Method and system for secure access of an electronic atomizing device

Abstract

The application discloses a kind of safety access method and system of electronic atomization device, and the application relates to the technical field of safety access, access feature is input to the preset safety access space to multiple levels, and determine the multiple access paths of user relative to electronic atomization device in the safety access space, according to the access mode corresponding to each access path and the use level of electronic atomization device Determine corresponding safety access event, improve the accuracy of safety access event.In the safety access event, the safety detection content and the use state of electronic atomization device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence;Electronic atomization device receives the authorization sequence, and sequentially executes the corresponding verification action to output the response result of each verification action, and dynamically configures the accessible content of electronic atomization device in combination with the working constraint condition of electronic atomization device, improve the accuracy of accessible content of electronic atomization device.

Description

Technical Field

[0001] This invention relates to the technical field of secure access, and more particularly to a secure access method and system for an electronic atomizing device. Background Technology

[0002] As an electronic device that heats and atomizes a medium for users to inhale, the safety and access control of electronic atomizing devices have always been a focus of attention in the industry. Since the atomizing medium of electronic atomizing devices usually contains ingredients such as nicotine, in order to prevent minors from stealing, non-owners from abusing, and safety accidents caused by improper operation, it is necessary to introduce a secure access mechanism in electronic atomizing devices to authenticate and control users' access behavior.

[0003] Existing technologies rely solely on single static biometric features or physical actions for identity verification. Due to the lack of multi-level access feature constraints, the system cannot finely divide and control the user's access path within the preset secure access space. It often has to adopt a coarse access control logic of "all pass" or "all deny". This single-dimensional identification is easily forged or stolen, and it cannot accurately distinguish the legitimate owner's operations under different intentions, affecting the accuracy of secure access events and resulting in low accuracy of the accessible content of electronic atomization devices. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a secure access method and system for electronic atomization devices.

[0005] This invention provides a secure access method for an electronic atomizing device, comprising: In response to the wake-up event of the electronic atomization device, the system acquires the user's multimodal access data, identifies the corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of the multimodal access data, and determines multi-level access characteristics in combination with the user's verification information. The multi-level access characteristics are input into a preset secure access space, and multiple access paths of the user relative to the electronic atomizing device are determined in the secure access space. The corresponding secure access events are determined according to the access mode corresponding to each access path and the usage level of the electronic atomizing device. In this security access event, the corresponding abnormal access content is marked, and the corresponding security detection content is determined by combining the usage scenario of the e-vaporization device and the previous access history. The security detection content and the usage status of the e-vaporization device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence. The electronic atomizing device receives the authorization sequence and executes the corresponding verification actions in sequence to output the response results of each verification action. It also dynamically configures the accessible content of the electronic atomizing device in combination with the device's operating constraints.

[0006] This invention provides a secure access system for an electronic atomizing device, which is applied to the aforementioned secure access method for the electronic atomizing device; the secure access system for the electronic atomizing device includes: The access data module is used to respond to the wake-up event of the electronic atomization device, acquire the user's multimodal access data, determine the corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of the multimodal access data, and determine multi-level access characteristics in combination with the user's verification information. The secure access module is used to input the multi-level access characteristics into the preset secure access space, and determine multiple access paths of the user relative to the electronic atomizing device in the secure access space. Based on the access mode corresponding to each access path and the usage level of the electronic atomizing device, the corresponding secure access event is determined. The authorization sequence module is used to mark the corresponding abnormal access content in the security access event, and determine the corresponding security detection content by combining the usage scenario of the electronic atomization device and the previous access history. The security detection content and the usage status of the electronic atomization device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence. The accessibility module is used by the electronic atomizing device to receive the authorization sequence and execute the corresponding verification actions in sequence to output the response results of each verification action, and dynamically configure the accessible content of the electronic atomizing device in combination with the working constraints of the electronic atomizing device.

[0007] Compared with the prior art, the beneficial effects of the present invention are: (1) In response to the wake-up event of the electronic atomizing device, the user’s multimodal access data is acquired. Based on the recognition of the multimodal access data, the corresponding dynamic behavioral characteristics and static biometric characteristics are determined, and the multi-level access characteristics are determined in combination with the user’s verification information. The multi-level access characteristics are input into the preset safe access space, and multiple access paths of the user relative to the electronic atomizing device are determined in the safe access space. The corresponding safe access events are determined according to the access mode corresponding to each access path and the usage level of the electronic atomizing device. Multi-level access characteristics are introduced to further control multiple access paths and improve the accuracy of safe access events.

[0008] (2) In the security access event, the corresponding abnormal access content is marked, and the corresponding security detection content is determined by combining the usage scenario of the electronic atomizing device and the previous access history. The security detection content and the usage status of the electronic atomizing device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence. The electronic atomizing device receives the authorization sequence and executes the corresponding verification actions in sequence to output the response results of each verification action. The accessible content of the electronic atomizing device is dynamically configured in combination with the working constraints of the electronic atomizing device, further controlling the security detection content. The response results of each verification action and the working constraints of the electronic atomizing device are fully considered, which improves the accuracy of the accessible content of the electronic atomizing device. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the secure access method for the electronic atomizing device in an embodiment of the present invention; Figure 2 This is a flowchart illustrating step S11 of the secure access method for the electronic atomizing device in an embodiment of the present invention. Figure 3 This is a flowchart illustrating step S12 of the secure access method for the electronic atomizing device in an embodiment of the present invention. Figure 4 This is a flowchart illustrating step S13 of the secure access method for the electronic atomizing device in an embodiment of the present invention. Figure 5 This is a flowchart illustrating step S14 of the secure access method for the electronic atomizing device in an embodiment of the present invention. Figure 6 This is a schematic diagram of the structural composition of the secure access system for the electronic atomizing device in an embodiment of the present invention. Detailed Implementation

[0010] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0011] Please see Figures 1 to 6 A secure access method for an electronic atomizing device, applied in a secure access scenario; the secure access method for the electronic atomizing device includes: Step S11: In response to the wake-up event of the electronic atomization device, acquire the user's multimodal access data, determine the corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of the multimodal access data, and determine the multi-level access characteristics in combination with the user's verification information; Step S12: Input the multi-level access feature into the preset secure access space, and determine multiple access paths of the user relative to the electronic atomizing device in the secure access space. Determine the corresponding secure access event based on the access mode corresponding to each access path and the usage level of the electronic atomizing device. Step S13: In this security access event, mark the corresponding abnormal access content, and determine the corresponding security detection content by combining the usage scenario of the electronic atomization device and the previous access history. Load the security detection content and the usage status of the electronic atomization device into the corresponding knowledge graph to output the corresponding authorization sequence. Step S14: The electronic atomizing device receives the authorization sequence and executes the corresponding verification actions in sequence to output the response results of each verification action, and dynamically configures the accessible content of the electronic atomizing device in combination with the working constraints of the electronic atomizing device.

[0012] refer to Figure 2 In step S11, the specific steps are as follows: S111: The user performs a wake-up operation on the electronic atomizing device and responds to the wake-up event of the electronic atomizing device. The electronic atomizing device integrates a multimodal sensing array, which collects the user's touch trajectory, voiceprint and fingerprint through the multimodal sensing array to form multimodal access data. S112: Perform spatiotemporal alignment on multimodal access data, and combine a fusion mechanism during the alignment process to determine the corresponding dynamic behavioral features and static biometric features. The dynamic behavioral features reflect the user's operating habits; the static biometric features reflect the user's inherent physiological attributes. S113: Obtain the user's verification information, which includes the user's registration information and historical interaction preference information. Perform deep concatenation and weighted mapping of dynamic behavioral features, static biometric features and verification information in the feature dimension to construct the corresponding multi-level access features. The multi-level access features include the identity confirmation layer, the behavioral intent layer and the permission matching degree.

[0013] In the embodiments of this application, the user performs a wake-up operation on the electronic atomizing device and responds to the wake-up event of the electronic atomizing device. The electronic atomizing device integrates a multimodal sensing array, which collects the user's touch trajectory, voiceprint and fingerprint through the multimodal sensing array to form multimodal access data and introduce a wake-up event.

[0014] At this time, the user performs a physical wake-up operation on the electronic atomizing device, which is specifically manifested as a set mechanical trigger action or physical contact action; in response to the wake-up operation, the microcontroller at the bottom of the electronic atomizing device generates a level transition or interrupt signal, thereby triggering a wake-up event; at the same time, the wake-up event serves as the unified time reference origin for multimodal data acquisition, initiating the subsequent sensing and acquisition process.

[0015] In response to a wake-up event, the multimodal sensing array integrated inside the electronic atomization device switches from a dormant state to an active state. The multimodal sensing array includes at least a touch sensing module for detecting the distribution and rate of change of surface capacitance, a miniature electret or MEMS microphone module for collecting acoustic features, and an RF or ultrasonic fingerprint sensing module for collecting epidermal ridge data. Each module achieves millisecond-level time synchronization based on the time reference origin to ensure strict alignment of multimodal data in time sequence.

[0016] Optionally, the touch sensing module is embedded in the thumb grip area of ​​the electronic atomizing device housing, and is distributed in a long strip shape to detect the capacitance change when the user's finger slides; the MEMS microphone module is set on the side wall of the air intake channel, 1 cm to 2 cm away from the mouthpiece, to avoid noise caused by direct airflow; the ultrasonic fingerprint sensing module is integrated into the power button or the fixed recognition area of ​​the housing, and its surface is kept flush with the housing.

[0017] To achieve time synchronization among multiple modules, the microcontroller of the electronic atomizing device generates a global synchronization pulse signal the instant it detects a level transition caused by a wake-up event. This signal is simultaneously sent to the three modules via a hardware interrupt line. Upon receiving the pulse, each module starts data acquisition within a ±1 millisecond error range, using its own internal clock as a reference. This ensures strict alignment of touch traces, voiceprints, and fingerprint images on the timeline.

[0018] To avoid resource conflicts or electrical interference caused by different sensors operating in parallel, the microcontroller employs a time-division multiplexing scheduling strategy: During the first 0 to 200 milliseconds after a wake-up event, bus bandwidth is preferentially allocated to the ultrasonic fingerprint sensing module to complete a full fingerprint image scan; between 200 and 700 milliseconds, the touch sensing module and the MEMS microphone module are alternately scheduled to acquire data at sampling rates of 100 Hz and 16 kHz, respectively; simultaneously, the power supply of each module is controlled by independent low-dropout regulators, and the microphone signal lines and the high-frequency drive lines of the fingerprint module are physically spaced at least 5 mm apart, and are placed on different layers on the printed circuit board and isolated with a complete ground plane to eliminate crosstalk. This combination of layout, synchronization, and scheduling ensures that the multimodal sensing array can stably and conflict-free acquire raw access data suitable for subsequent feature extraction during the user's natural grip and operation.

[0019] With the collaborative operation of the multimodal sensing array, multi-dimensional data of the user during the wake-up and holding process are collected in parallel: through the touch sensing module, the sliding coordinate sequence of the user's fingertip on the surface of the device shell and the change of the touch area are continuously recorded at a preset sampling frequency to form touch trajectory data; through the microphone module, the inherent lip and teeth airflow sound of the user's specific voice commands or operation process is collected and the acoustic frequency domain features are extracted to form voiceprint data; through the fingerprint sensing module, when the user naturally holds and touches the recognition area of ​​the device, the ridge topology and pore features of the finger skin are scanned and obtained to form hand fingerprint data. The time-series correlation data set containing touch trajectory data, voiceprint data and hand fingerprint data is packaged and constructed into multimodal access data.

[0020] Specifically, when the electronic atomization device is a Type A electronic cigarette, the Type A electronic cigarette is in standby sleep state. When the user takes the electronic cigarette out of his pocket and presses the fire button with his thumb or moves the airflow adjustment ring, this action is a wake-up operation. The micro switch or pressure sensor at the bottom of the fire button generates an electrical signal when it is subjected to force. The microcontroller responds to the electrical signal to trigger the wake-up event. At the same time, the system clock records this time as a timestamp T0, which serves as the time anchor point for subsequent voiceprint and fingerprint collection.

[0021] After a wake-up event occurs, the strip-shaped capacitive touch area on the outside of the Type A electronic cigarette, the ultrasonic fingerprint sensor integrated on the inhalation end, next to the mouthpiece, or on the ignition button, and the MEMS microphone that reuses the airflow sensing channel are instantly powered on; the microcontroller sends a synchronous clock signal to the above three modules to ensure that capacitive sampling, fingerprint scanning, and sound wave acquisition start synchronously on the same time reference, avoiding data misalignment caused by sensor response delay.

[0022] Multimodal data acquisition is essentially completed imperceptibly during the user's continuous action of "taking a cigarette - pressing the button - preparing to inhale": On the one hand, after pressing the ignition button, the user's thumb naturally slides down the surface of the cigarette to adjust the grip posture. The strip capacitive touch area captures the XY coordinate sequence of the thumb slide and the gradient of the change in the contact area of ​​the fingertip at a sampling rate of 100Hz, forming touch trajectory data that reflects the user's operating habits. For example, the sliding trajectory of a specific user presents a specific parabolic arc.

[0023] On the other hand, at the moment of pressing, the ultrasonic fingerprint sensor integrated on the surface of the ignition button penetrates the epidermis of the thumb, scans and extracts dermal ridge data and micropore distribution characteristics at a depth of 0.2mm, forming fingerprint data that reflects the user's inherent physiological attributes.

[0024] On the other hand, when users prepare to inhale, they utter short voices such as "power on" or habitually cough lightly before inhaling. The MEMS microphone collects the sound waves, and after front-end preprocessing, the fundamental frequency and Mel-frequency cepstral coefficients (MFCC) are extracted to form voiceprint data.

[0025] The microcontroller temporally binds the sliding trajectory coordinate set, fingerprint image feature matrix and voiceprint spectrum feature vector obtained at the same timestamp T0 to construct the multimodal access data for this access, and sends it to the subsequent step S112 for spatiotemporal alignment and feature extraction.

[0026] Furthermore, the multimodal access data is spatiotemporally aligned, and a fusion mechanism is used in the alignment process to determine the corresponding dynamic behavioral features and static biometric features. The dynamic behavioral features reflect the user's operating habits, while the static biometric features reflect the user's inherent physiological attributes, thus fully considering the spatiotemporal alignment.

[0027] At this point, for multimodal access data, the timestamp information carried by each modal data stream is extracted and uniformly calibrated using the time reference origin of the wake-up event to eliminate the hardware response delay differences between different sensor modules and achieve strict alignment in the time dimension. At the same time, a mapping relationship is established between the physical coordinates of the multimodal sensing array and the standard spatial model of the electronic atomization device. The coordinate set of the touch trajectory, the touch center point of the fingerprint sensor, and the time sequence of voiceprint acquisition are projected into the same spatial coordinate system to achieve spatial alignment, thereby constructing a temporally coherent and spatially related multimodal synchronous data matrix.

[0028] Optionally, the unified time reference origin recorded by the microcontroller when the wake-up event occurs is used as the time zero point. The timestamp of each sampling point in the touch trajectory data, the timestamp of each audio frame in the voiceprint data, and the acquisition completion timestamp of the fingerprint image data are subtracted from the known fixed response delay of each module to obtain the calibrated relative time. For sampling points that are not uniformly distributed on the time axis, a linear interpolation method is used to regenerate the data values ​​of each modality at a fixed time interval of 10 milliseconds, so that there are corresponding aligned data points in each 10-millisecond time slot for touch trajectory, voiceprint, and fingerprint data, thereby forming a time-synchronized data frame sequence.

[0029] In terms of spatial alignment, a right-handed Cartesian coordinate system is established with the center point of the power button in the 3D CAD model of the electronic atomizing device shell as the origin of the spatial coordinates. The sliding coordinates detected by the touch sensing module are mapped to the actual physical position in this coordinate system through a pre-calibrated affine transformation matrix. Similarly, the center point of the fingerprint image collected by the fingerprint sensing module is mapped to the coordinates of the center point of the power button. The sound source location estimated by the microphone sound source localization algorithm is projected onto a spatial cone-shaped area with the mouthpiece as the reference point, thereby ensuring that the data of all modes are described in the same spatial reference system.

[0030] In the spatiotemporally aligned multimodal synchronized data matrix, a fusion mechanism is introduced to decouple and reassemble features. This fusion mechanism is specifically manifested as a feature separation strategy based on attention weights, which dynamically converges time-varying feature components that change frequently with the time series and are strongly correlated with the operation process to form dynamic behavioral features. At the same time, steady-state feature components that remain relatively constant in the time dimension and represent physiological structures are statically extracted to form static biometric features. In this process, the fusion mechanism strengthens the continuity of dynamic features by calculating the temporal correlation between touch trajectory and voiceprint rhythm, and verifies the accuracy of static features by the spatial matching degree between fingerprint topology and touch area.

[0031] Optionally, a feature separation network based on a multi-head attention mechanism is used to process the spatiotemporally aligned multimodal synchronization data matrix. The velocity change sequence of the touch trajectory, the temporal features of the Mel frequency cepstral coefficients of the voiceprint, and the ridge orientation field features of the fingerprint are concatenated into a multidimensional feature tensor; the attention mechanism calculates a dynamic weight for each modality's feature component, which is determined by the variance of the feature over time and its correlation with the operation process.

[0032] Components that change drastically over time and are highly synchronized with the user's swiping and pressing rhythm are assigned higher dynamic weights and converged into a dynamic behavioral feature vector. This vector reflects the user's muscle memory and real-time operating habits during the current operation. Components that change smoothly over time and remain relatively stable at different acquisition points, such as the minutiae topology of fingerprints and the center frequency of voiceprint formants, are assigned lower dynamic weights and separated into a static biometric vector. This vector represents the user's inherent physiological attributes. During the separation process, the fusion mechanism also uses the temporal correlation between the swiping direction of the touch trajectory and the timing of voiceprint pronunciation to verify the consistency of dynamic behavioral features, and uses the degree of matching between the fingerprint center point and the touch swipe start point in spatial coordinates to verify the effectiveness of static biometric acquisition. Ultimately, this outputs two types of feature vectors that can be used for subsequent secure access to the space.

[0033] The extracted dynamic behavioral features are mapped to the representation space of user operating habits. Specifically, these dynamic behavioral features are jointly represented by the speed gradient of the touch trajectory, the frequency domain distribution of the pressing rhythm, and the dynamic fluctuation rate of the voiceprint pronunciation, reflecting the user's muscle memory and operational inertia during the holding, pressing, and vocalization process. The extracted static biometric features are mapped to the representation space of inherent physiological attributes. Specifically, these static biometric features are jointly represented by the fingerprint ridge topology and the formant frequency of the voiceprint vocal tract, reflecting the user's unchanging biological anatomy characteristics. This completes the accurate conversion of multimodal access data into dynamic and static features.

[0034] Specifically, the microcontroller of the Type A electronic cigarette acquires multimodal access data including touch trajectory, voiceprint, and fingerprint. Since the response time of the ultrasonic fingerprint module is 20ms, while the response time of the MEMS microphone and capacitive touch module is within 5ms, the system uses the time reference origin T0 of the "wake-up event: the instant the thumb presses the fire button" for time axis calibration, forcibly aligning the fingerprint data acquisition starting point to T0+20ms, and capturing touch and voiceprint data within the window from T0 to T0+500ms to achieve time alignment. At the same time, the 2cm×5cm capacitive touch coordinate set on the surface of the cigarette body and the fingerprint coordinates located at the center origin of the fire button are uniformly mapped to the three-dimensional CAD coordinate system of the Type A electronic cigarette, so that the "fingerprint pressing point" and the "touch sliding starting point" coincide in spatial geometry, forming a spatiotemporally synchronized data matrix.

[0035] The microcontroller's built-in feature fusion engine processes this spatiotemporal synchronized data matrix. When the user performs the action of "pressing the ignition button and sliding the finger down", the engine extracts the time-varying correlation weights of the "XY coordinate offset: displacement velocity" that changes rapidly over time in the capacitive touch trajectory, as well as the volume envelope and speech rate features of the "short voice: like taking a puff" collected by the microphone, through a long short-term memory network, and converges them into dynamic behavioral features.

[0036] Meanwhile, the engine extracts the steady-state weights of the dermal ridge bifurcation point (Minutiae) output by the fingerprint sensor and the formant frequency of the voiceprint that does not change with the speech rate through a spatial clustering algorithm, strips off their time-varying attributes, and converges them into static biometric features. During this process, the fusion mechanism will check the spatial offset between the "fingerprint pressing center" and the "touch area center of gravity". If the offset is extremely small, it is confirmed that the static feature is generated by the user's natural grip, thus improving anti-counterfeiting.

[0037] The extracted dynamic behavioral features characterize the specific operating habits of the e-cigarette user. For example, the user's habit is to "press the fire button hard, slide the thumb down quickly 3 cm with a slight pause, and make a short labiodental sound." This feature is jointly characterized by dynamic parameters such as [peak pressure = 2N, sliding speed = 15cm / s, voiceprint volume gradient = 5dB / ms]. The extracted static biometric features characterize the user's inherent physiological attributes. For example, the user's fingerprint feature point set and the third formant frequency of the voiceprint, approximately 2500Hz, are determined by their specific vocal tract length. This feature is jointly characterized by static parameters such as [fingerprint ridge topology vector, vocal tract formant feature matrix].

[0038] Therefore, by obtaining user verification information, which includes user registration information and historical interaction preference information, dynamic behavioral features, static biometric features and verification information are deeply concatenated and weighted along the feature dimension to construct corresponding multi-level access features. These multi-level access features include an identity confirmation layer, a behavioral intent layer and a permission matching degree, thus introducing multi-level access features.

[0039] At this point, pre-stored user verification information is retrieved from the secure storage area of ​​the electronic atomizing device or from the cloud server connected to it. This verification information specifically includes registration information representing the user's legitimate identity, as well as historical interaction preference information representing the user's past operational patterns. The registration information and historical interaction preference information are respectively encoded into feature vectors, mapping them from their original data form to a high-dimensional feature space with the same dimensions as dynamic behavioral features and static biometric features, forming registration feature vectors and preference feature vectors. This eliminates the dimensional differences between heterogeneous data and ensures the mathematical validity of subsequent feature operations.

[0040] The registration feature vector and preference feature vector generated by transforming dynamic behavioral feature vector, static biometric feature vector, and verification information are deeply concatenated along the feature channel dimension to form a full feature vector that integrates global information. The full feature vector is then input into a preset multilayer perceptron or attention-weighted network, and the weight coefficients of each sub-feature are dynamically assigned based on the current security strategy requirements of the electronic atomization device. Among them, high weights are assigned to static biometric features and registration information to ensure the rigidity of identity verification, while dynamic weights are assigned to dynamic behavioral features and preference information to capture the deviation of operational details. By performing a weighted nonlinear mapping on the full feature vector, a comprehensive feature matrix containing multi-dimensional attributes is output.

[0041] The comprehensive feature matrix is ​​deconstructed along the feature dimensions and mapped to three independent and related decision subspaces, thereby constructing a multi-level access feature with three dimensions: the weighted mapping result of static biometrics and registration information is mapped to the identity confirmation subspace, and the identity confirmation layer representing the user's legitimate identity confidence is output.

[0042] Secondly, the weighted mapping result of dynamic behavioral features and preference information is mapped to the behavioral intent subspace, and the behavioral intent layer is output, which represents the degree of consistency between the current operation and historical habits and the purpose of the operation. The output values ​​of the identity confirmation layer, the output values ​​of the behavioral intent layer and the security level parameters of the current device are jointly functioned and mapped to the permission decision subspace, and the quantified permission matching degree is output. In this way, the construction of multi-level access features from the bottom feature to the high-level access policy is completed.

[0043] Specifically, after the microcontroller of the Type A electronic cigarette completes the feature extraction of S112, it retrieves the user verification information from the device's built-in security encryption chip (SE). The registration information includes the user's initial input baseline fingerprint template vector, voiceprint feature vector, and authenticated adult age tag. The historical interaction preference information includes the user's statistical habit data over the past week, such as: the most frequently used heating power is 25 watts, the average single puff duration is 3 seconds, and the high-frequency usage period is 12:00-13:00 every day. The system performs unique thermal encoding and normalization processing on the above-mentioned adult tag, power preference, and time period preference, converting them into registration feature vectors and preference feature vectors of the same dimension as dynamic behavioral features and static biometric features.

[0044] The microcontroller performs deep splicing on the feature channel of the currently acquired "dynamic behavior features: such as sliding speed = 12cm / s and pressing time = 0.5s" and "static biometric features: such as fingerprint ridge vector". The microcontroller also extracts "registration feature vector: baseline fingerprint" and "preference feature vector: habitual 25 watts power" to form a full feature vector.

[0045] Since preventing minors from using e-cigarettes is a core concern, the attention-weighted network assigns an absolute high weight of 0.7 to static biometric features and registration features, and an auxiliary weight of 0.3 to dynamic behavioral features and preference features. After weighted mapping, if the current fingerprint highly overlaps with the baseline fingerprint and the swiping habits are consistent with the past, the comprehensive feature matrix will show a high confidence level. If the fingerprint has "partial overlap: such as blurring when wet" but the dynamic habits deviate significantly, the comprehensive feature matrix will show an abnormal distribution with local highs and local lows.

[0046] The comprehensive feature matrix is ​​deconstructed and mapped into multi-level access features: the identity verification layer outputs the label "legitimate adult owner" or "unknown user / suspected minor" based on the comparison results of static features and registration information; the behavioral intent layer determines the current intent based on the comparison results of dynamic features and preference information. For example, if the operation occurs at 23:00 at night and the pressing is extremely heavy and the sliding is extremely fast, deviating from the user's habit of light use at noon, then the "abnormal high-pressure suction intent" is output. The permission matching degree is jointly calculated based on the outputs of the first two layers. If the identity verification is "legitimate adult" and the intent is normal, then a high matching degree of 0.95 is output; if the identity verification is "unknown user" and the intent is abnormal, then a low matching degree of 0.2 is output; if the identity verification is "legitimate adult" but the intent is abnormal, such as the owner lending the device to a friend for trial use, then a medium matching degree of 0.6 is output. This quantified multi-level access feature will directly serve as the decisive parameter for determining the access path and usage level in the secure access space.

[0047] refer to Figure 3 In step S12, the specific steps are as follows: S121: Map the multi-level access features to a preset safe access space. In the safe access space, determine the user's set of interaction intentions relative to the electronic atomization device based on the vector distribution of the multi-level access features. Determine multiple access paths based on the iteration of the set of interaction intentions. Identify the access mode corresponding to each access path. The access mode is a continuous atomization mode, an intermittent atomization mode, or a parameter adjustment mode. S122: Obtain the current atomization medium of the electronic atomizing device, and determine the usage level of the electronic atomizing device by combining the basic information of the electronic atomizing device and the battery power. Input the access mode and usage level into the preset security decision tree and perform cross-matching to determine the corresponding security access events under each access path. The security access events include permission boundaries and risk factors.

[0048] In the embodiments of this application, multi-level access features are mapped to a preset secure access space. In the secure access space, the set of user interaction intentions relative to the electronic atomization device is determined based on the vector distribution of the multi-level access features. Multiple access paths are determined according to the iteration of the set of interaction intentions. The access mode corresponding to each access path is identified. The access mode is a continuous atomization mode, an intermittent atomization mode, or a parameter adjustment mode, thus introducing access modes.

[0049] At this point, the multi-level access features are used as input parameters and projected onto a pre-constructed high-dimensional secure access space on the e-vaporization device through a preset mapping function. This secure access space is a multi-dimensional topological space spanned by the identity confirmation dimension, behavioral intent dimension, and permission matching degree dimension. Multiple baseline interaction intent regions are predefined within the space. Based on the vector coordinate distribution of the multi-level access features in this space, the spatial distance or kernel function similarity between the features and the center point of each baseline interaction intent region is calculated. Several baseline interaction intents with similarity greater than a preset threshold are selected for union operation to determine the current user's set of interaction intents relative to the e-vaporization device.

[0050] Optionally, the secure access space can be constructed as follows: During the factory release or initialization phase of the electronic atomization device, multiple sets of registered access data from legitimate users are collected, and corresponding multi-level access features are extracted. The identity verification layer, behavioral intent layer, and permission matching degree value in these features are mapped to coordinate points in the space. Using these coordinate points as the core, a Gaussian mixture model is used to fit the probability density of the access behavior distribution of each legitimate user, thereby forming several continuous and statistically significant baseline interaction intent regions in the space. For example, the "regular inhalation region" corresponds to a subspace with an identity verification value of 0.9 or higher, a behavioral intent value of 0.8 or higher, and a permission matching degree value of 0.85 or higher, and the "parameter adjustment region" corresponds to a subspace with an identity verification value of 0.85 or higher, a behavioral intent value between 0.7 and 0.9, and a permission matching degree value between 0.7 and 0.85.

[0051] During real-time operation, the identity confirmation layer value, behavior intent layer value, and permission matching degree value in the multi-level access features output in step S113 are directly used as the three-dimensional coordinate point of the current user in the secure access space. The user's current set of interaction intents is determined by calculating the Euclidean distance between this coordinate point and the center point of each predefined baseline interaction intent area. Specifically, when the distance is less than the pre-set area radius threshold of 0.1, it is determined that the current user intent belongs to the corresponding baseline interaction intent area. If the distance of multiple areas is less than the threshold at the same time, they are merged to form an interaction intent set.

[0052] Within the secure access space, each element in the defined set of interactive intents is considered as a target node for path search. Combining the current hardware state and permission constraints of the electronic atomization device, a heuristic search algorithm or Markov decision process is used for iterative deduction. During the iteration process, the permission matching degree is used as the path cost weight, and the behavioral intent layer parameters are used as the path transition probability. The intent range is gradually narrowed through multi-step state transitions until it converges to a single deterministic target node. The sequence of state transition nodes experienced during the iterative deduction process is concatenated to generate multiple optional access paths from the current initial state to the final target state in the secure access space.

[0053] For each generated access path, the state transition sequence features and time span features traversed by the path in the safe access space are extracted, and these are input into a pre-trained path classifier for pattern recognition. Based on the output of the classifier, each access path is assigned to its corresponding access mode. At this point, when the path state sequence indicates that the heating component is activated for a long time and the air flow rate continuously exceeds the threshold, it is identified as a continuous atomization mode; when the path state sequence indicates that the heating component is activated and deactivated alternately and exhibits periodic pauses, it is identified as an intermittent atomization mode; when the path state sequence indicates that the heating component is in a deactivated state and the display module or adjustment module is in a high-frequency read / write state, it is identified as a parameter adjustment mode.

[0054] Specifically, the microcontroller of the Type A electronic cigarette will include a multi-level access feature vector containing "identity confirmation layer = legal adult, such as 0.9", "behavioral intent layer = high-pressure short-duration inhalation tendency, such as 0.7", and "permission matching degree = 0.85", which is mapped to a preset secure access space. Within this space, the overlap between this feature vector and the pre-stored "regular inhalation intent area", "device debugging intent area", and "low-power sleep intent area" is calculated. Since the behavioral intent layer indicates high-pressure short-duration inhalation and has a high permission matching degree, this vector falls on both the edge of the "regular inhalation intent area" and the core of the "high-pressure inhalation intent area". After union operation, the current user's interaction intent set is determined to be {regular inhalation intent, high-pressure inhalation intent}.

[0055] The system performs path iterative deduction within the secure access space for the set of interaction intentions. The "regular inhalation intention" node and the "high-pressure inhalation intention" node in the set are respectively used as candidate targets. The microcontroller performs Markov decision iteration based on the current "cartridge type: such as high nicotine concentration cartridge" as a constraint. Since the current permission matching degree is 0.85, which does not reach 1.0 absolute security, the iterative algorithm introduces the intermediate node of "power soft limit verification" as a path cost when deducing the path to "high-pressure inhalation intention", while the path to "regular inhalation intention" does not have this cost. After two rounds of iterative convergence, the system finally generates two access paths: Path 1 is "current state > unlock airflow > regular power heating > target state"; Path 2 is "current state > unlock airflow > power over-limit prompt > high-pressure limited heating > target state".

[0056] The microcontroller's built-in path classifier performs pattern recognition on the two paths mentioned above. For path 1, its state sequence features "fire button continuously pressed / airflow sensor continuously on > heating wire running at a constant power of 15W", with a long duration. The classifier identifies it as an intermittent atomization mode, consistent with the typical inhalation rhythm of an adult smoking a cigarette, with the heater intermittently deactivated between inhalations. For path 2, its state sequence includes "screen displays power limit icon > button anti-accidental touch lock > heating wire running at a limited 18W, despite user intent to apply high voltage, it is softly limited". The classifier, combined with its dynamic power adjustment logic, identifies it as a constrained continuous atomization mode, ensuring continuous output without dry burning under specific high voltage requirements. If the interaction intent set at this time is {adjust power parameters}, the generated path must pass through the "screen always on > scroll wheel adjustment action recognition" node. The classifier will directly identify it as a parameter adjustment mode. In this mode, the heating component of the type A electronic cigarette will be absolutely prohibited from activation, thus ensuring human-machine safety during the parameter setting stage.

[0057] Furthermore, the current atomization medium of the electronic atomizing device is obtained, and the usage level of the electronic atomizing device is determined by combining the basic information of the electronic atomizing device and the battery power. The access mode and usage level are input into the preset security decision tree and cross-matched to determine the corresponding security access events under each access path. The security access events include permission boundaries and risk factors, and the security access events are fully considered.

[0058] At this point, the attribute identification information of the currently loaded atomizing medium in the electronic atomizing device is read. This attribute identification information includes at least the concentration and viscosity coefficient of the medium components. Simultaneously, the basic information of the electronic atomizing device is retrieved, including the atomizer core material, atomizer core resistance, and airway structure parameters. The current battery output power and state of charge are also collected in real time. The aforementioned atomizing medium attributes, basic information, and battery power are input into a preset usage level evaluation model. Based on the thermodynamic matching relationship between the medium concentration and the atomizer core resistance, as well as the battery power's ability to support the atomization temperature, a quantitative value characterizing the current electronic atomizing device's overall output intensity and safety threshold is calculated. The corresponding usage level is then determined based on the preset range in which this value falls.

[0059] The access modes corresponding to each access path, along with the current usage level determined in this step, are used as binary input parameters and loaded into a preset security decision tree model. The branching paths from the root node to the leaf node of this security decision tree are alternately formed by the branching conditions of the access mode and the branching conditions of the usage level. The input parameters are traversed from top to bottom along the branching conditions of the decision tree. At each node split, a logical judgment is made based on the hierarchical combination relationship between the current "access mode: continuous fogging mode, intermittent fogging mode, or parameter adjustment mode" and the current usage level. The search space is gradually narrowed down to a specific leaf node, completing the cross-matching of access modes and usage levels under multi-dimensional constraints.

[0060] At each leaf node of the safety decision tree, a safety access event template corresponding to the branch path of that node is pre-set. When the cross-matching process converges to a target leaf node, the safety access event corresponding to that node is activated and exported. The safety access event is parsed to extract the permission boundary that represents the allowed operation limit under the current state. This permission boundary includes at least the maximum allowed heating power, the maximum duration of a single atomization, and the cumulative atomization port limit. At the same time, risk factors that represent potential hazards of the current combination are extracted. These risk factors include at least the probability of local overheating, battery over-discharge warning, and the risk of excessive atomization and dry burning of the medium. This completes the transformation from the underlying physical state and operational intent to the high-level safety control event.

[0061] Specifically, the Type A e-cigarette uses the identification resistor or NFC chip at the bottom of the cartridge to read the properties of the currently inserted atomizing medium as "high nicotine concentration (50mg / mL) and high viscosity vegetable glycerin (VG) base". It simultaneously retrieves the device's basic information as "porous ceramic atomizing core, resistance 0.8 ohms" and reads the maximum output power of 40W when the battery is fully charged. The microcontroller inputs the above parameters into the usage level evaluation model. Since the high nicotine concentration requires strict control of the amount of atomization per unit time, and the high viscosity VG requires higher power to be effectively atomized, but the 0.8 ohm resistance limits the maximum power, the model calculates that the thermodynamic matching value is medium intensity, and thus determines the current usage level of the e-cigarette as "limited medium power level (Level 2)".

[0062] Assuming the current user's access path corresponds to the "continuous vaping mode," if the user presses and holds the ignition button for continuous inhalation, the microcontroller inputs the "continuous vaping mode" and "restricted medium power level (Level 2)" into the safety decision tree. The decision tree determines the access mode as "continuous vaping" at the root node and enters the continuous vaping branch. At the second-level node, it determines the usage level as "Level 2." At this point, because the combination of "continuous vaping" and "high nicotine concentration" poses an extremely high risk of overdose, the decision tree's logic guides it to the most stringent safety control branch, ultimately converging to a specific leaf node labeled "continuous-restricted combination," completing the cross-matching.

[0063] The system extracts the pre-set security access event under the leaf node of the "continuous-restricted combination". This event clearly defines the permission boundaries: the maximum allowable heating power is forcibly limited to 15W, even though the battery supports 40W, to prevent the high viscosity medium from generating harmful aerosols due to instantaneous high temperature. The maximum duration of a single continuous atomization is strictly limited to 3 seconds to prevent excessive intake of nicotine under high concentrations. At the same time, this event clearly marks the risk factors in the current state: since the oil delivery speed of the high viscosity medium may not keep up with the consumption rate in continuous atomization mode, the risk factors are marked as "extremely high risk of burnt coil" and "risk of excessive nicotine intake". The generation of this security access event provides a precise permission baseline and risk context for marking abnormal content and loading the knowledge graph, ensuring that Type A e-cigarettes are in a controlled and safe state under any medium and operation combination.

[0064] refer to Figure 4 In step S13, the specific steps are as follows: S131: Dynamically identify the security access event, and in the process of identification, combine the anomaly detection mechanism to determine the abnormal access content that deviates from the historical behavior baseline, obtain the usage scenario of the electronic atomization device and the past access history, and combine the abnormal access content to generate security detection content for in-depth verification. S132: Collect the usage status of the electronic atomization device, which includes the current battery level and atomization chamber temperature. Use the safety detection content, the current battery level, and the atomization chamber temperature as corresponding nodes, and load them into the preset knowledge graph to trigger the corresponding graph reasoning. This will allow for association propagation and consistency verification along the topology of the knowledge graph, and then output the corresponding authorization sequence. This authorization sequence includes timing verification logic and condition triggering logic.

[0065] In the embodiments of this application, the security access event is dynamically identified, and during the identification process, an anomaly detection mechanism is combined to determine the abnormal access content that deviates from the historical behavior baseline, the usage scenario of the electronic atomization device and the previous access history are obtained, and security detection content for in-depth verification is generated in combination with the abnormal access content, thus introducing security detection content.

[0066] At this point, the security access event is dynamically identified in real time, and the real-time behavioral feature parameters and permission boundary constraint parameters contained in the event are extracted. At the same time, the historical behavior baseline, which is constructed by the statistics of the user's historical legitimate operations and is pre-stored in the security storage area of ​​the electronic atomization device, is called. The real-time behavioral feature parameters and the historical behavior baseline are input into the anomaly detection mechanism. The anomaly detection mechanism calculates the offset of the real-time parameters relative to the baseline distribution based on the sliding window algorithm. When the offset exceeds the preset adaptive dynamic threshold, it is determined that the current operation has an abnormal deviation, and the excess feature subset that caused the offset is extracted and marked as abnormal access content.

[0067] By integrating positioning, timing, and environmental sensors into the electronic atomization device, the current usage scenario of the device is obtained. This usage scenario includes at least time dimension tags, spatial geographic location tags, and environmental temperature and humidity parameters. Simultaneously, past access history is retrieved from a local log database or the cloud. This past access history includes historical anomaly trigger frequency, historical average atomization consumption, and historical operational preferences under specific scenarios. The abnormal access content is aligned with the obtained usage scenario and past access history on the time axis and spatial coordinates to construct a multi-dimensional context association matrix, thereby establishing the background conditions for the occurrence of abnormal deviations.

[0068] Based on a multidimensional contextual association matrix, risk tracing and scenario verification analysis are performed on abnormal access content. If past access history indicates that the abnormal deviation is an occasional event and the current usage scenario is in a high-risk area, or if the abnormal deviation characteristics highly match the operation fingerprint characteristics of a specific minor, then according to the attribute category of the abnormal access content, the corresponding proactive verification rule is matched from the preset detection strategy library. Based on the proactive verification rule, security detection content requiring users to provide additional biometric proof of liveness or physical interaction confirmation is generated. This security detection content specifically includes the detection modality type, challenge code logic, and verification pass threshold to achieve mandatory deep verification of abnormal access.

[0069] Specifically, the Type A electronic cigarette microcontroller dynamically identifies currently generated security access events, such as: permission boundaries with a maximum single duration of 3 seconds, and risk factors marked as "high risk of burnt coil and dry burning"; the system extracts the current user's real-time vaping behavior parameters and finds that the user has triggered the "intermittent atomization mode" 10 times in the past 60 seconds, with a single puff interval of less than 2 seconds; it calls the user's historical behavior baseline, which shows that the user is accustomed to slow-paced deep vaping, averaging only 5 vapings per hour. The anomaly detection mechanism, based on a sliding window calculation, finds that the current vaping frequency deviation has reached 4 times the baseline standard deviation, far exceeding the adaptive dynamic threshold of 2 times the standard deviation; based on this, the system determines that a serious abnormal deviation has occurred and marks the "high-frequency short vaping behavior" as abnormal access content.

[0070] The Type A electronic cigarette uses a built-in Wi-Fi probe and RTC clock to determine the current usage scenario as "weekday 14:00, outside of the high-risk period after school - located within 100 meters of a certain high school, a sensitive geofence". At the same time, it retrieves past access history records and finds that the device has triggered the anti-minor locking mechanism twice within the same time period in the past week. The system spatiotemporally correlates the abnormal access content of "high-frequency short puffs" with "sensitive scene around the high school" and "multiple historical records of minor interception", constructing a multi-dimensional context association matrix that is highly suspected of being used by a minor.

[0071] Based on this multi-dimensional contextual association matrix, the microcontroller performs risk tracing and determines that the current "high-frequency short vaping" is highly likely not operated by the device owner, but by a minor secretly vaping during a break between classes. Based on this analysis, the system matches the highest level of active verification rules and generates security detection content for in-depth verification. The specific requirement of this security detection content is "dynamic digital challenge code verification based on voiceprint," which requires the current operator to read aloud a randomly displayed 4-digit number on the screen within 5 seconds to verify the operator's "voiceprint physiological characteristics: whether it is an adult male device owner" and "liveness response capability: to prevent recording attacks." If this security detection content fails, subsequent steps will directly block the heating power supply of the atomizer, thereby achieving in-depth security interception of e-cigarettes in complex stolen scenarios.

[0072] Furthermore, the usage status of the electronic atomization device is collected, which includes the current battery level and atomization chamber temperature. The safety detection content, the current battery level, and the atomization chamber temperature are used as corresponding nodes and loaded into a preset knowledge graph to trigger corresponding graph reasoning. This allows for association propagation and consistency verification along the topology of the knowledge graph, ultimately outputting the corresponding authorization sequence. This authorization sequence includes timing verification logic and condition triggering logic, thus introducing the authorization sequence.

[0073] At this point, the current state of charge of the battery is collected by the built-in power meter of the electronic atomizing device to obtain the current power level, and the real-time physical temperature of the atomizing chamber is collected by the negative temperature coefficient thermistor or infrared temperature sensor integrated on the wall of the atomizing chamber to obtain the atomizing chamber temperature. The current power level and the atomizing chamber temperature are jointly represented as the usage status of the electronic atomizing device. The usage status and safety detection content are used as heterogeneous data sources, and are respectively materialized and attribute-assigned according to the preset ontology model to construct corresponding instance nodes, namely safety detection content nodes, power status nodes, and temperature status nodes. The instance nodes are then loaded and injected into the knowledge graph pre-stored in the local storage medium as the initial factual basis for graph reasoning.

[0074] Optionally, the knowledge graph is a directed graph structure that is pre-built and stored in the local secure storage area of ​​the electronic atomization device. Its nodes represent entities related to secure access, including but not limited to security detection content types, battery status ranges, temperature status ranges, user verification results, hardware execution actions, etc.; the directed edges between nodes represent the logical relationships between entities, specifically including four types: "triggering relationship", "mutual exclusion relationship", "dependency relationship" and "constraint relationship".

[0075] The knowledge graph is constructed as follows: During the design phase of the electronic atomization device, based on security access logic and thermodynamic physical laws, each entity node and its relational edges are manually defined. For example, the "Voiceprint verification passed" node is defined to point to the "Unlock heating permission" node through a "trigger relationship"; the "Atomization chamber temperature too high" node is defined to point to the "Immediately start heating" node through a "mutual exclusion relationship"; and the "Battery level below 10%" node is defined to point to the "Limit maximum output power to 15 watts" node through a "constraint relationship". All nodes and relational edges are stored in the form of resource description framework triples, forming a queryable and reasonable local knowledge graph.

[0076] In a knowledge graph, based on predefined relationships between entities, such as causal relationships, constraint relationships, and mutual exclusion relationships, the graph reasoning engine is triggered starting from the injected instance node. This reasoning engine performs association propagation along the topology of the knowledge graph, passing the attribute features of the initial node along the relationship edges to adjacent implicit nodes, thus activating potential rule paths.

[0077] During the propagation process, logical constraints are introduced for consistency verification. Specifically, a comparison is made between the expected action logic derived from the safety detection content and the physical limit logic allowed by the current power level and atomization chamber temperature. If a conflict exists, the propagation path is cut off and a suboptimal solution is sought by backtracking. If they are consistent, the current propagation path is confirmed as a valid inference link, thereby obtaining the set of actions that the device can perform under the current dual constraints of safety and physical conditions.

[0078] Based on the action set determined by the effective inference chain, the execution dependencies and sequence of each atomic action in the set are extracted. The atomic actions are arranged in the order of their timestamps to construct a sequence framework containing timing verification logic. This timing verification logic stipulates that the execution of a subsequent action can only be triggered if the verification of the preceding action is passed. At the same time, the power and temperature thresholds in the usage state are encapsulated into the sequence framework as pre-judgment criteria to construct condition triggering logic. This condition triggering logic stipulates that the execution of a specific action must meet specific physical state conditions. The timing verification logic and the condition triggering logic are combined and encoded to finally output a structured authorization sequence.

[0079] Specifically, the microcontroller of the Type A e-cigarette reads the current battery level as 15% using a coulomb counter, indicating a low battery state. It also reads the current atomizer temperature as 55°C using an NTC thermistor at the bottom of the atomizer chamber, indicating that the residual heat from recent continuous vaping has not dissipated. These two factors constitute the current usage state. Assuming the safety detection requirement is "requiring voiceprint dynamic digital challenge code verification," the system instantiates "voiceprint challenge verification," "15% battery level," and "55°C temperature" as three entity nodes and loads them into the local knowledge graph of the Type A e-cigarette.

[0080] The graph inference engine is triggered, propagating from the "Voiceprint Challenge Verification" node along the "Verification Passed" relation to the "Unlock Heating Permission" node; simultaneously, it propagates from the "Battery 15%" node along the "Support Capability" relation to the "Limit Maximum Output Power to 15W" node, and from the "Temperature 55℃" node along the "Thermal Safety Constraint" relation to the "Forced Heat Dissipation Waiting" node; during consistency verification, the system finds a logical conflict between "Unlock Heating Permission" and "Forced Heat Dissipation Waiting," meaning that the current temperature is too high and immediate heating is not allowed. Based on this, the inference engine cuts off the propagation path of direct heating, backtracks, and generates a new valid inference link: that is, it is necessary to first go through heat dissipation waiting, so that the temperature drops to a safe threshold, before heating can be unlocked at low power.

[0081] The microcontroller generates the final authorization sequence based on the above effective inference link. This authorization sequence strictly encapsulates the timing verification logic: Step 1 is "playing voice prompts and turning on the microphone to collect voiceprint challenge codes", Step 2 is "verifying voiceprints and digital content", Step 3 is "screen countdown for 30 seconds to allow for heat dissipation", and Step 4 is "starting atomization at 15W low power". At the same time, it encapsulates the condition triggering logic: the progression from Step 3 to Step 4 must meet the conditions [NTC detection temperature ≤ 40℃] and [battery level ≥ 10%]. If the countdown ends but the temperature is still above 40℃, Step 4 will not be triggered and the waiting time will be extended. Through this dual logic constraint of timing and conditions, the Type A electronic cigarette accurately achieves high-security access control in risk scenarios of low battery and high temperature residual heat, first forcing identity verification, then forcing physical cooling, and finally limiting power output.

[0082] refer to Figure 5 In step S14, the specific steps are as follows: S141: The safety control unit of the electronic atomizing device is marked. The safety control unit receives the authorization sequence and executes the corresponding feedback verification action, prompt verification action or preheating verification action in sequence according to the timing verification logic and condition triggering logic of the authorization sequence, thereby outputting the response result of each verification action. S142: Obtain the operating constraints of the electronic atomizing device, which include the current thermodynamic safety threshold and anti-dry burning content of the electronic atomizing device, and compare them with the response results of each verification action; if the response results meet the operating constraints, dynamically unlock the accessible content of the electronic atomizing device in the corresponding access mode, which includes the corresponding atomization program, atomization power limit, and atomization time.

[0083] In the embodiments of this application, a safety control unit of the electronic atomizing device is marked. The safety control unit receives an authorization sequence and executes the corresponding feedback verification action, prompt verification action, or preheating verification action in sequence according to the timing verification logic and condition triggering logic of the authorization sequence, thereby outputting the response result of each verification action, thus introducing the response result.

[0084] At this point, in the microcontroller underlying architecture of the electronic atomization device, a safety control unit that runs independently of the main control program is marked and activated. This safety control unit has the highest priority control over the hardware actuators and physically isolated storage space. The safety control unit receives the authorization sequence through the internal high-speed bus and parses the message structure of the authorization sequence. It then extracts the timestamp sorting rules of the timing verification logic, the state threshold criteria of the condition triggering logic, and the verification action instruction type corresponding to each execution step, forming an atomic execution task queue that can be scheduled sequentially.

[0085] The safety control unit schedules corresponding peripheral resources to perform verification actions in sequence according to the atomic execution task queue. During execution, it strictly follows the timing verification logic, that is, if the previous verification action does not output a response result or the response time does not meet the timestamp constraint, the scheduling of subsequent actions is frozen. At the same time, it strictly follows the condition triggering logic, that is, before triggering a specific action, it compares in real time whether the physical parameters collected by the underlying sensors meet the preset state threshold criteria. The verification actions specifically include: feedback verification actions that require the user to perform secondary biometric input or physical interaction confirmation, prompt verification actions that output warning information and operation guidance through visual or auditory modules, and preheating verification actions that control the heating components to preheat at a low power below the atomization threshold.

[0086] During the execution of feedback verification actions, prompt verification actions, or preheating verification actions, the safety control unit synchronously monitors the feedback channel or sensor bus of the corresponding peripheral device, collects the changes in physical quantities or user input data generated by the action execution, and generates the corresponding raw response signal. The raw response signal is compared with the pass criteria preset for the corresponding verification action to determine whether the execution status of the verification action is successful, failed, or timed out, and is quantified into a standard format response result for output. This response result serves as the sole criterion for triggering the next timing action or terminating the authorization sequence.

[0087] After the Type A electronic cigarette is powered on, the operating system kernel marks and wakes up an independent security control unit, such as the TrustZone security domain with an embedded hardware encryption engine. This unit takes over direct control of the airflow sensor, microphone, MOSFET switch, and OLED display. Upon receiving the authorization sequence generated in step S132, the security control unit parses the sequence to find that it contains three atomic tasks arranged in sequence: the first task: feedback verification action, requiring the input of a specific sliding trajectory within 3 seconds; the second task: preheating verification action, under the condition that the temperature is ≤40℃ and heated with 5W micropower for 2 seconds; and the third task: prompt verification action, displaying "Verification passed and can be inhaled". These three tasks are then pushed into the security scheduling queue.

[0088] The safety control unit begins to execute the following steps in sequence: performing the feedback verification action of the first task, the safety control unit sends an instruction to the touch IC to monitor screen swipe events, and according to the timing verification logic, if no valid touch interrupt signal is received within 3 seconds, the queue is directly locked and subsequent tasks are not executed; if the touch trajectory matches, a success response is output and the queue is advanced to the second task.

[0089] During the preheating verification action of the second task, the safety control unit reads the NTC temperature value and, according to the condition triggering logic, only outputs a PWM signal with an extremely low duty cycle to the MOSFET when the temperature is ≤40℃, so that the heating wire works at a low power of 5W for 2 seconds to preheat the atomizing chamber to avoid condensate splashing and improve the first puff taste of the e-cigarette; after the preheating is completed, the queue advances to the prompt verification action of the third task, and the safety control unit writes a specific character library to the OLED display memory and lights up the screen.

[0090] In response to the above actions, the safety control unit performs a rigorous judgment on the response result: For the first task, the touch IC returns the sliding coordinate sequence, the safety control unit calculates its spatial trajectory vector, compares it with the preset vector, and outputs the response result of "feedback verification successful".

[0091] For the second task, during the preheating process, the safety control unit monitors whether the MOSFET drain current reaches the rated value and whether the NTC temperature rise is within a reasonable range after 2 seconds, such as 2℃-5℃. If the current is zero, a "preheating fault open circuit" response is output; if the temperature rise is abnormally high, a "preheating dry burning warning" response is output. For the third task, the screen driver IC returns the write success register value and outputs a "completion prompt" response. All the above response results are packaged and encrypted. Only when the response results of the first and second tasks are both "successful" can the e-cigarette finally unlock the high-power atomization permission, ensuring that the Type A e-cigarette can only enter the vaping state under the premise that its identity is legal, its hardware is healthy, and its physical state is safe.

[0092] Furthermore, the operating constraints of the electronic atomizing device are obtained, which include the current thermodynamic safety threshold and anti-dry-burning measures of the electronic atomizing device, and compared with the response results of each verification action. If the response results meet the operating constraints, the accessible content of the electronic atomizing device in the corresponding access mode is dynamically unlocked. This accessible content includes the corresponding atomization program, atomization power limit, and atomization time, fully considering the accessible content.

[0093] At this point, the current operating constraints of the electronic atomizing device are obtained from the safe storage area and real-time sensing matrix. These operating constraints are hard boundaries set based on the device's physical limits and safety specifications. At this point, the thermodynamic safety threshold covering the highest tolerable temperature of the atomizing chamber and the heating element is extracted, along with the anti-dry-burning content calculated based on the current liquid guiding rate and liquid level of the oil guiding medium. This anti-dry-burning content includes at least the maximum continuous heating duration and the forced cooling intermittent duration. The thermodynamic safety threshold and the anti-dry-burning content are numerically deconstructed to form a set of safety upper limit parameters for subsequent logical judgments.

[0094] The response results output by each verification action are compared one by one with the set of safety upper limit parameters. The comparison process is as follows: the legitimacy of the identity and intent represented by the response results of the feedback verification action is taken as a prerequisite; the real-time temperature rise data and current feedback data extracted from the preheating verification action response results are compared with the thermodynamic safety threshold; and the calculated medium consumption rate is matched and verified with the limit parameters in the anti-dry burning content. If any response result exceeds the thermodynamic safety threshold or triggers the alarm logic of the anti-dry burning content, the comparison is deemed to have failed and a safety blocking command is generated. If all response results are within the allowable range of the set of safety upper limit parameters, the response results are deemed to meet the working constraints.

[0095] Once the response result is determined to meet the working constraints, the safety control unit dynamically generates and issues a hardware unlock command based on the permission baseline corresponding to the current "access mode: continuous atomization mode, intermittent atomization mode, or parameter adjustment mode" and in conjunction with the safety margin that has been verified. This unlock command switches the actuator of the electronic atomization device from the locked state to the controlled operation state, thereby opening up the corresponding accessible content. The accessible content specifically includes: the "atomization program" matching the current medium and mode: including the preheating curve and power ramp-up slope; the upper limit of atomization power calculated based on the thermodynamic safety margin and battery status; and the single or cumulative atomization duration strictly limited based on the anti-dry burning content, thereby achieving the maximum usable permission output under physical safety constraints.

[0096] Specifically, the safety control unit of the Type A electronic cigarette retrieves the current operating constraints from the built-in temperature control algorithm module. Since the currently loaded cartridge is a "low-resistance cartridge with a high VG: vegetable glycerin" content of 0.4 ohms, its wicking speed is relatively slow. Based on this, the system decomposes the following: the thermodynamic safety threshold is "the highest instantaneous temperature of the heating element must not exceed 280°C", and the anti-dry burning content is "the maximum continuous heating time must not exceed 8 seconds, and there must be a cooling interval of no less than 15 seconds between two puffs to allow for wicking". The above values ​​constitute the set of safety upper limit parameters for the current operation of the Type A electronic cigarette.

[0097] Assuming the "feedback verification action: such as sliding trajectory verification" response is valid, while the "preheating verification action: preheating with 5W low power for 2 seconds" response shows that the temperature rise rate during the preheating stage reaches 3℃ / millisecond, far exceeding the normal 1℃ / millisecond in a humid state, and the real-time temperature is approaching 250℃; the safety control unit compares this temperature rise rate and the extreme value of 250℃ with the "thermodynamic safety threshold: 280℃" and the anti-dry-burning logic, and determines that the current temperature rise is too fast and close to the limit, inferring that the oil-wicking cotton may be in a semi-dry-burning state, which triggers the warning logic of the anti-dry-burning content, and the comparison result is failure; if the preheating response result is "stable temperature rise: 1℃ / millisecond" and the temperature only reaches 150℃, then the comparison passes.

[0098] The system dynamically unlocks content: if the comparison fails, a semi-dry burning warning is issued. Even if identity verification is successful, the security control unit only dynamically unlocks extremely restricted access: the atomization program is forcibly switched to "micro-power pulse mode," the atomization power limit is limited to 10W, far below the user-set 40W, and the atomization time is limited to 2 seconds per cycle, sacrificing vapor production for the safety of preventing coil burn. Conversely, if the comparison passes completely and there is sufficient reserve, and the current access mode is "intermittent atomization mode," the security control unit dynamically unlocks complete access: the atomization program matches the user's "smooth boost curve," the atomization power limit is opened to 35W allowed by the cartridge, and the atomization time is opened to the anti-dry burning limit of 8 seconds. Through this dynamic unlocking mechanism that deeply binds the verification results with thermodynamics and anti-dry burning physical constraints, Type A e-cigarettes ensure legal access for users while eliminating the risk of high-temperature harmful substance release and device damage from the underlying hardware driver level.

[0099] Please see Figure 6 The secure access system for the electronic atomizing device is applied to the aforementioned secure access method for the electronic atomizing device; the secure access system for the electronic atomizing device includes: Access data module 21 is used to respond to the wake-up event of the electronic atomization device, acquire the user's multimodal access data, determine the corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of multimodal access data, and determine multi-level access characteristics in combination with the user's verification information. The secure access module 22 is used to input the multi-level access characteristics into a preset secure access space, and to determine multiple access paths of the user relative to the electronic atomizing device in the secure access space, and to determine the corresponding secure access events according to the access mode corresponding to each access path and the usage level of the electronic atomizing device. The authorization sequence module 23 is used to mark the corresponding abnormal access content in the security access event, and determine the corresponding security detection content in combination with the usage scenario of the electronic atomization device and the previous access history. The security detection content and the usage status of the electronic atomization device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence. The accessible module 24 is used for the electronic atomizing device to receive the authorization sequence and execute the corresponding verification actions in sequence to output the response results of each verification action, and dynamically configure the accessible content of the electronic atomizing device in combination with the working constraints of the electronic atomizing device.

[0100] It should be noted that although multiple modules are mentioned in the detailed description above, this division is not mandatory; in fact, according to the embodiments of this disclosure, the features and functions of two or more modules or described above can be embodied in one module; conversely, the features and functions of one module described above can be further divided into multiple modules to be embodied.

[0101] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein; this application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein; the specification and embodiments are to be considered exemplary only.

[0102] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A secure access method for an electronic atomizing device, characterized in that, include: In response to the wake-up event of the electronic atomization device, the system acquires the user's multimodal access data, identifies the corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of the multimodal access data, and determines multi-level access characteristics in combination with the user's verification information. The multi-level access characteristics are input into a preset secure access space, and multiple access paths of the user relative to the electronic atomizing device are determined in the secure access space. The corresponding secure access events are determined according to the access mode corresponding to each access path and the usage level of the electronic atomizing device. In this security access event, the corresponding abnormal access content is marked, and the corresponding security detection content is determined by combining the usage scenario of the e-vaporization device and the previous access history. The security detection content and the usage status of the e-vaporization device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence. The electronic atomizing device receives the authorization sequence and executes the corresponding verification actions in sequence to output the response results of each verification action. It also dynamically configures the accessible content of the electronic atomizing device in combination with the device's operating constraints.

2. The secure access method for the electronic atomizing device according to claim 1, characterized in that, In response to the wake-up event of the electronic atomization device, the system acquires the user's multimodal access data, determines the corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of the multimodal access data, and combines the user's verification information to determine multi-level access characteristics, including: The user performs a wake-up operation on the e-cigarette device and responds to the wake-up event of the e-cigarette device. The e-cigarette device integrates a multimodal sensing array, which collects the user's touch trajectory, voiceprint and fingerprint, thereby forming multimodal access data. Multimodal access data is spatiotemporally aligned, and a fusion mechanism is used in the alignment process to determine the corresponding dynamic behavioral features and static biometric features. The dynamic behavioral features reflect the user's operating habits, while the static biometric features reflect the user's inherent physiological attributes.

3. The secure access method for the electronic atomizing device according to claim 2, characterized in that, The process of responding to a wake-up event from an electronic atomization device, acquiring the user's multimodal access data, determining corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of the multimodal access data, and determining multi-level access characteristics in conjunction with the user's verification information, further includes: The system obtains user verification information, which includes user registration information and historical interaction preference information. Dynamic behavioral features, static biometric features, and verification information are deeply concatenated and weighted along the feature dimension to construct corresponding multi-level access features. These multi-level access features include an identity confirmation layer, a behavioral intent layer, and a permission matching degree.

4. The secure access method for the electronic atomizing device according to claim 1, characterized in that, The process involves inputting the multi-level access characteristics into a preset secure access space, determining multiple access paths for the user relative to the e-vaporization device within this secure access space, and determining corresponding secure access events based on the access mode corresponding to each access path and the usage level of the e-vaporization device, including: Multi-level access features are mapped to a preset secure access space. In the secure access space, the user's set of interaction intentions relative to the electronic atomization device is determined based on the vector distribution of the multi-level access features. Multiple access paths are determined based on the iteration of the set of interaction intentions. The access mode corresponding to each access path is identified. The access mode is either continuous atomization mode, intermittent atomization mode, or parameter adjustment mode.

5. The secure access method for the electronic atomizing device according to claim 4, characterized in that, The step of inputting the multi-level access characteristics into a preset secure access space, determining multiple access paths of the user relative to the electronic atomizing device within the secure access space, and determining corresponding secure access events based on the access mode corresponding to each access path and the usage level of the electronic atomizing device, further includes: The system obtains the current atomization medium of the electronic atomizing device and determines the usage level of the device by combining the device's basic information and battery power. The access mode and usage level are then input into a preset security decision tree and cross-matched to determine the corresponding security access events under each access path. These security access events include permission boundaries and risk factors.

6. The secure access method for the electronic atomizing device according to claim 1, characterized in that, In this secure access event, the corresponding abnormal access content is marked, and the corresponding security detection content is determined by combining the usage scenario of the e-cigarette device and the previous access history. The security detection content and the usage status of the e-cigarette device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence, including: The security access event is dynamically identified, and during the identification process, an anomaly detection mechanism is used to determine the abnormal access content that deviates from the historical behavior baseline. The usage scenarios of the electronic atomization device and the past access history are obtained, and security detection content for in-depth verification is generated by combining the abnormal access content.

7. The secure access method for the electronic atomizing device according to claim 6, characterized in that, In this secure access event, the corresponding abnormal access content is marked, and the corresponding security detection content is determined by combining the usage scenario of the e-cigarette device and the previous access history. The security detection content and the usage status of the e-cigarette device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence, which also includes: The usage status of the electronic atomizing device is collected, which includes the current battery level and atomization chamber temperature. The safety detection content, the current battery level, and the atomization chamber temperature are used as corresponding nodes and further loaded into a preset knowledge graph to trigger corresponding graph reasoning. This allows for association propagation and consistency verification along the topology of the knowledge graph, ultimately outputting a corresponding authorization sequence. This authorization sequence includes timing verification logic and condition triggering logic.

8. The secure access method for the electronic atomizing device according to claim 1, characterized in that, The electronic atomizing device receives the authorization sequence and sequentially executes the corresponding verification actions to output the response results of each verification action. It also dynamically configures the accessible content of the electronic atomizing device based on its operational constraints, including: The safety control unit of the electronic atomizing device is marked. The safety control unit receives the authorization sequence and executes the corresponding feedback verification action, prompt verification action or preheating verification action in sequence according to the timing verification logic and condition triggering logic of the authorization sequence, thereby outputting the response result of each verification action.

9. The secure access method for the electronic atomizing device according to claim 8, characterized in that, The electronic atomizing device receives the authorization sequence and sequentially executes the corresponding verification actions to output the response results of each verification action. It also dynamically configures the accessible content of the electronic atomizing device based on its operational constraints. The device further includes: The operating constraints of the electronic atomizing device are obtained, which include the current thermodynamic safety threshold and anti-dry-burning measures of the electronic atomizing device, and compared with the response results of each verification action. If the response results meet the operating constraints, the accessible content of the electronic atomizing device in the corresponding access mode is dynamically unlocked. The accessible content includes the corresponding atomization program, atomization power limit, and atomization time.

10. A secure access system for an electronic atomizing device, characterized in that, The secure access system for the electronic atomizing device is applied to the secure access method for the electronic atomizing device as described in any one of claims 1-9; the secure access system for the electronic atomizing device includes: The access data module is used to respond to the wake-up event of the electronic atomization device, acquire the user's multimodal access data, determine the corresponding dynamic behavioral characteristics and static biometric characteristics based on the recognition of the multimodal access data, and determine multi-level access characteristics in combination with the user's verification information. The secure access module is used to input the multi-level access characteristics into the preset secure access space, and determine multiple access paths of the user relative to the electronic atomizing device in the secure access space. Based on the access mode corresponding to each access path and the usage level of the electronic atomizing device, the corresponding secure access event is determined. The authorization sequence module is used to mark the corresponding abnormal access content in the security access event, and determine the corresponding security detection content by combining the usage scenario of the electronic atomization device and the previous access history. The security detection content and the usage status of the electronic atomization device are loaded into the corresponding knowledge graph to output the corresponding authorization sequence. The accessibility module is used by the electronic atomizing device to receive the authorization sequence and execute the corresponding verification actions in sequence to output the response results of each verification action, and dynamically configure the accessible content of the electronic atomizing device in combination with the working constraints of the electronic atomizing device.

Images