An electronic smoking set cartridge recognition method, system, device and electronic smoking set

Abstract

The application discloses an identification method, system and device of an electronic smoking set cartridge and the electronic smoking set. The identification method comprises the following steps: extracting the features of the original electrical signal generated in the heating process of the to-be-identified cartridge to obtain the original electrical features of the to-be-identified cartridge; performing normalization processing on the original electrical features based on the preset mean value and the preset variance and the dynamic mean value and the dynamic variance of the dynamic buffer sample to obtain the exclusive electromagnetic fingerprint of the to-be-identified cartridge; and inputting the exclusive electromagnetic fingerprint into the pre-trained cartridge identification model to obtain the cartridge type of the to-be-identified cartridge and / or whether the to-be-identified cartridge is in the best taste. The efficiency of cartridge identification is improved.

Description

Technical Field

[0001] This application belongs to the field of electronic cigarette atomization technology, and specifically relates to a method, system, device and electronic cigarette device for identifying e-cigarette cartridges. Background Technology

[0002] In the field of e-cigarette atomization technology, accurate identification of cartridge types is crucial. It can not only adapt the best heating curve to different cartridges to ensure a stable taste, but also distinguish the model, specifications, flavor attributes and manufacturers of cartridges, preventing the use of inferior or expired cartridges.

[0003] Current e-cigarette cartridge identification mainly relies on physical structure modification, external equipment assistance, or electrical signal detection. Some solutions use different combinations of conductive components on the cartridge to short-circuit the e-cigarette rod in series with a detection resistor, generating differentiated voltage values ​​that map to the cartridge type. Others integrate a memory chip into the cartridge to store information such as anti-counterfeiting measures and expiration dates, which are then read through dedicated contacts. Both of these solutions depend on physical contacts, are prone to oxidation and failure, and increase the cost of the cartridge.

[0004] Therefore, there is an urgent need for a new technical solution to identify e-cigarette cartridges. Summary of the Invention

[0005] To address the aforementioned issues, this application provides a method, system, device, storage medium, and electronic cigarette device for identifying e-cigarette cartridges.

[0006] Firstly, this application provides a method for identifying e-cigarette cartridges, the method comprising: The original electrical signal characteristics of the tobacco cartridge to be identified are obtained by extracting the features of the original electrical signal generated during the heating process. Based on the preset mean and preset variance, as well as the dynamic mean and dynamic variance of the dynamic buffer samples, the original electrical features are normalized to obtain the unique electromagnetic fingerprint of the e-cigarette cartridge to be identified. The unique electromagnetic fingerprint is input into the pre-trained cartridge recognition model to obtain the cartridge type and / or whether the cartridge is in the optimal flavor.

[0007] Furthermore, the extraction of features of the original electrical signals generated by the cigarette cartridge to be identified during the heating process includes: The original electrical signal is intercepted by a preset anti-leakage window to obtain multiple segments of electrical signal; Feature extraction is performed on multiple segments of electrical signals to obtain the original electrical features.

[0008] Furthermore, the original electrical characteristics include one or more of the following: Steady-state energy distribution characteristics; Morphological drift and bandwidth characteristics; Nonlinear distortion and harmonic characteristics; Temporal dynamic evolution characteristics.

[0009] Furthermore, the original electrical characteristics are normalized using the following formula:

[0010] in, Indicates the original electrical characteristics, This indicates a proprietary electromagnetic fingerprint. Indicates the mixed weighting coefficient. , This indicates the preset mean and preset variance. , This represents the dynamic mean and dynamic variance.

[0011] Furthermore, the identification method also includes: Determine whether the confidence level for the type of tobacco cartridge to be identified and / or whether it is in the optimal taste range is greater than the safety threshold. If the confidence level of the type of tobacco cartridge to be identified and / or whether it is in the best taste is greater than the safety threshold, the tobacco cartridge to be identified is stored in the dynamic buffer sample, and the dynamic mean and dynamic variance of the dynamic buffer sample are updated.

[0012] Furthermore, the cigarette cartridge recognition model is trained through the following steps: Multiple raw electrical signals generated during the heating process of each type of e-cigarette cartridge are obtained. Noise is injected into each raw electrical signal to obtain multiple noise-enhanced electrical signals, resulting in a set of noise-enhanced electrical signals for multiple e-cigarette cartridge types. For each noise-enhanced electrical signal in the noise-enhanced electrical signal set, label the cartridge type and / or whether it is in the optimal taste, obtain several training samples; The cartridge recognition model is trained using several training samples to determine the optimal model parameters.

[0013] Secondly, this application provides an identification system for e-cigarette cartridges, the identification system comprising: The feature extraction module is used to extract the features of the original electrical signals generated by the tobacco cartridge to be identified during the heating process, so as to obtain the original electrical features of the tobacco cartridge to be identified. The normalization module is used to normalize the original electrical features based on the preset mean and preset variance, as well as the dynamic mean and dynamic variance of the dynamic buffer samples, to obtain the unique electromagnetic fingerprint of the e-cigarette cartridge to be identified. The identification module is used to input a unique electromagnetic fingerprint into a pre-trained cartridge identification model to obtain the cartridge type and / or whether the cartridge is in optimal flavor.

[0014] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned identification method.

[0015] Fourthly, this application provides a computer storage medium storing a computer program, which, when executed by a processor, implements the aforementioned identification method.

[0016] Fifthly, this application provides an electronic cigarette device, which includes a power supply, an inductor coil, an electrical signal acquisition device, and the aforementioned identification system.

[0017] This application discloses a method, system, device, storage medium, and electronic cigarette device for identifying e-cigarette cartridges. By acquiring the original electrical characteristics of the cartridge to be identified, and normalizing the original electrical characteristics based on preset mean and preset variance, as well as the dynamic mean and dynamic variance of dynamic buffer samples, a unique electromagnetic fingerprint of the cartridge to be identified is obtained. The unique electromagnetic fingerprint is then identified using a pre-trained cartridge identification model. This method can quickly and accurately determine the cartridge type and / or whether the cartridge is in optimal flavor, and significantly reduces the identification cost.

[0018] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating a method for identifying an electronic cigarette cartridge according to an embodiment of this application is shown. Figure 2 A schematic diagram of the structure of an electronic cigarette cartridge identification system according to an embodiment of this application is shown; Figure 3 A schematic diagram of the structure of an electronic device according to an embodiment of this application is shown; Figure 4 A schematic diagram of the structure of an electronic cigarette device according to an embodiment of this application is shown. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] Figure 4 A schematic diagram of the structure of an electronic cigarette device according to an embodiment of this application is shown.

[0023] like Figure 4 As shown, when an e-cigarette cartridge is inserted into an inductor coil, different types of cartridges, due to their different shapes (e.g., flat or arched) and differences in material conductivity, generate measurable differences in the electrical signals, such as current and voltage waveforms, produced in an alternating magnetic field. Furthermore, even for the same type of cartridge, from the start of heating until reaching optimal flavor, the electrical signals will dynamically change over time due to differences in heating environment, heat dissipation conditions, and other factors.

[0024] In order to accurately identify the type of e-cigarette cartridge and whether it is at its optimal taste during the heating process, this application proposes a method for identifying e-cigarette cartridges in electronic cigarette devices.

[0025] It's understandable that when the e-liquid cartridge is at its optimal flavor level, the user will experience the best taste and a better overall experience when using it.

[0026] Figure 1 A flowchart illustrating a method for identifying an electronic cigarette cartridge according to an embodiment of this application is shown.

[0027] like Figure 1 As shown in the figure, an embodiment of this application provides a method for identifying e-cigarette cartridges, which includes: Step S101: Extract the features of the original electrical signals generated by the tobacco cartridge to be identified during the heating process to obtain the original electrical features of the tobacco cartridge to be identified; Step S102: Based on the preset mean and preset variance, as well as the dynamic mean and dynamic variance of the dynamic buffer sample, the original electrical features are normalized to obtain the unique electromagnetic fingerprint of the e-cigarette cartridge to be identified. Step S103: Input the exclusive electromagnetic fingerprint into the pre-trained tobacco cartridge recognition model to obtain the type of tobacco cartridge to be identified and / or whether it is in the best flavor.

[0028] Specifically, such as Figure 1As shown, in step S101, the original electrical signals generated by the tobacco cartridge to be identified during the heating process are collected, and the original electrical signals are extracted to obtain the original electrical features of the tobacco cartridge to be identified.

[0029] Specifically, such as Figure 4 As shown, the original electrical signals generated by the e-cigarette cartridge to be identified during the heating process are collected by an electrical signal acquisition device. For example, the electrical signal acquisition device collects the AC signal of the heating coil in real time through an ADC (Analog to Digital Converter) module. This signal is a timing signal.

[0030] Understandably, the original electrical signal reflects the electromagnetic response characteristics of the cartridge under the action of an alternating magnetic field.

[0031] Specifically, the method for identifying an electronic cigarette cartridge provided in this application is executed at regular intervals. During execution, the original electrical signal of a preset time prior to the execution time is collected. For example, if the identification method of this application is executed every 1 second, then the electrical signal of 1 second prior to the execution time is used as the original electrical signal of the cartridge to be identified.

[0032] Preferably, the extraction of features of the original electrical signals generated by the cigarette cartridge to be identified during the heating process includes: The original electrical signal is intercepted by a preset anti-leakage window to obtain multiple segments of electrical signal; Feature extraction is performed on multiple segments of electrical signals to obtain the original electrical features.

[0033] Specifically, such as Figure 1 As shown, after acquiring the original electrical signals generated by the e-cigarette cartridge to be identified during the heating process, feature extraction is performed on the original electrical signals.

[0034] Specifically, the original electrical signal is a continuous data stream. For example, the original electrical signal is a 1-second continuous data stream. The signal is intercepted by a preset anti-leakage window. A sliding time window is set to intercept the 1-second continuous data stream to obtain multiple segments of electrical signal.

[0035] It is worth noting that by setting a pre-defined anti-leakage window to reduce spectral leakage, subsequent feature extraction becomes more accurate.

[0036] Specifically, the preset leak-proof window can be a Hanning window, a Hamming window, or a Blackman window, but this application is not limited to these; other window functions with similar leak-proof effects can also be applied.

[0037] Preferably, the pre-installed leak-proof window is a Hanning window.

[0038] Specifically, feature extraction is performed on multiple segments of electrical signals to obtain the original electrical features.

[0039] For example, time-domain analysis, frequency-domain transformation, or time-frequency analysis can be performed on each segment of the extracted electrical signal to extract multi-dimensional original electrical features.

[0040] Preferably, the original electrical characteristics include one or more of the following: Steady-state energy distribution characteristics; Morphological drift and bandwidth characteristics; Nonlinear distortion and harmonic characteristics; Temporal dynamic evolution characteristics.

[0041] Specifically, the steady-state energy distribution characteristics are used to represent the energy intensity and peak density of the mapped signal at the core driving frequency and its response band. These characteristics can reflect the basic resistance and inductance characteristics of the heating element of the cartridge.

[0042] Specifically, the morphological drift and bandwidth characteristics are used to calculate the offset of the energy centroid of the response signal and the diffusion bandwidth. These characteristics reflect the deformation caused by the thermal expansion of the metal during heating and the rate of change of permeability with temperature.

[0043] Specifically, nonlinear distortion and harmonic characteristics are used to quantify the ratio of higher harmonic energy to fundamental energy. This characteristic reflects the nonlinear permeability of the heating element's metallic material, and the harmonic characteristics of different alloy materials vary significantly.

[0044] Specifically, the time-series dynamic evolution feature is used to characterize the rate of change of the time gradient of the signal, the transient response envelope, and the total energy of the statistical integral within the sliding window. This feature reflects the heat capacity and heat dissipation rate of the cartridge.

[0045] It is worth noting that the above four types of characteristics describe the electromagnetic response characteristics of the cartridge from the frequency domain, statistical domain, nonlinear domain, and time domain, respectively, and together constitute the original electrical characteristics of the cartridge.

[0046] Those skilled in the art will understand that the above four types of features can be used individually or in combination, and the recognition accuracy is higher when used in combination.

[0047] Specifically, such as Figure 1 As shown, after obtaining the original electrical characteristics of the e-cigarette cartridge to be identified through step S101, the original electrical characteristics are normalized in step S102 to obtain the unique electromagnetic fingerprint of the e-cigarette cartridge to be identified.

[0048] It is understandable that the extracted multidimensional raw electrical features differ in dimensions and numerical ranges, and directly feeding them into the cartridge recognition model will affect the recognition accuracy.

[0049] Preferably, the original electrical characteristics are normalized using the following formula:

[0050] in, Indicates the original electrical characteristics, This indicates a proprietary electromagnetic fingerprint. Indicates the mixed weighting coefficient. , This indicates the preset mean and preset variance. , This represents the dynamic mean and dynamic variance.

[0051] Specifically, the preset mean and preset variance are global statistical parameters calculated based on a large amount of sample data during the training phase of the cartridge recognition model. In implementing the cartridge recognition method for electronic cigarettes provided in this application, the preset mean and preset variance are fixed values.

[0052] Specifically, dynamic mean and dynamic variance represent the statistical parameters corresponding to the high-confidence recognition results stored in the dynamic buffer samples.

[0053] It is worth noting that the normalization process provided in this application embodiment integrates offline global benchmark parameters and real-time dynamic statistical parameters.

[0054] Offline baseline parameters provide a stable reference baseline, ensuring that the normalization results do not fluctuate drastically due to occasional noise. Meanwhile, dynamic statistical parameters track the current operating status in real time, such as component aging and changes in ambient temperature, ensuring that the normalization results always match the current operating conditions.

[0055] Through the above normalization processing mechanism, the normalized features effectively overcome the signal baseline offset problem caused by device aging and environmental drift at zero hardware cost.

[0056] Specifically, such as Figure 1 As shown, in step S103, the unique electromagnetic fingerprint of the tobacco cartridge to be identified obtained in step S102 is input into the pre-trained tobacco cartridge identification model. The tobacco cartridge identification model outputs the confidence level of the tobacco cartridge to be identified in each tobacco cartridge type. The tobacco cartridge type of the tobacco cartridge to be identified is determined according to the confidence level in each tobacco cartridge type. For example, the tobacco cartridge type corresponding to the highest confidence level is taken as the tobacco cartridge type of the tobacco cartridge to be identified.

[0057] Specifically, such as Figure 1As shown, in step S103, the unique electromagnetic fingerprint of the tobacco cartridge to be identified obtained in step S102 is input into the pre-trained tobacco cartridge identification model. The tobacco cartridge identification model outputs the confidence level of the tobacco cartridge to be identified being in the best taste and the confidence level of not being in the best taste. The confidence level of the best taste and the confidence level of not being in the best taste are compared to determine whether it is in the best taste. For example, the result corresponding to the larger confidence level of the best taste and the confidence level of not being in the best taste is taken as the identification result.

[0058] For example, in some embodiments of this application, the cartridge recognition model adopts a dual-branch output architecture: the first branch outputs the cartridge type, and the second branch outputs whether the cartridge is currently in the optimal flavor range. The two branches extract features through a shared backbone network, and each outputs classification results through independent fully connected layers and Softmax layers.

[0059] It is worth noting that when using the cigarette cartridge recognition model, no large dynamic computation graph is required. Pure C language hard-coded matrix operations enable it to achieve millisecond-level response even in an extremely limited environment with only a few KB of RAM.

[0060] Meanwhile, the cartridge type output by the cartridge recognition model can be used to distinguish cartridges of different models, specifications, and flavors, and the best taste judgment result output by the cartridge recognition model can guide users on whether to use it.

[0061] Specifically, the cigarette cartridge recognition model can use a convolutional neural network model.

[0062] Preferably, the identification method further includes: Determine whether the confidence level for the type of tobacco cartridge to be identified and / or whether it is in the optimal taste range is greater than the safety threshold. If the confidence level of the type of tobacco cartridge to be identified and / or whether it is in the best taste is greater than the safety threshold, the tobacco cartridge to be identified is stored in the dynamic buffer sample, and the dynamic mean and dynamic variance of the dynamic buffer sample are updated.

[0063] It is worth noting that, in order to improve the adaptability of the cigarette cartridge recognition model throughout its entire lifecycle, this application's embodiments introduce a high-confidence adaptive feedback mechanism.

[0064] Specifically, it determines whether the confidence level of the type of the tobacco cartridge to be identified is greater than the first safety threshold. If the maximum confidence level of the type of the tobacco cartridge to be identified is greater than the first safety threshold, the tobacco cartridge to be identified is stored in the dynamic buffer sample, and the dynamic mean and dynamic variance of the dynamic buffer sample are updated at the same time.

[0065] Specifically, the confidence level of whether the tobacco cartridge to be identified is in the optimal taste is greater than the second safety threshold. If the maximum confidence level of whether the tobacco cartridge to be identified is in the optimal taste is greater than the second safety threshold, the tobacco cartridge to be identified is stored in the dynamic buffer sample, and the dynamic mean and dynamic variance of the dynamic buffer sample are updated at the same time.

[0066] It is worth noting that the first safety threshold and the second safety threshold can be the same or different, and no specific limitation is made here. Preferably, the values ​​of the first safety threshold and the second safety threshold are in the range of 0.95-0.98.

[0067] For example, the dynamic buffer sample uses a sliding window mechanism to store the most recent N samples, where N typically ranges from 50 to 200. Whenever a new sample is added, the dynamic mean and dynamic variance of the dynamic buffer sample are recalculated and used for subsequent normalization processing.

[0068] This closed-loop feedback mechanism enables the normalized parameters to be updated adaptively as the equipment is used over time, thereby continuously overcoming feature drift caused by component aging and environmental changes.

[0069] Preferably, the cigarette cartridge recognition model is trained through the following steps: Multiple raw electrical signals generated during the heating process of each type of e-cigarette cartridge are obtained. Noise is injected into each raw electrical signal to obtain multiple noise-enhanced electrical signals, resulting in a set of noise-enhanced electrical signals for multiple e-cigarette cartridge types. For each noise-enhanced electrical signal in the noise-enhanced electrical signal set, label the cartridge type and / or whether it is in the optimal taste, obtain several training samples; The cartridge recognition model is trained using several training samples to determine the optimal model parameters.

[0070] Specifically, the steps for training the cartridge recognition model can be executed on the host computer. Based on different types of cartridges (such as cartridges of different models and materials), raw electrical signals are collected under multi-dimensional working conditions (such as different ambient temperatures and different battery voltages).

[0071] Specifically, to improve the generalization ability of the cigarette cartridge recognition model and prevent overfitting, adaptive Gaussian noise data enhancement is applied to the collected raw electrical signals. For example, the noise injection formula is:

[0072] Where x represents the original electrical signal, and x' represents the noise-enhanced electrical signal. This represents random noise that follows a normal distribution, with a mean of 0 and a standard deviation of . That is, the intensity of the injected random noise is adaptively positively correlated with the absolute amplitude of the original signal.

[0073] It is worth noting that this method is used when training the cigarette cartridge recognition model to simulate interference such as quantization errors of the acquisition device and power ripple that may be encountered in actual operation. By forcing the neural network to learn the essential evolution trend of the waveform rather than memorizing specific values, it effectively prevents the model from overfitting.

[0074] Specifically, each noise-enhanced electrical signal in the set of noise-enhanced electrical signals is labeled with the type of tobacco cartridge, and each noise-enhanced electrical signal in the set of noise-enhanced electrical signals is labeled with whether it is in the optimal taste, thus obtaining several training samples.

[0075] Specifically, for each enhanced noise-enhanced electrical signal sample, professional personnel or through experimental calibration methods label its corresponding cartridge type (such as cartridge A, cartridge B, etc.) and whether it is in the optimal taste range (such as class A: cartridge A normal heating range, class B: cartridge A optimal taste heating range, class C: cartridge B normal heating range, class D: cartridge B optimal taste heating range). After labeling, a labeled training sample set is formed.

[0076] Specifically, after obtaining the training samples, the cartridge recognition model is trained based on several training samples until the training cutoff condition is met, resulting in a trained cartridge recognition model, and the optimal model parameters of the cartridge recognition model are determined.

[0077] For example, the training sample set is input into the cigarette cartridge recognition model for forward propagation and backward error convergence training. The cigarette cartridge recognition model can use the cross-entropy loss function and iteratively optimize the network weights and bias parameters through the gradient descent algorithm.

[0078] After training is completed, global statistics are extracted, such as the preset mean μ_base, preset variance σ_base, network weights and bias parameters, and these global statistics are applied to the inference stage.

[0079] Figure 2 A schematic diagram of the structure of an electronic cigarette cartridge identification system according to an embodiment of this application is shown.

[0080] like Figure 2 As shown, this application provides an identification system for electronic cigarette cartridges, characterized in that the identification system includes: The feature extraction module 201 is used to extract the features of the original electrical signals generated by the tobacco cartridge to be identified during the heating process, so as to obtain the original electrical features of the tobacco cartridge to be identified. The normalization processing module 202 is used to normalize the original electrical features based on the preset mean and preset variance, as well as the dynamic mean and dynamic variance of the dynamic buffer sample, to obtain the unique electromagnetic fingerprint of the e-cigarette cartridge to be identified. The identification module 203 is used to input a unique electromagnetic fingerprint into a pre-trained tobacco cartridge identification model to obtain the type of tobacco cartridge to be identified and / or whether it is in the best flavor.

[0081] Figure 3 A schematic diagram of the structure of an electronic device according to an embodiment of this application is shown.

[0082] like Figure 3 As shown, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aforementioned identification method.

[0083] like Figure 4 As shown, this application provides an electronic cigarette device, including a power supply, an inductor coil, an electrical signal acquisition device, and the aforementioned identification system.

[0084] It is worth noting that when training the cartridge recognition model, offline training can be carried out on a host computer. The MCU (Microcontroller Unit) is placed in the electronic cigarette device as the recognition system. After the cartridge recognition model is trained, global statistics, such as preset mean, preset variance, network weights and bias parameters, are sent to the MCU for inference. Even in an extremely limited environment where the MCU only has a few KB of RAM, millisecond-level response can be achieved.

[0085] The present application provides a method, system, device, storage medium, and electronic cigarette device for identifying e-cigarette cartridges. By acquiring the original electrical characteristics of the cartridge to be identified, and normalizing the original electrical characteristics based on preset mean and preset variance, as well as the dynamic mean and dynamic variance of dynamic buffer samples, a unique electromagnetic fingerprint of the cartridge to be identified is obtained. The unique electromagnetic fingerprint is then identified using a pre-trained cartridge identification model. This method can quickly and accurately determine the cartridge type and / or whether the cartridge is in optimal flavor, and significantly reduces the identification cost.

[0086] Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for identifying e-cigarette cartridges, characterized in that, The identification method includes: The original electrical signal characteristics of the tobacco cartridge to be identified are obtained by extracting the features of the original electrical signal generated during the heating process. Based on the preset mean and preset variance, as well as the dynamic mean and dynamic variance of the dynamic buffer samples, the original electrical features are normalized to obtain the unique electromagnetic fingerprint of the e-cigarette cartridge to be identified. The unique electromagnetic fingerprint is input into the pre-trained cartridge recognition model to obtain the cartridge type and / or whether the cartridge is in the optimal flavor.

2. The identification method according to claim 1, characterized in that, The features of extracting the original electrical signals generated by the cigarette cartridge to be identified during the heating process include: The original electrical signal is intercepted by a preset anti-leakage window to obtain multiple segments of electrical signal; Feature extraction is performed on multiple segments of electrical signals to obtain the original electrical features.

3. The identification method according to claim 2, characterized in that, The original electrical characteristics include one or more of the following: Steady-state energy distribution characteristics; Morphological drift and bandwidth characteristics; Nonlinear distortion and harmonic characteristics; Temporal dynamic evolution characteristics.

4. The identification method according to claim 1, characterized in that, The original electrical characteristics are normalized using the following formula: in, Indicates the original electrical characteristics, This indicates a proprietary electromagnetic fingerprint. Indicates the mixed weighting coefficient. , This indicates the preset mean and preset variance. , This represents the dynamic mean and dynamic variance.

5. The identification method according to claim 1, characterized in that, The identification method further includes: Determine whether the confidence level for the type of tobacco cartridge to be identified and / or whether it is in the optimal taste range is greater than the safety threshold. If the confidence level of the type of tobacco cartridge to be identified and / or whether it is in the best taste is greater than the safety threshold, the tobacco cartridge to be identified is stored in the dynamic buffer sample, and the dynamic mean and dynamic variance of the dynamic buffer sample are updated.

6. The identification method according to claim 1, characterized in that, Train the cigarette cartridge recognition model using the following steps: Multiple raw electrical signals generated during the heating process of each type of e-cigarette cartridge are obtained. Noise is injected into each raw electrical signal to obtain multiple noise-enhanced electrical signals, resulting in a set of noise-enhanced electrical signals for multiple e-cigarette cartridge types. For each noise-enhanced electrical signal in the noise-enhanced electrical signal set, label the cartridge type and / or whether it is in the optimal taste, obtain several training samples; The cartridge recognition model is trained using several training samples to determine the optimal model parameters.

7. A system for identifying cartridges in an electronic cigarette device, characterized in that, The identification system includes: The feature extraction module is used to extract the features of the original electrical signals generated by the tobacco cartridge to be identified during the heating process, so as to obtain the original electrical features of the tobacco cartridge to be identified. The normalization module is used to normalize the original electrical features based on the preset mean and preset variance, as well as the dynamic mean and dynamic variance of the dynamic buffer samples, to obtain the unique electromagnetic fingerprint of the e-cigarette cartridge to be identified. The identification module is used to input a unique electromagnetic fingerprint into a pre-trained cartridge identification model to obtain the cartridge type and / or whether the cartridge is in optimal flavor.

8. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the identification method according to any one of claims 1-6.

9. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, implements the identification method according to any one of claims 1-6.

10. An electronic cigarette device, characterized in that, The electronic cigarette device includes a power supply, an inductor coil, an electrical signal acquisition device, and the identification system as described in claim 7.

Images