A system and method based on intelligent analysis of atomized liquid components

Abstract

The present invention belongs to the technical field of atomized liquid component analysis. The present invention discloses a system and method based on intelligent analysis of atomized liquid components, including: multi-parameter sampling of atomized liquid to obtain pyrolysis data; temperature gradient analysis to generate a pyrolysis spectrum; identification of trace markers to generate a feature library; analysis of component evolution trajectories to generate a dynamic evolution spectrum; partitioning and demarcation processing to generate an atomization characteristic spectrum; thermodynamic analysis to generate a reaction spectrum; analysis of component diffusion to generate quantitative data; construction of a sensory response prediction engine; intelligent analysis to generate response characteristic data; evaluation to generate a quality report; optimization processing to generate recipe suggestions and implement precise blending. This method achieves precise analysis and evaluation of atomized liquid components through multi-dimensional analysis, providing a scientific basis for product optimization and effectively balancing safety and taste experience.

Description

Technical Field

[0001] The present invention relates to the technical field of atomized liquid component analysis, and more specifically, to a system and method based on intelligent analysis of atomized liquid components. Background Art

[0002] With the emergence of a wide variety of aerosol liquid products, the complexity and diversity of their composition have also increased. Currently, aerosol liquid composition analysis primarily relies on traditional chemical analysis methods such as gas chromatography-mass spectrometry and high-performance liquid chromatography. While accurate and reliable, these methods have significant drawbacks. Traditional methods are unable to detect in real time the thermal degradation products generated during high-temperature aerosolization and their dynamic changes. When the product operates at different power levels and in different usage modes, the chemical components in the aerosol liquid undergo complex thermochemical reactions, generating a variety of intermediate and final products. The composition and content of these products differ significantly from the initial aerosol liquid. Traditional analytical methods often only provide static sample composition data and fail to capture dynamic changes during actual use. Existing technologies struggle to identify and quantify ultra-trace specific thermal degradation markers. Some specific fragrance ingredients (such as certain aldehydes and ketones) generate extremely trace, yet bioactive, specific degradation products during the aerosolization process. Although these products are present at extremely low levels, they may have unique physiological effects on the user. For example, vanillin may generate 4-methylguaiacol at a specific temperature. This substance can cause respiratory irritation reactions in specific people at extremely low concentrations. However, due to its extremely low content and co-elution with other components, it is difficult to accurately separate and quantify it using traditional analytical methods. Traditional analytical methods cannot establish an accurate correlation between the fingerprint of the nebulized liquid components and the subjective feelings of the user. Different users may have completely different subjective evaluations of the same nebulized liquid. This difference is partly due to differences in individual sensory sensitivity and partly due to the complex interactions of trace components in the nebulized liquid. The existing technology lacks a method that can correlate objective chemical analysis with the user's subjective experience, which leads to a large amount of blind and repetitive work in the development of nebulized liquid formulas.

[0003] In view of this, the present invention proposes a system and method based on intelligent analysis of atomized liquid components to solve the above problems. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned objectives, the present application provides a method for intelligent analysis of atomized liquid components, comprising:

[0005] Step S1: sampling the atomized liquid under multi-parameter atomization conditions to obtain raw pyrolysis data; performing temperature gradient deconstruction analysis based on the raw pyrolysis data to generate a multi-dimensional pyrolysis spectrum;

[0006] Step S2: performing precise identification of trace markers based on the multidimensional pyrolysis spectrum to generate a trace marker feature library; performing multidimensional component evolution trajectory analysis based on the trace marker feature library to generate a dynamic component evolution spectrum;

[0007] Step S3: performing partition demarcation processing on the atomization process according to the dynamic component evolution spectrum to generate a partition atomization characteristic spectrum; performing component-temperature interaction thermodynamic analysis based on the partition atomization characteristic spectrum to generate a component thermodynamic reaction map;

[0008] Step S4: Analyze the component diffusion dynamics based on the partitioned atomization characteristic spectrum to generate component diffusion quantitative data; construct a sensory response prediction engine based on the component thermodynamic reaction map to obtain a sensory response prediction engine; transmit the component diffusion quantitative data to the sensory response prediction engine in real time for personalized sensory experience intelligent analysis to generate sensory response characteristic data; perform global atomization liquid characteristics intelligent evaluation based on the sensory response characteristic data to generate an atomization liquid quality evaluation report; optimize the safety and taste balance of the original pyrolysis data based on the atomization liquid quality evaluation report to generate optimized formula recommendations, and perform precise preparation of the atomization liquid through the optimized formula recommendations.

[0009] The present application provides a system for intelligent analysis of atomized liquid components, which is used to perform the above-mentioned intelligent analysis method of atomized liquid components, including:

[0010] Multi-dimensional thermal spectrum construction module: multi-parameter atomization condition sampling of the atomized liquid is performed to obtain raw pyrolysis data; temperature gradient deconstruction analysis is performed based on the raw pyrolysis data to generate a multi-dimensional pyrolysis spectrum;

[0011] Dynamic trajectory analysis module: accurately identifies trace markers based on multi-dimensional pyrolysis spectra and generates a trace marker feature library; analyzes multi-dimensional component evolution trajectories based on the trace marker feature library and generates a dynamic component evolution spectrum;

[0012] Thermodynamic interaction modeling module: The atomization process is partitioned and demarcated according to the dynamic component evolution spectrum to generate the partitioned atomization characteristic spectrum; based on the partitioned atomization characteristic spectrum, the component-temperature interaction thermodynamic analysis is performed to generate the component thermodynamic reaction map;

[0013] Intelligent evaluation and optimization module: Analyze the component diffusion dynamics based on the partitioned atomization characteristic spectrum to generate component diffusion quantitative data; construct a sensory response prediction engine based on the component thermodynamic reaction map to obtain a sensory response prediction engine; transmit the component diffusion quantitative data to the sensory response prediction engine in real time for personalized sensory experience intelligent analysis to generate sensory response characteristic data; perform global intelligent evaluation of atomized liquid characteristics based on the sensory response characteristic data to generate an atomized liquid quality evaluation report; optimize the safety and taste balance of the original pyrolysis data based on the atomized liquid quality evaluation report to generate optimized formula recommendations, and execute precise blending of the atomized liquid through the optimized formula recommendations.

[0014] The technical effects and advantages of the intelligent analysis system and method of atomized liquid components of the present invention are as follows:

[0015] The present invention achieves a comprehensive characterization of the pyrolysis components produced during the atomization process through multi-parameter atomization condition sampling and temperature gradient deconstruction analysis, and can convert the complex pyrolysis process into a visual multi-dimensional pyrolysis spectrum, thereby enhancing the monitoring accuracy of the changes in the composition of the atomized liquid. By constructing a trace marker feature library and generating a dynamic component evolution spectrum, it can effectively identify and track ultra-trace active ingredients, achieve accurate quantitative analysis of bioactive substances, ensure that trace components with potential effects on human health are fully monitored, greatly improve the accuracy of the safety assessment of the atomized liquid, and avoid the problem of trace components being easily overlooked in traditional analysis methods. Through the partitioning and demarcation processing of the atomization process and the component-temperature interactive thermodynamic analysis, the system can fully grasp the thermodynamic behavior characteristics of the atomized liquid components in different temperature ranges, respond to the composition change laws under different atomization conditions, and improve the system's analytical ability and accuracy for complex thermodynamic changes in the atomization process through thermal degradation critical point analysis and molecular reorganization activity analysis. The introduction of reaction path network maps and nonlinear thermodynamic behavior analysis realizes multi-level information fusion of complex reaction systems, providing a theoretical basis for component safety assessment. By analyzing the ingredient diffusion dynamics and building a sensory response prediction engine, the diffusion characteristics of the atomized ingredients are analyzed in real time, the possible sensory experience is predicted, and the impact of different ingredient combinations on the user's senses is dynamically evaluated to ensure a high-quality sensory experience when the atomization conditions change. Based on the automated evaluation and analysis of individual perception difference data, personalized quality assessment reports can be automatically generated and the formula can be optimized according to the real-time atomization characteristics, thereby improving the system's ability to adapt to the needs of different users, significantly enhancing the user experience, and ensuring product safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Schematic diagram of an intelligent analysis method of atomized liquid components according to the present invention;

[0017] Figure 2This is a schematic diagram of an intelligent analysis system based on atomized liquid components according to the present invention. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] Example 1

[0020] See also Figure 1 As shown, this embodiment provides an intelligent analysis method based on the composition of atomized liquid, including: step S1: performing multi-parameter atomization condition sampling on the atomized liquid to obtain original pyrolysis data; performing temperature gradient deconstruction analysis based on the original pyrolysis data to generate a multi-dimensional pyrolysis spectrum;

[0021] Step S2: performing precise identification of trace markers based on the multidimensional pyrolysis spectrum to generate a trace marker feature library; performing multidimensional component evolution trajectory analysis based on the trace marker feature library to generate a dynamic component evolution spectrum;

[0022] Step S3: performing partition demarcation processing on the atomization process according to the dynamic component evolution spectrum to generate a partition atomization characteristic spectrum; performing component-temperature interaction thermodynamic analysis based on the partition atomization characteristic spectrum to generate a component thermodynamic reaction map;

[0023] Step S4: Analyze the component diffusion dynamics based on the partitioned atomization characteristic spectrum to generate component diffusion quantitative data; construct a sensory response prediction engine based on the component thermodynamic reaction map to obtain a sensory response prediction engine; transmit the component diffusion quantitative data to the sensory response prediction engine in real time for personalized sensory experience intelligent analysis to generate sensory response characteristic data; perform global atomization liquid characteristics intelligent evaluation based on the sensory response characteristic data to generate an atomization liquid quality evaluation report; optimize the safety and taste balance of the original pyrolysis data based on the atomization liquid quality evaluation report to generate optimized formula recommendations, and perform precise preparation of the atomization liquid through the optimized formula recommendations.

[0024] Preferably, step S1 includes:

[0025] Perform dynamic sampling of atomized liquid under multi-parameter atomization conditions to generate raw pyrolysis data;

[0026] Classify the pyrolysis product types based on the original pyrolysis data to generate pyrolysis product type characteristic data;

[0027] Perform multi-stage pyrolysis kinetic characterization based on the pyrolysis product type characteristic data to generate pyrolysis kinetic characterization data;

[0028] Temperature gradient deconstruction analysis is performed based on the pyrolysis kinetic characterization data to generate a multidimensional pyrolysis spectrum.

[0029] In this embodiment of the present invention, an adjustable atomizer is used with multiple atomization parameters, including atomization temperature (200°C-350°C, in 5°C increments), heating power (10W-50W, in 5W increments), and airflow rate (10-30 mL / s, in 2 mL / s increments). Dynamic sampling of the atomized liquid is performed. During the sampling process, pyrolysis products generated by the atomized liquid at different temperatures are captured using pyrolysis mass spectrometry and gas chromatography-mass spectrometry. Each temperature point is sampled for 30 seconds, with mass spectrometry data recorded every 5 seconds to generate time-series pyrolysis data. Auxiliary parameters such as resistance changes and power output fluctuations of the atomizer are also recorded to ensure precise control of sampling conditions. All collected data are stored in order of temperature and time, generating a raw pyrolysis dataset encompassing the complete pyrolysis process. The collected raw pyrolysis data is classified and processed, and the pyrolysis products are categorized into multiple categories based on molecular weight, structural characteristics, and chemical properties, including alcohols, aldehydes, ketones, acids, esters, aromatic compounds, and heterocyclic compounds. The characteristic ion fragments in the mass spectrometry data (such as mass-to-charge ratio: 31 for alcohols, mass-to-charge ratio: 43 for ketones, and mass-to-charge ratio: 91 for aromatics) were used for preliminary classification. The characteristic data of the pyrolysis product type were generated for each type of compound based on its frequency and intensity at different temperatures. For example, at 250°C, the characteristic peak intensity of the detected alcohol compounds was 12,000 counts, ketones were 8,500 counts, and aromatics were 3,200 counts; while at 300°C, these values became 9,500, 12,800, and 7,600 counts, respectively, indicating that the increase in temperature led to a decrease in alcohols and an increase in ketones and aromatic compounds. For each type of pyrolysis product, its formation kinetic parameters at different temperature stages were analyzed, including formation rate, peak temperature, half-life, etc. The apparent activation energy and pre-exponential factor of each type of compound were calculated using the Arrhenius equation to evaluate the temperature sensitivity of the pyrolysis reaction. For example, the apparent activation energy of alcohol compounds is 145 kJ / mol, and the pre-exponential factor is / s; while the apparent activation energy of ketones is 120kJ / mol, and the pre-exponential factor is / s, indicating that ketones are more readily formed at lower temperatures. Based on the thermodynamic data, a curve plotting the reaction rate constant versus temperature was constructed, comparing the relative production of various compounds at different temperatures to generate thermodynamic characterization data. Using temperature gradient deconstruction technology, the thermodynamic characterization data was converted into a three-dimensional spectrum, with the x-axis representing temperature (200°C to 350°C), the y-axis representing retention time or mass-to-charge ratio, and the z-axis representing signal intensity. Mathematical processing methods such as Fourier transform and wavelet transform were used to reduce noise and enhance resolution in the raw spectrum, extracting the fine structure of the temperature-response relationship. In the spectrum, color gradients or contour lines represent the relative abundance of each compound at different temperatures, with darker areas indicating higher abundance and lighter areas indicating lower abundance. Spectral data from each temperature point was integrated to generate a multidimensional thermodynamic spectrum covering the entire temperature range. This spectrum visually displays a complete profile of the composition of the atomized liquid as it changes with temperature, providing a comprehensive data foundation for subsequent trace marker identification.

[0030] Preferably, step S2 includes:

[0031] Extract pyrolysis intensity features based on multi-dimensional pyrolysis spectra to generate pyrolysis intensity spectrum data;

[0032] Extract molecular fragment spectrum features based on multi-dimensional pyrolysis spectra to generate molecular fragment spectrum data;

[0033] Using a preset multi-dimensional spectrum deconvolution algorithm to perform initial trace marker identification analysis on the pyrolysis intensity spectrum data and the molecular fragmentation spectrum data to generate initial trace marker data;

[0034] Perform low signal-to-noise ratio signal enhancement processing based on initial trace marker data to generate a trace marker feature library;

[0035] Multi-dimensional component evolution trajectory analysis is performed based on the trace marker feature library to generate a dynamic component evolution spectrum.

[0036] In this embodiment, total ion current (TIC) data (the sum of the mass-to-charge ratio) for each temperature point is extracted from a multidimensional pyrolysis spectrum to quantify the overall intensity of the pyrolysis reaction at different temperatures. TIC values for each temperature point (e.g., 200°C, 205°C, 210°C, and up to 350°C) are normalized to eliminate the effects of sample size variation. A smoothing algorithm is used to process the TIC curve to reduce the influence of random noise. The first and second derivatives of the TIC are calculated at each temperature point to identify inflection points and acceleration characteristics of the pyrolysis intensity change. These inflection points typically correspond to the initiation or completion temperatures of key chemical reactions, such as the onset temperature of thermal cracking of the main component or the temperature of complete pyrolysis. For example, a significant peak in the first derivative of the TIC at 270°C indicates that the pyrolysis reaction rate reaches its maximum at this temperature; a shift from positive to negative in the second derivative at 320°C indicates that the pyrolysis rate begins to slow. The pyrolysis intensity data for all temperature points are integrated to generate a pyrolysis intensity spectrum, which contains a quantitative relationship between atomization temperature and pyrolysis intensity. The mass spectrometric data from the multidimensional pyrolysis spectra were meticulously analyzed, extracting the signal intensity variations at different temperatures for each mass-to-charge ratio (MCR) to generate molecular fragmentation spectra. Characteristic molecular fragments, such as MCR: 31 (characteristic fragments of alcohols), MCR: 43 (characteristic fragments of ketones), and MCR: 91 (characteristic fragments of aromatics), were focused on, and their intensity profiles were recorded over the temperature range of 200°C to 350°C. Multivariate statistical methods such as principal component analysis (PCA) or non-negative matrix factorization (NMF) were used to reduce the data dimensionality and extract the main modes of variation. Peak shape analysis was performed on the temperature-intensity curves of each characteristic fragment, extracting parameters such as peak height, peak width, peak area, and peak asymmetry. These parameters characterize the pyrolysis behavior of the molecular fragments. For example, a MCR: 31 exhibits a broad peak with an area of 8500 units in the low-temperature range (200-250°C), while a sharp peak with an area decreasing to 3200 units in the high-temperature range (300-350°C) indicates rapid decomposition of alcohols at high temperatures. Data from all characteristic fragments are integrated to generate molecular fragmentation spectra, describing the fragmentation patterns of various molecular structures during pyrolysis. A pre-designed multidimensional spectral deconvolution algorithm, combining pyrolysis intensity spectrum data with molecular fragmentation spectrum data, is used to identify trace markers. Based on a nonlinear least-squares approach, the deconvolution algorithm can separate complex mixed spectra into the contributions of individual components and is particularly well-suited for processing overlapping spectral signals. Deconvolution parameters are set: the maximum number of iterations is 1000, the convergence threshold is 0.001, and a fitting residual error of less than 5% is considered effective. The algorithm processes the spectral data at each temperature point, extracting potential trace marker signals, even at extremely low levels (as low as 0.1% of the total signal). For each potential marker identified by deconvolution, its mass spectral signature is matched against a reference library, and a match score (0 to 100) is calculated. A match score exceeding 85% is considered a reliable identification.Through deconvolution, dozens of potential trace markers were identified from the raw spectra, including common flavoring ingredients such as nicotine, vanillin, and menthol, as well as several potential pyrolysis byproducts such as toluene, naphthalene, and benzopyrene. For each identified trace marker, its relative concentration change at different temperatures was recorded to generate initial trace marker data. Low signal-to-noise ratio signals in the initial trace marker data were enhanced to improve the accuracy of trace component detection. Adaptive wavelet transform was used to estimate and suppress background noise, using wavelet basis functions tailored to the noise characteristics of different frequency regions. A signal-to-noise ratio threshold of 3:1 was set, and signals below this threshold were enhanced through cumulative averaging (over 100 scans). Mathematical enhancement techniques, such as correlation integration and peak detection algorithms, were applied to particularly weak signals (with a signal-to-noise ratio below 2:1) to improve signal discernibility. Isotope ratio verification was used to verify the authenticity of trace marker signals by examining the isotopic peak ratios of elements such as carbon, hydrogen, and oxygen, eliminating instrument noise or background interference. For example, the 13C / 12C ratio of carbon should be close to 1.1%. Any significant deviation from the measured ratio is considered a suspicious signal. After signal enhancement, a trace marker signature library is constructed, containing each marker's retention time, mass spectral characteristics, characteristic ion fragments, occurrence patterns at different temperatures, and co-occurrence relationships with other components. Each marker entry in the signature library includes information such as a unique ID (e.g., MK001), chemical name, molecular formula, mass-to-charge ratio, maximum response temperature, and apparent activation energy. Based on this trace marker signature library, the concentration trajectory of each marker at different temperatures is tracked to analyze its generation, transformation, and disappearance patterns. For each key marker, a temperature-concentration curve is plotted, indicating its onset temperature, maximum concentration temperature, and complete disappearance temperature. Transformation relationships between markers are analyzed to identify precursor-product pairs. For example, if precursor A begins to decrease at 250°C while product B begins to increase, this indicates a possible A→B transformation. Multivariate statistical methods (such as cluster analysis and principal component analysis) were used to explore correlations between markers and identify co-variation groups, which often reflect common chemical reaction pathways. For example, cluster analysis revealed that markers MK003, MK007, and MK012 formed a co-variation group, with a synchronous increase between 270 and 290°C, suggesting that they may originate from the decomposition of a common precursor. The evolutionary trajectories of all markers were integrated to construct a complete dynamic composition evolution spectrum. This spectrum illustrates the evolution of each component in the atomized liquid from its initial state to complete pyrolysis, providing a data foundation for subsequent partitioning of the atomization process.

[0037] Preferably, low signal-to-noise ratio signal enhancement processing is performed based on the initial trace marker data to generate a trace marker feature library, comprising:

[0038] performing a bioactivity threshold analysis based on the initial trace marker data to generate bioactivity threshold data, wherein the bioactivity threshold data includes physiological response triggering concentration data and a marker specificity index;

[0039] Perform ultra-trace active ingredient enrichment processing based on physiological response triggering concentration data to generate ultra-trace active ingredient data;

[0040] Perform bioactive marker correlation processing based on ultra-trace active ingredient data and marker specificity index to generate bioactive marker data;

[0041] The initial trace marker data is subjected to marker screening processing according to the bioactive marker data to generate a trace marker feature library.

[0042] In this embodiment of the present invention, the bioactivity thresholds of each component in the initial trace marker data were analyzed to determine the lowest concentration level capable of eliciting a physiological response in humans. A database of bioactivity thresholds for each marker was established through literature review and experimental verification. For flavoring markers (e.g., vanillin and menthol), their taste threshold (the lowest concentration that can be perceived, typically in the ppb level) and odor threshold (the lowest concentration that can be detected by olfaction) were recorded. For example, the taste threshold for vanillin is 2 ppb, and the odor threshold is 0.5 ppb; the taste threshold for menthol is 5 ppb, and the odor threshold is 1 ppb. For potentially harmful markers (e.g., formaldehyde and benzo[a]pyrene), their health impact thresholds, including short-term exposure limits (STELs) and long-term exposure limits (TWAs), were recorded. For example, the STEL for formaldehyde is 2 ppm, and the TWA is 0.75 ppm; the TWA for benzo[a]pyrene is 0.2 μg / m³. Based on the chemical structure and receptor affinity of the marker, a marker specificity index was calculated to quantify its binding ability to specific biological receptors. The specificity index is scored on a scale of 0 to 100, with higher scores indicating greater specificity. For example, the specificity index for nicotine and nicotinic acetylcholine receptors is 95, the specificity index for menthol and TRPM8 cold receptors is 87, and the specificity index for vanillin and vanillin receptors is 76. Combining these data, bioactivity threshold data are generated, including the physiological trigger concentration and specificity index for each marker, providing a threshold reference for subsequent analysis. Based on the physiological trigger concentration data, ultra-trace components with concentrations below the detection limit but potentially bioactive are enriched. Solid-phase microextraction (SPME) is employed, using extraction tips of varying polarity (e.g., PDMS, carboxen, and DVB) to selectively adsorb the aerosol. The extraction conditions are: temperature 300°C, adsorption time 10 minutes, and desorption time 2 minutes. The pyrolysis products are condensed and enriched using a cold trap (-20°C), and then released gradually by programmed temperature ramping (5°C / min from -20°C to 300°C). High-sensitivity analysis of enriched samples using selected ion monitoring (SIM) mode increases instrument sensitivity by 100 times that of conventional full-scan mode. For example, for trace nicotine metabolites, characteristic ions with mass-to-charge ratios (m / s) of 84, 133, and 161 are monitored, achieving a detection limit of 0.1 ppb. Through these enrichment and detection techniques, a variety of ultra-trace active ingredients, including nicotine alkaloids, polycyclic aromatic hydrocarbons, and carbonyl compounds, are extracted from the initial marker data. These ingredients, despite their extremely low concentrations (typically at the ppb or ppt level), possess significant bioactivity due to their high specificity. Ultra-trace active ingredient data are generated, recording their precise concentrations and temperature ranges of occurrence. The ultra-trace active ingredient data are combined with the marker specificity index to analyze the bioactivity correlations between different markers. Multivariate correlation analysis and receptor simulation software are used to predict the binding patterns and synergistic effects between the markers and biological receptors.Marker combination models were developed for taste perceptions (e.g., sweet, sour, bitter, salty, and umami), olfactory perceptions (e.g., floral, woody, tobacco, and fruity), and other sensations (e.g., cool, warm, and spicy). For example, for the "cooling" experience, menthol (specificity index 87) and menthene (specificity index 65) exhibited a strong synergistic effect, with a 3:1 ratio producing the strongest cooling effect. Nicotine (specificity index 95) and certain pyrazine compounds (specificity indexes 73 to 82) exhibited complementary effects, enhancing the integrity of tobacco sensory profiles. Based on the results of the correlation analysis, bioactive marker data was generated, encompassing synergistic relationships, antagonistic relationships, and threshold shift effects between key markers, providing a basis for subsequent sensory experience prediction. Initial trace marker data was screened using the bioactive marker data to construct a comprehensive trace marker signature library. Screening rules were established: markers with a specificity index greater than 50 and concentrations above the physiological trigger concentration were retained; markers with a specificity index greater than 80 were retained even at concentrations slightly below the trigger concentration; and marker combinations with synergistic effects were retained, even if individual components did not meet the criteria, as long as the combined effect was significant. These rules were used to identify markers with significant impacts on the sensory properties and physiological effects of the aerosol solution, while eliminating interfering components with insignificant bioactivity. The retained markers were classified and cataloged, establishing a hierarchy based on chemical structure (e.g., alcohols, ketones, aldehydes) and bioactivity type (e.g., aroma contribution, flavor contribution, sensory stimulation, etc.). Each marker was assigned a unique signature code, recording detailed information such as its chemical properties, spectral characteristics, and bioactivity parameters. This ultimately generated a trace marker signature library containing dozens to hundreds of trace components with significant impacts on aerosol solution properties, providing a comprehensive marker reference system for subsequent analysis.

[0043] Preferably, step S3 includes:

[0044] Perform partitioning and demarcation processing on the atomization process according to the dynamic component evolution spectrum to generate the partitioned atomization characteristic spectrum;

[0045] Detect the thermodynamic stability limit based on the partitioned atomization characteristic spectrum and generate thermodynamic stability data;

[0046] Composition-temperature interaction thermodynamic analysis is performed based on thermodynamic stability data to generate a composition thermodynamic reaction map.

[0047] In an embodiment of the present invention, the dynamic component evolution spectrum is analyzed to find the key transition points in the atomization process, and the entire atomization process is divided into several characteristic areas. An inflection point detection algorithm (such as the second-order derivative method) is used to identify mutation points in the component spectrum, which usually correspond to changes in the reaction mechanism. Through cluster analysis, the natural partitioning of the atomization process is determined, for example: preheating zone (200, 230] ℃: the main components remain stable, and only a small amount of volatile components are released; mild pyrolysis zone (230, 270] ℃: low-boiling point components evaporate rapidly, and some compounds begin to pyrolyze; moderate pyrolysis zone (270, 310] ℃: the main components are pyrolyzed in large quantities to generate a variety of secondary pyrolysis products; deep pyrolysis zone (310, 350] ℃: high molecular weight polycyclic compounds and low molecular weight fragments are formed, and more harmful substances are produced at the same time. Calculate the characteristic parameters of each area, such as the rate of change of the main components, the number of types of newly generated substances, the pyrolysis intensity, etc. For example, in the mild In the pyrolysis zone, 27 newly generated substances were detected, and the degradation rate of the main components was 15%; in the deep pyrolysis zone, the number of newly generated substances increased to 56, and the degradation rate of the main components reached 85%. The component spectrum of each partition was reconstructed to highlight the characteristic components and transformation relationships of the area, and a partitioned atomization characteristic spectrum was generated, which included the typical component spectrum and transformation characteristics of each temperature range. Thermodynamic stability analysis was performed on each temperature range in the partitioned atomization characteristic spectrum to determine the stability boundaries of each component at different temperatures. The thermal degradation starting temperature (T-onset) and complete degradation temperature (T-complete) of each key component were determined to construct a temperature-stable The thermal stability curve is shown in Figure 1. For example, the T-onset of glycerol is 240°C and the T-complete is 275°C; the T-onset of propylene glycol is 225°C and the T-complete is 260°C. The relationship between the thermal degradation rate constant k and the temperature T is calculated, and the activation energy Ea is determined using the Arrhenius equation. A higher activation energy indicates greater thermal stability. For example, the Ea of glycerol is 125 kJ / mol, and that of propylene glycol is 115 kJ / mol. Based on the results of thermodynamic analysis, a safe atomization temperature range is determined for each component, i.e., within this temperature range, the component remains stable or degrades in a controlled manner without producing harmful byproducts. Generate thermodynamic stability data, including stability parameters, safe temperature ranges, and degradation kinetic constants for each component. Based on this thermodynamic stability data, analyze the interactions and temperature dependence of different components during pyrolysis. Construct a component-temperature interaction matrix to quantify the strength of the interaction between each pair of components at different temperatures. For example, the interaction strength between glycerol and nicotine at 250°C is 0.75 (a positive value indicates mutual promotion of degradation), while the interaction strength between glycerol and hydroxymethylfurfural at the same temperature is -0.4 (a negative value indicates mutual inhibition of degradation). Using pyrolysis mass spectrometry, track the decomposition paths of each component in the mixture and identify pyrolysis intermediates and final products.For example, at 270°C, the concentration of propionaldehyde produced by the mixture of glycerol and propylene glycol is 30% higher than when they are pyrolyzed separately, indicating a synergistic decomposition effect between the two. Draw a reaction energy diagram, indicate the activation energy and reaction heat of the key transformation steps, and help understand the reaction rate and product distribution. For example, the energy barrier of nicotine N-oxidation reaction is 110kJ / mol, while the energy barrier of demethylation reaction is 130kJ / mol, indicating that N-oxidation is the dominant reaction pathway at lower temperatures. Integrating the above thermodynamic analysis results, a component thermodynamic reaction map is generated, which details the decomposition pathways, interactions and product distribution of each component in the atomized liquid at different temperatures, providing a thermodynamic basis for subsequent sensory response predictions.

[0048] Preferably, thermodynamic stability limit detection is performed based on the partitioned atomization characteristic spectrum to generate thermodynamic stability data, including:

[0049] Perform thermal degradation critical point analysis based on the partitioned atomization characteristic spectrum to generate thermal degradation critical point data;

[0050] Perform molecular recombination activity analysis based on the partitioned atomization characteristic spectrum to generate molecular recombination activity data;

[0051] The composition-temperature phase transition matrix is constructed based on the thermal degradation critical point data and the molecular reorganization activity data to generate the composition-temperature phase transition matrix;

[0052] Thermodynamic stability limits are detected based on the composition-temperature phase transition matrix to generate thermodynamic stability data.

[0053] In an embodiment of the present invention, the thermal degradation behavior of each main component in the partitioned atomization characteristic spectrum is analyzed to determine the critical transition point of its thermal stability. For each key component, the three key temperature points of its thermal degradation are determined by temperature scanning experiments: thermal degradation onset temperature (T-onset), that is, the temperature at which the component begins to decompose significantly; thermal degradation acceleration temperature (T-acceleration), that is, the temperature at which the decomposition rate increases significantly; thermal degradation complete temperature (T-complete), that is, the temperature at which the component is completely decomposed. These temperature points are accurately measured using differential thermogravimetric analysis (DTG) and thermogravimetric-mass spectrometry (TG-MS) with a resolution of ±1°C. For example, for glycerol, T-onset = 245°C, T-acceleration = 262°C, and T-complete = 278°C were measured; for propylene glycol, T-onset = 232°C, T-acceleration = 248°C, and T-complete = 265°C were measured. The thermal degradation kinetic parameters of each component near the critical point are calculated, including the zero-order reaction rate constant , first-order reaction rate constant and activation energy Ea. For example, glycerol at T-acceleration temperature =0.028 / s, Ea=125kJ / mol. Plot the thermal degradation rate-temperature curve, mark the critical point position, and calculate the temperature gradient and reaction gradient between each critical point. Generate thermal degradation critical point data, including the critical temperature value of each component, the corresponding degradation rate and kinetic parameters. Study the recombination activity of molecular fragments during the atomization process, and analyze the ability of thermal decomposition products to form new compounds. Use high-resolution mass spectrometry technology to monitor the newly generated molecular ions during the atomization process, especially focusing on compounds with molecular weights greater than the original components, which usually come from the recombination reactions of molecular fragments. For example, at 280°C, a newly generated substance with a mass-to-charge ratio of 284.1362 was detected, which was determined by accurate mass analysis to be a product formed by aldol condensation of two aldehyde fragments. For each temperature point, calculate the molecular recombination index (MRI), which is defined as the ratio of the number of newly generated compounds to the number of original degradation products. For example, at 250°C, the MRI is 0.15, indicating that approximately 15 out of every 100 original degradation products participate in the recombination reaction; at 300°C, the MRI rises to 0.42, indicating that high temperature promotes molecular recombination activity. Isotope labeling experiments (e.g., using 13C-labeled components) can track the migration and recombination pathways of specific molecular fragments and identify key recombination mechanisms. For example, 13C-labeling experiments showed that at 290°C, the rate of addition reactions of propylene glycol carbon chain fragments with aromatic compounds is 2.3 times higher than direct oxidation. Molecular recombination activity data are generated, including the molecular recombination index, the main recombination reaction types, and the relative abundance of recombination products at different temperatures. Combining the thermal degradation critical point data with the molecular recombination activity data, a composition-temperature phase transition matrix is constructed to describe the phase transition behavior of each component in the atomized liquid under varying temperatures. The rows of the matrix represent different components, the columns represent temperature points (ranging from 200°C to 350°C, with a step size of 5°C), and the matrix elements represent the phase transition state encoding under corresponding conditions. Phase transition states are coded as follows: 0 indicates a stable state (no significant change); 1 indicates mild decomposition (<10% degradation); 2 indicates moderate decomposition (10% to 50% degradation); 3 indicates severe decomposition (50% to 90% degradation); and 4 indicates complete decomposition (>90% degradation). For each temperature point, a total phase transition index (TPI) is calculated as the weighted average of the phase transition states of all components. For example, at 270°C, glycerin has a phase transition state of 3 (severe decomposition), propylene glycol has a phase transition state of 4 (complete decomposition), and nicotine has a phase transition state of 2 (moderate decomposition), resulting in a total TPI of 3.1, indicating a severe decomposition state at this temperature. Cluster analysis identifies groups of components with similar behavior within the phase transition matrix, providing a basis for formulation optimization. A component-temperature phase transition matrix is generated to comprehensively describe the phase transition behavior of each component in the aerosol liquid as it changes temperature. Based on this component-temperature phase transition matrix, the thermodynamic stability of the aerosol liquid is comprehensively assessed and its stability limits are determined.Key stability indicators are defined: safe vaporization temperature range (SVTR), the temperature range over which each major component remains stable or undergoes controlled degradation; critical instability temperature (CIT), the lowest temperature at which the system begins to produce significant harmful substances; and optimal vaporization temperature (OVT), the temperature at which optimal sensory properties are released within a safe range. These indicators are calculated based on phase change matrix data. For example, a formulation has an SVTR of 235–265°C, a CIT of 282°C, and an OVT of 255°C. A thermal stability coefficient (TSC) is calculated for each component, defined as the percentage of temperatures within the SVTR range at which the component remains stable (phase transition state ≤ 2). For example, a TSC of 0.75 for glycerol indicates that the component remains relatively stable at 75% of the temperature range. Synergistic effects between different components are analyzed for thermal stability; some components may exhibit higher or lower stability in a mixed state compared to their individual states. For example, experimental data shows that adding 5% glycerol esters can increase the T-onset of propylene glycol by 8°C, demonstrating positive synergy. Generate thermodynamic stability data, including the safe temperature range of the atomized liquid, the stability coefficient of each component, the unstable critical point, and the synergistic effect parameters between components, laying the foundation for subsequent thermodynamic analysis and sensory prediction.

[0054] Preferably, a composition-temperature interaction thermodynamic analysis is performed based on the thermodynamic stability data to generate a composition thermodynamic reaction map, including:

[0055] Perform multi-dimensional thermodynamic parameter extraction on thermodynamic stability data to generate a thermodynamic parameter set including Gibbs free energy change rate, activation energy distribution, and entropy change gradient;

[0056] Construct component interaction potential energy field based on thermodynamic parameter set to generate component interaction potential energy data;

[0057] Performing temperature sensitivity classification processing on component interaction potential energy data to generate temperature sensitivity classification data;

[0058] Nonlinear thermodynamic behavior analysis is performed based on temperature sensitivity classification data to generate nonlinear thermodynamic behavior data including critical phase transition points, reaction path weights, and energy barrier heights;

[0059] Perform temperature-component synergistic effect analysis on nonlinear thermodynamic behavior data and generate a temperature-component synergistic effect matrix;

[0060] Conduct micro-thermodynamic equilibrium analysis based on the temperature-component synergistic effect matrix to generate micro-equilibrium data;

[0061] Tracing thermodynamic reaction paths based on microscopic equilibrium data to generate reaction path network maps;

[0062] The reaction path network map and nonlinear thermodynamic behavior data are subjected to multi-level information fusion processing to generate a component thermodynamic reaction map.

[0063] In an embodiment of the present invention, key thermodynamic parameters were extracted from thermodynamic stability data to construct a thermodynamic framework for describing the pyrolysis process of the atomized liquid. The Gibbs free energy change rate of each major component at each temperature point was calculated to characterize the temperature sensitivity and spontaneity of the reaction. For example, in the range of 250 to 260°C, the Gibbs free energy change rate of glycerol was -2.3 kJ / (mol·K), indicating that increasing temperature significantly promoted the spontaneity of its decomposition reaction. By measuring reaction rates at different temperatures and combining them with the Arrhenius equation, the activation energy distribution of each major reaction was calculated, and an activation energy spectrum was plotted. The change in reaction equilibrium constant at different temperatures was calculated using the van't Hoff equation to assess the effect of temperature on the equilibrium. A thermodynamic parameter set containing comprehensive thermodynamic descriptive parameters was generated, providing a theoretical basis for subsequent analysis. Based on the extracted thermodynamic parameter set, a potential energy field model was constructed to describe the interactions between the components in the atomized liquid. Molecular simulation techniques (such as molecular dynamics simulation or Monte Carlo simulation) were used to calculate the interaction energies between the components in the mixed system, including hydrogen bonding, van der Waals forces, and electrostatic interactions. For example, simulations revealed that the average interaction energy between glycerol and aromatic compounds is -18.5 kJ / mol (negative values indicate attractive interactions). A component-component interaction matrix was constructed to quantify the interaction strength between each pair of components and identify strong interactions (interaction energy < -25 kJ / mol) and weak interactions (interaction energy > -10 kJ / mol). The effect of temperature on the interaction potential energy was analyzed, and the rate of change of interaction energy with temperature was calculated. For example, the interaction energy between nicotine and propylene glycol decreased at a rate of 0.15 kJ / (mol·°C) between 230 and 290°C, indicating that increasing temperature weakens the interaction between them. Based on quantum chemical calculations, the breakage and formation energies of key chemical bonds were analyzed to predict the dominant reaction pathways during pyrolysis. Component interaction potential energy data were generated to describe the energy relationship network and temperature dependence of the components in the aerosol liquid. The temperature sensitivity of the component interaction potential energy data was evaluated, and the components were ranked according to the temperature sensitivity of their pyrolysis behavior. The temperature sensitivity grading standard is set as follows: Level 1 (low sensitivity): within a temperature change range of 10°C, the thermal decomposition rate changes by <10%; Level 2 (medium sensitivity): within a temperature change range of 10°C, the thermal decomposition rate changes by [10%, 30%]; Level 3 (high sensitivity): within a temperature change range of 10°C, the thermal decomposition rate changes by [30%, 50%]; Level 4 (extremely high sensitivity): within a temperature change range of 10°C, the thermal decomposition rate changes by >50%. Each major component is evaluated. For example, the temperature sensitivity of propylene glycol in the range of 240 to 250°C is Level 3 (high sensitivity), while it drops to Level 1 (low sensitivity) in the range of 270 to 280°C, indicating that its decomposition at high temperatures is almost complete.The overall temperature sensitivity index (TSI) is calculated for each temperature point, taking the weighted average of the sensitivity of each component. For example, at 245°C, the overall TSI is 2.8, indicating that the atomized liquid is highly sensitive to temperature changes at this temperature. Transition points in temperature sensitivity are analyzed; these typically correspond to the onset temperatures of key thermochemical reactions. Temperature sensitivity grading data is generated, providing a basis for precise control of atomization temperature. Based on this temperature sensitivity grading data, the nonlinear thermodynamic behavior of the atomized liquid during pyrolysis is analyzed in depth. Critical phase transition points are identified on the temperature-reaction rate curve; these points indicate a sudden change in the reaction mechanism or a shift in the dominant reaction. For example, in a certain formulation, 260°C was identified as a critical critical point. Below this temperature, unimolecular decomposition reactions primarily occur, while above this temperature, free radical chain reactions dominate. Through kinetic modeling, weight coefficients for different reaction pathways are calculated to quantify the contribution of each pathway to the overall pyrolysis process. For example, at 270°C, the decomposition pathways of propylene glycol include dehydration (weight 0.45), C-C bond cleavage (weight 0.30), and oxidation (weight 0.25). Determine the energy barrier heights for key reactions and predict the effects of temperature changes on reaction selectivity. For example, the energy barrier for propylene glycol dehydration is 112 kJ / mol, while the energy barrier for C-C bond cleavage is 125 kJ / mol, indicating that dehydration is more dominant at low temperatures. Analyze nonlinear characteristics of reaction kinetics, such as autocatalysis, reaction inhibition, and induction period phenomena. Generate nonlinear thermodynamic behavior data, including information such as critical phase transition points, reaction path weights, and energy barrier heights, to reveal the complex kinetics of atomized liquid pyrolysis. Analyze the synergistic effects between temperature changes and component interactions to investigate the synergistic behavior of different component combinations under specific temperature conditions. Construct a temperature-component synergy matrix, where rows represent different temperature points, columns represent component pairs, and matrix elements represent synergy coefficients. The synergy coefficient is defined as the ratio of the actual reaction rate in the mixed state to the sum of the reaction rates of the individual components. A value greater than 1 indicates positive synergy (promoting effect), while a value less than 1 indicates negative synergy (inhibiting effect). For example, the synergy coefficient for glycerol and glycerides at 255°C is 1.35, indicating that the decomposition rate of glycerol increases by 35% when mixed. Through experiments and modeling, we identify the temperature ranges and ingredient combinations with the most synergistic effects. For example, studies have found that nicotine and certain aldehydes exhibit the strongest negative synergy between 265 and 275°C (synergy coefficient 0.72), reducing aldehyde formation. We analyze the molecular mechanisms underlying these synergistic effects, such as free radical capture and stabilization of reaction intermediates. We generate a temperature-ingredient synergy matrix to inform formulation optimization. Based on this temperature-ingredient synergy matrix, we analyze the microscopic thermodynamic equilibrium state of each component in the atomized liquid. For each temperature point, we calculate the total free energy and chemical potential distribution of the system to determine the thermodynamic equilibrium conditions.For example, at 258°C, the total free energy of the system reaches a local minimum of -1245 kJ / mol, indicating that the system is approaching a stable state at this temperature. The activity and fugacity coefficients of each component at different temperatures are analyzed to assess their degree of deviation from thermodynamic equilibrium. For example, in the high-temperature region (>300°C), the activity coefficient of nicotine increases from 1.15 to 1.85, indicating that its thermodynamic behavior deviates increasingly from the ideal solution. Phase equilibrium theory is used to analyze the partitioning behavior of each component in the liquid and gas phases, calculating the temperature-dependent variation of the partition coefficient. For example, the gas-liquid partition coefficient of vanillin is 2.8 at 240°C, but increases to 4.5 at 280°C, indicating that high temperatures promote its migration into the gas phase. Microequilibrium data are generated to describe the thermodynamic equilibrium characteristics of each component in the atomized liquid at different temperatures. Based on this microequilibrium data, the thermodynamic reaction paths of each component in the atomized liquid are traced, and a complete reaction network is constructed. Reaction flow analysis techniques are used to identify primary and secondary reaction pathways and calculate the fluxes of each pathway. For example, analysis shows that propylene glycol degrades primarily through a pathway involving dehydration to propionaldehyde and subsequent oxidation to propionic acid, accounting for 65% of the total degradation flux. A reaction pathway diagram was constructed, connecting all major components and intermediates into a network, with arrow width representing the magnitude of the reaction flux. For example, the glycerol pathway diagram shows that the flux of the decomposition pathway to formaldehyde is 3.2 times that of the decomposition pathway to propionaldehyde. Thermodynamically controlled pathways (determined by the Gibbs free energy change) and kinetically controlled pathways (determined by the activation energy) were labeled to analyze the effect of temperature on pathway selection. For example, at low temperatures (<250°C), glycerol decomposition is primarily kinetically controlled, preferentially forming products with low activation energy; whereas at high temperatures (>280°C), it is primarily thermodynamically controlled, preferentially forming products with smaller Gibbs free energy changes. A reaction pathway network diagram was generated to visually demonstrate the reaction transformation relationships during the pyrolysis of the atomized liquid. The reaction pathway network diagram was combined with nonlinear thermodynamic behavior data to construct a comprehensive component thermodynamic reaction map using multi-level information fusion technology. A hierarchical data structure is used to link macroscopic reaction behaviors (such as pyrolysis rate and product distribution) with microscopic reaction mechanisms (such as bond cleavage mechanisms and free radical reactions). Four dimensions of information are integrated into the map: temperature, composition, reaction type, and time, creating a multidimensional visualization. Reaction type is indicated by color coding (e.g., blue for hydrolysis, red for oxidation), line thickness indicates reaction rate, and node size indicates substance content. Thermodynamic stability data is integrated to indicate the safe temperature range and critical instability point for each component. For example, the safe temperature limit for glycerol is marked on the map as 255°C; exceeding this temperature will rapidly generate harmful aldehydes. An interactive design allows on-demand viewing of detailed information for specific temperatures, components, or reaction types. A component thermodynamic reaction map is generated, comprehensively displaying the thermodynamic behavior and reaction network of the atomized liquid across the entire temperature range, providing a theoretical basis for subsequent sensory response prediction and formulation optimization.

[0064] Preferably, step S4 includes:

[0065] Analyze the component diffusion dynamics based on the partitioned atomization characteristic spectrum to generate component diffusion quantitative data;

[0066] Conduct multi-dimensional sensory impact factor evaluation based on the component thermodynamic reaction map, generate sensory impact factor data, and design perception response decision nodes based on the sensory impact factor data;

[0067] Based on the preset sensory feature extraction algorithm, a neural perceptual mapping is constructed to associate sensory experience with physiological response for sensory response decision nodes and component thermodynamic reaction maps, thereby obtaining a sensory response prediction engine.

[0068] The component diffusion quantification data is transmitted in real time to the sensory response prediction engine for intelligent analysis of personalized sensory experience to generate sensory response feature data;

[0069] Perform intelligent evaluation of global atomized liquid characteristics based on sensory response characteristic data and generate an atomized liquid quality assessment report;

[0070] Based on the atomized liquid quality assessment report, the original pyrolysis data is optimized for safety and taste balance, and optimized formula recommendations are generated. The atomized liquid is then precisely formulated using the optimized formula recommendations.

[0071] In this embodiment of the present invention, the diffusion dynamics of atomized liquid components in the vapor phase were analyzed based on the partitioned atomization characteristic spectrum. Pulsed laser-induced fluorescence (PLIF) technology was used to measure the spatial distribution and diffusion rate of flavoring molecules in the vapor phase. Experimental conditions were as follows: an atomization temperature range of 200 to 350°C, in 10°C intervals; a measurement distance of 0 to 10 cm, in 1 cm intervals; and a sampling time of 0 to 5 seconds, in 0.5 second intervals. Based on the measured data, the diffusion coefficient D of each key component was calculated. This coefficient characterizes the diffusion rate of the molecule in the vapor phase. For example, at 280°C, the diffusion coefficient of nicotine was 0.082 cm² / s, and the diffusion coefficient of vanillin was 0.065 cm² / s. The Fick diffusion equation was used to simulate the concentration profiles of each component at different temperatures and times. For example, simulations showed that nicotine diffused to a distance of 4.5 cm in 2 seconds, while vanillin diffused only to 3.8 cm. The half-life distance (the distance required for the concentration to drop to half of its initial value) and time to peak (the time required to reach maximum concentration) of each component were calculated. For example, the half-life distance of nicotine is 3.2 cm and the time to peak is 1.8 seconds, while the half-life distance of menthol is 2.5 cm and the time to peak is 1.2 seconds, indicating that the minty sensation is perceived more quickly but lasts shorter. The influence of temperature on diffusion behavior was analyzed in conjunction with atomization profile data. For example, experimental data showed that the diffusion coefficient of nicotine increased by approximately 8% with every 10°C increase in temperature, while the diffusion coefficient of vanillin increased by approximately 12%. Quantitative component diffusion data was generated, including the diffusion parameters and spatiotemporal distribution characteristics of each component at different temperatures, providing a kinetic basis for sensory experience analysis. Based on the component thermodynamic reaction maps, the impact of each component in the atomized liquid on the sensory experience was comprehensively evaluated. A multi-dimensional assessment was conducted across the five basic tastes (sweet, sour, bitter, salty, and umami), the three-dimensional aroma (top, middle, and base notes), and tactile properties (such as irritation, cooling, warmth, and smoothness). A database of sensory impact parameters for each ingredient was established through human sensory testing and literature data. For example, the sweetness impact coefficient of vanillin was 0.85 (out of a maximum score of 1.0), the top note aroma impact coefficient was 0.65, and the pungency impact coefficient was 0.15. Menthol's cooling impact coefficient was 0.92, and the middle note aroma impact coefficient was 0.78. Multiple regression analysis was used to quantify the nonlinear relationship between ingredient concentration and sensory intensity, and a mathematical model was established. For example, the sweetness perception model for a particular spice ingredient is: sweetness = 0.85 × log(concentration) + 0.23, indicating that perceived intensity is proportional to the logarithm of concentration. Sensory synergies and antagonisms between ingredients were analyzed. For example, experimental data showed that menthol's cooling sensation was enhanced by 25% when combined with certain sweeteners, while it was reduced by 15% when combined with certain sour agents. Based on the evaluation results, sensory impact factor data was generated, and corresponding sensory response decision nodes were designed. Each decision node represents a specific sensory attribute (such as "cooling sensation", "sweetness", "throat feeling", etc.) and contains the influence weights of relevant components and threshold judgment rules.For example, the rule for the "cooling" decision node is: If the menthol concentration is >5 ppm and the ambient temperature is <28°C, the cooling score is 0.8 times the menthol concentration coefficient plus 0.2 times the temperature coefficient. Based on a pre-set sensory feature extraction algorithm and designed sensory response decision nodes, a neural perceptual mapping model is constructed to link the composition of aerosol liquids with human sensory experience. Using deep learning architectures such as multi-layer perceptrons (MLPs) or convolutional neural networks (CNNs), the neural network model is trained using component data from thermodynamic reaction maps as input and sensory experience scores as output. The neural network model's structural design is as follows: the input layer contains n neurons, corresponding to the concentrations of n key components and their interaction parameters; the hidden layer uses a three-layer structure, with each layer containing 64, 32, and 16 neurons, respectively, using the ReLU activation function; and the output layer contains m neurons, corresponding to the intensity scores of m sensory attributes. The model is trained using historical sensory evaluation data, including composition data and corresponding human sensory scores for 300 different aerosol liquid formulations at different temperatures. 80% of the data was used for training and 20% for validation, using the mean squared error (MSE) as the loss function and the Adam optimizer for parameter optimization. During training, emphasis was placed on identifying components with low concentrations but high sensory impact, ensuring the model's ability to accurately predict the sensory contributions of trace components. After training and validation, the model achieved 87% prediction accuracy on the test set, with an average sensory intensity prediction error of less than 0.15 (out of a maximum score of 1.0). After the neural perceptual mapping was completed, it was integrated into a sensory response prediction engine, which predicts corresponding sensory experience characteristics in real time based on the input component data. The generated component diffusion quantification data is fed into the sensory response prediction engine in real time for personalized sensory experience analysis. High-speed caching technology is used for data transmission, ensuring millisecond-level response speeds to meet real-time analysis requirements. The prediction engine calculates sensory experience changes over time based on the input component diffusion dynamics data and environmental parameters such as temperature and airflow. For example, the engine can predict that at an atomization temperature of 280°C, the primary sensations from inhalation will be the cooling sensation and top notes of mint within 0.5 seconds, the sweetness and middle notes within 1 to 2 seconds, and the tobacco aroma and base notes within 2 to 4 seconds. The engine can adjust prediction parameters for different user groups (e.g., age, gender, smoking history, etc.) to provide personalized sensory experience predictions. For example, for long-term smokers, the engine will appropriately increase the threshold parameters for certain flavors to reflect changes in their sensory sensitivity. Sensory response feature data is generated, including a sensory intensity curve over time, a sensory attribute distribution map, and a satisfaction prediction index, providing a basis for atomizer liquid quality assessment. Based on this sensory response feature data, a comprehensive quality assessment of the atomizer liquid is conducted, taking into account safety, sensory experience, and stability of use.A quality assessment system for aerosol liquids was designed, encompassing the following core indicators: Safety Index (SI): This assesses the generation of hazardous substances during the aerosolization process, with a maximum score of 100, with scores below 60 indicating unsafe, between 60 and 80 indicating basic safety, and above 80 indicating high safety. Sensory Balance (SB): This assesses the coordination of sensory attributes, with a maximum score of 100, with scores below 60 indicating unbalanced, between 60 and 80 indicating basic harmony, and above 80 indicating high harmony. Durability (DS): This assesses the duration and stability of the sensory experience, with a maximum score of 100. Thermal Sensitivity (TS): This assesses the sensitivity of the aerosol liquid to temperature changes, with lower scores being better. Overall Satisfaction (OS): This provides an overall evaluation based on a weighted average of the above indicators. Aerosol liquid samples were assigned a score based on predefined scoring rules. For example, a sample received an SI score of 75 (basic safety), an SB score of 82 (highly balanced), a DS score of 68 (moderately durable), a TS score of 42 (moderately sensitive), and an OS score of 72 (good). The scoring results were analyzed to identify key areas requiring improvement. For example, the sample's low TS score indicates that it is sensitive to temperature fluctuations and requires the addition of a stabilizer to improve heat resistance. A quality assessment report for the atomized liquid is generated, including detailed scoring data, improvement suggestions, and a recommended temperature range. Based on the atomized liquid quality assessment report, optimization strategies are proposed for safety risks and sensory balance issues identified in the original pyrolysis data. Regarding safety issues, the pyrolysis data is analyzed for the formation temperatures and precursors of hazardous substances (such as aldehydes and polycyclic aromatic hydrocarbons) and risk mitigation methods are proposed. For example, if a significant amount of glycerol is converted to acrolein above 280°C, it is recommended to reduce the glycerol content and add antioxidants, or to control the recommended atomization temperature below 270°C. Regarding sensory balance issues, the proportional relationship between various sensory attributes is analyzed and adjustment plans are proposed. For example, if the cooling sensation is too strong and the sweetness is insufficient, it is recommended to reduce the menthol content and increase the proportion of vanillin or maltitol. Based on the thermodynamic reaction map, the effects of various optimization plans are predicted and the optimal one is selected. For example, simulations predict that adding 2% antioxidants to the base formula and adjusting the glycerol / propylene glycol ratio from 30:70 to 20:80 can increase the SI by 12 points, while only slightly affecting the sensory balance (SB reduction of 3 points). Optimized formulation recommendations are generated, including specific ingredient adjustment ratios, recommended atomization temperature ranges, and expected results. Based on these recommendations, the atomizer liquid formulation is adjusted to achieve precise blending. Formulation adjustments can be implemented, such as adding antioxidants (such as vitamin E, at a concentration of 0.5% to 2%) to reduce aldehyde formation; adding stabilizers (such as glycerides, at a concentration of 1% to 5%) to improve thermal stability; and adjusting the base solvent ratio (such as glycerol / propylene glycol from 30:70 to 20:80) to optimize safety and taste. Flavoring adjustments can also be made, such as reducing the menthol content (from 1.2% to 0.8%) to reduce the excessive cooling effect; and increasing the vanillin content (from 0.5% to 0.7%) to enhance sweetness.Nicotine concentration is controlled within a safe range, and the most stable nicotine salt form is selected based on thermodynamic data. The adjusted formula is then tested in small batches to verify its performance meets expectations. If satisfactory, the atomizer liquid is precisely blended and the final formula is finalized. If any deficiencies remain, the formula is further fine-tuned until the target quality standard is achieved.

[0072] Preferably, a neural perceptual mapping is constructed based on a preset sensory feature extraction algorithm to associate sensory experience with physiological response for sensory response decision nodes and component thermodynamic reaction maps, so as to obtain a sensory response prediction engine, including:

[0073] Based on the preset sensory feature extraction algorithm, a neural perception mapping architecture is designed to associate sensory experience with physiological response at the sensory response decision node, generating a sensory response prediction framework.

[0074] Conduct individual perception difference analysis based on component thermodynamic reaction maps to generate individual perception difference data;

[0075] Based on individual perception difference data, the sensory response prediction framework is used to learn the correlation weights between sensory experience and physiological response to generate a sensory response prediction engine.

[0076] In this embodiment of the present invention, a neural perceptual mapping architecture is constructed to connect the composition of atomized liquid with human sensory experience, based on a pre-designed sensory feature extraction algorithm. This sensory feature extraction algorithm includes the following core components: a multidimensional sensory feature quantization module, which encodes various sensory attributes (such as sweetness, sourness, cooling sensation, and irritation) into standardized numerical vectors; a temporal sensory experience model, which describes the temporal evolution of sensory experience; and a component-sensor association matrix, which records the contribution weight of each component to different sensory attributes. A layered neural network architecture was designed, including the following: an input layer, which receives parameters such as ingredient concentration, temperature, and time. The number of neurons is determined by these input parameters (typically 50 to 100). A feature extraction layer, employing a multi-layer convolutional network (CNN) architecture, extracts implicit features from the ingredient combination. This layer comprises three convolutional layers, each with 64, 128, and 64 kernels. A sensory mapping layer, employing a fully connected network architecture, maps the extracted features into sensory space. This layer comprises three fully connected layers, each with 128, 64, and 32 neurons. A temporal prediction layer, employing a recurrent neural network (RNN) or LSTM architecture, predicts the temporal evolution of sensory experience. This layer comprises two layers of LSTM units, each with 16 neurons. An output layer, which generates a multidimensional sensory experience prediction, comprises a number of neurons equal to the number of predefined sensory attributes. A dedicated neural network branch was designed for each sensory response decision node, forming a multi-task learning architecture. For example, the "cold sensation" decision node is connected to a dedicated hidden layer with 32 neurons, while the "throat sensation" decision node is connected to another dedicated hidden layer with 24 neurons. The loss function of the network is designed as a multi-task weighted loss: ,in It is The prediction loss of sensory attributes, It is The sensory weights of each sensory attribute are initially set to be evenly distributed. A sensory response prediction framework is generated, including a complete neural network structure, parameter initialization scheme, and training strategy, providing the infrastructure for subsequent personalized training. Data from the component thermodynamic reaction map is analyzed to study the differences in sensory perception between individuals. Based on research data on sensory receptor gene polymorphisms, the population is divided into different types of perception abilities. For example, based on the TAS2R38 genotype, the population is divided into "taste sensitive type" (strongly perceiving the bitterness of PTC) and "taste insensitive type" (weakly perceiving the bitterness of PTC); based on the TRPM8 gene polymorphism, the population is divided into "high cold perception type" and "low cold perception type." A sensory difference test experiment is designed and executed, inviting 100 volunteers with different backgrounds (age, gender, smoking history, etc.) to conduct blind test ratings on standard aerosol liquid samples. The test included: basic taste recognition testing (recognition thresholds for the five flavors of sweet, sour, bitter, salty, and umami); special sensory testing (perception intensity of cool, hot, and pungent sensations); aroma recognition testing (ability to identify top, middle, and base notes); and overall preference ratings (preference ratings for different atomization temperatures and ingredient combinations). The test data was analyzed to calculate the difference coefficients between different groups across sensory dimensions. For example, people with a "high coolness" perception had a 45% higher perception intensity for menthol than those with a "low coolness" perception; and long-term smokers had a 2.8-fold higher taste threshold for nicotine than non-smokers. Based on the difference analysis results, users were categorized into 5 to 10 groups with typical perceptual patterns, and standardized perceptual parameter sets were established for each group. This generated individual perceptual difference data, including sensory thresholds, perceptual intensity coefficients, and preference patterns for each user type, providing a basis for personalized neural network training. Based on this individual perceptual difference data, a pre-designed sensory response prediction framework was trained to learn the sensory-physiological response association patterns for different user groups. A weight learning strategy was designed to adjust the neural network's association weights for different user groups. For example, for users with a high sense of coolness, the weight of the menthol-cooling pathway is increased; for chronic smokers, the weight of the nicotine-stimulation pathway is decreased. The network is trained using supervised learning methods, using labeled data collected from sensory testing as the training set. The training set consists of sensory ratings of 50 different formulations at five different temperatures for each user group, totaling 1,250 training samples. Each sample set contains input features (ingredient concentration, temperature, etc.) and an output label (sensory rating). Training parameter settings include: batch size = 32, initial learning rate = 0.001, learning rate decay strategy (10% reduction every 50 epochs), maximum number of training rounds = 500, and early stopping criterion: no improvement in validation set loss for 10 consecutive rounds. Dedicated weight sets are trained for each user group, forming a multi-model ensemble system. By integrating the predictive capabilities of the models from each group, personalized predictions are achieved for specific users.For example, the system can automatically select or fuse the most appropriate prediction model based on the user's basic information (age, gender, smoking history, etc.). After training, the model was evaluated, achieving an average prediction accuracy of 91% on the test set, with the average sensory intensity prediction error reduced to 0.08 (out of a maximum score of 1.0). This generated a complete sensory response prediction engine that accepts component thermodynamic reaction data and diffusion kinetics data as input and outputs personalized sensory experience predictions, including sensory intensity changes over time, harmony scores for each sensory attribute, and overall satisfaction predictions.

[0077] This embodiment achieves a comprehensive characterization of the pyrolysis components produced during the atomization process through multi-parameter atomization condition sampling and temperature gradient deconstruction analysis. It can convert the complex pyrolysis process into a visual multi-dimensional pyrolysis spectrum, thereby enhancing the monitoring accuracy of the changes in the composition of the atomized liquid. By constructing a trace marker feature library and generating a dynamic component evolution spectrum, it can effectively identify and track ultra-trace active ingredients, achieve accurate quantitative analysis of bioactive substances, ensure that trace components with potential impacts on human health are fully monitored, greatly improve the accuracy of the atomized liquid safety assessment, and avoid the problem of trace components being easily overlooked in traditional analysis methods. Through the atomization process partitioning and component-temperature interactive thermodynamic analysis, the system can fully grasp the thermodynamic behavior characteristics of the atomized liquid components in different temperature ranges, respond to the composition change laws under different atomization conditions, and improve the system's analytical ability and accuracy for complex thermodynamic changes in the atomization process through thermal degradation critical point analysis and molecular reorganization activity analysis. The introduction of reaction path network diagrams and nonlinear thermodynamic behavior analysis realizes multi-level information fusion of complex reaction systems, providing a theoretical basis for component safety assessment. By analyzing the ingredient diffusion dynamics and building a sensory response prediction engine, the diffusion characteristics of the atomized ingredients are analyzed in real time, the possible sensory experience is predicted, and the impact of different ingredient combinations on the user's senses is dynamically evaluated to ensure a high-quality sensory experience when the atomization conditions change. Based on the automated evaluation and analysis of individual perception difference data, personalized quality assessment reports can be automatically generated and the formula can be optimized according to the real-time atomization characteristics, thereby improving the system's ability to adapt to the needs of different users, significantly enhancing the user experience, and ensuring product safety.

[0078] Example 2

[0079] See also Figure 2 As shown, for the parts not described in detail in this embodiment, please refer to the description of Example 1. A system for intelligent analysis of atomized liquid components is provided, including:

[0080] Multi-dimensional thermal spectrum construction module: multi-parameter atomization condition sampling of the atomized liquid is performed to obtain raw pyrolysis data; temperature gradient deconstruction analysis is performed based on the raw pyrolysis data to generate a multi-dimensional pyrolysis spectrum;

[0081] Dynamic trajectory analysis module: accurately identifies trace markers based on multi-dimensional pyrolysis spectra and generates a trace marker feature library; analyzes multi-dimensional component evolution trajectories based on the trace marker feature library and generates a dynamic component evolution spectrum;

[0082] Thermodynamic interaction modeling module: The atomization process is partitioned and demarcated according to the dynamic component evolution spectrum to generate the partitioned atomization characteristic spectrum; based on the partitioned atomization characteristic spectrum, the component-temperature interaction thermodynamic analysis is performed to generate the component thermodynamic reaction map;

[0083] Intelligent evaluation and optimization module: Analyze the component diffusion dynamics based on the partitioned atomization characteristic spectrum to generate component diffusion quantitative data; construct a sensory response prediction engine based on the component thermodynamic reaction map to obtain a sensory response prediction engine; transmit the component diffusion quantitative data to the sensory response prediction engine in real time for personalized sensory experience intelligent analysis to generate sensory response characteristic data; perform global intelligent evaluation of atomized liquid characteristics based on the sensory response characteristic data to generate an atomized liquid quality evaluation report; optimize the safety and taste balance of the original pyrolysis data based on the atomized liquid quality evaluation report to generate optimized formula recommendations, and execute precise blending of the atomized liquid through the optimized formula recommendations.

[0084] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will be able to modify the technical solutions described in the foregoing embodiments or to substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

[0085] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0086] In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0087] In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0088] In the description of the present invention, “several” means one or more, and “a large number” means two or more.

[0089] Throughout this specification, references to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0090] The formulas in this manual are all dimensionless and calculated using numerical values. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field based on actual conditions.

[0091] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A method for intelligent analysis of atomized liquid components, characterized in that: include: Step S1: sampling the atomized liquid under multi-parameter atomization conditions to obtain original pyrolysis data; Perform temperature gradient deconstruction analysis based on the original pyrolysis data to generate a multi-dimensional pyrolysis spectrum; Step S2: performing precise identification of trace markers based on the multidimensional pyrolysis spectrum to generate a trace marker feature library; performing multidimensional component evolution trajectory analysis based on the trace marker feature library to generate a dynamic component evolution spectrum; Step S3: performing partitioning and demarcation processing on the atomization process according to the dynamic component evolution spectrum to generate a partitioned atomization characteristic spectrum; Conduct component-temperature interaction thermodynamic analysis based on the partitioned atomization characteristic spectrum to generate a component thermodynamic reaction map; Step S4: Analyze the component diffusion dynamics based on the partitioned atomization characteristic spectrum to generate component diffusion quantitative data; A sensory response prediction engine is constructed based on the component thermodynamic reaction map to obtain a sensory response prediction engine; component diffusion quantification data is transmitted to the sensory response prediction engine in real time to perform personalized sensory experience intelligent analysis and generate sensory response characteristic data; Based on the sensory response characteristic data, the system intelligently evaluates the global characteristics of the atomized liquid and generates an atomized liquid quality assessment report. Based on the atomized liquid quality assessment report, the original pyrolysis data is optimized for safety and taste balance, and optimized formula recommendations are generated. The atomized liquid is then precisely formulated based on the optimized formula recommendations. Step S2 includes: Extract pyrolysis intensity features based on multi-dimensional pyrolysis spectra to generate pyrolysis intensity spectrum data; Extract molecular fragment spectrum features based on multi-dimensional pyrolysis spectra to generate molecular fragment spectrum data; Using a preset multi-dimensional spectrum deconvolution algorithm to perform initial trace marker identification analysis on the pyrolysis intensity spectrum data and the molecular fragmentation spectrum data to generate initial trace marker data; Perform low signal-to-noise ratio signal enhancement processing based on initial trace marker data to generate a trace marker feature library; Analyze the multi-dimensional component evolution trajectory based on the trace marker feature library to generate a dynamic component evolution spectrum; The low signal-to-noise ratio signal enhancement processing is performed based on the initial trace marker data to generate a trace marker feature library, including: performing a bioactivity threshold analysis based on the initial trace marker data to generate bioactivity threshold data, wherein the bioactivity threshold data includes physiological response triggering concentration data and a marker specificity index; Perform ultra-trace active ingredient enrichment processing based on physiological response triggering concentration data to generate ultra-trace active ingredient data; Perform bioactive marker correlation processing based on ultra-trace active ingredient data and marker specificity index to generate bioactive marker data; The initial trace marker data is subjected to marker screening processing according to the bioactive marker data to generate a trace marker feature library.

2. The intelligent analysis method based on atomized liquid composition according to claim 1, characterized in that: Step S1 includes: Perform dynamic sampling of atomized liquid under multi-parameter atomization conditions to generate raw pyrolysis data; Classify the pyrolysis product types based on the original pyrolysis data to generate pyrolysis product type characteristic data; Perform multi-stage pyrolysis kinetic characterization based on the pyrolysis product type characteristic data to generate pyrolysis kinetic characterization data; Temperature gradient deconstruction analysis is performed based on the pyrolysis kinetic characterization data to generate a multidimensional pyrolysis spectrum.

3. The method for intelligent analysis of atomized liquid components according to claim 1, characterized in that: Step S3 includes: Perform partitioning and demarcation processing on the atomization process according to the dynamic component evolution spectrum to generate the partitioned atomization characteristic spectrum; Detect the thermodynamic stability limit based on the partitioned atomization characteristic spectrum and generate thermodynamic stability data; Composition-temperature interaction thermodynamic analysis is performed based on thermodynamic stability data to generate a composition thermodynamic reaction map.

4. The method for intelligent analysis of atomized liquid components according to claim 3, characterized in that: The thermodynamic stability limit detection based on the partitioned atomization characteristic spectrum to generate thermodynamic stability data includes: Perform thermal degradation critical point analysis based on the partitioned atomization characteristic spectrum to generate thermal degradation critical point data; Perform molecular recombination activity analysis based on the partitioned atomization characteristic spectrum to generate molecular recombination activity data; The composition-temperature phase transition matrix is constructed based on the thermal degradation critical point data and the molecular reorganization activity data to generate the composition-temperature phase transition matrix; Thermodynamic stability limits are detected based on the composition-temperature phase transition matrix to generate thermodynamic stability data.

5. The method for intelligent analysis of atomized liquid components according to claim 3, characterized in that: The composition-temperature interaction thermodynamic analysis is performed based on the thermodynamic stability data to generate a composition thermodynamic reaction map, including: Perform multi-dimensional thermodynamic parameter extraction on thermodynamic stability data to generate a thermodynamic parameter set including Gibbs free energy change rate, activation energy distribution, and entropy change gradient; Construct component interaction potential energy field based on thermodynamic parameter set to generate component interaction potential energy data; Performing temperature sensitivity classification processing on component interaction potential energy data to generate temperature sensitivity classification data; Nonlinear thermodynamic behavior analysis is performed based on temperature sensitivity classification data to generate nonlinear thermodynamic behavior data including critical phase transition points, reaction path weights, and energy barrier heights; Perform temperature-component synergistic effect analysis on nonlinear thermodynamic behavior data and generate a temperature-component synergistic effect matrix; Conduct micro-thermodynamic equilibrium analysis based on the temperature-component synergistic effect matrix to generate micro-equilibrium data; Tracing thermodynamic reaction paths based on microscopic equilibrium data to generate reaction path network maps; The reaction path network map and nonlinear thermodynamic behavior data are subjected to multi-level information fusion processing to generate a component thermodynamic reaction map.

6. The method for intelligent analysis of atomized liquid components according to claim 1, characterized in that: The step S4 comprises: Analyze the component diffusion dynamics based on the partitioned atomization characteristic spectrum to generate component diffusion quantitative data; Conduct multi-dimensional sensory impact factor evaluation based on the component thermodynamic reaction map, generate sensory impact factor data, and design perception response decision nodes based on the sensory impact factor data; Based on the preset sensory feature extraction algorithm, a neural perceptual mapping is constructed to associate sensory experience with physiological response for sensory response decision nodes and component thermodynamic reaction maps, thereby obtaining a sensory response prediction engine. The component diffusion quantification data is transmitted in real time to the sensory response prediction engine for intelligent analysis of personalized sensory experience to generate sensory response feature data; Perform intelligent evaluation of global atomized liquid characteristics based on sensory response characteristic data and generate an atomized liquid quality assessment report; Based on the atomized liquid quality assessment report, the original pyrolysis data is optimized for safety and taste balance, and optimized formula recommendations are generated. The atomized liquid is then precisely formulated using the optimized formula recommendations.

7. The method for intelligent analysis of atomized liquid components according to claim 6, characterized in that: The sensory response prediction engine is obtained by constructing a neural perception mapping system that associates sensory experience with physiological response based on a preset sensory feature extraction algorithm for sensory response decision nodes and component thermodynamic reaction maps, including: Based on the preset sensory feature extraction algorithm, a neural perception mapping architecture is designed to associate sensory experience with physiological response at the sensory response decision node, generating a sensory response prediction framework. Conduct individual perception difference analysis based on component thermodynamic reaction maps to generate individual perception difference data; Based on individual perception difference data, the sensory response prediction framework is used to learn the correlation weights between sensory experience and physiological response to generate a sensory response prediction engine.

8. A system for intelligent analysis of atomized liquid components, which is used to implement the method for intelligent analysis of atomized liquid components according to any one of claims 1 to 7, characterized in that: include: Multi-dimensional thermal spectrum construction module: multi-parameter atomization condition sampling of the atomized liquid to obtain raw pyrolysis data; Perform temperature gradient deconstruction analysis based on the original pyrolysis data to generate a multi-dimensional pyrolysis spectrum; Dynamic trajectory analysis module: accurately identifies trace markers based on multi-dimensional pyrolysis spectra and generates a trace marker feature library; analyzes multi-dimensional component evolution trajectories based on the trace marker feature library and generates a dynamic component evolution spectrum; Thermodynamic interaction modeling module: performs partitioning and demarcation processing of the atomization process according to the dynamic component evolution spectrum to generate the partitioned atomization characteristic spectrum; Conduct component-temperature interaction thermodynamic analysis based on the partitioned atomization characteristic spectrum to generate a component thermodynamic reaction map; Intelligent evaluation and optimization module: analyzes component diffusion dynamics based on partitioned atomization characteristic spectra and generates component diffusion quantitative data; A sensory response prediction engine is constructed based on the component thermodynamic reaction map to obtain a sensory response prediction engine; component diffusion quantification data is transmitted to the sensory response prediction engine in real time to perform personalized sensory experience intelligent analysis and generate sensory response characteristic data; Based on the sensory response characteristic data, an intelligent evaluation of the global atomized liquid characteristics is performed to generate an atomized liquid quality evaluation report. Based on the atomized liquid quality evaluation report, the original pyrolysis data is optimized for safety and taste balance, and optimized formula recommendations are generated. The atomized liquid is then precisely prepared using the optimized formula recommendations.

Images