A detection method and system for electrothermal medical beauty equipment

Abstract

The present invention belongs to the field of electromagnetic variable measurement and discloses a detection method and system for electrothermal medical cosmetic equipment; the system comprises a bionic skin model and a main testing center; the skin model is constructed based on multi-layer composite materials and dynamic control technology, a high-resolution camera array is deployed on the outside of the bionic skin model, and a high-precision electro-magnetic-thermal composite sensor array is integrated inside to collect multi-dimensional data and transmit it to a host computer and the main testing center; a pre-inspection safety station is used to perform stable pre-inspection on the output of the cosmetic equipment; the fusion detection module uses a multi-branch shunt method to generate a multimodal data report, uses a machine learning model to conduct in-depth learning on the multimodal data report, and outputs quantitative indicators. The qualified cosmetic equipment is determined based on the quantitative indicators and sent to the host computer, which is used for visual interaction and data storage to realize the detection of electrothermal medical cosmetic equipment.

Description

Technical Field

[0001] The present invention relates to the field of electromagnetic variable measurement, and more specifically, to a detection method and system for electrothermal medical beauty equipment. Background Art

[0002] When testing electrothermal medical beauty equipment, experiments cannot be conducted directly on real people, as this poses certain risks. However, preliminary experiments are required before electrothermal medical beauty equipment is put into mass production. Therefore, a bionic skin dynamic simulation platform is designed to be used as a high-precision bionic test system for testing electrothermal medical beauty equipment. The system simulates the biophysical properties of human skin of different age groups (such as thickness, thermal conductivity, resistivity, water content, elastic modulus, etc.). However, general testing only detects electrothermal properties, and rarely considers magnetic changes.

[0003] Secondly, due to the complex texture of human skin, electricity, heat, and magnetism do not act on the skin independently. Electricity and magnetism will couple, and electricity and heat will also couple. Therefore, the measurement of electricity and magnetism alone will have certain limitations and cannot fully reflect the effects of beauty equipment. Moreover, the measurement equipment may be far from the actual physical and chemical reactions of the skin.

[0004] In view of this, the present invention proposes a detection system for electrothermal medical cosmetic equipment to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solutions: a detection system for electrothermal medical cosmetic equipment, comprising a bionic skin model and a main detection center;

[0006] The bionic skin model is constructed based on multi-layer composite materials and dynamic control technology. It uses a three-layer gradient structure for bionic skin simulation, and the layers are bonded by gradient bonding technology. An external control system is used to simulate the skin state under different environmental conditions. The integrated bionic microcirculatory physiological evolution is deployed, including embedded photosensitive materials, microfluidic chips and integrated graphene heating layers.

[0007] A high-resolution camera array is deployed on the outside of the bionic skin model, and a high-precision electro-magnetic-thermal composite sensor array is integrated inside to form a three-dimensional sensor network, which collects multi-dimensional data and transmits it to the host computer and the main testing center.

[0008] The main testing center includes a pre-inspection safety station and a fusion detection module. The pre-inspection safety station is used to perform stable pre-inspection of the output of beauty equipment. The fusion detection module uses a multi-branch diversion method to perform multi-interactive fusion processing on the collected multi-dimensional data, generating a multimodal data report, pre-defining quantitative detection indicators for beauty equipment, and using a machine learning model to deeply learn the multimodal data report to output quantitative indicators. The machine learning model also introduces a causal chain, which adopts a hierarchical dynamic selection architecture and is divided into a bottom causal layer, a middle causal layer, and a high causal layer.

[0009] The qualification of the beauty equipment is determined based on the quantitative indicators and sent to the host computer, which is used for visual interaction and data storage.

[0010] Preferably, the method for constructing the bionic skin model based on multi-layer composite materials and dynamic control technology includes:

[0011] A three-layer gradient structure is used for bionic skin simulation, including the epidermis, dermis, and subcutaneous tissue layers. The thickness of each layer is dynamically adjusted according to age, and the layers are bonded together using gradient bonding technology.

[0012] The epidermis uses a polymer to simulate the stratum corneum, and a hydrophobic coating is applied to the surface to simulate the skin barrier function. The dermis uses a hydrogel matrix to simulate the density and elastic modulus of collagen fibers. The subcutaneous tissue layer uses fat to simulate gel to regulate water content and thermal conductivity. An external control interface is set up and connected to an external control system to simulate skin conditions under different environmental conditions. Photosensitive materials are embedded in the epidermis or dermis, and the area and concentration of the photosensitive materials are predefined.

[0013] Deploy integrated biomimetic microcirculatory physiological evolution, including embedding a microfluidic chip and an integrated graphene heating layer between the dermis and subcutaneous tissue layers. The microfluidic chip is used to simulate the capillary network, with the network density adjusted according to age group and the fluid medium set to simulate blood, and the flow rate is adjusted through programming. The integrated graphene heating layer is used to simulate the electrical and thermal response of skin tissue;

[0014] The microfluidic chip is driven by an external peristaltic pump and dynamically simulates the thermal conductivity characteristics of the skin under different ambient temperatures or motion conditions by adjusting the flow rate and fluid temperature. A Peltier effect heating / cooling device is used to control the initial temperature of the bionic skin. A micro humidifier and desiccant are used to adjust the water content of the bionic skin.

[0015] Simulating electrical properties on bionic skin, including setting different resistivity and conductivity based on the characteristics of the skin layer;

[0016] The microfluidic flow rate and skin temperature are preset according to needs to simulate the state of the skin under different environmental conditions, and the parameters of the bionic skin are dynamically adjusted according to the output of the beauty device.

[0017] Preferably, the dynamically adjusting the parameters of the bionic skin according to the output of the cosmetic device includes:

[0018] Work outputs scheduled for aesthetic devices include radiofrequency, ultrasound, laser, electrical current stimulation, and hot and cold therapies;

[0019] In response to the heat, current, electromagnetic or mechanical vibrations output by the beauty device, the high molecular polymers and hydrophobic coatings in the epidermis undergo deformation or changes in surface properties; the hydrogel matrix in the dermis undergoes changes in elastic modulus and water content; the fat-mimicking gel in the subcutaneous tissue layer undergoes changes in thermal conductivity and water content; the microcirculation dynamically adjusts the flow rate and fluid temperature through an external control system; and the capillary network density is dynamically adjusted through an external control system to simulate the dynamic process of skin tissue repair or aging.

[0020] Simulate the resistivity and conductivity of each layer of bionic skin; the sensor array in the electromagnetic-thermal composite sensor array collects the multi-dimensional data output by the beauty device acting on the bionic skin model in real time;

[0021] The preset biochemical protection mechanism includes lowering the initial temperature or increasing the water content through a Peltier effect device or a micro humidifier when the temperature of the dermis is too high, so as to simulate the self-protection of the skin; if the mechanical vibration of the subcutaneous tissue layer is too strong, the microfluidic flow rate is reduced through an external control system to simulate the stress relaxation of the skin tissue.

[0022] Preferably, the integrated deployment method of the high-precision electromagnetic-magnetic-thermal composite sensor array includes:

[0023] A high-resolution camera array is deployed on the outside of the bionic skin model to comprehensively monitor changes on its surface. Micro strain sensors are embedded in the dermis and subcutaneous tissue layers to measure local stress changes caused by heat.

[0024] For resistance, current and voltage, electrical variable sensors are combined with micro-electrode arrays and impedance analyzers for measurement; wide-band sweep frequency technology is introduced;

[0025] A TMR magnetoresistive sensor array is used to measure the magnetic field strength and magnetic flux changes. For the temperature of each layer, a thermal field sensor uses a thermocouple array to measure the temperature. A FLIRRepton 3.5 infrared thermal imaging camera is used to collect the surface temperature distribution map in real time.

[0026] The sensor array is evenly distributed in the epidermis, dermis and subcutaneous tissue layers of the bionic skin with a spacing of 5mm, forming a three-dimensional sensing network. The number of sensors is determined according to the size of the test area. It is connected to the data acquisition card through a multiplexer. The data acquisition card uses a high-precision analog-to-digital converter and transmits data in real time to the host computer and the main test center via USB or Wi-Fi.

[0027] Preferably, the pre-inspection safety station is used to perform a stable pre-inspection on the output of the beauty device, and the method includes:

[0028] The pre-inspection safety station uses a high-precision analog-to-digital converter (ADC) combined with a fast Fourier transform algorithm to calculate the total harmonic distortion (THD) of the PWM signal through spectrum analysis.

[0029] If it detects that the duty cycle fluctuation exceeds 0.5% or the THD exceeds 3%, the host computer will issue a warning and switch the mode or shut down the beauty device.

[0030] Preferably, the module uses a multi-branching method to perform multi-interaction fusion processing on the collected multi-dimensional data to generate a multimodal data report, and the method includes:

[0031] The report includes 3D electric field distribution, 3D magnetic field distribution, 3D thermal field distribution and key demonstration characteristics;

[0032] The three-dimensional electric field distribution map is obtained by collecting spatial distribution data of electrical parameters using a microelectrode array and constructing a three-dimensional electric field distribution map of the bionic skin model using a Kriging interpolation method or a finite element analysis algorithm. The electrical parameters include resistance, current, and voltage.

[0033] The method for obtaining a three-dimensional magnetic field distribution map is to denoise the magnetic field intensity and magnetic flux changes collected by the TMR sensor array. Then, a magnetic tomography algorithm is used based on Maxwell's equations and finite element analysis to invert the spatial distribution of the magnetic field in the bionic skin model. The algorithm inputs the magnetic field intensity and magnetic flux changes, and outputs a three-dimensional magnetic field distribution map. Abnormal areas in the magnetic field distribution map are identified using gradient analysis and their spatial locations are marked.

[0034] The method for obtaining the three-dimensional thermal field distribution map is to fuse the point measurement data of the thermocouple array with the surface temperature distribution map of the FLIRRepton3.5 infrared thermal imaging through the Bayesian estimation algorithm to generate a three-dimensional thermal field distribution map;

[0035] Key demo features include:

[0036] The complex impedance of the graphene heating layer was measured using a microelectrode array and an impedance analyzer. The impedance spectrum was analyzed using Nyquist and Bode plots to extract the equivalent circuit parameters of the graphene heating layer. A time series dynamic change curve was established based on the sampling frequency.

[0037] The temperature of each layer is measured at a fixed frequency using a thermocouple array, and the thermal conductivity and thermal diffusivity of the epidermis, dermis, and subcutaneous tissue layers are calculated using the Fourier heat conduction equation.

[0038] A finite element analysis algorithm was used to establish a thermal conduction model of the bionic skin model. The input parameters included the thermal conductivity of each layer, water content, microfluidic flow rate, and fluid temperature. The output was the heat conduction path and heat flux density distribution.

[0039] For the data collected by the TMR magnetoresistive sensor array and the microelectrode array, the Kalman filter algorithm is used to eliminate noise and interference. An electromagnetic field coupling model is established based on Maxwell's equations. Through the coupling model, key parameters are extracted, and a time series dynamic change curve is established according to the sampling frequency.

[0040] The finite element analysis algorithm is used to establish a thermal-mechanical coupling model of the bionic skin model. The input parameters include the thermal expansion coefficient, elastic modulus and thermal stress distribution of each layer of material, and the output is the deformation distribution map and stress concentration area.

[0041] Preferably, the method for determining whether a beauty device is qualified based on quantitative indicators includes:

[0042] The testing of beauty devices is set to include functional evaluation and safety evaluation, and quantitative indicators of functional evaluation and safety evaluation are defined according to the model of beauty devices;

[0043] For multimodal data reports, a hierarchical dynamic selection architecture is used to extract descriptive primitives to represent data features. Predefined machine learning models are used to learn these primitives and output quantitative indicators, introducing causal chains into the machine learning models.

[0044] Calculate the deviation between the quantitative indicator and the predefined effect indicator;

[0045] Deploy a multi-objective comprehensive evaluation, taking a weighted average of the quantitative indicators to obtain a comprehensive score. Predefine hard constraints for the comprehensive score and deviation value. If the beauty device meets all hard constraints, it is judged as qualified and can be put into mass production; otherwise, it is judged as unqualified and requires further design improvement or re-testing.

[0046] Among them, the causal chain adopts a hierarchical dynamic selection architecture, which is divided into the bottom causal layer, the middle causal layer and the high causal layer;

[0047] In the underlying causal layer, the heat output by the beauty device affects the surface temperature distribution map captured by the FLIR Lepton3.5 infrared thermal imaging through the deformation of the high molecular polymer and hydrophobic coating in the epidermis, thereby affecting the three-dimensional thermal field distribution map. The current stimulation changes the elastic modulus and water content of the hydrogel substrate in the dermis, affecting the complex impedance measured by the microelectrode array, thereby affecting the three-dimensional electric field distribution map. The electromagnetic wave changes the thermal conductivity of the fat-simulated gel in the subcutaneous tissue layer, affecting the magnetic field intensity and magnetic flux measured by the TMR magnetoresistive sensor array, thereby affecting the three-dimensional magnetic field distribution map. Mechanical vibration reduces the microfluidic flow rate through an external control system, affecting the temperature values of each layer measured by the thermocouple array, thereby affecting the three-dimensional thermal field distribution map.

[0048] In the middle causal layer, epidermal deformation measures local stress changes through micro-strain sensors, affecting the deformation distribution diagram and stress concentration areas of the thermal-mechanical coupling model. Changes in the elastic modulus and water content of the dermis measure the temperature values of each layer through a thermocouple array, affecting the three-dimensional thermal field distribution diagram. Changes in the flow rate and fluid temperature of the microfluidic chip adjust the density of the capillary network through an external control system, affecting the thermal conduction characteristics. The three-dimensional thermal field distribution diagram is then affected by fusing the thermocouple array data with the infrared thermal imaging data through a Bayesian estimation algorithm. When the temperature of the dermis is too high, the initial temperature or water content is adjusted through a Peltier effect device or a micro-humidifier, affecting the thermal conduction characteristics, and thus affecting the three-dimensional thermal field distribution diagram.

[0049] In the high-level causal layer, the electrical parameters of the three-dimensional electric field distribution map affect the three-dimensional magnetic field distribution map through the Maxwell equations. The abnormal areas of the three-dimensional magnetic field distribution map are identified through the gradient analysis method and affect the three-dimensional electric field distribution map through feedback from the electromagnetic field coupling model. The three-dimensional thermal field distribution map calculates the thermal conductivity and thermal diffusion coefficient through the Fourier heat conduction equation, affecting the heat conduction path and heat flux density distribution of the heat conduction model, and thus affecting the deformation distribution map and stress concentration area of the thermal-mechanical coupling model.

[0050] Preferably, the method of extracting description primitives for representing data features using a hierarchical dynamic selection architecture includes:

[0051] For the numerical data in the multimodal data report, the data were normalized and processed using a multilayer perceptron.

[0052] For the curve data in the multimodal data report, dynamic time warping is first performed to synchronize the time slices with the three-dimensional graph data. A time series Transformer structure is used to output a sequence of feature vectors in the time dimension.

[0053] For three-dimensional graph data in multimodal data reports, numerical data is used as the physical benchmark for cross-modal alignment. The numerical data is expanded into a query vector for the attention mechanism, and temporal features and spatial features are spliced as keys and values to construct a joint spatiotemporal representation. A dual-channel parallel architecture is developed based on sparsity detection. If the sparsity is >70%, CNN local feature extraction is prioritized. If the sparsity is <30%, PCA global dimensionality reduction is enabled, and a gradient penalty term is used to constrain feature redundancy.

[0054] For CNN, the channel importance score is added after each convolutional layer, and the channels with importance scores lower than the expected performance are closed, and octree adaptive sampling is used;

[0055] The features extracted from numerical data, curve data and three-dimensional graph data are recorded as description primitives.

[0056] Preferably, the method for visualization interaction and data storage of the host computer includes:

[0057] The host computer is developed based on an open technology stack and is used to provide a user interface, including setting detection parameters, displaying real-time data, and generating detection reports, supporting data visualization and data storage.

[0058] Preferably, a method for detecting an electrothermal medical cosmetic device comprises:

[0059] Step S1: Construct a bionic skin model. The thickness of each layer and the density of the capillary network are dynamically adjusted according to age groups. An external control interface is set up to connect to an external control system to simulate skin conditions under different environmental conditions. The area and concentration of the photosensitive material, blood flow rate, and temperature are predefined. A Peltier effect device is used to control the initial temperature. A micro humidifier and desiccant are used to adjust the moisture content.

[0060] Step S2: Collect multi-dimensional data through a high-resolution camera array and an electromagnetic-magnetic-thermal composite sensor array, and transmit it to the host computer and the main testing center;

[0061] Step S3: Perform a stable pre-inspection on the beauty device output through a pre-inspection safety station;

[0062] Step S4: The main testing center uses a multi-branch diversion method to perform multi-interaction fusion processing on the multi-dimensional data and generate a multimodal data report;

[0063] Step S5: A hierarchical dynamic selection architecture is used to extract descriptive primitives. A machine learning model is used to learn and output quantitative indicators. Causal chain analysis is introduced to calculate the deviation between the quantitative indicators and predefined effect indicators. A multi-objective comprehensive evaluation is performed to determine whether the equipment meets the hard constraints. If qualified, it can be mass-produced. Otherwise, the design needs to be improved or re-tested.

[0064] Step S6: The host computer performs visual interaction.

[0065] The technical effects and advantages of the detection system of the electrothermal medical beauty equipment of the present invention are as follows:

[0066] 1. This bionic skin model, constructed using multi-layer composite materials and dynamic control technology, simulates the biophysical properties of human skin, including thickness, thermal conductivity, resistivity, water content, and elastic modulus, through a three-layer gradient structure (epidermis, dermis, and subcutaneous tissue). The thickness of each layer and the density of the capillary network can be dynamically adjusted according to age. Combined with a microfluidic chip to simulate blood circulation and a graphene heating layer to simulate electrical and thermal responses, the model realistically reproduces the dynamic reactions of skin under different environmental conditions. Compared to traditional testing systems that use only a single material or a static model, this bionic skin model more closely resembles the physical and chemical properties of real human skin, avoiding the risks of direct testing on the human body. It also significantly improves the authenticity and reliability of test results, providing a more scientific basis for the development and mass production of electrothermal medical cosmetic devices.

[0067] 2. A high-precision electro-magnetic-thermal composite sensor array is integrated into the bionic skin model, forming a three-dimensional sensor network. This network can collect multidimensional data from beauty devices acting on the bionic skin in real time. Traditional detection systems often focus on a single parameter (such as electricity or heat), ignoring multi-field coupling effects such as electro-magnetic, thermal-mechanical, and so on. This design, through multi-dimensional data collection, comprehensively reflects the combined effects of beauty devices on the skin, overcoming the limitations of traditional detection methods and providing more comprehensive data support for device performance evaluation.

[0068] 3. A multi-branching approach uses multi-interaction fusion processing on the collected multidimensional data. This is combined with a hierarchical dynamic selection architecture to extract descriptive primitives. A machine learning model incorporates causal chain analysis, enabling accurate assessment of the functionality and safety of beauty devices. Compared to traditional detection systems that rely on manual analysis or simple statistics, this design, through data fusion and intelligent analysis, automatically identifies abnormal areas in the data (such as magnetic field anomalies and thermal stress concentrations). It then reveals the inherent correlation between device output and skin reactions through a causal chain (bottom, middle, and top layers). This approach not only improves assessment accuracy but also significantly reduces human error, providing scientific guidance for optimized device design.

[0069] In summary, this method addresses existing issues such as insufficient detection accuracy, single-source data, neglected coupling effects, and potential safety risks. Its technical effectiveness and advantages not only enhance the scientific nature and reliability of electrothermal medical cosmetic device testing but also provide strong technical support for device R&D, optimization, and mass production, demonstrating significant innovation and practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1This is a schematic structural diagram of a detection system for an electrothermal medical beauty device according to the present invention;

[0071] Figure 2 This is a schematic diagram of the flow of a detection system for an electrothermal medical beauty device of the present invention;

[0072] Figure 3 This is a schematic diagram of a detection method for an electrothermal medical beauty device of the present invention. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] Example 1

[0075] See also Figure 1 、 Figure 2 and Figure 3 As shown, this embodiment is a detection system for electrothermal medical beauty equipment, the main design contents of which are:

[0076] Methods for constructing bionic skin models based on multilayer composite materials and dynamic control technology include:

[0077] A three-layer gradient structure is used for bionic skin simulation, including the epidermis, dermis, and subcutaneous tissue layers. The thickness of each layer is dynamically adjusted according to age, and the layers are bonded together using gradient bonding technology.

[0078] For example:

[0079] 20-30 years old: epidermis 0.1mm, dermis 1.5mm, subcutaneous tissue 3.0mm.

[0080] 40+ years old: epidermis 0.15mm, dermis 1.2mm, subcutaneous tissue 2.5mm.

[0081] Gradient bonding technology is used to achieve seamless transition between layers, ensuring continuity of thermal conduction and electrical properties.

[0082] The epidermis uses a high molecular weight polymer (such as polydimethylsiloxane (PDMS)) to simulate the stratum corneum, and a hydrophobic coating is applied to the surface to simulate the skin barrier function. The dermis uses a hydrogel matrix (such as polyethylene glycol (PEG)) to simulate the density and elastic modulus of collagen fibers. The subcutaneous tissue layer uses a fat-simulated gel (such as a silicone-based composite material) to regulate the water content and thermal conductivity. An external control interface is set up and connected to an external control system to simulate the skin state under different environmental conditions. Photosensitive materials (such as particles simulating melanin) are embedded in the epidermis or dermis, and the area and concentration of the photosensitive material are predefined.

[0083] For example:

[0084] 20-30 years old: Collagen fiber density 120mg / cm³, elastic modulus 2.5MPa; water content 70%, thermal conductivity 0.25W / m·K, low concentration of photosensitive material is set at the corners of the eyes and cheekbones.

[0085] 40+ years old: Collagen fiber density 60mg / cm³, elastic modulus 0.5MPa; water content 50%, thermal conductivity 0.18W / m·K, high concentration of photosensitive material is set at the corners of the eyes and cheekbones.

[0086] Through material formulation and processing techniques (such as 3D printing and microfluidic perfusion), dynamic matching of parameters of each layer is achieved. Using external control systems (such as temperature control modules and humidity regulation modules), the water content, thermal conductivity, and elastic modulus of the bionic skin are adjusted in real time to simulate the skin state under different environmental conditions.

[0087] Deploy integrated biomimetic microcirculatory physiological evolution, including embedding a microfluidic chip and an integrated graphene heating layer between the dermis and subcutaneous tissue layers. The microfluidic chip is used to simulate the capillary network. The microfluidic channel is fabricated using photolithography technology, with a channel diameter of 50-100μm, and the network density is adjusted according to age group (higher density in young skin, lower density in mature skin). The fluid medium (physiological saline or biomimetic fluid containing red blood cells) is set to simulate blood, and the flow rate is programmable (range 0-15mL / min). The integrated graphene heating layer is used to simulate the electrical and thermal response of skin tissue.

[0088] Driven by an external peristaltic pump, the microfluidic chip simulates the absorption and diffusion of heat by blood circulation. By adjusting the flow rate and fluid temperature (flow rate range 0-15mL / min, temperature range 20°C-40°C), the thermal conductivity characteristics of skin under different ambient temperatures or motion conditions are dynamically simulated. A Peltier effect heating / cooling device is used to control the initial temperature of the bionic skin (range 20°C-40°C) with an accuracy of ±0.1°C, allowing simulation of different ambient temperatures or initial skin conditions. A micro-humidifier and desiccant are used to adjust the water content of the bionic skin (range 30%-80%), simulating the changes in skin properties under different humidity environments.

[0089] For example:

[0090] Thermal conductivity is dynamically matched via thermal diffusivity:

[0091] Young skin: Thermal diffusivity 0.25W / m·K, simulating efficient heat diffusion.

[0092] Mature skin: Thermal diffusivity is 0.18 W / m·K, simulating lower heat diffusion capacity.

[0093] Simulating electrical properties on bionic skin, including setting different resistivity and conductivity based on the characteristics of the skin layer;

[0094] For example:

[0095] For resistivity:

[0096] The resistivity of the epidermis is relatively high (simulating the insulating properties of the stratum corneum), about 10 5 Ω·m.

[0097] The resistivity of the dermis and subcutaneous tissue layer is low (simulating tissue with a high water content), about 10²Ω·m.

[0098] For electrical conductivity: The electrical conductivity of each layer can be regulated by doping conductive particles (such as carbon nanotubes and graphene) into the hydrogel matrix, fat-mimicking gel and polymer.

[0099] Using external electric fields or ion injection technology, the resistivity and conductivity of the bionic skin are dynamically adjusted to simulate the changes in the electrical properties of the skin under different humidity or electrical stimulation.

[0100] The microfluidic flow rate and skin temperature are preset according to needs to simulate the state of the skin under different environmental conditions (exercise or rest), and the parameters of the bionic skin are dynamically adjusted according to the output of the beauty device to simulate the operation of the beauty device in different usage scenarios.

[0101] Dynamically adjust the parameters of the bionic skin based on the output of the beauty device, including:

[0102] Work outputs scheduled for aesthetic devices include radiofrequency, ultrasound, laser, electrical current stimulation, and hot and cold therapies;

[0103] Beauty devices typically have multiple operating modes, including radio frequency (RF), ultrasound, laser, current stimulation, hot and cold therapy, etc. The output of these modes acts on the bionic skin model in different forms of energy (such as electromagnetic energy, thermal energy, mechanical energy, etc.).

[0104] The working mechanism of various mode outputs is as follows:

[0105] 1. Radiofrequency mode: Radiofrequency devices output high-frequency alternating current (usually in the range of 0.3MHz to 10MHz), which acts on the skin through electrodes, generating electromagnetic fields and Joule heat.

[0106] Acting on bionic skin model:

[0107] Electromagnetic field effect: The electromagnetic field generated by the radiofrequency (RF) acts on the epidermis, dermis, and subcutaneous tissue layers of the bionic skin. Because each layer has different resistivity and conductivity (predetermined by material design), the electromagnetic field causes changes in charge distribution, simulating the electrical response of skin tissue.

[0108] Thermal Effects: Joule heating generates heat in the dermis and subcutaneous tissue layers, and heat distribution is affected by the thermal conductivity and water content of each layer. An integrated, high-precision electromagnetic-thermal hybrid sensor array measures thermal field distribution (via thermocouple arrays and infrared thermal imaging) and electrical parameter changes (via microelectrode arrays) in real time.

[0109] Microcirculation simulation: The radiofrequency thermal effect drives the fluid (simulating blood) in the microfluidic chip through an external peristaltic pump, adjusts the flow rate and fluid temperature, and dynamically simulates the accelerated microcirculation of the skin after heating.

[0110] 2. Ultrasonic mode: The ultrasonic device outputs high-frequency mechanical vibrations (usually in the range of 1MHz to 10MHz), which act on the skin through the transducer to generate mechanical waves and local thermal effects.

[0111] Acting on bionic skin model:

[0112] Mechanical Wave Effect: Ultrasonic vibrations act on the various layers of the bionic skin, particularly the hydrogel matrix (simulating collagen fibers) in the dermis and the fat-simulating gel in the subcutaneous tissue. These vibrations induce minute deformations within the material, simulating the mechanical response of skin tissue.

[0113] Thermal Effects: Ultrasound waves generate heat due to tissue impedance during propagation, particularly at the interface between the dermis and subcutaneous tissue. Thermocouple arrays and infrared thermal imaging measure heat distribution in real time.

[0114] Microcirculation simulation: The mechanical vibration of ultrasound will simulate the vibration effect of the capillary network through the microfluidic chip, and the flow rate may be adjusted through an external control system to simulate the enhancement of blood circulation.

[0115] 3. Laser mode: The laser device outputs light energy of a specific wavelength (such as 532nm, 1064nm, etc.), which acts on the skin through photothermal or photochemical effects.

[0116] Acting on bionic skin model:

[0117] Photothermal effect: Laser energy is absorbed by the high-molecular-weight polymer in the epidermis (simulating the stratum corneum) and the hydrogel matrix in the dermis, converting it into heat. The heat distribution is monitored in real time using a thermocouple array and infrared thermal imaging.

[0118] Photochemical effect: By embedding photosensitive materials (such as particles that simulate melanin) into the epidermis or dermis to simulate skin aging or pigmentation, the laser action will trigger a photochemical reaction, simulating the pigment decomposition or tissue repair of skin tissue.

[0119] Microcirculation simulation: The laser thermal effect will adjust the flow rate and fluid temperature of the microfluidic chip through an external control system to simulate the enhancement of local blood circulation.

[0120] 4. Current stimulation mode: The current stimulation device outputs low-frequency or medium-frequency current (such as microcurrent, iontophoresis, etc.), which acts on the skin through electrodes to promote tissue repair or substance penetration.

[0121] Acting on bionic skin model:

[0122] Electrical Effect: Electric current flows through the epidermis, dermis, and subcutaneous tissue layers. Due to the different resistivity and conductivity of each layer, the current distribution varies. A microelectrode array measures changes in resistance, current, and voltage in real time.

[0123] Material penetration simulation: If iontophoresis is simulated, a solution of simulated drugs or cosmetic ingredients can be coated on the surface of the epidermis, and the material penetration can be simulated through the gradient adhesive layer under the action of electric current.

[0124] Microcirculation simulation: Current stimulation will regulate the flow rate of the microfluidic chip through an external control system, simulating the enhancement of blood circulation.

[0125] 5. Hot and cold therapy mode: changes the skin surface temperature through heating or cooling modules (such as Peltier effect devices).

[0126] Acting on bionic skin model:

[0127] Thermal Effect: The heating mode increases the surface temperature of the bionic skin, which is then transferred to the dermis and subcutaneous tissue layers through thermal conduction. Thermocouple arrays and infrared thermal imaging monitor heat distribution in real time.

[0128] Cooling effect: The cooling mode will lower the surface temperature of the bionic skin and may adjust the water content through micro humidifiers and desiccants to simulate the contraction or moisturizing response of the skin in a low temperature environment.

[0129] Microcirculation simulation: Hot and cold stimulation will adjust the flow rate and fluid temperature of the microfluidic chip through an external control system to simulate the dynamic changes of blood circulation.

[0130] The bionic skin model, through its multi-layer composite material design, dynamic control technology and integrated high-precision sensor array, can autonomously respond to the output of the beauty device. The specific response mechanism is as follows:

[0131] The heat, current, electromagnetic or mechanical vibrations output by beauty devices can cause deformation or changes in surface properties of the epidermal polymers and hydrophobic coatings. For example, heat can cause changes in the surface tension of the hydrophobic coating, simulating the dynamic adjustment of the skin's barrier function. Current can change the resistivity of the epidermal layer, simulating the skin's electrical barrier function.

[0132] The hydrogel matrix of the dermis produces changes in elastic modulus and water content; for example, the thermal effect of radiofrequency or ultrasound may cause water migration inside the hydrogel, simulating the thermoplastic changes of collagen fibers; current stimulation may change the ion distribution of the hydrogel, simulating tissue repair.

[0133] The fat-mimicking gel in the subcutaneous tissue layer produces changes in thermal conductivity and water content; for example, the thermal effect of laser or radiofrequency may cause local "melting" of the fat-mimicking gel or evaporation of water, simulating the thermal response of subcutaneous fat.

[0134] Microcirculation dynamically regulates flow rate and fluid temperature through external control systems; for example, the thermal effect of radiofrequency or laser may increase the flow rate through program control, simulating the acceleration of blood circulation after the skin is heated; the cold therapy mode may reduce the flow rate, simulating the slowdown of blood circulation in the skin at low temperatures.

[0135] The capillary network density of the microfluidic chip can be preset based on age (e.g., higher density in younger skin, lower density in older skin). The output of the cosmetic device (e.g., ultrasound or electrical stimulation) could dynamically adjust the capillary network density through an external control system, simulating the dynamic process of skin tissue repair or aging.

[0136] The resistivity and conductivity of each layer of biomimetic skin are simulated; a microelectrode array measures these changes in real time. For example, electrical stimulation can cause the resistivity of the epidermis to decrease, simulating a temporary weakening of the skin's barrier function; while radiofrequency electromagnetic fields can increase the conductivity of the dermis, simulating electrical activation of skin tissue. Based on these changes in electrical parameters, an external control system dynamically adjusts the microfluidic flow rate or the electrical properties of the material layer, simulating the skin's dynamic response to electrical stimulation.

[0137] The electromagnetic-thermal hybrid sensor array (including microelectrode arrays, thermocouple arrays, fluxgate sensors, infrared thermal imaging, etc.) collects multidimensional data (resistance, current, voltage, magnetic field, temperature, etc.) output by the beauty device acting on the bionic skin model in real time. This data is transmitted to the host computer via a multiplexer (MUX) and a high-precision analog-to-digital converter (ADC).

[0138] The preset biochemical protection mechanism includes: when the output of the beauty device (radio frequency) causes the temperature of the dermis to be too high, the initial temperature is reduced or the water content is increased through a Peltier effect device or a micro humidifier to simulate the skin's self-protection; if the output of the ultrasonic device causes the mechanical vibration of the subcutaneous tissue layer to be too strong, the microfluidic flow rate is reduced through an external control system to simulate the stress relaxation of the skin tissue.

[0139] Methods for integrating a high-precision electro-magnetic-thermal composite sensor array into a bionic skin model include:

[0140] To achieve real-time acquisition of multi-dimensional data, a high-precision electromagnetic-thermal composite sensor array is integrated into the bionic skin model.

[0141] A high-resolution camera array is deployed on the outside of the bionic skin model to comprehensively monitor surface changes (e.g., mechanical deformation) of the bionic skin model; the deformation measurement accuracy is 0.01mm.

[0142] Embedding micro strain sensors (e.g., fiber Bragg grating sensors, FBGs) in the dermis and subcutaneous tissue layers to measure local stress changes caused by heat. The strain measurement range is ±5000 με with a resolution of 1 με.

[0143] For resistance, current and voltage, the electrical variable sensor is combined with a micro-electrode array and an impedance analyzer for measurement; for example, the measurement range is: resistance 10Ω-10 6Ω, current 0.1mA-100mA, voltage 0.1V-50V; wide-band frequency sweep technology is introduced to measure the impedance changes of the graphene heating layer and evaluate the dynamic response of its electrical and thermal properties. A wide frequency sweep of 10Hz to 1MHz covers the low-frequency (current stimulation) and high-frequency (RF) output ranges commonly found in beauty devices. The sweep resolution is 0.1Hz, ensuring high-precision capture of impedance changes.

[0144] Use TMR magnetoresistive sensor arrays to measure magnetic field strength and magnetic flux changes; for example, the measurement range: magnetic field strength 0.01μT-1mT, accuracy ±0.1μT.

[0145] The thermal field sensor uses a thermocouple array to measure the temperature of each layer; for example, the measurement range is 0°C-100°C with an accuracy of ±0.1°C. A FLIRR Lepton 3.5 infrared thermal imaging camera is used to collect real-time surface temperature distribution maps. The resolution is 160×120 pixels, and the measurement range is -10°C-140°C with an accuracy of ±0.5°C. The thermal imaging layer is overlaid on the surface of the bionic skin to collect real-time surface temperature distribution maps.

[0146] The sensor array is evenly distributed across the epidermis, dermis, and subcutaneous tissue layers of the bionic skin at 5mm spacing, forming a three-dimensional sensor network. The number of sensors is determined by the size of the test area; for example, 400 sensor nodes are deployed within a 10cm x 10cm test area. These nodes are connected to a data acquisition card via a multiplexer (MUX) with a sampling frequency of 100Hz. The data acquisition card uses a high-precision analog-to-digital converter (ADC) with a resolution of 16 bits. The data is transmitted in real time via USB or Wi-Fi to a host computer and the main testing center for subsequent analysis and visualization.

[0147] Traditional testing methods focus solely on electrical variables and thermal field data, but often overlook the magnetic variations caused by electrical heating. Therefore, this solution innovatively incorporates magnetic variable measurement to assess the electromagnetic compatibility of beauty devices during operation and the electromagnetic properties of graphene materials. This not only enhances comprehensive testing but also provides data support for electromagnetic optimization design of next-generation beauty devices.

[0148] Case: Radiofrequency beauty equipment acting on a bionic skin model

[0149] Beauty equipment output: The radio frequency device outputs at a frequency of 1MHz and a power of 50W, acting on the surface of the bionic skin model.

[0150] Acting on bionic skin model:

[0151] Electromagnetic field effect: The radiofrequency electromagnetic field penetrates the epidermis and acts on the dermis and subcutaneous tissue layers. Because the hydrogel matrix of the dermis has high electrical conductivity, the electromagnetic field causes charge redistribution within the dermis. The microelectrode array measures the changes in resistance and current in real time.

[0152] Thermal Effect: RF energy is converted into heat in the dermis and subcutaneous tissue layers. Thermocouple arrays measured dermal temperature increases of up to 42°C and subcutaneous tissue temperature increases of up to 40°C. Infrared thermal imaging generated a surface temperature distribution map, showing that heat is concentrated in the dermis.

[0153] Microcirculatory response: The external control system automatically increases the flow rate of the microfluidic chip (from 5 μL / min to 10 μL / min) based on the heat distribution data, simulating accelerated blood circulation to dissipate heat.

[0154] Autonomous response of the bionic skin model:

[0155] Material-level reactions: The hydrogel matrix in the dermis expands due to heat, reducing its elastic modulus and simulating the thermoplastic changes of collagen fibers. The fat-simulated gel in the subcutaneous tissue layer evaporates due to heat, reducing its thermal conductivity and simulating the thermal barrier function of subcutaneous fat.

[0156] Dynamic control: After the host computer analyzes the sensor data, if it detects that the temperature of the dermis exceeds the safety threshold (for example, 45°C), it will lower the initial temperature of the bionic skin through the Peltier effect device, or increase the water content of the dermis through a micro humidifier to simulate the skin's self-protection mechanism.

[0157] Electrical property adjustment: The radiofrequency electromagnetic field increases the electrical conductivity of the dermis, and the microelectrode array measures a decrease in resistance of approximately 20%. The host computer adjusts the microfluidic flow rate based on this change, simulating the enhanced blood circulation in the skin tissue caused by electrical stimulation.

[0158] The main testing center includes a pre-inspection safety station, which is used to perform stable pre-inspection on the output of beauty equipment;

[0159] Beauty devices (such as those using radiofrequency or current stimulation) typically output high-frequency pulse-width modulation (PWM) signals to control the intensity and pattern of energy output. The pre-inspection safety station assesses the stability and accuracy of the device's output by capturing and analyzing the duty cycle, frequency, and waveform characteristics of the PWM signal in real time.

[0160] The pre-inspection safety station uses a high-precision analog-to-digital converter (ADC) (16-bit resolution, sampling frequency over 100Hz), combined with a fast Fourier transform algorithm, to analyze duty cycle fluctuations of PWM signals in real time with an accuracy of 0.1%. For example, for a 1MHz PWM signal output by a radio frequency device, the system can detect a slight change in duty cycle from 50% to 50.1%.

[0161] Spectral analysis calculates the total harmonic distortion (THD) of the PWM signal to assess the uniformity of the device's output energy and ensure the signal quality of the beauty device. A THD of ≤ 3% indicates low harmonic interference in the device's output signal, preventing unintended electrical or thermal effects on the bionic skin model.

[0162] If it detects that the duty cycle fluctuation exceeds 0.5% or the THD exceeds 3%, the host computer will issue a warning, indicating that the device may have an output instability problem, and switch the mode or shut down the beauty device to protect the skin tissue.

[0163] The electro-magnetic-thermal composite sensor array is a core component of the bionic skin dynamic simulation platform. It is used to collect, analyze, and process multi-dimensional data (electrical, magnetic, and thermal parameters) output by beauty devices acting on the bionic skin model in real time. This high-precision sensor array collects data, enabling comprehensive monitoring and feedback of the bionic skin model's dynamic response.

[0164] Perform multi-modal fusion processing on the collected multi-dimensional data, that is, integrate electrical parameters (resistance, current, voltage), magnetic parameters (magnetic field strength, magnetic flux change) and thermal parameters (temperature, thermal conductivity) to generate a comprehensive multimodal data report. The methods include:

[0165] The report includes 3D electric field distribution, 3D magnetic field distribution, 3D thermal field distribution and key demonstration characteristics;

[0166] The method for obtaining the three-dimensional electric field distribution map is to collect the spatial distribution data of electrical parameters through a microelectrode array (distributed in the epidermis, dermis and subcutaneous tissue layer), and use the Kriging interpolation method or finite element analysis algorithm to construct a three-dimensional electric field distribution map of the bionic skin model. The electrical parameters include resistance, current and voltage; the map can intuitively display the spatial distribution of electric field intensity, current density and resistivity.

[0167] In current stimulation beauty device testing, the system uses electric field distribution maps to evaluate the penetration depth and uniformity of current in a bionic skin model, thereby verifying the effectiveness of the beauty device.

[0168] The spatial distribution of magnetic fields is used to monitor and analyze the magnetic fields output by beauty devices (such as the alternating magnetic fields generated by radiofrequency devices) and the magnetic response of bionic skin models (such as magnetic flux changes and magnetic flux leakage distribution). The high-frequency electromagnetic fields output by beauty devices (such as radiofrequency devices) generate alternating magnetic fields, which may affect the electrical and thermal properties of bionic skin models. Tunnel magnetoresistance (TMR) sensor arrays are used to measure magnetic field strength and magnetic flux changes in real time, with a sensitivity of 1nT, capable of capturing weak magnetic flux leakage signals.

[0169] The (32-channel) TMR magnetoresistive sensor array acquires magnetic field parameters with a measurement range of ±100 μT and a resolution of 1 nT, making it suitable for detecting alternating magnetic fields generated by RF devices (typically in the 10 nT to 10 μT range). The sampling frequency is 100 Hz, with a resolution of 16 bits. Magnetic field parameters include magnetic field strength (B) and magnetic flux change (dB / dt).

[0170] During RF beauty device testing, a TMR sensor array monitors the alternating magnetic field distribution generated by the RF electromagnetic field. For example, if the magnetic field intensity at the dermis junction exceeds 50μT (which may cause tissue overheating), the device may be susceptible to irritation.

[0171] In order to intuitively analyze the spatial distribution of the magnetic field in the bionic skin model, a magnetic tomography algorithm was used to construct a three-dimensional magnetic field distribution map based on the measured magnetic field intensity and magnetic flux changes, and to locate the abnormal magnetic leakage area.

[0172] The method for obtaining a three-dimensional magnetic field distribution map is to denoise the magnetic field intensity and magnetic flux changes (magnetic field data) collected by the TMR sensor array. For example, the wavelet transform algorithm is used to remove high-frequency noise to ensure data quality. The magnetic tomography algorithm is used based on Maxwell's equations and finite element analysis to invert the spatial distribution of the magnetic field in the bionic skin model. The algorithm inputs the magnetic field intensity and magnetic flux changes, and outputs a three-dimensional magnetic field distribution map (resolution 1mm). The gradient analysis method is used to identify abnormal areas in the magnetic field distribution map (such as sudden changes in magnetic field intensity or leakage magnetic areas) and mark their spatial positions; abnormal leakage magnetic areas are located: abnormal leakage magnetic areas may correspond to uneven output of beauty equipment or internal material defects of the bionic skin model.

[0173] During RF beauty device testing, a three-dimensional magnetic field distribution map is used to assess the penetration depth and uniformity of the RF electromagnetic field. For example, if an abnormal magnetic leakage area (a sudden change of more than 20% in magnetic field intensity) is detected at the junction of the dermis and skin, the device may have an uneven output.

[0174] Thermal parameter analysis is used to monitor and analyze in real time the heat output of beauty devices (such as the thermal effects generated by radio frequency, laser, and ultrasound) and the thermal response of bionic skin models (such as temperature distribution and thermal conductivity characteristics); the thermal response of bionic skin models is a key indicator for evaluating the safety and effectiveness of beauty devices.

[0175] The method for obtaining a three-dimensional thermal field distribution map is to fuse the point measurement data of the thermocouple array with the surface temperature distribution map of the FLIRRepton3.5 infrared thermal imaging through a Bayesian estimation algorithm to generate a three-dimensional thermal field distribution map; the fused thermal field map has a resolution of 1mm and can accurately reflect the conduction, diffusion and dissipation process of heat in the bionic skin model.

[0176] During laser beauty device testing, a 3D thermal field map is used to assess the penetration depth and uniformity of laser heat within a bionic skin model. For example, if the dermis temperature exceeds 45°C (potentially causing tissue damage), the device is deemed to be over-energy and exceeding the specified limit.

[0177] Key demo features include:

[0178] The complex impedance (including resistance and reactance) of the graphene heating layer is measured using a microelectrode array and an impedance analyzer. The impedance resolution reaches 0.1Ω, capable of detecting small impedance changes caused by temperature, humidity, or mechanical vibration. The impedance spectrum is analyzed using Nyquist and Bode plots to extract the equivalent circuit parameters of the graphene heating layer (such as resistance, capacitance, and inductance). A time-series dynamic change curve is created based on the sampling frequency to evaluate the stability of its electrical performance.

[0179] During testing of RF or current-stimulated beauty devices, impedance spectroscopy is used to monitor the electrical response of the graphene heating layer. For example, if the RF device output causes the graphene heating layer to heat up, the impedance may drop by approximately 5% (due to increased conductivity caused by increased temperature), simulating the heat loss effect of the skin.

[0180] The thermal conductivity properties of the bionic skin model (such as thermal conductivity and thermal diffusivity) change dynamically depending on the heat output from the cosmetic device, the moisture content of the material layer, and the microfluidic flow rate. Dynamic analysis of thermal conductivity can assess the heat conduction behavior in the bionic skin model.

[0181] The temperature of each layer is measured at a fixed frequency using a thermocouple array. The thermal conductivity (unit: W / m·K) and thermal diffusivity (unit: m² / s) of the epidermis, dermis, and subcutaneous tissue layers are calculated using the Fourier heat conduction equation.

[0182] A finite element analysis algorithm was used to establish a thermal conduction model for the bionic skin model. The input parameters included the thermal conductivity of each layer (material), water content, microfluidic flow rate, and fluid temperature. The output was the heat conduction path (curve) and heat flux density distribution (distribution map).

[0183] In the testing of hot and cold therapy cosmetic devices, dynamic analysis of heat conduction characteristics is used to evaluate the thermal effects of hot and cold stimuli on the bionic skin model. For example, if the thermal diffusivity of the subcutaneous tissue layer is detected to be too low in cooling mode (potentially leading to tissue frostbite), the device can be used to evaluate the thermal effects of hot and cold stimuli on the bionic skin model.

[0184] The electromagnetic field output by beauty devices typically contains both electric and magnetic components, which can produce coupling effects (such as electromagnetic induction and eddy currents) in bionic skin models. To fully assess the impact of electromagnetic fields, a coupling analysis of the magnetic and electric fields was performed.

[0185] For data collected by the TMR magnetoresistive sensor array and microelectrode array, a Kalman filter algorithm is used to eliminate noise and interference, improving data accuracy. An electromagnetic field coupling model is established based on Maxwell's equations to calculate the interaction between electric and magnetic fields in the bionic skin model. For example, the alternating magnetic field output by the radiofrequency device may induce eddy currents in the dermis, resulting in localized thermal effects. Through the coupling model, key parameters such as eddy current density (unit: A / m²), electromagnetic induction intensity (unit: V / m), and local thermal power density (unit: W / m³) are extracted, and a time-series dynamic change curve is established based on the sampling frequency.

[0186] During RF beauty device testing, electromagnetic field coupling analysis is used to assess the thermal impact of eddy currents on the bionic skin model. For example, if the eddy current density in the dermis is too high, it may cause overheating of the skin tissue.

[0187] Heat output from cosmetic devices can cause thermal stress and mechanical deformation (e.g., expansion and contraction) in the material layer of the bionic skin model, simulating the thermomechanical response of skin tissue. The effects of thermal stress and mechanical deformation on the structural stability of the bionic skin model are assessed using a sensor array and mechanical modeling.

[0188] A finite element analysis algorithm was used to establish a thermal-mechanical coupling model of the bionic skin model. The input parameters included the thermal expansion coefficient, elastic modulus, and thermal stress distribution of each layer of material, and the output was a deformation distribution map and stress concentration areas.

[0189] During RF beauty device testing, thermal stress and mechanical deformation analysis are used to assess the structural impact of heat on the bionic skin model. For example, if excessive thermal deformation of the dermis is detected (potentially leading to tissue damage), the host system will reduce the microfluidic flow rate through an external control system to simulate stress relaxation in the skin tissue.

[0190] The testing of beauty devices is set to include functional evaluation and safety evaluation, and quantitative indicators of functional evaluation and safety evaluation are defined according to the model of beauty devices;

[0191] For example, quantitative indicators for functional assessment include: dermal temperature increase (target: 40-45°C) and collagen fiber density increase of 10% in radiofrequency mode; subcutaneous fat decomposition rate of 5% and mechanical vibration intensity of 2Pa in ultrasound mode; and optical response intensity of epidermal photosensitive materials of 3 and pigment decomposition rate of 7% in laser mode. Quantitative indicators for safety assessment include: epidermal temperature not exceeding 45°C to avoid burns; dermal stress not exceeding 2 MPa to avoid collagen fiber breakage; subcutaneous vibration intensity not exceeding 0.5 m / s² to avoid fat tissue damage; and electromagnetic field strength meeting International Commission on Non-Ionizing Radiation Protection (ICNIRP) standards.

[0192] Functional evaluations comprehensively assess the device's performance under different operating modes (e.g., thermal, electrical, and mechanical responses of radiofrequency, ultrasound, and laser devices to the skin) through multimodal data reporting, comparing this performance with the intended cosmetic effects (e.g., firming, repair, and anti-aging). Safety evaluations, based on the dynamic response of a bionic skin model, verify that the device's output remains within a safe range to avoid damage to skin tissue (e.g., overheating, excessive mechanical vibration, and electromagnetic interference).

[0193] You can also add device stability, repeatability, environmental adaptability and user experience optimization.

[0194] For multimodal data reports, a hierarchical dynamic selection architecture is used to extract descriptive primitives to represent data features. Predefined machine learning models are used to learn these primitives and output quantitative indicators. Causal chains are introduced into the machine learning models (to analyze causal reasoning between data).

[0195] Calculate the deviation between the quantitative indicator and the predefined effect indicator;

[0196] Deploy a multi-objective comprehensive evaluation, perform a weighted average of the quantitative indicators to obtain a comprehensive score, and pre-define hard constraints for the comprehensive score and deviation value (such as a safety score SS, SS ≥ 90 points, and a temperature deviation value CPS ≤ 3). If the beauty device meets all hard constraints, it is judged as qualified and can be put into mass production; otherwise, it is judged as unqualified and requires further design improvement or re-testing.

[0197] The input data of the causal chain comes from multimodal data reports, including three-dimensional electric field distribution maps, three-dimensional magnetic field distribution maps, three-dimensional thermal field distribution maps and key demonstration characteristics; it adopts a hierarchical dynamic selection architecture, which is divided into the bottom causal layer, the middle causal layer and the high causal layer;

[0198] In the underlying causal layer, the heat output by the beauty device affects the surface temperature distribution map captured by the FLIR Lepton3.5 infrared thermal imaging through the deformation of the high molecular polymer and hydrophobic coating in the epidermis, thereby affecting the three-dimensional thermal field distribution map. The current stimulation changes the elastic modulus and water content of the hydrogel substrate in the dermis, affecting the complex impedance measured by the microelectrode array, thereby affecting the three-dimensional electric field distribution map. The electromagnetic wave changes the thermal conductivity of the fat-simulated gel in the subcutaneous tissue layer, affecting the magnetic field intensity and magnetic flux measured by the TMR magnetoresistive sensor array, thereby affecting the three-dimensional magnetic field distribution map. Mechanical vibration reduces the microfluidic flow rate through an external control system, affecting the temperature values of each layer measured by the thermocouple array, thereby affecting the three-dimensional thermal field distribution map.

[0199] In the middle causal layer, epidermal deformation measures local stress changes through micro-strain sensors, affecting the deformation distribution diagram and stress concentration areas of the thermal-mechanical coupling model. Changes in the elastic modulus and water content of the dermis measure the temperature values of each layer through a thermocouple array, affecting the three-dimensional thermal field distribution diagram. Changes in the flow rate and fluid temperature of the microfluidic chip adjust the density of the capillary network through an external control system, affecting the thermal conduction characteristics. The three-dimensional thermal field distribution diagram is then affected by fusing the thermocouple array data with the infrared thermal imaging data through a Bayesian estimation algorithm. When the temperature of the dermis is too high, the initial temperature or water content is adjusted through a Peltier effect device or a micro-humidifier, affecting the thermal conduction characteristics, and thus affecting the three-dimensional thermal field distribution diagram.

[0200] In the high-level causal layer, the electrical parameters of the three-dimensional electric field distribution map affect the three-dimensional magnetic field distribution map through the Maxwell equations. The abnormal areas of the three-dimensional magnetic field distribution map are identified through the gradient analysis method and affect the three-dimensional electric field distribution map through feedback from the electromagnetic field coupling model. The three-dimensional thermal field distribution map calculates the thermal conductivity and thermal diffusion coefficient through the Fourier heat conduction equation, affecting the heat conduction path and heat flux density distribution of the heat conduction model, and thus affecting the deformation distribution map and stress concentration area of the thermal-mechanical coupling model.

[0201] An electro-magnetic-thermal-mechanical coupling model is established through the finite element analysis algorithm. The input parameters include thermal conductivity, water content, microfluidic flow rate, fluid temperature, thermal expansion coefficient, elastic modulus, thermal stress distribution, resistivity, conductivity, magnetic field strength and magnetic flux change, and a comprehensive causal influence diagram is output. The timing dynamic change curve is extracted through the timing Transformer structure, the key causal nodes are identified through the attention mechanism, and the noise and interference are eliminated through the Kalman filter algorithm. The causal chain output is verified by the pre-inspection safety station. If the duty cycle fluctuation exceeds 0.5% or the THD exceeds 3%, the host computer will issue a warning and adjust the bionic skin model parameters for functional and safety evaluation of beauty equipment.

[0202] Because, in multimodal data reports, data are expressed in too many forms, including not only numerical values but also curves and charts, so layered processing is required to improve the efficiency of feature extraction.

[0203] Methods for extracting descriptive primitives to represent data features using a hierarchical dynamic selection architecture include:

[0204] For numerical data in multimodal data reports (such as resistivity and thermal conductivity), they are processed using a multi-layer perceptron (MLP) after normalization;

[0205] For curve data in multimodal data reports (such as time series dynamic change curves), dynamic time warping (DTW) is first performed to synchronize the time slices with the three-dimensional graph data. A temporal Transformer structure is used to capture long-term dependencies in the curve data (such as temperature fluctuation cycles and transient changes in electromagnetic fields) through a self-attention mechanism, and the feature vector sequence in the time dimension is output.

[0206] For three-dimensional graph data in multimodal data reports (such as three-dimensional thermal field distribution maps), numerical data is used as the physical benchmark for cross-modal alignment. Because it has clear biomedical significance (such as "average epidermal temperature is 42°C"), it can effectively anchor the scale reference system of spatiotemporal data. The numerical data is expanded into a query vector (Query) for the attention mechanism, representing the core indicators of skin condition. Temporal features (temporal curve evolution pattern) and spatial features (3D field distribution gradient) are spliced as keys (Key) and values (Value) to construct a joint spatiotemporal representation. A dual-channel parallel architecture is developed based on sparsity detection. If the sparsity is >70%, CNN local feature extraction is prioritized (preserving spatial correlation). If the sparsity is <30%, PCA global dimensionality reduction is enabled (retaining 95% of the principal components of the variance). The sparsity judgment limit can be adaptively adjusted. Sparsity = total number of elements / number of non-zero elements. A gradient penalty term is used to constrain feature redundancy. , θij is the angle between eigenvectors i and j, representing the correlation between features (θij→0 indicates high correlation), λ is the regularization coefficient (usually 0.01∼0.1), which controls the penalty intensity. By penalizing highly correlated feature pairs, the network is forced to learn orthogonal feature representations, solving the information redundancy problem in multimodal data.

[0207] For CNN, the channel importance score is added after each convolution layer, and the channels with importance scores lower than the expected performance are closed. For example, the importance score , dynamic close < 0.1·max( ) channel, reducing the amount of calculation by 40%, using octree adaptive sampling, where represents the batch size, Represents the feature map of the c-th channel in the n-th sample;

[0208] For example, for a 3D thermal field distribution map, the sampling rate , maintaining high resolution in areas with large temperature gradients and downsampling in flat areas, reducing the amount of calculation by 65%; is the temperature gradient modulus (unit: °C / mm), 1 / 8 means retaining 1 out of every 8 voxels (downsampling rate 87.5%), and 1 means full resolution sampling (no downsampling);

[0209] The features extracted from numerical data, curve data and three-dimensional graph data are recorded as description primitives.

[0210] For example:

[0211] Extract description primitives for different modal data. For example:

[0212] Electric field data: Extract the peak value, mean value, variance, etc. of resistivity, conductivity, and current density distribution.

[0213] Magnetic field data: Extract magnetic field strength, gradient of magnetic flux change, spatial location of abnormal areas, etc.

[0214] Thermal field data: Extract thermal conductivity, thermal diffusivity, temperature gradient, heat flux, etc.

[0215] Mechanical data: Extract stress concentration areas, maximum values of deformation distribution, changes in elastic modulus, etc.

[0216] The machine learning model can be based on RNN, deep neural network, etc. Use pre-designed data for pre-training to obtain a machine learning model that meets the training standards.

[0217] The host computer is developed based on an open technology stack and is used to provide a user interface (GUI), including setting detection parameters, displaying real-time data, and generating detection reports. It supports data visualization (such as thermal field distribution diagrams and electromagnetic timing curves) and data storage.

[0218] The open technology stack includes open source systems (Linux kernel, Android framework), cross-platform development tools (Python, LabVIEW), and standardized interfaces (Python library ecosystem, LabVIEW hardware driver).

[0219] Specifically, it includes an Android intelligent interaction layer (based on the Linux kernel) and a Python-LabVIEW hybrid development framework, supporting modular expansion and cross-platform deployment. It recommends an open heterogeneous system architecture for top-level design, a cross-platform development kit (Python / LabVIEW) for the development layer, and a Linux / Android dual-mode runtime environment for the runtime layer.

[0220] For example, a 3D magnetic field distribution map can be presented as a heat map or vector map, with color depth representing magnetic field strength and arrows indicating magnetic field direction. Users can select a specific area through the interactive interface to view detailed magnetic field parameters (such as magnetic field strength, magnetic flux change rate, or time series curve expression).

[0221] Example 2

[0222] See also Figure 1 and Figure 3 As shown, for the parts not described in detail in this embodiment, please refer to the description of Example 1. A detection method for an electrothermal medical beauty device is provided, comprising:

[0223] Step S1: Construct a biomimetic skin model. The thickness of each layer and the density of the capillary network are dynamically adjusted based on age. An external control interface is provided to connect to an external control system to simulate skin conditions under different environmental conditions. The area and concentration of the photosensitive material, blood flow rate, and temperature are predefined. A Peltier effect device is used to control the initial temperature, and a micro-humidifier and desiccant are used to adjust the moisture content. The external control system dynamically adjusts the biomimetic skin parameters (such as microfluidic flow rate, skin temperature, and capillary network density) to simulate skin conditions under different environmental conditions. Furthermore, pre-defined biochemical protection mechanisms are implemented. For example, if the dermis temperature is too high, a Peltier effect device or micro-humidifier is used to reduce the temperature or increase the moisture content. If mechanical vibration in the subcutaneous tissue is too strong, the microfluidic flow rate is reduced to simulate the skin's self-protection and stress relaxation mechanisms. This dynamic control capability enables the system to adapt to the operating modes of different cosmetic devices (such as radiofrequency, ultrasound, and hot and cold therapy) and realistically simulate the skin's reactions under extreme conditions, significantly improving the system's adaptability and practicality.

[0224] Step S2: Collect multi-dimensional data through a high-resolution camera array and an electromagnetic-magnetic-thermal composite sensor array, and transmit it to the host computer and the main testing center;

[0225] Step S3: The beauty device output is pre-inspected for stability using a pre-inspection safety station. A high-precision analog-to-digital converter (ADC) combined with a fast Fourier transform algorithm is used to calculate the total harmonic distortion (THD) of the PWM signal. If duty cycle fluctuations exceeding 0.5% or THD exceeding 3% are detected, the host computer issues a warning and initiates a mode change or device shutdown. This function effectively avoids detection errors or safety hazards caused by unstable device output, ensuring the reliability and safety of the testing process. It is particularly suitable for device verification before mass production.

[0226] Step S4: The main testing center uses a multi-branch diversion method to perform multi-interaction fusion processing on the multi-dimensional data and generate a multimodal data report;

[0227] Step S5: A hierarchical dynamic selection architecture is used to extract descriptive primitives. A machine learning model is used to learn and output quantitative indicators. Causal chain analysis is introduced to calculate the deviation between the quantitative indicators and predefined effect indicators. A multi-objective comprehensive evaluation is performed to determine whether the equipment meets the hard constraints. If qualified, it can be put into mass production; otherwise, the design needs to be improved or retested. Through the multi-objective comprehensive evaluation, the quantitative indicators are weighted averaged to obtain a comprehensive score, and hard constraints are predefined for the score and deviation value to determine whether the equipment meets the functional and safety requirements. If qualified, it can be put into mass production; otherwise, the design needs to be improved or retested. This standardized evaluation method avoids the arbitrariness of subjective judgment in traditional testing, provides an objective and scientific decision-making basis for equipment mass production, and significantly improves production efficiency and product quality.

[0228] Step S6: The host computer performs visual interaction. The user-friendly interface design and powerful data management functions allow R&D personnel to intuitively analyze test results and optimize equipment design. It also facilitates long-term data storage and traceability, meeting the needs of industrial production and quality control.

[0229] Example 3

[0230] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the detection system of the electrothermal medical cosmetic device provided above is realized.

[0231] Since the electronic device described in this embodiment is an electronic device used to implement a detection system for an electrothermal medical cosmetic device in the embodiment of this application, those skilled in the art will be able to understand the specific implementation of the electronic device of this embodiment and its various variations based on the detection system for an electrothermal medical cosmetic device described in the embodiment of this application. Therefore, how the electronic device implements the method in the embodiment of this application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the detection system for an electrothermal medical cosmetic device in the embodiment of this application, it falls within the scope of protection to be provided by this application.

[0232] The above formulas are all dimensionless and numerical calculations. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0233] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for users of ordinary skill in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A detection system for electrothermal medical beauty equipment, characterized in that: Including bionic skin model and main testing center; The bionic skin model is constructed based on multi-layer composite materials and dynamic control technology. It uses a three-layer gradient structure to simulate bionic skin, and the layers are bonded using gradient bonding technology. And through the external control system to simulate the skin state under different environmental conditions; deploy integrated bionic microcirculatory physiological evolution, including embedded photosensitive materials, microfluidic chips and integrated graphene heating layer; A high-resolution camera array is deployed on the outside of the bionic skin model, and a high-precision electro-magnetic-thermal composite sensor array is integrated inside to form a three-dimensional sensor network, which collects multi-dimensional data and transmits it to the host computer and the main testing center. The main testing center includes a pre-inspection safety station and a fusion detection module. The pre-inspection safety station is used to perform stable pre-inspection of the output of beauty equipment. The fusion detection module uses a multi-branch diversion method to perform multi-interactive fusion processing on the collected multi-dimensional data, generating a multimodal data report, pre-defining quantitative detection indicators for beauty equipment, and using a machine learning model to deeply learn the multimodal data report to output quantitative indicators. The machine learning model also introduces a causal chain, which adopts a hierarchical dynamic selection architecture and is divided into a bottom causal layer, a middle causal layer, and a high causal layer. The method for generating a multimodal data report includes: The report includes 3D electric field distribution, 3D magnetic field distribution, 3D thermal field distribution and key demonstration characteristics; The three-dimensional electric field distribution map is obtained by collecting spatial distribution data of electrical parameters using a microelectrode array and constructing a three-dimensional electric field distribution map of the bionic skin model using a Kriging interpolation method or a finite element analysis algorithm. The electrical parameters include resistance, current, and voltage. The method for obtaining a three-dimensional magnetic field distribution map is to denoise the magnetic field intensity and magnetic flux changes collected by the TMR sensor array. Then, a magnetic tomography algorithm is used based on Maxwell's equations and finite element analysis to invert the spatial distribution of the magnetic field in the bionic skin model. The algorithm inputs the magnetic field intensity and magnetic flux changes, and outputs a three-dimensional magnetic field distribution map. Abnormal areas in the magnetic field distribution map are identified using gradient analysis and their spatial locations are marked. The method for obtaining the three-dimensional thermal field distribution map is to fuse the point measurement data of the thermocouple array with the surface temperature distribution map of the FLIRRepton3.5 infrared thermal imaging through the Bayesian estimation algorithm to generate a three-dimensional thermal field distribution map; The qualification of the beauty equipment is determined based on the quantitative indicators and sent to the host computer, which is used for visual interaction and data storage.

2. The detection system of the electrothermal medical beauty device according to claim 1, characterized in that: The method for constructing the bionic skin model based on multi-layer composite materials and dynamic control technology includes: A three-layer gradient structure is used for bionic skin simulation, namely the epidermis, dermis and subcutaneous tissue layers; the thickness of each layer is dynamically adjusted according to age, and the layers are bonded using gradient bonding technology; The epidermis uses a polymer to simulate the stratum corneum, and a hydrophobic coating is applied to the surface to simulate the skin barrier function. The dermis uses a hydrogel matrix to simulate the density and elastic modulus of collagen fibers. The subcutaneous tissue layer uses fat to simulate gel to regulate water content and thermal conductivity. An external control interface is set up and connected to an external control system to simulate skin conditions under different environmental conditions. Photosensitive materials are embedded in the epidermis or dermis, and the area and concentration of the photosensitive materials are predefined. Deploy integrated biomimetic microcirculatory physiological evolution, including embedding a microfluidic chip and an integrated graphene heating layer between the dermis and subcutaneous tissue layers. The microfluidic chip is used to simulate the capillary network, with the network density adjusted according to age group and the fluid medium set to simulate blood, and the flow rate is adjusted through programming. The integrated graphene heating layer is used to simulate the electrical and thermal response of skin tissue; The microfluidic chip is driven by an external peristaltic pump and dynamically simulates the thermal conductivity characteristics of the skin under different ambient temperatures or motion conditions by adjusting the flow rate and fluid temperature. A Peltier effect heating / cooling device is used to control the initial temperature of the bionic skin. A micro humidifier and desiccant are used to adjust the water content of the bionic skin. Simulating electrical properties on bionic skin, including setting different resistivity and conductivity based on the characteristics of the skin layer; The microfluidic flow rate and skin temperature are preset according to needs to simulate the state of the skin under different environmental conditions, and the parameters of the bionic skin are dynamically adjusted according to the output of the beauty device.

3. The detection system of the electrothermal medical beauty device according to claim 2, characterized in that: The method of dynamically adjusting the parameters of the bionic skin according to the output of the beauty device includes: Work outputs scheduled for aesthetic devices include radiofrequency, ultrasound, laser, electrical current stimulation, and hot and cold therapies; In response to the heat, current, electromagnetic or mechanical vibrations output by the beauty device, the high molecular polymers and hydrophobic coatings in the epidermis undergo deformation or changes in surface properties; the hydrogel matrix in the dermis undergoes changes in elastic modulus and water content; the fat-mimicking gel in the subcutaneous tissue layer undergoes changes in thermal conductivity and water content; the microcirculation dynamically adjusts the flow rate and fluid temperature through an external control system; and the capillary network density is dynamically adjusted through an external control system to simulate the dynamic process of skin tissue repair or aging. Simulate the resistivity and conductivity of each layer of bionic skin; the sensor array in the electric-magnetic-thermal composite sensor array collects the multi-dimensional data output by the beauty device acting on the bionic skin model in real time; The preset biochemical protection mechanism includes lowering the initial temperature or increasing the water content through a Peltier effect device or a micro humidifier when the temperature of the dermis is too high, so as to simulate the self-protection of the skin; if the mechanical vibration of the subcutaneous tissue layer is too strong, the microfluidic flow rate is reduced through an external control system to simulate the stress relaxation of the skin tissue.

4. The detection system of the electrothermal medical beauty device according to claim 3, characterized in that: The integrated deployment method of the high-precision electromagnetic-magnetic-thermal composite sensor array includes: A high-resolution camera array is deployed on the outside of the bionic skin model to comprehensively monitor changes on its surface. Micro strain sensors are embedded in the dermis and subcutaneous tissue layers to measure local stress changes caused by heat. For resistance, current and voltage, electrical variable sensors are combined with micro-electrode arrays and impedance analyzers for measurement; wide-band sweep frequency technology is introduced; A TMR magnetoresistive sensor array is used to measure the magnetic field strength and magnetic flux changes. For the temperature of each layer, a thermal field sensor uses a thermocouple array to measure the temperature. A FLIRRepton 3.5 infrared thermal imaging camera is used to collect the surface temperature distribution map in real time. The sensor array is evenly distributed in the epidermis, dermis and subcutaneous tissue layers of the bionic skin with a spacing of 5mm, forming a three-dimensional sensing network. The number of sensors is determined according to the size of the test area. It is connected to the data acquisition card through a multiplexer. The data acquisition card uses a high-precision analog-to-digital converter and transmits data in real time to the host computer and the main test center via USB or Wi-Fi.

5. The detection system of the electrothermal medical beauty device according to claim 4, characterized in that: The pre-inspection safety station is used to perform a stable pre-inspection on the output of the beauty device, and the method includes: The pre-inspection safety station uses a high-precision analog-to-digital converter (ADC) combined with a fast Fourier transform algorithm to calculate the total harmonic distortion (THD) of the PWM signal through spectrum analysis. If it detects that the duty cycle fluctuation exceeds 0.5% or the THD exceeds 3%, the host computer will issue a warning and switch the mode or shut down the beauty device.

6. The detection system of the electrothermal medical beauty device according to claim 5, characterized in that: The method for obtaining the key demonstration characteristics includes: The complex impedance of the graphene heating layer was measured using a microelectrode array and an impedance analyzer. The impedance spectrum was analyzed using Nyquist and Bode plots to extract the equivalent circuit parameters of the graphene heating layer. A time series dynamic change curve was established based on the sampling frequency. The temperature of each layer is measured at a fixed frequency using a thermocouple array, and the thermal conductivity and thermal diffusivity of the epidermis, dermis, and subcutaneous tissue layers are calculated using the Fourier heat conduction equation. A finite element analysis algorithm was used to establish a thermal conduction model of the bionic skin model. The input parameters included the thermal conductivity of each layer, water content, microfluidic flow rate, and fluid temperature. The output was the heat conduction path and heat flux density distribution. For the data collected by the TMR magnetoresistive sensor array and the microelectrode array, the Kalman filter algorithm is used to eliminate noise and interference. An electromagnetic field coupling model is established based on Maxwell's equations. Through the coupling model, key parameters are extracted, and a time series dynamic change curve is established according to the sampling frequency. The finite element analysis algorithm is used to establish a thermal-mechanical coupling model of the bionic skin model. The input parameters include the thermal expansion coefficient, elastic modulus and thermal stress distribution of each layer of material, and the output is the deformation distribution map and stress concentration area.

7. The detection system for electrothermal medical cosmetic equipment according to claim 6, characterized in that: The method for determining whether a beauty device is qualified by calculating based on quantitative indicators includes: The testing of beauty devices is set to include functional evaluation and safety evaluation, and quantitative indicators of functional evaluation and safety evaluation are defined according to the model of beauty devices; For multimodal data reports, a hierarchical dynamic selection architecture is used to extract descriptive primitives to represent data features. Predefined machine learning models are used to learn these primitives and output quantitative indicators, introducing causal chains into the machine learning models. Calculate the deviation between the quantitative indicator and the predefined effect indicator; Deploy a multi-objective comprehensive evaluation, taking a weighted average of the quantitative indicators to obtain a comprehensive score. Predefine hard constraints for the comprehensive score and deviation value. If the beauty device meets all hard constraints, it is judged as qualified and can be put into mass production; otherwise, it is judged as unqualified and requires further design improvement or re-testing. Among them, the causal chain adopts a hierarchical dynamic selection architecture, which is divided into the bottom causal layer, the middle causal layer and the high causal layer; In the underlying causal layer, the heat output by the beauty device model affects the three-dimensional thermal field distribution through the deformation of the high molecular polymer and hydrophobic coating in the epidermis. The electric current stimulates the changes in the elastic modulus and water content of the hydrogel substrate in the dermis, affecting the three-dimensional electric field distribution. The electromagnetic field affects the magnetic field intensity and magnetic flux changes through the changes in the thermal conductivity of the fat-simulated gel in the subcutaneous tissue layer, thereby affecting the three-dimensional magnetic field distribution. Mechanical vibration reduces the microfluidic flow rate through the external control system, affecting the temperature values of each layer, thereby affecting the three-dimensional thermal field distribution. In the middle causal layer, epidermal deformation affects the deformation distribution and stress concentration areas of the thermal-mechanical coupling model through local stress changes. Changes in the elastic modulus and water content of the dermis affect the three-dimensional thermal field distribution through the temperature values of each layer. Changes in the flow rate and fluid temperature of the microfluidic chip affect the thermal conduction characteristics through the density of the capillary network, and then affect the three-dimensional thermal field distribution through the fusion of thermocouple array data and infrared thermal imaging data through the Bayesian estimation algorithm. When the temperature of the dermis is too high, the initial temperature or water content is adjusted through a Peltier effect device or humidifier to affect the thermal conduction characteristics, which in turn affects the three-dimensional thermal field distribution. In the high-level causal layer, the electrical parameters of the three-dimensional electric field distribution map affect the three-dimensional magnetic field distribution map through the Maxwell equations. The abnormal areas of the three-dimensional magnetic field distribution map are identified through the gradient analysis method and affect the three-dimensional electric field distribution map through feedback from the electromagnetic field coupling model. The three-dimensional thermal field distribution map calculates the thermal conductivity and thermal diffusion coefficient through the Fourier heat conduction equation, affecting the heat conduction path and heat flux density distribution of the heat conduction model, and thus affecting the deformation distribution map and stress concentration area of the thermal-mechanical coupling model.

8. The detection system for electrothermal medical cosmetic equipment according to claim 7, characterized in that: The method of extracting description primitives for representing data features using a hierarchical dynamic selection architecture includes: For the numerical data in the multimodal data report, the data were normalized and processed using a multilayer perceptron. For the curve data in the multimodal data report, dynamic time warping is first performed to synchronize the time slices with the three-dimensional graph data. A time series Transformer structure is used to output a sequence of feature vectors in the time dimension. For three-dimensional graph data in multimodal data reports, numerical data is used as the physical benchmark for cross-modal alignment. The numerical data is expanded into a query vector for the attention mechanism, and temporal features and spatial features are spliced as keys and values to construct a joint spatiotemporal representation. A dual-channel parallel architecture is developed based on sparsity detection. If the sparsity is >70%, CNN local feature extraction is prioritized. If the sparsity is <30%, PCA global dimensionality reduction is enabled, and a gradient penalty term is used to constrain feature redundancy. For CNN, the channel importance score is added after each convolutional layer, and the channels with importance scores lower than the expected performance are closed, and octree adaptive sampling is used; The features extracted from numerical data, curve data and three-dimensional graph data are recorded as description primitives.

9. The detection system of the electrothermal medical cosmetic device according to claim 8, characterized in that: The method for visual interaction and data storage of the host computer includes: The host computer is developed based on an open technology stack and is used to provide a user interface, including setting detection parameters, displaying real-time data, and generating detection reports, supporting data visualization and data storage.

10. A detection method for electrothermal medical beauty equipment, applied to the detection system for electrothermal medical beauty equipment according to any one of claims 1 to 9, characterized in that: The detection method of the electrothermal medical cosmetic device includes: Step S1: Construct a bionic skin model. The thickness of each layer and the density of the capillary network are dynamically adjusted according to age groups. An external control interface is set up to connect to an external control system to simulate skin conditions under different environmental conditions. The area and concentration of the photosensitive material, blood flow rate, and temperature are predefined. A Peltier effect device is used to control the initial temperature. A micro humidifier and desiccant are used to adjust the moisture content. Step S2: Collect multi-dimensional data through a high-resolution camera array and an electromagnetic-magnetic-thermal composite sensor array, and transmit it to the host computer and the main testing center; Step S3: Perform a stable pre-inspection on the beauty device output through a pre-inspection safety station; Step S4: The main testing center uses a multi-branch diversion method to perform multi-interaction fusion processing on the multi-dimensional data and generate a multimodal data report; Step S5: A hierarchical dynamic selection architecture is used to extract descriptive primitives. A machine learning model is used to learn and output quantitative indicators. Causal chain analysis is introduced to calculate the deviation between the quantitative indicators and predefined effect indicators. A multi-objective comprehensive evaluation is performed to determine whether the equipment meets the hard constraints. If qualified, it can be mass-produced. Otherwise, the design needs to be improved or re-tested. Step S6: The host computer performs visual interaction.

Images