2℃ and the minimum was 0.7℃, representing the temperature fluctuation range. The temperature fluctuation exceeds the preset range; take , Substitute the values ​​into the formula to calculate the heat loss power: .

[0034] In step S14, the heat loss power is converted to obtain a power compensation value, which is then superimposed with the second output power to obtain the third output power.

[0035] It should be noted that the core of the power conversion of heat loss power is to convert the quantified heat loss into a corresponding power compensation value, which is used to offset the heat loss caused by temperature fluctuations and ensure that the temperature of the skin around the eyes is stable near the comfort baseline. The heat loss power (unit: W) is used for power conversion using linear conversion logic, and a power loss conversion coefficient is introduced. (Dimensionless, preset to 1.0-1.2, preferably 1.1 based on the energy conversion efficiency compensation coefficient of the product's heating element), the conversion formula is: in This is the power compensation value (unit: W). The sign of the power compensation value is determined by the direction of heat conduction. When heat is lost from the skin to the outside, It is a positive value. A positive value indicates the presence of calories; if an abnormal accumulation of calories occurs ( (negative value) A negative value is used to reduce power and decrease heat input; For users' heat exchange efficiency.

[0036] It should be noted that the core calculation formula for the third output power is the superposition of the power compensation value and the second output power, and the formula is as follows: ,in This is the third output power (unit: W). The second output power (in W) obtained in step S12. This is the power compensation value (unit: W). It also helps avoid problems such as insufficient power failing to compensate for heat loss, or power overload causing skin discomfort.

[0037] For example, step S13 calculates the heat loss power. Take the power loss conversion factor Substituting into the conversion formula yields the power compensation value. The second output power obtained in step S12 The third output power is obtained by superimposing the two. If the heat loss power ,but After superposition .

[0038] In step S15, the third output power is executed to control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix. Based on the skin temperature matrix, the power of the heating element is continuously adjusted, and the change in the power matrix over time is recorded to obtain an optimized power command sequence, including: After executing the third output power, the initial skin temperature matrix and the initial ambient temperature are recorded, and the current power is collected at a preset frequency. These are used together as the device adjustment record, and the historical adjustment record that matches the device adjustment record is queried from the controller. The historical adjustment records and the instantaneous skin temperature matrix are input into a pre-trained adaptive learning model to generate a fused feature vector; The fused feature vector is input into a pre-trained stable state evaluation model for comparison to determine whether the skin has entered a stable state. If yes, then record the skin temperature matrix at this time to obtain the steady-state distribution of skin temperature; if no, then adjust the third output power according to the preset gradient power value, and return to the step of executing the third output power until the skin enters a stable state. Adjust the power matrix of the heating element until the steady-state distribution of the skin temperature becomes uniform, record the change process of the power matrix over time, and obtain the optimized power command sequence; It should be noted that the device adjustment record consists of three core data parts: the initial skin temperature matrix, the initial ambient temperature, and the current power. The initial skin temperature matrix is ​​the periorbital skin temperature matrix collected at the initial moment of executing the third output power. Its structure is consistent with the instantaneous skin temperature matrix, being a single-row matrix (row vector), defined as follows: For the first Instantaneous skin temperature matrix at each moment, Indicates the number of sensors (4-6, preferably 4); initial ambient temperature. The ambient temperature collected at the initial moment of the third output power is consistent with the ambient temperature collection standards in steps S11 and S12; the current power is the real-time collected eye mask heating power, and the collection frequency is consistent with the sensor temperature collection frequency (5Hz). The controller locally stores historical adjustment records. Each record contains the skin initial temperature matrix, ambient initial temperature, initial power, and subsequent power adjustment process corresponding to the device adjustment record. During querying, a feature matching algorithm is used to filter out historical records whose deviations from the parameters in the current device adjustment record do not exceed a preset threshold, which serve as the basis for subsequent fusion analysis.

[0039] It should be noted that the pre-trained adaptive learning model is a hybrid neural network (HNN) that integrates a temporal coding network and a feature fusion network. The input layer of this network receives two types of heterogeneous data: the historical adjustment record branch receives a temporalized power adjustment process, including the initial skin temperature matrix, initial ambient temperature, initial power, and subsequent power adjustment sequences; the instantaneous skin temperature matrix branch receives a real-time collected single-row temperature matrix time series. The historical adjustment record branch is connected to a Long Short-Term Memory (LSTM) network to extract trend features and temporal dependencies in power adjustment. The instantaneous skin temperature matrix branch is connected to a one-dimensional convolutional layer using a 3×3 convolutional kernel and the ReLU activation function to extract spatial features and rate of change features of the temperature distribution. The outputs of the two branches are fused through a concatenation layer and then connected to two fully connected layers, ultimately outputting a fixed-dimensional fused feature vector. The training of this model is supervised learning. The training data comes from a large amount of historical adjustment records and corresponding skin temperature change data stored locally on the controller. The known historical adjustment records and the instantaneous skin temperature matrix are used as input, and a comprehensive evaluation index representing the system's thermal state, verified by physical rules, is used as the target output. Using mean squared error (MSE) as the loss function, the Adam optimizer was used for iterative training on a large amount of data until the model converged. The trained model was able to generate a fusion feature vector that combined historical experience with real-time status from new device adjustment records and real-time skin temperature matrices.

[0040] It should be noted that the pre-trained steady-state evaluation model is used to compare and analyze the fused feature vectors to determine whether the skin around the eyes has entered a stable state. This model internally presets a steady-state feature threshold, and the core criterion is the temperature values ​​of each sensor in the instantaneous skin temperature matrix compared to the comfort temperature baseline value. The absolute value of the temperature deviation is less than 0.8℃, and the absolute value of the temperature change rate within five consecutive sampling periods is less than 0.02℃ / s. The fused feature vector is input into the model and compared with the preset stable state features. If the above criteria are met, the skin is determined to have entered a stable state; otherwise, it is determined not to have entered a stable state, and the third output power needs to be adjusted according to the preset gradient power value. The preset gradient power value is set to 0.1W-0.3W, preferably 0.2W, and the adjustment direction is determined based on the temperature deviation: when the temperature is below... Then increase the gradient power, and the temperature is higher than The gradient power is then reduced, and the process returns to the step of executing the third output power, repeating the acquisition and judgment process until the skin enters a stable state. After the skin enters a stable state, the skin temperature matrix at this time is recorded as the steady-state distribution of skin temperature. Subsequently, the heating element power matrix is ​​adjusted using a proportional feedback control algorithm. Based on the temperature deviation of each region in the steady-state distribution of skin temperature, the power of the corresponding heating region is fine-tuned until the steady-state distribution of skin temperature reaches uniformity (the temperature deviation of each region does not exceed 0.1℃). The entire process of power matrix change over time is recorded to obtain the optimized power command sequence. The heating element power matrix is ​​a two-dimensional matrix, with the number of rows equal to the total number of samplings (within the current time range) and the number of columns equal to the number of sensors (heating regions, preferably 4). It is smoothed by a moving average (the sliding window length is 5 timestamps) to ensure smooth power changes.

[0041] It should be noted that the pre-trained stable state evaluation model is a feature comparison network (FCN) based on distance metric learning. The input layer of this network receives a fixed-dimensional fused feature vector, generated by the aforementioned adaptive learning model, containing six pieces of information: historical power adjustment trend, current temperature deviation, temperature change rate, temperature distribution variance, number of power adjustments, and power adjustment magnitude. A fully connected layer is then connected, using a sigmoid activation function to map the fused feature vector to the stable state probability space. The model is trained using supervised learning. Training data comes from a large amount of historical adjustment records and corresponding stable state annotation data stored locally on the controller. The fused feature vector generated from known historical adjustment records is used as input, and the label value (0 or 1) representing whether the skin has entered a stable state, verified by physical rules, is used as the target output. A binary cross-entropy loss function is used, and the Adam optimizer is employed for iterative training on a large amount of data until the model converges. The trained model can predict whether the skin around the eyes has entered a stable state from the new fused feature vector, providing a basis for subsequent power adjustment decisions.

[0042] For example, the third output power is executed. Record the initial skin temperature matrix (4 sensors), initial ambient temperature The system collects current power at a frequency of 5Hz, which is used as a record for device adjustments. Matching historical adjustment records are retrieved from the controller, and historical records with initial ambient temperature of 18.0℃-18.2℃ and initial power of 4.0W-4.5W are selected as the fusion basis. The sensor collects real-time skin temperature at a frequency of 5Hz to obtain the time... Instantaneous skin temperature matrix The historical adjustment records and the instantaneous skin temperature matrix are input into the adaptive learning model to generate a fused feature vector. The fused feature vectors are input into the steady-state assessment model, at which point the temperature of each region is compared with the comfort temperature baseline. The deviations were all less than 0.8℃, and the absolute value of the temperature change rate was 0.015℃ / s < 0.02℃ / s for five consecutive sampling periods. The skin was then determined to have entered a stable state, and the skin temperature matrix at this point was recorded as the steady-state skin temperature distribution. The heating element power matrix was adjusted, and a proportional feedback control algorithm (proportional coefficient) was used to address the small deviations in each region of the steady-state distribution. Fine-tune the power until the temperature deviation in each region does not exceed 0.1℃; continuously collect data for 20 seconds (5Hz×20 seconds) to obtain a power matrix with 100 rows (number of collections) and 4 columns. After moving average filtering, a smooth optimized power command sequence is obtained.

[0043] In step S16, according to the optimized power command sequence, the closed-loop control gain of each heating element is adjusted and locked to obtain the final temperature control scheme, including: According to the optimized power command sequence, electrical energy is allocated to the heating element to obtain an energy allocation matrix; The power distribution matrix is ​​mapped and calculated using the impedance characteristic matrix of the preset heating element to obtain the required PWM duty cycle for each heating element; The conduction time of each heating element power transistor is controlled according to the PWM duty cycle, and the heat flux density on the surface of each heating element is collected at a preset frequency to obtain the instantaneous heat flux density distribution. The instantaneous heat flux density distribution is input into a preset comfort temperature field model for thermodynamic simulation to generate a predicted sensory temperature; Based on the predicted sensory temperature, the closed-loop control gain of each heating element is adjusted and locked to form the final temperature control scheme.

[0044] It should be noted that the power allocation matrix is ​​the core carrier for quantifying and allocating the time-series power values ​​of the optimized power command sequence according to the heating area dimension. Its dimension is consistent with the number of heating elements (4-6, preferably 4), and it is a single-row matrix (row vector), defined as: For example, For the first The heating element in the first The power allocation value (in W·s) for each timestamp. The power allocation value follows a linear conversion relationship between time-series power and electrical energy, and the calculation formula is as follows: ,in To optimize the power command sequence, the first... The heating element Target power (in W) for each timestamp. The duration of a single control cycle (in seconds) ensures precise correspondence between power distribution and power commands, adapting to the varying power demands of different heating zones. Power distribution values ​​are calculated separately for each heating element and then combined into a vector. .

[0045] It should be noted that the preset impedance characteristic matrix of the heating element is the core basis for hardware characteristic mapping, and the matrix dimension is... r( (4) The number of heating elements is specified, and the impedance parameters of each heating element in different power ranges are stored, defined by the formula: , among which For the first The heating element corresponds to the first The impedance values ​​(in Ω) for each power range are then determined. The target power needs to be converted to the corresponding duty cycle command. For the first... The required duty cycle for each heating element is: in This is the bus voltage supplied to the heating element, a constant value determined by the hardware circuitry, and its unit is... This formula is derived from the electric power formula, ensuring that the duty cycle is linearly related to the target power, and its value range is limited to between 0 and 1. express Impedance value mapped to the corresponding power range This ensures precise matching between the modulation frequency and power distribution, as well as hardware impedance, reducing power conversion losses and adapting to the operating characteristics of the heating element.

[0046] Get the duty cycle Then, it is compared with a preset high-frequency carrier signal, such as 2kHz, to generate the final pulse width modulation signal. This signal directly controls the on and off states of the corresponding heating element power transistor. On-time With modulation period The relationship is After obtaining the duty cycle, it is compared with a preset fixed-frequency carrier signal (e.g., 2kHz) to generate the corresponding PWM waveform, which is then controlled by adjusting the on-time of the power transistor. This allows for precise adjustment of the target power, thereby avoiding flickering and noise issues caused by low-frequency modulation, ensuring smooth heating of the heating element, and achieving precise tracking of the target power.

[0047] It should be noted that the heat flux density is collected at a preset sampling frequency (5Hz), consistent with the sensor sampling frequency mentioned earlier. The heat flux density data of the heating element surface is obtained in real time through the heat flux density sensor, and the instantaneous heat flux density distribution is in the form of a single-row matrix. ,in For the first Instantaneous heat flux density of the surface of each heating element (unit: W / m³) 2 This provides accurate heat flow data support for subsequent thermodynamic simulations.

[0048] It should be noted that the preset comfort temperature field model is a simulation model built based on thermodynamic finite element analysis. In order to describe the diffusion process of heat after entering the skin in the biological heat conduction equation, this boundary condition acting on the surface needs to be transformed into a source term acting on the micro-elemental body inside the skin. This transformation is based on the physical fact that heat is distributed in the superficial layers of the skin. When heat flows in from the surface, it does not remain on the surface, but is rapidly conducted and accumulated to the thin layer of tissue below. From a macroscopic modeling perspective, it can be assumed that the heat provided by the eye mask is completely absorbed by a very shallow feature depth below the skin surface. The heat is absorbed by the inner tissue layer and uniformly released within this thin layer. Based on this assumption, the heat flux density acting on the surface can be... Equivalent to the volumetric heat generation rate within this thin layer For the ll-th heating region, the transformation relationship is determined by the following formula: in, This is the equivalent depth of heat distribution in the skin, measured in meters (m). It's an empirical parameter based on the physiological structure of the skin around the eyes, typically on the order of millimeters. After this conversion, the heat originally acting on the boundary... This becomes a computational domain source term acting within the skin. The unit also changed accordingly from W / m 2 Convert to W / m 3 .

[0049] After the conversion is complete, the body heat production rate will be... As the source term, it is substituted into the transient heat conduction equation for solution. The core simulation formula is: in The temperature field of the skin tissue is a function of spatial location and time, and its unit is °C. Time, in seconds. Thermal diffusivity of skin tissue (unit: mm) 2 / s), with typical values ​​ranging from 0.05 to 0.30 mm. 2 / s, which varies depending on skin dryness and body part; here, we take 0.15mm. 2 For example, / s; The temperature Laplace operator (unit: °C / m) 2 ), Skin tissue density (unit: kg / m³) 3 ), 1000~1200 kg / m 3 Here, we take 1100 kg / m³. 3 For example; Let J be the specific heat capacity of skin tissue (unit: J / (kg·℃)). The specific heat capacity of skin tissue is 3500~4000 J / (kg·℃), and 3600 J / (kg·℃) is taken as an example here. The second term on the right side of the equation is the volumetric heat source term, which quantifies the contribution of the heat per unit volume generated by the eye mask heating element inside the skin to the rate of temperature change.

[0050] It should be noted that the thermal diffusivity is usually expressed in mm. 2 / s is given in units, but it needs to be converted to SI units (m) for simulation calculations. 2 / s, to ensure dimensional consistency.

[0051] By solving this equation, the model can calculate the temperature changes at various points inside the skin throughout the entire simulation period, thus obtaining a detailed skin temperature field. This temperature field This is an intermediate computational component that describes the complete physical process of heat transfer. Finally, the model is based on this refined temperature field. The distribution characteristics of the temperature sensory data are analyzed, and combined with human thermal perception (e.g., considering the sensitivity weight of the eye area to temperature and the smoothness of temperature changes), a scalar value representing the user's overall thermal sensation is calculated, which is the predicted sensory temperature. . It is not the physical temperature of a point on the skin, but a perceived temperature value that is closer to the user's actual thermal perception than the skin surface temperature alone, providing a core basis for adjusting the gain of subsequent closed-loop control.

[0052] It should be noted that the predicted sensory temperature and the comfort temperature baseline are calculated by adjusting and locking the closed-loop control gain based on the predicted sensory temperature. The absolute value of the difference is used to obtain the target temperature deviation. Preset deviation tolerance The value range is 0.2℃-0.5℃, with 0.5℃ being preferred.

[0053] like Then the closed-loop control gain needs to be adjusted to optimize the control effect. The adjustment method is as follows: the gradient descent method is used to fine-tune the PID gain coefficient, with each adjustment step being 0.05, and the adjustment direction is determined by the sign of the deviation: if > This indicates that the predicted temperature is too high, and the proportional coefficient should be appropriately reduced to decrease the power response intensity. Meanwhile, the integral and derivative coefficients can be temporarily kept unchanged or adjusted proportionally. < If the scaling factor is too high, then increase it. After adjustment, re-execute the heat flux density acquisition, temperature field simulation, and predicted sensory temperature calculation, repeating the above steps until... At this point, the adjusted gain coefficient is locked.

[0054] like If the current power command has reached the comfort requirement, the current closed-loop control gain is locked, i.e., the proportional, integral, and derivative coefficients of the PID controller, and combined with parameters such as the power distribution matrix and PWM duty cycle, a final temperature control scheme is formed.

[0055] In one optional implementation, based on the predicted sensory temperature, the closed-loop control gain of each heating element is adjusted and locked to form a final temperature control scheme, including: The absolute value of the difference between the predicted sensory temperature and the comfort temperature baseline is calculated to obtain the target temperature deviation; If the target temperature deviation is greater than the preset deviation tolerance, the closed-loop control gain coefficient of the heating element is adjusted. If the target temperature deviation is less than the preset deviation tolerance, the optimized power output sequence is determined to have passed verification, the closed-loop control gain coefficient is locked, and the final temperature control scheme is formed.

[0056] When adjusting the closed-loop control gain, prioritize adjusting the proportional gain, as it has the most significant impact on response speed. Set the adjustment step size to 5% of the current proportional gain value (or a fixed step size of 0.05), and adjust in the direction of... > If the proportionality coefficient decreases by 5%, then the proportionality coefficient decreases by 5%. < If the proportionality coefficient increases by 5%, then after each adjustment, the heat flux density is collected again, the temperature field is simulated, and the predicted sensory temperature is calculated again, until... If adjusting the proportional gain still fails to meet the deviation tolerance, the integral and derivative gain coefficients can be further fine-tuned, with the adjustment step size set to 2% of the current value. All gain coefficients must be kept within the preset threshold range, with the proportional gain coefficient at 0.5~1.2, the integral gain coefficient at 0.05~0.2, and the derivative gain coefficient at 0.02~0.1, to prevent system oscillation.

[0057] In summary, the implementation process of this invention takes into account individual differences, environmental influences, and temperature stability, effectively solving the problems of inaccurate temperature control and poor comfort of traditional eye masks, and achieving personalized and precise temperature regulation around the eyes.

[0058] Reference Figure 2 The second embodiment of the present invention provides an intelligent control system for a multifunctional eye mask, comprising: The thermal inertia assessment module is used to continuously acquire the skin temperature around the eyes and the current ambient temperature from the eye mask sensor, and input them into the pre-trained thermal inertia assessment model for analysis to obtain the heat exchange efficiency and the individual thermal inertia coefficient. The two-factor correction module is used to query a preset coefficient-power mapping table based on the individual thermal inertia coefficient to obtain a first output power and a power amplification factor, and adjust the first output power based on the current ambient temperature and the power amplification factor to obtain a second output power; The heat loss calculation module is used to execute the second output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain a skin temperature sequence, query a preset comfort temperature reference table according to the skin temperature sequence to obtain a comfort temperature reference value, and calculate the heat loss power. A heat conduction compensation module is used to convert the heat loss power into a power compensation value, and then superimpose it with the second output power to obtain a third output power. The differential allocation module is used to execute the third output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix, continuously adjust the power of the heating element according to the skin temperature matrix, record the change process of the power matrix over time, and obtain an optimized power command sequence. The simulation verification module is used to adjust and lock the closed-loop control gain of each heating element according to the optimized power command sequence to obtain the final temperature control scheme.

[0059] It should be noted that the intelligent control system for a multifunctional eye mask provided in this embodiment of the invention is used to execute all the process steps of the intelligent control method for a multifunctional eye mask in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0060] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0061] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A smart control method for a multifunctional eye mask, characterized in that, include: The initial skin temperature and current ambient temperature are continuously acquired from the eye mask sensor and input into a pre-trained thermal inertia assessment model for analysis to obtain the heat exchange efficiency and the individual thermal inertia coefficient. The first output power and power amplification factor are obtained by querying the preset coefficient-power mapping table based on the individual thermal inertia coefficient. The first output power is then adjusted based on the current ambient temperature and the power amplification factor to obtain the second output power. The second output power is executed to control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain a skin temperature sequence. The skin temperature sequence is then queried from a preset comfort temperature reference table to obtain a comfort temperature reference value, and the heat loss power is calculated. The heat loss power is converted to obtain a power compensation value, which is then superimposed with the second output power to obtain the third output power. The third output power is executed to control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix. The power of the heating element is continuously adjusted according to the skin temperature matrix, and the power matrix is ​​recorded over time to obtain an optimized power command sequence. Based on the optimized power command sequence, the closed-loop control gain of each heating element is adjusted and locked to obtain the final temperature control scheme.

2. The intelligent control method for a multifunctional eye mask as described in claim 1, characterized in that, The process involves continuously acquiring the skin temperature around the eyes and the current ambient temperature from the eye mask sensor, and inputting this data into a pre-trained thermal inertia assessment model for analysis to obtain heat exchange efficiency and individual thermal inertia coefficients, including: Within a preset time period, the temperature of the skin around the eyes and the current ambient temperature are continuously acquired from the eye mask sensor; Record the fluctuation range of skin temperature around the eyes over time as thermal sensitivity data, and calculate the dynamic change value of the current ambient temperature. The thermosensitive response data is input into the pre-trained thermal inertia assessment model to analyze the heat conduction rate and obtain the heat exchange efficiency. Based on the heat exchange efficiency and the dynamic change value, the thermal hysteresis effect is calculated to obtain the individual thermal inertia coefficient.

3. The intelligent control method for a multifunctional eye mask as described in claim 1, characterized in that, The process of querying a preset coefficient-power mapping table based on the individual thermal inertia coefficient to obtain a first output power and a power amplification factor, and adjusting the first output power based on the current ambient temperature and the power amplification factor to obtain a second output power includes: The corresponding first output power is obtained by querying the preset coefficient-power mapping table based on the individual thermal inertia coefficient. When the individual thermal inertia coefficient is greater than the preset coefficient threshold, the corresponding power amplification factor is obtained by querying the coefficient-power mapping table. Multiplying the power amplification factor by the first output power yields the enhanced power; The current ambient temperature is converted into an additional power value, and then smoothed to obtain the environmental compensation power. The enhanced power and the environmental compensation power are superimposed to obtain the second output power.

4. The intelligent control method for a multifunctional eye mask as described in claim 1, characterized in that, The step of querying a preset comfort temperature reference table based on the skin temperature sequence to obtain a comfort temperature reference value and calculating the heat loss power includes: The instantaneous temperature rise rate calculated based on the skin temperature sequence is matched with a preset comfort temperature reference value table to obtain a comfort temperature reference value. When the absolute value of the difference between the skin temperature sequence and the comfort temperature reference value exceeds a preset fluctuation threshold, the extreme value distribution characteristics of the absolute value of the difference on the time axis are extracted, and the temperature fluctuation amplitude is calculated. When the temperature fluctuation exceeds the preset temperature fluctuation range, the heat loss power is calculated based on the temperature fluctuation. The preset comfort temperature reference value table stores the steady-state temperature of the skin around the eyes that is perceived as comfortable by the user, corresponding to different heating rate ranges.

5. The intelligent control method for a multifunctional eye mask as described in claim 1, characterized in that, The third output power is executed to control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix. Based on this instantaneous skin temperature matrix, the power of the heating element is continuously adjusted, and the change in the power matrix over time is recorded to obtain an optimized power command sequence, including: After executing the third output power, the initial skin temperature matrix and the initial ambient temperature are recorded, and the current power is collected at a preset frequency. These are used together as the device adjustment record, and the historical adjustment record that matches the device adjustment record is queried from the controller. The historical adjustment records and the instantaneous skin temperature matrix are input into a pre-trained adaptive learning model to generate a fused feature vector; The fused feature vector is input into a pre-trained stable state evaluation model for comparison to determine whether the skin has entered a stable state. If yes, then record the skin temperature matrix at this time to obtain the steady-state distribution of skin temperature; if no, then adjust the third output power according to the preset gradient power value, and return to the step of executing the third output power until the skin enters a stable state. Adjust the power matrix of the heating element until the steady-state distribution of the skin temperature becomes uniform, record the change process of the power matrix over time, and obtain the optimized power command sequence; The pre-trained stable state evaluation model internally presets at least two stable state modes and their corresponding allowable fluctuation deviation ranges. It can match the corresponding state mode according to the fused feature vector and calculate the degree of deviation from the corresponding stable state mode to obtain the fluctuation deviation value. When the fluctuation deviation value is less than the preset allowable fluctuation deviation range, it is determined that the skin state has entered a stable state.

6. The intelligent control method for a multifunctional eye mask as described in claim 1, characterized in that, The step of adjusting and locking the closed-loop control gain of each heating element according to the optimized power command sequence to form the final temperature control scheme includes: According to the optimized power command sequence, electrical energy is allocated to the heating element to obtain an energy allocation matrix; The power distribution matrix is ​​mapped and calculated using the impedance characteristic matrix of the preset heating element to obtain the required PWM duty cycle for each heating element; The conduction time of each heating element power transistor is controlled according to the PWM duty cycle, and the heat flux density on the surface of each heating element is collected at a preset frequency to obtain the instantaneous heat flux density distribution. The instantaneous heat flux density distribution is input into a preset comfort temperature field model for thermodynamic simulation to generate a predicted sensory temperature; Based on the predicted sensory temperature, the closed-loop control gain of each heating element is adjusted and locked to form the final temperature control scheme.

7. The method as described in claim 6, characterized in that, The step of adjusting and locking the closed-loop control gain of each heating element based on the predicted sensory temperature to form the final temperature control scheme includes: The absolute value of the difference between the predicted sensory temperature and the comfort temperature baseline is calculated to obtain the target temperature deviation; If the target temperature deviation is greater than the preset deviation tolerance, the closed-loop control gain coefficient of the heating element is adjusted. If the target temperature deviation is less than the preset deviation tolerance, the optimized power output sequence is determined to have passed verification, the closed-loop control gain coefficient is locked, and the final temperature control scheme is formed.

8. An intelligent control system for a multifunctional eye mask, characterized in that, include: The thermal inertia assessment module is used to continuously acquire the initial skin temperature and the current ambient temperature from the goggle sensor and input them into the pre-trained thermal inertia assessment model for analysis to obtain the heat exchange efficiency and the individual thermal inertia coefficient. The two-factor correction module is used to query a preset coefficient-power mapping table based on the individual thermal inertia coefficient to obtain a first output power and a power amplification factor, and adjust the first output power based on the current ambient temperature and the power amplification factor to obtain a second output power; The heat loss calculation module is used to execute the second output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain a skin temperature sequence, query a preset comfort temperature reference table according to the skin temperature sequence to obtain a comfort temperature reference value, and calculate the heat loss power. A heat conduction compensation module is used to convert the heat loss power into a power compensation value, and then superimpose it with the second output power to obtain a third output power. The differential allocation module is used to execute the third output power, control the eye mask sensor to measure the skin temperature around the eyes at a preset frequency to obtain an instantaneous skin temperature matrix, continuously adjust the power of the heating element according to the skin temperature matrix, record the change process of the power matrix over time, and obtain an optimized power command sequence. The simulation verification module is used to adjust and lock the closed-loop control gain of each heating element according to the optimized power command sequence to obtain the final temperature control scheme.

Images