An adaptive control system and method for an intelligent hot-air comb

Abstract

The application relates to the technical field of intelligent control, and discloses an adaptive regulation and control system and method of an intelligent hot air comb, which comprises a power interaction module, a multi-modal sensing module, a high-frequency response execution module and a central intelligent control module. The multi-modal sensing module collects mechanical resistance vectors, hair dielectric properties and thermodynamic temperature data in real time; the central intelligent control module deduces a dynamic glass transition temperature based on the hair dielectric properties and calculates a phase delay time in combination with a heat conduction model. The system generates a control time sequence according to the mechanical resistance characteristics, controls a pulse heater to output a heat pulse first, triggers a piezoelectric actuating mechanism to produce vibration disentangling after the hair is softened for the phase delay time, and then controls a pneumatic adjusting fan to cool and shape. Through the cooperation of thermal energy and mechanical energy and rheological heat dose control, the application solves the knot breaking problem, maintains the constant hair enthalpy, and improves the styling consistency and safety.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to an adaptive control system and method for an intelligent hot air comb. Background Technology

[0002] Currently, personal image management is receiving increasing attention, and hot air combs, as heat-setting devices that integrate drying, straightening, and curling functions, are widely used in daily life. These devices primarily utilize the comb's tooth structure to apply physical tension to the hair strands, and combine this with hot airflow to alter the internal hydrogen bond structure of the hair to achieve shaping. During the combing process, hair strands often become tangled due to changes in surface friction coefficient, electrostatic adsorption, or their own curling characteristics. How to solve the problem of combing resistance while achieving hairstyle shaping is a key focus in this field.

[0003] Existing electric hair combing devices typically employ an open-loop control strategy, integrating a PTC heating element and a fixed-frequency motor. During operation, the device continuously outputs hot air at preset levels, relying on heat to reduce the modulus of the hair strands. Some products are equipped with a mechanical vibration module to assist in untangling, using a motor to drive the comb body to generate continuous physical vibrations, attempting to separate tangled hair strands using inertial force. After the user selects a fixed level (high, medium, low, etc.) via a physical switch, the device outputs heat and vibrates at a constant frequency until the power is turned off.

[0004] However, conventional techniques have limitations in handling knots and maintaining consistent styling. Traditional vibration untangling doesn't consider the viscoelastic transition characteristics of hair strands, often applying high-frequency mechanical force before the hair has softened from heat. At this stage, the glassy hair is brittle, and forced vibration can easily cause breakage or cuticle peeling. Simultaneously, constant power heating cannot respond to changes in the user's combing speed. Combing too slowly leads to excessive heat buildup in certain areas, causing thermal damage, while rapid combing results in insufficient enthalpy injection, leading to springback. Existing systems also lack quantitative sensing of the hair's dielectric properties, such as moisture content, and a single control logic cannot achieve precise matching of thermal and mechanical energy under different hair types.

[0005] Therefore, the present invention provides an adaptive control system and method for an intelligent hot air comb to address the shortcomings of the prior art. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an adaptive control system and method for an intelligent hot air comb, which solves the problems of existing hot air combs causing breakage due to forced untangling because they do not consider the hair softening sequence, and the inability to dynamically match the heat dosage according to the combing speed and hair quality, resulting in inconsistent styling effects and heat damage.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides an adaptive control system for an intelligent hot air comb, employing the following technical solution: An adaptive control system for an intelligent hot air comb includes a power interaction module, a multimodal sensing module, a high-frequency response execution module, and a central intelligent control module. The power interaction module connects to each module and provides operating voltage and drive current; The multimodal sensing module collects physical environment data in real time during the combing process and outputs mechanical resistance vector, hair dielectric properties data, and thermodynamic temperature data. The high-frequency response execution module includes a pulse heater, a piezoelectric actuator, and a pneumatic regulating fan. The central intelligent control module performs adaptive parameter optimization for different hair quality states, which is used to dynamically map the dynamic glass transition temperature of the current hair quality based on the real-time changing dielectric properties data of the hair strand, and adaptively calculate the phase delay time required for heat conduction by combining the thermodynamic temperature data and the dynamic glass transition temperature. The central intelligent control module also includes a thermomechanical phase-locked untying control loop. This loop generates a dynamic timing control signal when a pre-tangle state is detected based on the mechanical resistance vector. It controls the pulse heater to output a thermal pulse first. After the phase delay time required for the hair to complete the phase transition from the glassy state to the elastic state, the pulse heater adaptively triggers the piezoelectric actuator to generate mechanical vibration, which then drives the pneumatic regulating fan to output cold air.

[0009] By adopting the above technical solution, this invention solves the technical problem of traditional hair combing equipment easily causing breakage or thermal damage in the process of hair tangling by utilizing a multi-physics field synergistic mechanism. Specifically: This system, based on the time-temperature equivalence principle of polymer materials science, accurately inverts the current moisture content and corresponding dynamic glass transition temperature of the hair using the dielectric properties data. Instead of immediately applying mechanical vibration upon detecting resistance, the system first outputs a thermal pulse. Utilizing the calculated phase delay time, it ensures sufficient time for heat energy to conduct from the hair surface to the center, allowing the hair to undergo a phase transition from a glassy state to a highly elastic state at the microstructure level. At the moment when the hair material modulus decreases and flexibility increases, the mechanical vibration of the piezoelectric actuator is precisely triggered, achieving untangling with minimal mechanical stress and avoiding brittle fracture caused by forced combing in a cold state. The subsequent immediate cold air output rapidly freezes the hair morphology, closing the cuticles. This sequential synergy of thermal and mechanical energy improves the smoothness of combing and reduces physical damage.

[0010] Preferably, the multimodal sensing module acquires the physical environment data in the following ways: The multimodal sensing module uses a six-axis torque sensor to detect the force components along three orthogonal directions and the torque components rotating around the axis on the comb teeth, and synthesizes them to obtain the mechanical resistance vector that characterizes the magnitude and direction of combing resistance. The multimodal sensing module uses a dielectric constant detection probe to detect changes in dielectric constant or impedance in the comb electrode circuit, quantifies the real-time moisture content of the hair, and uses the real-time moisture content as the dielectric property data of the hair. The multimodal sensing module uses a temperature monitor to detect the airflow temperature in the air duct and the conduction temperature at the interface between the comb teeth and the hair strands, as the thermodynamic temperature data.

[0011] By adopting the above technical solutions, the system constructs a comprehensive perception dimension. The six-axis torque sensor can distinguish between the complex torque caused by hair tangling and normal combing tension; the dielectric constant detection probe utilizes the dielectric response characteristics of water molecules as polar molecules to achieve non-destructive detection of the internal water content of hair, providing a key material parameter basis for the establishment of thermodynamic models; temperature monitoring of the contact interface ensures the accuracy of the boundary conditions for heat conduction calculations, thereby guaranteeing the accuracy of subsequent control logic.

[0012] Preferably, the internal logic architecture of the central intelligent control module includes: A state space construction unit is used to receive data from the multimodal sensing module and generate a state vector describing environmental features. A physical property deduction unit is specifically used to dynamically map the dynamic glass transition temperature based on the dielectric properties data of the hair strand. A strategy decision unit, specifically configured to analyze the state vector to identify the pre-knotting state and determine the diversion of the control strategy; A phase coordination control unit is specifically designed to perform adaptive calculation of the phase delay time based on thermal conduction logic.

[0013] By adopting the above technical solution, the modularization and hierarchical nature of the control logic are achieved. The raw sensor data is transformed into standardized state vectors, enabling the system to be compatible with changes in different hair types and environments. The physical property inference unit makes the previously unobservable material property, namely the dynamic glass transition temperature, explicit, providing a physical basis for control decisions and avoiding the blindness of relying solely on empirical thresholds for control.

[0014] Preferably, the process by which the state space construction unit constructs the state vector for identifying the pre-knotting state includes: Perform time difference operation on the mechanical resistance vector to calculate the rate of change of tension per unit time and obtain the tension gradient index, which is used to characterize the abrupt change trend of combing resistance; Obtain the real-time combing speed of the comb teeth along the combing direction; The mechanical resistance vector, the tension gradient index, the real-time moisture content, the thermodynamic temperature data, and the real-time combing speed are mapped to the same normalized dimension range and combined to generate the state vector.

[0015] By adopting the above technical solution, the system introduces the tension gradient index, which can capture the dynamic pattern of combing resistance changing over time. Combined with the normalization processing of multi-dimensional data such as speed and humidity, standardized input data with rich physical meaning is provided to the strategy decision-making unit, improving the robustness of recognizing the pre-knotting state.

[0016] Preferably, the logic of the physical property deduction unit dynamically mapping the dynamic glass transition temperature includes: A Gaussian process regression model is constructed with the real-time moisture content in the dielectric properties data of the hair as the input variable. The Gaussian process regression model is pre-set with a training set describing the rheological properties of hair under different humidity conditions. The similarity covariance feature between the current input real-time moisture content and the training set samples is calculated using a kernel function. The predicted temperature required for hair to transition from a glassy state to a viscous flow state under the current moisture content is calculated using a probability distribution, and this predicted temperature is defined as the dynamic glass transition temperature.

[0017] By employing the above technical solution, a Gaussian process regression model is used to fit and predict the nonlinear relationship between hair moisture content and its dynamic glass transition temperature. This method not only provides predicted temperature values ​​but also assesses the uncertainty of the prediction through covariance characteristics, thereby maintaining the robustness of the extrapolation results even with high data noise and ensuring the reliability of the thermodynamic control benchmark.

[0018] Preferably, the process by which the strategy decision unit identifies the pre-knotting state and triggers the thermomechanical phase-locked untying control loop includes: The time series of the state vector is input into the long short-term memory network model, and the output is a probability value representing the risk of mechanical knotting. When the probability value is greater than the preset safety threshold, it is determined to be the pre-knotting state, and the thermodynamic phase-locked untying control loop is officially activated, and the phase coordination control unit is called to perform the untying operation. When the probability value is less than or equal to the preset safety threshold, it is determined to be a smooth shaping state, triggering rheological shaping thermal dose control.

[0019] By employing the above technical solution, the system possesses the capability to extract temporal features. The Long Short-Term Memory (LSTM) network model can capture the dynamic pattern of changes in combing resistance over time, thus identifying the minute resistance fluctuations before the formation of a knot as the precursor state of knotting. By distinguishing between the precursor state of knotting and the smooth shaping state, the system can adopt two distinct control strategies: fixed-point untying and dynamic isoenthalpy control, balancing untying efficiency and shaping consistency.

[0020] Preferably, the process by which the phase coordination control unit adaptively calculates the phase delay time includes: Set the target pulse temperature of the pulse heater and obtain the current ambient temperature; Calculate the first temperature difference between the target pulse temperature and the current ambient temperature, and the second temperature difference between the target pulse temperature and the dynamic glass transition temperature, and perform a natural logarithmic operation on the ratio of the first temperature difference to the second temperature difference; The result of the natural logarithm operation is multiplied by the ratio of the square of the hair's geometric radius to the effective thermal diffusivity of the hair material to obtain the time required for heat energy to be conducted from the surface of the hair to the center and soften, which is used as the phase delay time.

[0021] By employing the above technical solution, this invention applies the analytical method of unsteady-state thermal conduction in cylinders to hair combing control. This calculation logic quantitatively describes the rate of heat propagation in the radial direction of the hair strand. By calculating the logarithmic ratio between the target temperature, ambient temperature, and the dynamic glass transition temperature, and combining this with the hair radius and thermal diffusivity, the time required for the hair center to reach the softening temperature is accurately determined. This ensures that when mechanical vibration is triggered, the hair strands are already in a viscoelastic state suitable for deformation, minimizing mechanical damage.

[0022] Preferably, when the smooth styling state is determined, the strategy decision unit performs adaptive parameter optimization for styling consistency, and the strategy decision unit includes power dynamic following logic: Based on the principle of energy conservation, and according to the preset target enthalpy density and hair linear density, combined with the combing speed collected in real time, the heating power required to maintain a constant heat energy per unit length of hair is calculated. Establish a feedback mechanism based on the difference between the measured surface temperature of hair strands and the optimal plastic deformation temperature; The output power of the pulse heater is dynamically adjusted so that the output power increases linearly with the increase of the combing speed, thereby compensating for insufficient energy intake caused by the shortened heating time.

[0023] By adopting the above technical solution, the problem of uneven heating of traditional hot air combs at different combing speeds is solved. By introducing the concept of enthalpy density, the system can automatically compensate the heating power according to the user's combing speed. The power increases linearly when the combing speed is fast and decreases when the combing speed is slow, ensuring that the heat energy received by each section of hair is constant, thereby guaranteeing the consistency of the styling effect and preventing thermal carbonization caused by excessive local dwell time or styling failure caused by excessive speed.

[0024] Preferably, the strategy decision unit further includes a reinforcement learning iterative update mechanism, the process of which the reinforcement learning iterative update mechanism continuously optimizes the accuracy of the adaptive parameter optimization includes: At the end of the control cycle, an instant reward value is calculated. The calculation logic of the instant reward value is set to be positively correlated with the magnitude of the decrease in the modulus of the mechanical resistance vector during the combing process, and includes a penalty term for the thermodynamic temperature data exceeding a safety threshold. The weight parameters in the policy network are updated in reverse using the instant reward value, thereby optimizing the system's action selection when faced with similar state vectors in the next cycle.

[0025] By adopting the above technical solution, the system is endowed with the ability to learn and evolve online. Through a reinforcement learning mechanism, the system can automatically adjust control parameters based on the actual detangling effect (i.e., the reduction in resistance) and safety (i.e., whether the temperature exceeds the limit). This allows the intelligent hot air comb to gradually adapt to the hair characteristics and combing habits of specific users with increased usage, continuously optimizing its control strategy.

[0026] Secondly, the present invention provides an adaptive control method for an intelligent hot air comb, employing the following technical solution: An adaptive control method for an intelligent hot air comb, applied to the above system, includes the following steps: A multi-dimensional state space, including the mechanical resistance vector and the dielectric properties of the hair strand, is constructed using a multi-modal sensing module. The central intelligent control module dynamically maps the current dynamic glass transition temperature of the hair based on the dielectric properties data of the hair, and adaptively calculates the phase delay time by combining thermodynamic temperature data. Determine the current combing status. If the pre-knotting state is detected, activate the thermodynamic phase-locked untangling control loop, control the pulse heater to output a heat pulse, and after the phase delay time, adaptively trigger the piezoelectric actuator to vibrate, and then output cold air to lock. If a smooth styling state is detected, the rheological styling heat dose control is activated, and the heating power is dynamically adjusted according to the combing speed to maintain a constant enthalpy received per unit length of hair.

[0027] By adopting the above technical solution, the method provided by the present invention achieves precise intervention on the physical state of hair through closed-loop control of perception, deduction, decision-making and execution; This method not only monitors the mechanical and thermal state of hair strands in real time, but also dynamically adjusts the timing of thermal and mechanical energy intervention based on the derived dynamic glass transition temperature. When dealing with tangles, it strictly follows the physical process of softening, untangling, and then styling, effectively avoiding damage caused by rough combing. In daily styling, this method ensures consistent heat application, improving user experience and hair health protection.

[0028] This invention provides an adaptive control system and method for an intelligent hot air comb. It has the following beneficial effects: 1. This invention constructs a time-series synergistic mechanism for thermal and mechanical energy by using a central intelligent control module to deduce the dynamic glass transition temperature based on the dielectric properties of hair and calculating the phase delay time using a heat conduction model. This mechanism, for different knotting conditions, controls the pulse heater to start first, and only after the phase delay time has elapsed and the hair has undergone a phase transition to a highly elastic state, is the piezoelectric actuator triggered to generate vibration. This technique effectively reduces mechanical stress concentration caused by the brittle characteristics of hair in low-humidity or cold conditions, thereby improving combing efficiency while reducing physical damage such as hair breakage. 2. This invention employs a rheological styling heat dose control strategy to solve the problem of uneven heating caused by differences in combing speed. Based on real-time collected movement speed data, the system dynamically and linearly adjusts the output power of the pulse heater to ensure a constant enthalpy density received per unit length of hair. This technical feature guarantees that hair strands receive consistent plastic deformation energy at different combing speeds, avoiding heat accumulation damage at low speeds and insufficient styling effect at high speeds, thus achieving stable styling quality. 3. This invention utilizes a multimodal sensing module to collect real-time data on the dielectric properties and thermodynamic temperature of hair strands, enabling adaptive parameter adjustment for different hair types. The system converts biophysical indicators such as hair moisture content into quantifiable control parameters, automatically matching the most suitable heating temperature curve and the cooling and styling logic of the pneumatically adjustable fan. This not only optimizes the curling or straightening effects for specific hair types but also assists users in achieving the desired hairstyle effect while minimizing thermal damage through precise temperature control and suggested operating time output. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the overall architecture of the intelligent hot air comb adaptive control system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the overall process of the intelligent hot air comb adaptive control method according to an embodiment of the present invention.

[0030] Among them, 10. Multimodal sensing module; 11. Six-axis torque sensor; 12. Dielectric constant detection probe; 13. Temperature monitor; 20. Central intelligent control module; 21. State space construction unit; 22. Physical property inference unit; 23. Strategy decision-making unit; 24. Phase coordination control unit; 30. High-frequency response execution module; 31. Pulse heater; 32. Piezoelectric actuation mechanism; 33. Pneumatic regulating fan; 40. Power interaction module. Detailed Implementation

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] See attached document Figure 1 This invention provides an adaptive control system for an intelligent hot air comb. Physically, the system includes a handheld handle assembly, a comb head assembly, and an internal circuit control motherboard. Electrically and logically, the system comprises a multimodal sensing module 10, a central intelligent control module 20, a high-frequency response execution module 30, and a power interaction module 40. These modules work collaboratively to form a closed-loop control circuit, enabling adaptive and precise control of thermal and mechanical energy during the combing process.

[0033] The central intelligent control module 20, serving as the system's computation and decision-making core, establishes electrical connections with the multimodal sensing module 10, the high-frequency response execution module 30, and the power interaction module 40. The central intelligent control module 20 is connected to the multimodal sensing module 10 via a high-speed signal bus to receive raw sensor data with a high sampling rate. The central intelligent control module 20 is also connected to the high-frequency response execution module 30 via multiple control signal lines to send control commands containing heating power, vibration frequency, and phase delay parameters. The power interaction module 40 is connected to the multimodal sensing module 10, the central intelligent control module 20, and the high-frequency response execution module 30 via a power management bus, providing stable operating voltage and drive current to each module and processing user input signals and tactile feedback signals.

[0034] The multimodal sensing module 10 is configured as the sensing front end of the system to collect physical environment data in real time during the combing process. The multimodal sensing module 10 is arranged at key physical nodes of the comb head assembly and the hand handle assembly, and can simultaneously acquire the mechanical resistance vector of the comb teeth, the dielectric properties of the contact hair strands, and real-time thermodynamic temperature data, and convert the above analog physical quantities into digital signals and transmit them to the central intelligent control module 20.

[0035] The high-frequency response execution module 30 is configured as the system's execution terminal, used to output physical actions according to control commands. Physically integrated within the comb head assembly and air duct structure, the high-frequency response execution module 30 possesses millisecond-level thermal response and high-frequency mechanical vibration capabilities. It receives control strategy vectors from the central intelligent control module 20 and converts them into specific pulse heating currents, piezoelectric drive voltages, and fan speed control signals, thereby applying controlled thermal, mechanical, and fluid fields to the hair strands.

[0036] The power interaction module 40 includes a rechargeable battery pack, a power management circuit, and a haptic feedback oscillator. The power management circuit monitors the battery status and dynamically allocates power according to the system load, ensuring voltage stability when the high-frequency response execution module 30 outputs instantaneous pulses. The haptic feedback oscillator is physically connected to the inner wall of the handheld handle assembly and generates mechanical vibration cues that can be perceived by the user's hand when the central intelligent control module 20 determines that a specific shaping task has been completed or detects a specific operating state.

[0037] See attached document Figure 1 The multimodal sensing module 10, as the data acquisition front end of the system, is physically integrated and distributed at the connection between the comb teeth and the handle, as well as on the surface of the comb teeth. The multimodal sensing module 10 includes a six-axis torque sensor 11, a dielectric constant detection probe 12, and a temperature monitor 13, which are used to simultaneously collect mechanical data, hair physical property data, and thermodynamic environment data.

[0038] A six-axis torque sensor 11 is rigidly mounted between the root drive shaft of the comb assembly and the mechanical coupling interface of the hand handle assembly. The six-axis torque sensor 11 can simultaneously detect the force components along the three orthogonal directions (X, Y, and Z axes) and the torque components rotating around these three axes. When the comb teeth move among the hair strands and encounter resistance, the reaction force generated by the comb assembly is directly transmitted to the sensing element of the six-axis torque sensor 11 via the drive shaft. The six-axis torque sensor 11 converts the aforementioned mechanical deformation into an analog voltage signal or a digital signal, thereby outputting three-dimensional tension vector data containing information on the magnitude and direction of combing resistance. This three-dimensional tension vector data reflects the tightness of the hair entanglement and the instantaneous tension exerted on the hair by the combing action.

[0039] The dielectric constant detection probe 12 mainly consists of a microelectrode array arranged on the surface of the comb teeth. In specific implementations, the comb teeth are made of insulating, high-temperature resistant material, and the conductive metal layer of the dielectric constant detection probe 12 is attached to the side contact area of ​​the comb teeth through an inlay process or electroplating process. When a hair strand slides over or contacts the surface of the comb teeth, the hair strand acts as a medium, filling the spaces between the electrodes of the dielectric constant detection probe 12, forming an equivalent capacitance or impedance loop. The dielectric constant detection probe 12 detects changes in the dielectric constant or impedance in the loop by applying a weak high-frequency AC signal to the electrodes. Since the dielectric constant of water is much higher than that of keratin, this detection value is used to quantify the real-time moisture content of the hair strand.

[0040] The temperature monitor 13 includes at least one set of negative temperature coefficient thermistors or thermocouples. The first sensing end of the temperature monitor 13 is located within the air outlet duct of the high-frequency response execution module 30 to detect the real-time temperature of the hot airflow. The second sensing end of the temperature monitor 13 is embedded inside a selected reference comb tooth and close to the tooth tip to detect the conduction temperature at the interface between the comb tooth material and the hair. The temperature monitor 13 converts the collected resistance change signal into a temperature value signal and transmits this signal to the central intelligent control module 20, which can serve as feedback for calorific value calculation. The multimodal sensing module 10 amplifies, filters, and performs analog-to-digital conversion on the raw signals collected by the six-axis torque sensor 11, dielectric constant detection probe 12, and temperature monitor 13 through its internal signal conditioning circuit, and then sends the signals to the next-level processing unit via a high-speed serial bus.

[0041] See attached document Figure 1 The central intelligent control module 20, at the physical level, consists of an embedded computing platform including a microcontroller unit, a digital signal processor, or a field-programmable gate array. Internally, the central intelligent control module 20 integrates non-volatile memory and random access memory to store control algorithm code, neural network weight parameters, and historical state data. At the logical function level, the central intelligent control module 20 includes a state space construction unit 21, a physical attribute inference unit 22, a strategy decision-making unit 23, and a phase coordination control unit 24. These units interact via an internal data bus, collaboratively completing the calculation process from sensor data input to control command output.

[0042] The state space construction unit 21, serving as a data preprocessing stage, is connected to both the multimodal sensing module 10 and the strategy decision unit 23. The state space construction unit 21 is configured to receive raw sampled data streams from sensors, including a three-dimensional tension vector signal, a dielectric constant signal, and a temperature signal. The state space construction unit 21 uses a digital filter to remove high-frequency noise and baseline drift from the raw signal. The state space construction unit 21 performs a differential operation on the filtered three-dimensional tension vector to calculate the rate of change of tension per unit time to obtain a tension gradient index, which characterizes the abrupt change in combing resistance. The state space construction unit 21 normalizes the processed instantaneous tension, tension gradient, moisture content, and current combing speed, and then combines them to generate a state vector describing the current physical environment of the system.

[0043] The physical property deduction unit 22 is connected to the state space construction unit 21 and the strategy decision-making unit 23. The physical property deduction unit 22 internally stores Gaussian process regression model parameters or multidimensional mapping tables describing the rheological properties of keratin materials. Using the moisture content data of the hair strand as the main input variable, the physical property deduction unit 22 calculates the dynamic glass transition temperature of the hair strand in real time based on the plasticizing effect principle of polymers. The dynamic glass transition temperature value output by the physical property deduction unit 22 is not fixed but fluctuates dynamically with changes in hair humidity. This value is transmitted to the strategy decision-making unit 23 as a critical reference threshold for subsequent thermal pulse control.

[0044] The strategy decision-making unit 23 is an inference engine based on a deep reinforcement learning algorithm, connected to the state space construction unit 21, the physical property inference unit 22, and the phase coordination control unit 24. The strategy decision-making unit 23 receives the state vector and the dynamic glass transition temperature as environmental observations. By running a pre-trained policy network model, the strategy decision-making unit 23 searches for optimal control parameters in a preset action space and outputs an action command vector containing the thermal pulse power amplitude, piezoelectric vibration frequency, and vibration amplitude. The decision objective of the strategy decision-making unit 23 is to maximize comb throughput with minimal mechanical work consumption while ensuring that the hair temperature does not exceed the thermal damage threshold.

[0045] The phase coordination control unit 24, acting as the system's timing controller, connects the strategy decision unit 23 and the high-frequency response execution module 30. The phase coordination control unit 24 is responsible for resolving the synchronous coupling problem of thermal and mechanical energy application on the time axis. Based on a thermal conduction physical model, the phase coordination control unit 24 calculates the penetration time required for heat to conduct from the hair surface to the axis and sets this penetration time as the phase delay time. Accordingly, the phase coordination control unit 24 generates two trigger signals with precise time differences: the first is a heating trigger signal used to initiate a thermal pulse; the second is a vibration trigger signal, which lags behind the first signal by the phase delay time. The phase coordination control unit 24 sends these two time-locked signals to the pulse heater 31 and the piezoelectric actuator 32 respectively, ensuring that the timing of mechanical vibration intervention precisely corresponds to the moment when the hair material modulus decreases to the plastic window period.

[0046] See attached document Figure 1 The high-frequency response execution module 30, as the physical output terminal of the system, is configured to receive control commands from the central intelligent control module 20 and output controlled thermal fluid field and mechanical vibration field. The high-frequency response execution module 30 is physically integrated into the internal cavity of the comb head assembly, and its main components include a pulse heater 31, a piezoelectric actuator 32, and a pneumatic regulating fan 33.

[0047] The pulse heater 31 is constructed using low-heat-capacity thick-film heating technology, comprising an alumina ceramic substrate, a resistive paste layer printed on the substrate surface, and an outermost glass glaze insulating and thermally conductive layer. The pulse heater 31 is fixedly installed at the center of the air outlet duct. The thickness of its ceramic substrate is set between 0.3 mm and 0.6 mm to minimize thermal inertia and achieve a millisecond-level temperature response rate. The input of the pulse heater 31 is connected to the central intelligent control module 20 via a power drive circuit. When a heating trigger signal is received, the pulse heater 31 can convert electrical energy into heat energy in a short time and rapidly increase the temperature of the air flowing over its surface through forced convection, forming a thermal pulse airflow with a steep rise edge.

[0048] The piezoelectric actuation mechanism 32 consists of a multilayer piezoelectric ceramic stacked actuator and a mechanical coupling bracket. The piezoelectric actuation mechanism 32 is located at the mechanical connection interface between the comb plate assembly and the handle frame. One end of the multilayer piezoelectric ceramic stacked actuator is rigidly fixed to the internal support structure of the handle frame, while the other end is tightly fitted or mechanically fixed to the root of the comb plate assembly. When the central intelligent control module 20 applies an alternating driving voltage of a specific frequency and amplitude to the piezoelectric actuation mechanism 32, the piezoelectric ceramic material generates micron-level axial expansion and contraction deformation based on the inverse piezoelectric effect. This mechanical deformation is transmitted to the comb plate assembly through the coupling bracket, driving all the comb teeth to generate high-frequency micro-amplitude vibrations along or perpendicular to the comb tooth axis. This vibration is used to reduce the static friction coefficient between the hair strands and the comb teeth.

[0049] The pneumatic regulating fan 33 includes a high-speed brushless DC motor, a centrifugal impeller, and a hydrodynamic shroud. The pneumatic regulating fan 33 is installed in the air inlet section of the handle assembly, and its outlet is connected to the heating chamber containing the pulse heater 31 via a sealed air duct. The brushless DC motor receives pulse width modulation signals from the central intelligent control module 20 to adjust the impeller's rotational speed, thereby controlling the volumetric flow rate and velocity of the output airflow. The pneumatic regulating fan 33 is configured to continue operating at high speed when the pulse heater 31 stops working, to deliver ambient-temperature cool air to the comb tooth area for rapid cooling and styling of the hair. Furthermore, the air duct structure of the pneumatic regulating fan 33 includes diverting blades to evenly guide the airflow into the gaps between each comb tooth.

[0050] See attached document Figure 2 This invention provides an adaptive control method for an intelligent hot air comb. This method operates within the aforementioned central intelligent control module 20, and achieves real-time closed-loop control of the combing process by controlling the multimodal sensing module 10 and the high-frequency response execution module 30. The method includes a cyclical processing procedure from steps S1 to S6, which can dynamically adjust the output strategy of thermal and mechanical energy under different combing conditions.

[0051] In step S1, the system performs real-time construction of a multi-dimensional state space. The central intelligent control module 20 controls the multimodal sensing module 10 to simultaneously acquire tension data from the six-axis torque sensor 11, moisture content data from the dielectric constant detection probe 12, and temperature data from the temperature monitor 13. The central intelligent control module 20 performs noise reduction processing on the acquired raw data, extracts tension gradient features, and combines them with the current combing speed to generate a current-moment state vector containing the physical and motion states of the hair strands.

[0052] In step S2, the system performs a Gaussian process regression-based deduction of the physical properties of the hair. The central intelligent control module 20 uses a preset nonparametric regression model, with the current moisture content data as the input variable, to deduce the dynamic glass transition temperature of the hair material under the current humidity environment. This dynamic glass transition temperature characterizes the thermodynamic critical point at which the hair transitions from a highly elastic state to a viscous flow state, serving as a key temperature threshold in subsequent control strategies.

[0053] In step S3, the system performs knotting precursor identification and strategy diversion. The central intelligent control module 20 inputs the current state vector generated in step S1 into a long short-term memory network or a threshold logic classifier to identify whether the current combing resistance characteristics match the knotting precursor pattern. If it is determined that the current state is in a knotting precursor state, the system proceeds to step S4 to execute the phase-locked knot-resolving control process. If it is determined that the current state is smooth, the system proceeds to step S5 to execute the rheological modeling thermal dosage control process.

[0054] In step S4, the system performs thermomechanical phase-locked untangling control. When a risk of tangling is detected, the central intelligent control module 20 calculates the target temperature of the heat pulse and the phase delay time required for heat energy to be conducted to the hair shaft. The central intelligent control module 20 first controls the pulse heater 31 to emit a high-energy heat pulse to soften the surface of the hair. After waiting for the phase delay time, it controls the piezoelectric actuator 32 to start high-frequency mechanical vibration, using the time window of hair modulus reduction to untangle the entanglement. Subsequently, it controls the pneumatic regulating fan 33 to output cold air to lock the microstructure.

[0055] In step S5, the system performs rheological styling heat dosage control. During normal combing, the central intelligent control module 20 calculates the required cumulative enthalpy based on the desired curl level. The central intelligent control module 20 dynamically adjusts the average power of the pulse heater 31 according to the current combing speed to ensure a constant heat energy received per unit length of hair, maintaining the hair temperature within the optimal plastic deformation range.

[0056] In step S6, the system performs iterative updates to the reinforcement learning model. After completing a control action, the central intelligent control module 20 calculates a reward value based on the decrease in tension gradient and the thermal damage assessment results. The central intelligent control module 20 uses this reward value to update the policy network weights in the policy decision unit 23 to optimize the action selection at the next moment. The system then returns to step S1 to enter the next control cycle until a task completion signal is detected.

[0057] See attached document Figure 2In step S1, the system performs real-time construction of the multi-dimensional state space. This step is mainly completed by the state space construction unit 21 inside the central intelligent control module 20 in collaboration with the multimodal sensing module 10. This step aims to transform the raw data from dispersed and heterogeneous physical sensors into standardized digital features that can be processed by a computer, providing a unified data input basis for subsequent physical attribute inference and strategy decision-making.

[0058] The state space construction unit 21 first sends a synchronous acquisition trigger signal to the multimodal sensing module 10. Upon receiving the trigger signal, the six-axis torque sensor 11 samples the mechanical resistance experienced by the comb teeth and outputs a raw analog signal containing force components in three orthogonal directions and torque components in three directions. Simultaneously, the dielectric constant detection probe 12 samples the high-frequency impedance in the comb tooth electrode circuit and outputs a raw electrical signal characterizing the dielectric properties of the hair. The temperature monitor 13 samples the air temperature at the air duct opening and the contact temperature on the comb tooth surface and outputs a thermodynamic temperature signal. The analog-to-digital conversion circuit inside the multimodal sensing module 10 converts all the above analog signals into digital signals and transmits them to the state space construction unit 21 via a high-speed bus.

[0059] After receiving the raw digital signal, the state space construction unit 21 first preprocesses the data using a low-pass digital filter with a preset cutoff frequency to filter out high-frequency noise caused by mechanical vibration or electromagnetic interference, and calibrates the baseline drift of the six-axis torque sensor 11 to obtain the calibrated real-time tension vector. Subsequently, the state space construction unit 21 calculates the tension modulus based on the calibrated real-time tension vector and performs a feature extraction operation on the tension gradient. The logic of this extraction operation is as follows: calculate the difference between the tension modulus value of the current sampling period and the tension modulus value of the previous sampling period, and divide this difference by the duration of the sampling period to obtain the tension gradient value. This tension gradient value is used to characterize the degree of change in combing resistance over time and is a key dynamic indicator for identifying whether the hair is about to enter a knotted state.

[0060] After completing the calculation of individual data, the state space construction unit 21 acquires the current combing speed data. This combing speed data can be obtained based on the displacement integral of the six-axis torque sensor 11, or through the inertial measurement unit integrated in the handle assembly. The state space construction unit 21 uses a max-min normalization algorithm to map the real-time tension modulus value, tension gradient value, hair moisture content value, real-time temperature value, and combing speed value to a dimensionless range of zero to one.

[0061] Finally, the state space construction unit 21 arranges and combines the normalized values ​​in a preset order to generate the state vector at the current moment. This state vector is a multi-dimensional array that contains complete information describing the current physical environment of the system, clearly defining the instantaneous state of the hair in the thermodynamic and mechanical fields. The state space construction unit 21 stores the generated state vector in a cache and passes it as an input parameter to the physical property inference unit 22 and the policy decision unit 23 for subsequent steps.

[0062] See attached document Figure 2 In step S2, the system performs a physical property deduction of hair strands based on Gaussian process regression. This step is independently executed by the physical property deduction unit 22 within the central intelligent control module 20. The core logic of this step is based on the hygroscopic plasticizing mechanism of keratin materials, namely, the intervention of water molecules will disrupt the hydrogen bond network between keratin molecular chains, increase the mobility of polymer chain segments, and thus lower the critical temperature for the material to transition from the glassy state to the elastic state. Since the glass transition temperature of hair strands exhibits highly nonlinear variation characteristics under different moisture contents, and is subject to uncertainty due to individual hair quality differences, traditional linear fitting cannot meet the requirements of precise control. The physical property deduction unit 22 adopts a non-parametric Bayesian method, namely the Gaussian process regression model, to map the normalized moisture content data obtained in step S1 to the thermodynamic temperature domain.

[0063] The physical property inference unit 22 pre-stores a training set constructed based on a large amount of rheological experimental data, which includes hair rheological test data under different humidity environments. The physical property inference unit 22 uses the current moisture content value as the test input point and calculates the similarity between this test input point and all sample points in the training set using the covariance function, i.e., the kernel function. Through this similarity measure, the physical property inference unit 22 constructs a covariance vector and, combined with the target value vector of the training set, predicts the posterior mean of the dynamic glass transition temperature of the hair in the current state through matrix operations. This posterior mean is considered the minimum thermodynamic condition for the hair to enter a plastic deformation state at the current moment.

[0064] The regression model formula used by physical property deduction unit 22 to calculate the dynamic glass transition temperature is as follows:

[0065] ;

[0066] in, The predicted value of the dynamic glass transition temperature obtained from the simulation; This represents a column vector of covariances, whose elements are composed of the covariances between the current moisture content input value calculated by the kernel function and the input value of each sample in the training set. The covariance matrix represents the training set, which characterizes the degree of correlation between the sample points in the training set. The variance represents the observation noise and is used to filter measurement noise introduced by the sensor. Represents the identity matrix; This represents the observed target vector in the training set, which is the set of known true glass transition temperatures corresponding to different water contents. Represents the matrix transpose operation; This represents the matrix inversion operation.

[0067] After completing the above calculations, the physical property deduction unit 22 will obtain the predicted value of the dynamic glass transition temperature. The output is sent to the strategy decision unit 23. This predicted value serves as a dynamic floating threshold in subsequent steps to determine whether the thermal pulse energy meets the standard, ensuring that the applied heat energy is just enough to soften the hair without causing excessive heat accumulation. This process does not rely on a fixed temperature setting, but rather uses adaptive thermodynamic parameter definition based on the real-time microscopic physical state of the hair.

[0068] See attached document Figure 2 In step S3, the system performs pre-knotting sign identification and strategy routing. This step is specifically executed by the strategy decision-making unit 23 within the central intelligent control module 20. Its purpose is to accurately identify the mechanical interaction nature of the current combing process based on the multi-dimensional state space data output in step S1, and accordingly determine the execution path of subsequent control loops. This step divides the combing state into two categories—pre-knotting sign state and smoothing shaping state—by analyzing the temporal variation characteristics of physical parameters.

[0069] The strategy decision-making unit 23 first allocates a time sliding window in memory to store the state vector sequence of N consecutive sampling periods from the state space construction unit 21. The length of this time sliding window is set to cover the typical time span from the hair contacting the comb teeth to the occurrence of elastic deformation. The strategy decision-making unit 23 uses the state vector sequence within this time sliding window as input data and transmits it to the internally integrated long short-term memory network model. This long short-term memory network model is trained offline and has the ability to capture the correlation between abnormal fluctuations in tension gradient and nonlinear decay of combing speed, and can distinguish between normal combing resistance and deadlock resistance caused by hair entanglement.

[0070] The Long Short-Term Memory (LSTM) network model iteratively processes the input time-series data, extracting temporal dependency features through internal forgetting, input, and output gates, and outputting a probability value between zero and one via a fully connected layer. This probability value represents the confidence level of the current hair strand's risk of mechanical tangling. The strategy decision unit 23 compares this probability value with a preset safety threshold. When the probability value is greater than the safety threshold, the strategy decision unit 23 determines that the current state is a precursor to tangling; when the probability value is less than or equal to the safety threshold, the strategy decision unit 23 determines that the current state is a smooth styling state.

[0071] After classifying the states, the strategy decision unit 23 executes logical routing based on the judgment results. If the state is determined to be a pre-tangling state, the strategy decision unit 23 immediately generates a high-priority untangling interruption command, transfers system control to step S4, and initiates the thermomechanical phase-locked untangling control process. At this time, the system's control objective switches to releasing physical entanglement under minimum tension. If the state is determined to be a smooth styling state, the strategy decision unit 23 generates a styling maintenance command, transfers system control to step S5, and initiates the rheological styling heat dosage control process. At this time, the system's control objective switches to applying precisely measured enthalpy to the hair strands to achieve plastic shaping. Through this routing mechanism, the system can respond to combing obstruction within milliseconds.

[0072] See attached document Figure 2 In step S4, the system executes thermomechanical phase-locked untangling control. This step is handled by the phase coordination control unit 24 within the central intelligent control module 20, which is responsible for specific parameter calculations and timing scheduling. When the strategy decision unit 23 identifies a tangling risk in the preceding steps and issues an untangling interruption command, the system enters this high-priority control branch. The core logic of this step lies in utilizing the time lag characteristic of heat transfer to precisely control the timing of mechanical vibration intervention, ensuring that the comb teeth apply untangling force only during the brief window period when the microstructure of the hair material undergoes a softening transformation, thereby reducing the risk of hair breakage while achieving efficient separation.

[0073] The phase coordination control unit 24 first reads the real-time ambient temperature collected by the multimodal sensing module 10 and the predicted dynamic glass transition temperature derived by the physical property inference unit 22. Based on the current combing resistance level, the phase coordination control unit 24 sets the target pulse temperature of the pulse heater 31. This target pulse temperature is set to a specific value higher than the predicted dynamic glass transition temperature to provide a sufficient thermal potential energy gradient. Subsequently, the phase coordination control unit 24 calculates the phase delay time required for heat energy to be conducted from the hair surface to the physical central axis. This phase delay time determines the hysteresis between mechanical action and thermal action.

[0074] The phase coordination control unit 24 calculates the phase delay time using the following thermal conduction hysteresis formula:

[0075] ;

[0076] in, This represents the calculated phase delay time, in seconds. This represents the average geometric radius of a hair strand, and this value is stored in the memory as a system preset parameter; The effective thermal diffusivity of hair material represents the rate at which heat energy propagates within this medium. The target pulse temperature represents the output airflow of the pulse heater 31. This represents the current ambient temperature collected by temperature monitor 13; This represents the predicted dynamic glass transition temperature obtained in step S2 above; Represents the natural logarithm operation.

[0077] After calculating the control parameters, the phase coordination control unit 24 sends a pulse trigger signal to the pulse heater 31. In response to this signal, the pulse heater 31 outputs a burst of hot air at maximum power, which rapidly coats the hair surface and establishes an inward thermal flux. At this time, the piezoelectric actuator 32 remains stationary to avoid applying mechanical shear force before the hair core has softened.

[0078] The timer inside the phase coordination control unit 24 starts running. The timer continues until the calculated phase delay time is reached. At this point, the temperature distribution across the cross-section of the hair strand has met the conditions for the core region to reach or exceed the dynamic glass transition temperature, and the hair strand as a whole enters a viscoelastic state with a low Young's modulus. At this moment, the phase coordination control unit 24 immediately sends a high-frequency drive signal to the piezoelectric actuation mechanism 32. The piezoelectric actuation mechanism 32 drives the comb teeth to generate micro-amplitude vibrations at a specific frequency. Since the hair strand is in a relaxed state at this time, the vibration of the comb teeth can effectively overcome the static friction between the hair strands, untangling the entangled nodes under low stress.

[0079] After a preset untangling cycle of continuous mechanical vibration, the phase coordination control unit 24 simultaneously sends a stop heating command to the pulse heater 31 and a maximum airflow command to the pneumatic regulating fan 33. The pneumatic regulating fan 33 outputs a high-speed cold airflow at ambient temperature to force convection cooling of the hair strands, i.e., pneumatic quenching. This process rapidly reduces the hair temperature below the dynamic glass transition temperature, rebuilds the hydrogen bond network between keratin molecules, thereby locking in the straightened state after untangling and preventing secondary deformation or rebound of the hair strands due to residual heat.

[0080] See attached document Figure 2In step S5, the system performs rheological styling heat dose control. This step is initiated after the strategy decision unit 23 determines that the current styling state is smooth, and is mainly controlled by the central intelligent control module 20 directly regulating the power output of the pulse heater 31. This step aims to solve the problem of uneven heating of hair strands caused by changes in combing speed in traditional constant power heating methods. By establishing a real-time mapping relationship between enthalpy and motion state, it ensures that the heat energy per unit volume obtained by the hair strands remains constant at different combing speeds, thereby maintaining the consistency of rheological properties.

[0081] The central intelligent control module 20 first reads the target styling parameters set by the user, which correspond to the target enthalpy density required for the hair to undergo plastic deformation. Then, the central intelligent control module 20, in conjunction with the real-time combing speed collected by the multimodal sensing module 10, calculates the instantaneous heating power required to achieve the target enthalpy density. This process is based on the principle of energy conservation, treating the combing process as an integral heating process of the hair as a continuous medium. The system must ensure that the heat flux injected within any differential time interval matches the rate at which the medium flows through the heating zone.

[0082] The central intelligent control module 20 calculates the real-time output power of the pulse heater 31 using the following power dynamic tracking formula:

[0083] ;

[0084] in, The real-time target output power of the pulse heater 31 is represented in watts. This represents the target enthalpy increment required per unit length of hair based on the target hairstyle; this value is a preset constant. The average linear density of a hair bundle is expressed in kilograms per meter. This represents the real-time tangential combing speed along the combing direction calculated by the state space construction unit 21; The coefficient representing the thermal coupling efficiency from the heating element to the hair strand is used to compensate for heat loss. Represents the proportional feedback gain coefficient; This represents the set optimal plastic deformation temperature; This represents the real-time surface temperature of the hair strands, collected by temperature monitor 13.

[0085] After calculating the real-time target output power, the central intelligent control module 20 adjusts the duty cycle of the pulse heater 31 using pulse width modulation technology. When the real-time tangential combing speed is detected... When the power is increased, the system linearly increases the output power to compensate for insufficient energy intake caused by the shortened heating time; when a slowdown or stagnation in combing speed is detected, the system quickly reduces the power or cuts off the heating to prevent excessive heat accumulation in local areas and causing thermal damage. This dynamic power tracking mechanism ensures that regardless of the speed of the user's operation, the amount of heat received by the hair along the axial direction is always maintained within the predetermined rheological window.

[0086] See attached document Figure 2 In step S6, the system performs iterative updates to the reinforcement learning model. This step is executed by the strategy decision unit 23 within the central intelligent control module 20 at the end of each control cycle, aiming to establish a closed-loop optimization mechanism between action execution and environmental feedback. The strategy decision unit 23 dynamically adjusts the weight parameters within the control algorithm by quantitatively evaluating the actual effects of the control strategies in the preceding steps, thereby enabling the system to gradually adapt to the hair characteristics and styling habits of specific users as the number of uses increases.

[0087] The strategy decision-making unit 23 first collects environmental state data after the end of the current control cycle, and then retrieves environmental state data before the start of the cycle and the action commands executed during the cycle. The strategy decision-making unit 23 calculates the current immediate reward value based on multi-objective optimization principles. This calculation process comprehensively considers two mutually constraining evaluation dimensions: one is the combing efficiency dimension, which is reflected in the reduction of mechanical resistance experienced by the comb teeth; the other is the hair health dimension, which is reflected in the cumulative amount of heat load borne by the hair. The system aims to maximize the tension reduction while minimizing thermal damage.

[0088] Strategy decision-making unit 23 calculates the immediate reward value using the following reward function formula:

[0089] ;

[0090] in, This represents the immediate reward value calculated for the current control cycle, which is used to quantitatively evaluate the quality of the current action sequence. The coefficient representing the weight of untying efficiency is used to set the degree of importance the system places on eliminating mechanical resistance; The tension modulus, representing the start time of the control cycle, is acquired by the six-axis torque sensor 11 and processed by the state space construction unit 21. The tension modulus represents the moment the control cycle ends; This represents the reference tension constant used for normalization; A coefficient representing the thermal damage penalty weight, used to set the system's penalty level for overheating risk; Represents the duration of the current control cycle; Represents the time variable The real-time temperature of the hair surface is constantly collected by the temperature monitor 13; This represents the preset safe temperature threshold for hair heat. Represents a time integral infinitesimal element; This represents the maximum value function, ensuring that the heat penalty term is calculated only when the temperature exceeds the safe threshold.

[0091] Calculate the instant reward value Then, the policy decision unit 23 substitutes it into the value function update logic of the reinforcement learning algorithm. The policy decision unit 23 calculates the time difference error between the actual reward and the expected reward generated by the current action, and updates the synaptic weights in the policy network using gradient descent backpropagation based on this error. If the immediate reward value... If the value is positive and large, the strategy decision unit 23 increases the probability weight of selecting the action sequence again under similar conditions; if the immediate reward value... When the value is negative, the strategy decision unit 23 suppresses the tendency to choose that action sequence under similar conditions. Through this continuous iterative update, the control model inside the strategy decision unit 23 gradually converges to the optimal strategy for the current hair quality characteristics, realizing the migration from a general initialization model to a personalized special model.

[0092] The application scenario expansion scheme and technical effects provided by this invention are described as follows:

[0093] Different initial configurations for different hair types are crucial to ensuring the system's basic safety and effectiveness from the outset. Upon initial startup or upon receiving user input via an external interface, the central intelligent control module 20 initializes its internal algorithms by calling different preset parameter sets based on the hair's physical classification. For fine, soft hair, due to its small cross-sectional area and low keratin density, it has a low heat capacity and is prone to thermal denaturation. Therefore, the central intelligent control module 20 lowers the baseline glass transition temperature in the physical property deduction unit 22, setting a lower thermal safety threshold. Simultaneously, the strategy decision unit 23 sets the tension trigger threshold of the six-axis torque sensor 11 to a low-sensitivity range to ensure that the knot protection mechanism is triggered with minimal mechanical resistance, preventing breakage due to insufficient tensile strength of the hair strands. Furthermore, the initial vibration amplitude of the piezoelectric actuator 32 in the high-frequency response execution module 30 is limited to a low-energy range to avoid excessive mechanical shear force causing physical damage to fragile hair strands.

[0094] Conversely, for coarse hair, due to its thicker medulla and tightly closed cuticles, the rate of heat conduction to the hair core is slower, and it has a higher tensile modulus. The central intelligent control module 20 increases the initial heating power reference and extends the basic heat conduction lag time calculated by the phase coordination control unit 24 to reserve a sufficient heat diffusion time window. The strategy decision unit 23 sets the tension trigger threshold to a high range, allowing the system to avoid interruption when applying a large combing torque, thereby effectively penetrating thick hair strands. The piezoelectric actuator 32 is configured to operate in a high-frequency, high-amplitude mode to generate sufficient kinetic energy to overcome the strong friction coefficient between coarse hair strands. This differentiated initialization strategy based on physical properties provides the subsequent reinforcement learning model with a starting state within a safe convergence region, shortening the learning cycle for the model to adapt to specific users.

[0095] In summary, the core technical features of this invention address several key challenges. First, the dynamic glass transition temperature estimation technique based on Gaussian process regression solves the thermodynamic control problem of hair strands under unsteady moisture content conditions. Traditional isothermal control neglects the significant impact of moisture as a plasticizer on the mobility of polymer chains, easily leading to excessively high temperatures causing keratin denaturation in wet hair or insufficient temperatures causing styling failure in dry hair. This invention, by real-time monitoring of the dielectric constant and using a nonlinear regression model to precisely locate the current glass transition critical point, achieves on-demand heat distribution. This control method ensures at the microscopic level that hair strands are heat-treated only at the minimum energy level required to reach the viscoelastic transition, thereby maximizing the preservation of internal moisture and protein structural integrity while achieving effective plastic deformation.

[0096] Secondly, the thermomechanical phase-locked untangling control technology overcomes the mechanical breakage problem caused by forced pulling in traditional combing methods by decoupling the thermal softening effect and the mechanical separation effect in the time domain. The phase coordination control unit 24 accurately calculates the heat conduction lag time, ensuring that the piezoelectric actuation mechanism 32 only intervenes during the low modulus window period after the temperature in the core area of ​​the hair reaches the glass transition temperature. This precise timing utilizes the physical relaxation characteristics of the material near the phase transition point, reducing the external mechanical work required for the untangling process. Combined with the subsequent pneumatic quenching and locking action, the system can quickly freeze the polymer chains straightened by thermomechanical action in a new conformation, preventing the high-elasticity rebound caused by residual heat, thereby improving the smoothness retention time and combing efficiency of the hair.

Claims

1. An adaptive control system for an intelligent hot air comb, characterized in that, include: A power interaction module (40) connects to each module and provides operating voltage and drive current; The multimodal sensing module (10) collects physical environment data during the combing process in real time and outputs mechanical resistance vector, hair dielectric properties data and thermodynamic temperature data; The high-frequency response execution module (30) includes a pulse heater (31), a piezoelectric actuator (32), and a pneumatic regulating fan (33). The central intelligent control module (20) performs adaptive parameter optimization for different hair quality states, which is used to dynamically map the dynamic glass transition temperature of the current hair quality based on the real-time changing dielectric properties data of the hair strand, and adaptively calculate the phase delay time required for heat conduction by combining the thermodynamic temperature data and the dynamic glass transition temperature. The central intelligent control module (20) also has a thermomechanical phase-locked untying control loop. The thermomechanical phase-locked untying control loop is used to generate a dynamic timing control signal when the pre-knotting state is identified based on the mechanical resistance vector. It controls the pulse heater (31) to output a heat pulse first. After the phase delay time required for the hair to complete the phase transition from the glass state to the elastic state, it adaptively triggers the piezoelectric actuator (32) to generate mechanical vibration, and then drives the pneumatic regulating fan (33) to output cold air.

2. The adaptive control system of the intelligent hot air comb according to claim 1, characterized in that, The multimodal sensing module (10) collects the physical environment data in the following ways: The multimodal sensing module (10) uses a six-axis torque sensor (11) to detect the force components along three orthogonal directions and the torque components rotating around the axis on the comb teeth, and synthesizes the mechanical resistance vector that characterizes the magnitude and direction of combing resistance. The multimodal sensing module (10) uses the dielectric constant detection probe (12) to detect the change in dielectric constant or impedance in the comb electrode circuit, quantifies the real-time moisture content of the hair, and uses the real-time moisture content as the dielectric property data of the hair. The multimodal sensing module (10) uses a temperature monitor (13) to detect the airflow temperature in the air duct and the conduction temperature at the interface between the comb teeth and the hair strands, as the thermodynamic temperature data.

3. The adaptive control system of the intelligent hot air comb according to claim 2, characterized in that, The internal logic architecture of the central intelligent control module (20) includes: State space construction unit (21) is used to receive data from the multimodal perception module (10) and generate a state vector describing environmental features; Physical property deduction unit (22), the physical property deduction unit (22) is specifically used to complete the dynamic mapping of the dynamic glass transition temperature based on the dielectric property data of the hair; Strategy decision unit (23), which is specifically used to analyze the state vector to identify the pre-knotting state and determine the diversion of the control strategy; Phase coordination control unit (24) is specifically designed to perform adaptive calculation of the phase delay time based on thermal conduction logic.

4. The adaptive control system of the intelligent hot air comb according to claim 3, characterized in that, The process by which the state space construction unit (21) constructs the state vector for identifying the pre-knotting state includes: Perform time difference operation on the mechanical resistance vector to calculate the rate of change of tension per unit time and obtain the tension gradient index, which is used to characterize the abrupt change trend of combing resistance; Obtain the real-time combing speed of the comb teeth along the combing direction; The mechanical resistance vector, the tension gradient index, the real-time moisture content, the thermodynamic temperature data, and the real-time combing speed are mapped to the same normalized dimension range and combined to generate the state vector.

5. The adaptive control system of the intelligent hot air comb according to claim 3, characterized in that, The logic of the physical property deduction unit (22) dynamically mapping the dynamic glass transition temperature includes: A Gaussian process regression model is constructed with the real-time moisture content in the dielectric properties data of the hair as the input variable. The Gaussian process regression model is pre-set with a training set describing the rheological properties of hair under different humidity conditions. The similarity covariance feature between the current input real-time moisture content and the training set samples is calculated using a kernel function. The predicted temperature required for hair to transition from a glassy state to a viscous flow state under the current moisture content is calculated using a probability distribution, and this predicted temperature is defined as the dynamic glass transition temperature.

6. The adaptive control system of the intelligent hot air comb according to claim 3, characterized in that, The process by which the strategy decision unit (23) identifies the pre-knotting state and triggers the thermomechanical phase-locked untying control loop includes: The time series of the state vector is input into the long short-term memory network model, and the output is a probability value representing the risk of mechanical knotting. When the probability value is greater than the preset safety threshold, it is determined to be the pre-knotting state, and the thermodynamic phase-locked untying control loop is officially activated, and the phase coordination control unit (24) is called to perform the untying operation. When the probability value is less than or equal to the preset safety threshold, it is determined to be a smooth shaping state, triggering rheological shaping thermal dose control.

7. The adaptive control system of the intelligent hot air comb according to claim 3, characterized in that, The process by which the phase coordination control unit (24) adaptively calculates the phase delay time includes: Set the target pulse temperature of the pulse heater (31) and obtain the current ambient temperature; Calculate the first temperature difference between the target pulse temperature and the current ambient temperature, and the second temperature difference between the target pulse temperature and the dynamic glass transition temperature, and perform a natural logarithmic operation on the ratio of the first temperature difference to the second temperature difference; The result of the natural logarithm operation is multiplied by the ratio of the square of the hair's geometric radius to the effective thermal diffusivity of the hair material to obtain the time required for heat energy to be conducted from the surface of the hair to the center and soften, which is used as the phase delay time.

8. The adaptive control system of the intelligent hot air comb according to claim 6, characterized in that, When the smooth styling state is determined, the strategy decision unit (23) performs adaptive parameter optimization for styling consistency. The strategy decision unit (23) includes power dynamic following logic: Based on the principle of energy conservation, and according to the preset target enthalpy density and hair linear density, combined with the combing speed collected in real time, the heating power required to maintain a constant heat energy per unit length of hair is calculated. Establish a feedback mechanism based on the difference between the measured surface temperature of hair strands and the optimal plastic deformation temperature; The output power of the pulse heater (31) is dynamically adjusted so that the output power increases linearly with the increase of the combing speed, thereby compensating for insufficient energy intake caused by the shortened heating time.

9. The adaptive control system of the intelligent hot air comb according to claim 6, characterized in that, The policy decision unit (23) further includes a reinforcement learning iterative update mechanism, the process of which continuously optimizes the accuracy of the adaptive parameter optimization includes: At the end of the control cycle, an instant reward value is calculated. The calculation logic of the instant reward value is set to be positively correlated with the magnitude of the decrease in the modulus of the mechanical resistance vector during the combing process, and includes a penalty term for the thermodynamic temperature data exceeding a safety threshold. The weight parameters in the policy network are updated in reverse using the instant reward value, thereby optimizing the system's action selection when faced with a similar state vector in the next cycle.

10. An adaptive control method for an intelligent hot air comb, characterized in that, An adaptive control system for an intelligent hot air comb as described in any one of claims 1-9 includes the following steps: A multi-dimensional state space containing mechanical resistance vectors and hair dielectric properties data is constructed using a multi-modal sensing module (10). The central intelligent control module (20) dynamically maps the current dynamic glass transition temperature of the hair based on the dielectric properties data of the hair, and adaptively calculates the phase delay time by combining the thermodynamic temperature data. Determine the current combing state. If a pre-knotting state is detected, activate the thermodynamic phase-locked unknotting control loop, control the pulse heater (31) to output a heat pulse, and after the phase delay time, adaptively trigger the piezoelectric actuator (32) to vibrate, and then output cold air to lock. If a smooth styling state is detected, the rheological styling heat dose control is activated, and the heating power is dynamically adjusted according to the combing speed to maintain a constant enthalpy received per unit length of hair.

Images