Artificial intelligence acquisition method for actual temperature of aerosol generating substrate in heating smoking set and heating smoking set

Abstract

The invention discloses an artificial intelligence obtaining method for the actual temperature of an aerosol generating substrate in a heating smoking set and the heating smoking set. The method comprises the steps that the actual temperature of a first temperature sensor and the actual temperature of a second temperature sensor are obtained; inputting the actual temperature of the first temperature sensor and the actual temperature of the second temperature sensor into a pre-constructed actual temperature acquisition model of the aerosol generation matrix to obtain the actual temperature of the aerosol generation matrix; wherein the actual temperature acquisition model of the aerosol generation matrix is constructed through machine learning by taking a test data group pair as a training sample and taking the actual temperature of the aerosol generation matrix as a target. A highly integrated and intelligent solution is constructed, the temperature sensor network is effectively deployed, the temperature information of multiple positions of the smoking set is captured in real time so as to obtain the real-time temperature change of tobacco, and a data foundation is laid for accurate control.

Description

Technical Field

[0001] The present invention relates to an artificial intelligence acquisition method for the actual temperature of an aerosol generating matrix in a heating smoking device and the heating smoking device, belonging to the technical field of heating smoking device control. Background Art

[0002] As countries gradually strengthen cigarette control and people pay more and more attention to health, a type of heated tobacco devices for heating cigarettes has become increasingly popular in recent years. It is mainly a product that allows users to inhale after nicotine and other substances in tobacco are turned into steam through means such as atomization. In the continuous evolution of heated cigarette technology, achieving high-precision temperature control is not only a breakthrough in technology, but also a reflection of a deep understanding of consumer experience. In order to ensure that the taste, aroma and nicotine delivery of each puff of smoke remain highly consistent, various existing technologies continue to try to improve the measurement accuracy and control accuracy of temperature sensors, in order to give consumers a stable and satisfactory smoking experience. Although existing technologies have made certain progress in the temperature control of heated tobacco devices, they usually rely on fixed heating curves and lack adaptability to different tobacco characteristics and actual use environment conditions.

[0003] However, these existing techniques ignore two fundamental issues: First, the temperature control curves of current heated tobacco devices are all better heating curves that are gradually explored by a certain heated cigarette on a certain heated tobacco device based on manual smoking in a normal environment (room temperature of 25°C). Once a heated tobacco device is replaced with a different cigarette, or a cigarette is replaced with a heated tobacco device, a new heating curve needs to be explored again.

[0004] Second, the object controlled by the heated smoking device is always the temperature of the temperature sensor near the heater, rather than the actual temperature of the tobacco segment of the heated cigarette. For the centrally heated smoking device in which the heater and the temperature sensor are directly inserted into the tobacco segment during use, the temperature sensor and the tobacco segment are already very close, but there is still a temperature difference of more than 10°C; for the circumferentially heated smoking device, since the tobacco segment is inside the cylindrical heater and the temperature sensor is often attached to the outside of the cylindrical heater, there will be a greater temperature difference, some even exceeding 30°C. Therefore, once the smoking device is taken to a special environment (winter in the north or summer in the south), the deviation between the temperature of the temperature sensor and the actual temperature of the tobacco will fluctuate greatly, making the optimal heating curve artificially explored in a conventional environment not the optimal heating curve in actual use.

[0005] Therefore, it is necessary to conduct research on the above two basic issues and find a more scientific method to obtain the actual temperature of the aerosol generating matrix in the heated smoking device, so as to control the temperature more accurately. Summary of the invention

[0006] The purpose of the present invention is to overcome the deficiencies in the prior art and to provide an artificial intelligence method for obtaining the actual temperature of an aerosol generating substrate in a heated smoking device and a heated smoking device.

[0007] In order to achieve the above purpose / solve the above technical problems, the present invention is implemented by adopting the following technical solutions: In a first aspect, the present invention provides an artificial intelligence method for obtaining the actual temperature of an aerosol-generating substrate in a heated smoking device, comprising: Acquire the actual temperature of the first temperature sensor and the actual temperature of the second temperature sensor; Inputting the actual temperature of the first temperature sensor and the actual temperature of the second temperature sensor into a pre-constructed aerosol generation substrate actual temperature acquisition model to obtain the aerosol generation substrate actual temperature; Among them, the aerosol generating matrix actual temperature acquisition model is constructed by machine learning with test data group pairs as training samples and the aerosol generating matrix actual temperature as the target. The test data group includes the first temperature sensor test temperature, the second temperature sensor test temperature and the aerosol generating matrix test temperature.

[0008] Furthermore, the test data group also includes a humidity sensor testing humidity.

[0009] Furthermore, the test data set also includes timing information and derivative information of the first temperature sensor test temperature, the second temperature sensor test temperature, the sensor test humidity and / or the aerosol generating substrate test temperature.

[0010] Furthermore, the aerosol generating substrate actual temperature acquisition model is based on a neural network model.

[0011] Furthermore, the neural network model includes 3 hidden layers, and each hidden layer includes 128 neurons.

[0012] Furthermore, the hidden layer uses a ReLU activation function.

[0013] Furthermore, in the machine learning process, mean square error is selected as the loss function, and Adam optimizer is selected as the optimizer.

[0014] Furthermore, after constructing the aerosol generating substrate actual temperature acquisition model, the method further includes: optimizing the aerosol generating substrate actual temperature acquisition model.

[0015] Furthermore, optimizing the aerosol generation matrix actual temperature acquisition model includes: hyperparameter tuning, model integration and model pruning.

[0016] In a second aspect, the present invention provides a heated smoking device, comprising a shell, a heating body, a power source, a main control circuit, a first temperature sensor and a second temperature sensor, wherein the shell defines a heating chamber for receiving an aerosol generating substrate, the heating body is located in the heating chamber and can be in thermal contact with the received aerosol generating substrate, the first temperature sensor is closer to the heating body relative to the aerosol generating substrate, the second temperature sensor is closer to the outer surface of the shell relative to the heating body, and the main control circuit is configured to control the heated smoking device to obtain the actual temperature of the aerosol generating substrate according to the following steps: The method comprises: obtaining an actual temperature of the first temperature sensor and an actual temperature of the second temperature sensor; Inputting the actual temperature of the first temperature sensor and the actual temperature of the second temperature sensor into a pre-constructed aerosol generation substrate actual temperature acquisition model to obtain the aerosol generation substrate actual temperature; Among them, the aerosol generating matrix actual temperature acquisition model is constructed by machine learning with test data group pairs as training samples and the aerosol generating matrix actual temperature as the target. The test data group includes the first temperature sensor test temperature, the second temperature sensor test temperature and the aerosol generating matrix test temperature. The first temperature sensor is used to obtain the temperature at the first position, and the second temperature sensor is used to obtain the temperature at the second position.

[0017] Compared with the prior art, the present invention has the following beneficial effects: The present invention has constructed a highly integrated and intelligent solution. By effectively deploying a temperature sensor network, it can capture temperature information at multiple locations of the smoking device in real time to obtain real-time temperature changes of the tobacco itself, laying a data foundation for precise control. In the future, when developing the optimal heating curve for heated cigarettes, the study of the optimal heating curve of the tobacco itself can be separated from the study of how the smoking device measures the temperature of the tobacco itself, avoiding the dilemma of having to re-explore the optimal heating curve for any combination of cigarettes and smoking devices, greatly improving the efficiency and quality of research and development. For any cigarette, it is only necessary to obtain the optimal heating curve of the tobacco itself, and the heated smoking device can operate on the automatic control heater, so that the actual heating temperature of the tobacco is heated according to the heating curve. Compared with the prior art, the present invention can dynamically adjust the heating strategy according to the actual temperature of the tobacco, thereby improving the uniformity and efficiency of tobacco heating.

[0018] In addition, by optimizing the algorithm and streamlining the code level, redundant code is removed, and the load and power consumption of the processor and memory are greatly reduced under a certain temperature control accuracy. These comprehensive optimization measures can effectively improve the energy efficiency of the system, extend the battery life of the embedded heating tobacco device, and reduce the hardware cost of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a schematic cross-sectional view of an aerosol generating device and an aerosol generating article; Figure 2 is a schematic diagram of the longitudinal distribution of thermocouples of the aerosol generating product for testing; Figure 3 is a schematic diagram of the longitudinal distribution of thermocouples of the test aerosol generating device; Figure 4 is a schematic diagram of the circumferential distribution of thermocouples of the test aerosol generating device; Figure 5 It is a flow chart of a method for obtaining the actual temperature of an aerosol-generating substrate in a heated smoking device; The reference numerals are described as follows: Aerosol generating products: 100 Aerosol generating matrix: 110 Support structure: 120 Cooling structure: 130 Filter structure: 140 Aerosol generating device: 200 Battery and control components: 210 Heating chamber: 220 Heating element: 230 Thermocouples for cigarette testing: 111, 112, 113 Thermocouples for testing the inner wall of the heating element: 221, 222, 223 Thermocouples for testing the outer wall of the heating body: 224, 225, 226 Thermocouples for shell testing: 241, 242, 243, 241a, 241b, 241c Humidity sensor: 290 Multi-channel recorder: 300 DETAILED DESCRIPTION

[0020] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0021] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", and the like may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0022] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0023] like Figure 1 As shown, the aerosol generating article 100 includes a smoking section 110, a supporting structure 120, a cooling structure 130 and a filtering structure 140 from the distal end to the proximal end. The aerosol generating device 200 as a heated smoking device includes a power supply and a main control circuit 210, a heating body 220, a heating chamber 230 for receiving an aerosol generating substrate and a housing 240. The aerosol generating device 200 is also provided with a first temperature sensor, a second temperature sensor and a humidity sensor, and their specific functions and positions need to be determined and optimized based on test data and machine learning results. The heating body 220 is a heating resistance wire and is arranged on the outside of a heating tube with good thermal conductivity, so that the heat is evenly distributed around the smoking section 110.

[0024] Test equipment like Figure 2As shown, the cigarette used as the test aerosol generating product 100 is inserted into three thermocouples 111, 112, and 113 for cigarette testing in the center and at the upper, middle, and lower longitudinal positions respectively. The thermocouples are connected to the multi-channel recorder 300 (GL240 thermometer + Japanese graphtec GL240 insulated multi-channel data recorder) signal, so as to realize the temperature of different positions inside the cigarette during the heating process. If necessary, the number of thermocouples can be increased or decreased. A larger number of thermocouples can improve the accuracy of machine learning results, but it will also bring the trouble of building an overly complex model structure and processing massive data during the use of smoking tools.

[0025] like Figure 3 As shown, the heated tobacco device as the test aerosol generating device 200 is first inserted with a plurality of cigarette test thermocouples, wherein three thermocouples 221, 222, and 223 are located on the inner wall of the heating body (heating tube) and are located at the upper, middle, and lower longitudinal positions, respectively; three thermocouples 224, 225, and 226 are located on the outer wall of the heating body (heating tube) and are located at the upper, middle, and lower longitudinal positions, respectively; and three thermocouples 241, 242, and 243 are located on the shell test thermocouples and are located at the upper, middle, and lower longitudinal positions, respectively. In addition, as Figure 4 As shown, for a non-circumferentially symmetrical structure, the number of thermocouples can be increased to further obtain temperature information related to the shape of the heating smoking device, for example, thermocouples 241a, 241b, and 241c are respectively arranged at asymmetric positions of the shell. Similarly, the number of thermocouples can be increased or decreased when necessary. Thermocouples 221, 222, and 223 are closer to the aerosol generating matrix 110 relative to the heating body 220, thermocouples 224, 225, and 226 are closer to the heating body 220 relative to the aerosol generating matrix 110, and thermocouples 241, 242, and 243 are closer to the outer surface of the shell relative to the heating body. A temperature field gradient is formed between these thermocouple groups. The temperature field gradient is related to the structure, material properties, heating temperature, etc. of the heating smoking device itself, but it is not a simple linear relationship. Therefore, it is necessary to construct an artificial intelligence model and use machine learning to discover the temperature field gradient law, so that the thermocouples on the smoking device can be used to more accurately obtain the temperature of the aerosol generating matrix, rather than the prior art that regards the thermocouple temperature on the smoking device as the temperature of the aerosol generating matrix.

[0026] In addition, a temperature sensor 290 may be provided at a location such as inside the heating chamber 230 to increase the dimension of the machine learning model parameters.

[0027] Reference Figure 5 For testing and training: Testing Process The test experiment adopts a three-factor mixed experimental design: (ambient temperature: 0-40℃, according to actual weather conditions) × (body temperature: 70-250℃, thermal engine data is recorded every 30℃) × (ambient humidity: 10%-80%, according to actual weather conditions). In order to make the ambient temperature and ambient humidity have more possibilities, the present invention carried out test experiments in three cities (Haikou, Chengdu, and Harbin) with large differences in ambient temperature and humidity, in different time periods of early spring weather, including early morning, morning, noon, afternoon, and evening.

[0028] Prepare the test smoking device and the thermocouples on the test cigarettes according to Table 1, insert the test cigarettes into the test smoking device, connect the recording equipment (GL240 thermometer + Japan Graphtec GL240 insulated multi-channel data recorder), start the heating smoking device and then perform manual inhalation (heat each cigarette for 210 seconds and take 7 puffs), and the recording equipment will automatically record the timing, the real-time temperature and humidity of each thermocouple, and the real-time voltage and power of the smoking device.

[0029] Under the same ambient temperature and humidity, the experiment was conducted with one room temperature startup test and two continuous tests after the machine was warmed up. A fully charged test smoking device can undergo three rounds of experiments.

[0030] The test data can be recorded according to Table 1 (automatically recorded by computer).

[0031] Table 1: Test data record table

[0032] Machine Learning with Test Data The multiple test data groups obtained through the test are preprocessed (step A2), including: Derivative processing A21: Supplementary calculation of the change rate (derivative) of temperature and humidity at each time series based on the real-time temperature value; Outlier processing A22: Use the interquartile range (IQR) method to detect and remove outliers to remove data noise; Data normalization A23: Apply Z-score normalization to temperature and voltage data to eliminate dimensional differences and accelerate model convergence; Time series feature extraction A24: Introducing time lag features, such as the temperature change rate and humidity change rate, to enhance the model's ability to capture time dependencies. This step ensures that the model can fully understand the dynamic behavior of temperature and humidity over time.

[0033] In addition, in order to verify the effectiveness of the model, it is necessary to use unseen data sets to evaluate the prediction performance and stability of the model. The specific settings include: Dataset partitioning A31: Divide the dataset into training set, validation set, and test set to ensure the generalization ability of the model on different data; Evaluation indicators were established A32: Mean square error (MSE) and root mean square error (RMSE) were used as evaluation indicators to measure the prediction accuracy of the model.

[0034] Taking the test data set as training samples and the actual temperature of the aerosol generating matrix as the target, machine learning is performed to build an aerosol generating matrix actual temperature acquisition model (step B). The model is based on a multi-layer deep neural network (DNN) and can be implemented in PyTorch. Its structure is as follows: Input layer M1: Design the input layer according to the number of preprocessed data features, such as temperature, humidity, duration and its derivatives; Hidden layer M2: Three hidden layers are used, each layer contains 128 neurons, and uses ReLU (Rectified Linear Unit) activation function. This design can provide sufficient model complexity to help capture the complex nonlinear relationship in the temperature field; Output layer M3: A single output node used to predict the actual temperature of the aerosol generating matrix. The output layer uses a linear activation function to adapt to the prediction task of continuous values.

[0035] The training process includes selecting a suitable loss function, optimization algorithm, regularization technique, and setting training parameters. The specific steps are as follows: Loss function T1: The mean square error (MSE) is used as the loss function, which directly reflects the difference between the predicted temperature and the actual temperature; Optimizer selection T2: Use the Adam optimizer, which has the advantage of adaptive learning rate and is suitable for processing large-scale parameter optimization tasks; Regularization technology T3: Introducing L2 regularization to reduce the risk of overfitting of the model and maintain the generalization ability of the model; Batch size and number of iterations T4: Set the batch size to 32 and the number of iterations to 1000. This configuration strikes a balance between memory usage and training speed and ensures stable convergence of the model.

[0036] The experimental results section will record the performance of the model on the training set and the validation set in detail, including the error reduction curve, the prediction accuracy under different operating conditions, and the final performance on the test set.

[0037] In order to further optimize the above-mentioned aerosol generation matrix actual temperature acquisition model (step C), the following optimization method is adopted: Hyperparameter tuning C1: Use grid search or Bayesian optimization to tune hyperparameters and find the optimal learning rate, batch size, regularization coefficient and other parameters; Model ensemble C2: Use ensemble learning methods, such as random forests, combined with neural networks to improve overall prediction accuracy; Model pruning C3: Reduce redundant neurons and connections through model pruning technology, and improve the operation efficiency and deployment convenience of the model.

[0038] Through the above optimization method, the optimal number and location of temperature and humidity sensors can be selected, so as to achieve the simplest model structure and data processing consumption while meeting the accuracy requirements (the difference between the predicted temperature and the actual temperature does not exceed 1°C).

[0039] Through the above-mentioned machine learning and optimization processes, it was found that the first temperature sensor at position 226 and the second temperature sensor at position 241 should be set, and a neural network model with three hidden layers and 128 neurons in each layer that can be run on the main control circuit of the handheld heating tobacco device should be adopted to obtain the actual heating temperature of the tobacco with sufficient accuracy.

[0040] The model can be deployed to actual heated tobacco products and applied by following the steps below: Preparation of the heated smoking device D1: manufacturing the heated smoking device in a conventional manner, including a housing, a heating body, a power supply, a main control circuit, a first temperature sensor and a second temperature sensor; Model export D2: Export the trained model to a deployable format; Device integration D3: Integrate the exported model into the main control circuit; Equipment operation D4: After inserting the cigarette, the heating device is started, and the optimal heating curve of the cigarette is selected and read according to the type of cigarette; the main control circuit control system obtains the temperature of the first temperature sensor and the second temperature sensor of the equipment in real time, inputs them into the model, and obtains the actual temperature of the tobacco in the cigarette, thereby increasing or decreasing the heating power of the heating body, so that the actual temperature of the tobacco meets the optimal heating curve of the cigarette.

[0041] The present invention has built a highly integrated and intelligent solution. By effectively deploying a temperature sensor network, it can capture temperature information at multiple locations of the smoking device in real time to obtain real-time temperature changes of the tobacco itself, laying a data foundation for precise control. In the future, when developing the optimal heating curve for heated cigarettes, the study of the optimal heating curve of the tobacco itself can be separated from the study of how the smoking device measures the temperature of the tobacco itself, avoiding the dilemma of having to re-explore the optimal heating curve for any combination of cigarettes and smoking devices, greatly improving the efficiency and quality of research and development. For any cigarette, it is only necessary to obtain the optimal heating curve of the tobacco itself, and the heated smoking device can operate on the automatic control heater so that the actual heating temperature of the tobacco is heated according to the heating curve.

[0042] In addition, by optimizing the algorithm and streamlining the code level, redundant code is removed, and the load and power consumption of the processor and memory are greatly reduced under a certain temperature control accuracy. These comprehensive optimization measures can effectively improve the energy efficiency of the system, extend the battery life of the embedded heating tobacco device, and reduce the hardware cost of the product.

[0043] In a typical configuration of the present invention, the main control circuit includes one or more processors (CPU), input / output interface, network interface and memory. The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium. Computer-readable media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. The information can be a computer-readable instruction, a data structure, a module of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk-read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0044] The terms and expressions used herein are for descriptive purposes only, and the present invention should not be limited to these terms and expressions. The use of these terms and expressions does not mean to exclude any equivalent features of the illustrations and descriptions (or parts thereof), and it should be recognized that various modifications that may exist should also be included in the scope of the claims. Other modifications, variations and substitutions may also exist. Accordingly, the claims should be deemed to cover all such equivalents.

[0045] Similarly, it should be pointed out that although the present invention has been described with reference to the current specific embodiments, ordinary technicians in this technical field should realize that the above embodiments are only used to illustrate the present invention, and various equivalent changes or substitutions may be made without departing from the spirit of the present invention. Therefore, as long as the changes and modifications to the above embodiments are within the scope of the essential spirit of the present invention, they will fall within the scope of the claims of the present invention.

Claims

1. A method for obtaining the actual temperature of an aerosol generating substrate in a heated smoking device by artificial intelligence, wherein the heated smoking device comprises a shell, a heating body, a power supply, a main control circuit, a first temperature sensor and a second temperature sensor, wherein the shell defines a heating chamber for receiving the aerosol generating substrate, the heating body is located in the heating chamber and can be in thermal contact with the aerosol generating substrate, and wherein: The first temperature sensor is closer to the heating body than the aerosol generating substrate, and the second temperature sensor is closer to the outer surface of the housing than the heating body; The main control circuit is configured to control the heating smoking device to obtain the actual temperature of the aerosol generating substrate according to the following steps: The method comprises: obtaining an actual temperature of the first temperature sensor and an actual temperature of the second temperature sensor; Inputting the actual temperature of the first temperature sensor and the actual temperature of the second temperature sensor into a pre-constructed aerosol generation substrate actual temperature acquisition model to obtain the aerosol generation substrate actual temperature; Among them, the aerosol generating matrix actual temperature acquisition model is constructed by machine learning with test data group pairs as training samples and the aerosol generating matrix actual temperature as the target. The test data group includes the first temperature sensor test temperature, the second temperature sensor test temperature and the aerosol generating matrix test temperature. The first temperature sensor is used to obtain the temperature at the first position, and the second temperature sensor is used to obtain the temperature at the second position.

2. The method for obtaining the actual temperature of the aerosol generating substrate in the heated smoking device by artificial intelligence according to claim 1, characterized in that: The test data set also includes a humidity sensor testing humidity.

3. The method for obtaining the actual temperature of the aerosol-generating substrate in the heated smoking device according to claim 2, characterized in that: The test data set also includes timing information and derivative information of the first temperature sensor test temperature, the second temperature sensor test temperature, sensor test humidity and / or the aerosol generating substrate test temperature.

4. The method for obtaining the actual temperature of the aerosol generating substrate in the heated smoking device by artificial intelligence according to claim 3, characterized in that: The aerosol generating substrate actual temperature acquisition model is based on a neural network model.

5. The method for obtaining the actual temperature of the aerosol generating substrate in the heated smoking device by artificial intelligence according to claim 4, characterized in that: The neural network model includes 3 hidden layers, and each hidden layer includes 128 neurons.

6. The method for obtaining the actual temperature of the aerosol-generating substrate in the heated smoking device by artificial intelligence according to claim 5, characterized in that: The hidden layer uses the ReLU activation function.

7. The method for obtaining the actual temperature of the aerosol-generating substrate in the heated smoking device according to claim 6, characterized in that: include: In the machine learning process, mean square error is selected as the loss function and Adam optimizer is selected as the optimizer.

8. The method for obtaining the actual temperature of the aerosol-generating substrate in the heated smoking device according to claim 7, characterized in that: After constructing the aerosol generating substrate actual temperature acquisition model, the method further includes: optimizing the aerosol generating substrate actual temperature acquisition model.

9. The method for obtaining the actual temperature of the aerosol generating substrate in the heated smoking device by artificial intelligence according to claim 8, characterized in that: Optimizing the actual temperature acquisition model of the aerosol generating matrix includes: hyperparameter tuning, model integration and model pruning.

10. A heated smoking device, comprising a shell, a heating body, a power source, a main control circuit, a first temperature sensor and a second temperature sensor, wherein the shell defines a heating chamber for receiving an aerosol generating substrate, the heating body is located in the heating chamber and can be in thermal contact with the received aerosol generating substrate, the first temperature sensor is closer to the heating body relative to the aerosol generating substrate, and the second temperature sensor is closer to the outer surface of the shell relative to the heating body, characterized in that: The main control circuit is configured to control the heating smoking device to obtain the actual temperature of the aerosol generating substrate according to the following steps: The method comprises: obtaining an actual temperature of the first temperature sensor and an actual temperature of the second temperature sensor; Inputting the actual temperature of the first temperature sensor and the actual temperature of the second temperature sensor into a pre-constructed aerosol generation substrate actual temperature acquisition model to obtain the aerosol generation substrate actual temperature; Among them, the aerosol generating matrix actual temperature acquisition model is constructed by machine learning with test data group pairs as training samples and the aerosol generating matrix actual temperature as the target. The test data group includes the first temperature sensor test temperature, the second temperature sensor test temperature and the aerosol generating matrix test temperature. The first temperature sensor is used to obtain the temperature at the first position, and the second temperature sensor is used to obtain the temperature at the second position.

Images