Tobacco product flavor optimization method and system based on multispectral sensing system

Abstract

The embodiment of the invention provides a tobacco product flavor optimization method and system based on a multispectral sensing system, and belongs to the technical field of tobacco flavor optimization. The optimization method comprises the following steps: acquiring flow velocity, temperature and humidity of tobacco shreds in a tobacco shred drying process; inputting the flow velocity, the temperature and the humidity into a multi-dimensional parameter flavor generation model to calculate the generation rate of the aromatic compound; and constructing a flavor compound distribution diagram according to the generation rates of the aromatic compounds and the volatile components so as to optimize cut tobacco drying conditions. Through dynamic monitoring and modeling of key parameters such as the flow rate, the temperature and the humidity of the tobacco shreds, generation and release of aromatic substances can be more accurately regulated and controlled, and therefore the flavor and taste of tobacco products are effectively improved. And a flavor compound distribution diagram is used for displaying the generation condition of aromatic substances in each region under different temperature, humidity and flow velocity conditions, and the operation parameters of the cut tobacco drying equipment are adjusted according to the distribution diagram, so that the uniformity and consistency of the generation of the flavor compounds are ensured.

Description

[0001] Technical Field

[0002] The present invention relates to the technical field of tobacco flavor optimization, and particularly to a method and system for optimizing the flavor of tobacco products based on a multispectral sensing system. Background Art

[0003] In the production process of tobacco products, flavor and taste are key factors determining product quality. As an important link in tobacco processing, drying of tobacco leaves plays a crucial role in the drying degree of cut tobacco, the generation and retention of aromatic substances. Traditional drying processes of tobacco leaves rely on fixed equipment parameters, such as heating temperature, fan speed, and drying time of tobacco leaves, and these parameters are usually not adjusted dynamically during the production process. Due to changes in environmental conditions, characteristics of cut tobacco, and equipment operating status, this method of fixed parameters is likely to result in inconsistent flavor and taste of the produced tobacco products, thereby affecting the market performance and user experience of the products.

[0004] Currently, although some drying processes of tobacco leaves have introduced temperature and humidity sensors to monitor the environment, these systems can usually only collect static data such as temperature or humidity singly, and it is difficult to achieve dynamic control during the complex drying process. The generation of flavor compounds in tobacco not only depends on temperature and humidity, but also is closely related to the flow rate of cut tobacco, heat distribution, and the generation kinetics of aromatic substances. Due to the complex coupling relationship between these multi-dimensional parameters, traditional single-parameter control methods are difficult to effectively optimize the flavor generation and taste regulation during the entire drying process. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide a method and system for optimizing the flavor of tobacco products based on a multispectral sensing system, which solves the problem of dynamically monitoring key parameters such as the flow rate, temperature, and humidity of cut tobacco to precisely control the generation and release of aromatic substances, thereby effectively improving the flavor and taste of tobacco products.

[0006] To achieve the above purpose, the embodiments of the present invention provide a method for optimizing the flavor of tobacco products based on a multispectral sensing system, and the optimization method includes:

[0007] Obtaining the flow rate, temperature, and humidity of cut tobacco during the drying process of tobacco leaves;

[0008] Inputting the flow rate, temperature, and humidity into a multi-dimensional parameter flavor generation model to calculate the generation rate of aromatic compounds;

[0009] Constructing a flavor compound distribution map according to the generation rate of the aromatic compounds to optimize the drying conditions.

[0010] Optionally, obtaining the flow rate, temperature, and humidity of cut tobacco during the drying process of tobacco leaves in real time includes:

[0011] Collect visible light images of cut tobacco through a visible light sensor, use image algorithms to extract the brightness of each pixel point, and use the flow light algorithm for continuous visible light images to obtain the flow rate of cut tobacco;

[0012] Obtain infrared images of cut tobacco through an infrared sensor to obtain the temperature of cut tobacco;

[0013] Construct a polynomial fitting model of brightness-temperature-humidity according to formula (1),

[0014] H(I, T) = a 0 + a 1 I + a 2 T + a 3 IT + a 4 I 2 + a 5 T 2 , (1)

[0015] where a 0 , a 1 , a 2 , a 3 , a 4 , a 5 are fitting coefficients obtained from experimental data, I is the brightness, T is the temperature, and H(I, T) is the humidity value;

[0016] Input the brightness and temperature into the polynomial fitting model of brightness-temperature-humidity to obtain the humidity value of cut tobacco;

[0017] Dynamically adjust the humidity value according to environmental factors and flow rate.

[0018] Optionally, dynamically adjusting the humidity value according to environmental factors and flow rate includes:

[0019] Dynamically adjust the humidity value according to formulas (3) to (4),

[0020] k H = 1 + α·(T ambient - T), (3)

[0021] H adjusted (I, T) = k H ·H(I, T), (4)

[0022] where k H is the humidity response coefficient, α is the humidity response adjustment coefficient, T ambient is the environmental temperature, and H adjusted (I, T) is the corrected humidity value;

[0023] Dynamically adjust the corrected humidity value according to formula (5),

[0024] H final (I, T, V) = H adjusted (I, T)·(1 - k V ·|V(x, y, t)|), (5)

[0025] where H final (I, T, V) is the final humidity value, k V is the flow rate correction coefficient, and V(x, y, t) is the flow rate of the cut tobacco.

[0026] Optionally, the flow rate, temperature, and humidity of the cut tobacco during the cut tobacco drying process are obtained in real time, including:[[]]

[0027] Construct a neural network to optimize the brightness-temperature-humidity model;

[0028] Determine the loss function of the neural network optimized brightness-temperature-humidity model according to formula (6),

[0029]

[0030] where MSE is the loss function, is the actual humidity, is the predicted humidity;

[0031] Input the brightness and temperature into the neural network optimized brightness-temperature-humidity model to obtain the humidity value.

[0032] Optionally, collect the visible light image of the cut tobacco through a visible light sensor, use an image algorithm to extract the brightness of each pixel point, and use the optical flow algorithm for continuous visible light images to obtain the flow rate of the cut tobacco, including:[[]]

[0033] Obtain the brightness of a single pixel point according to formulas (7) to (8),

[0034] I(x, y, t) = I(x + Δx, y + Δy, t + Δt), (7)

[0035]

[0036] where I(x, y, t) is the brightness of the pixel at the position (x, y, t), and Δx, Δy, Δt are the movement amounts of the pixel between individual image frames, is the brightness gradient of the pixel point in the x direction, is the brightness gradient of the pixel point in the y direction, is the change rate of the pixel point over time, u is the velocity component of the pixel point in the x direction, and v is the velocity component of the pixel point in the y direction;

[0037] Obtain the brightness matrix of the neighborhood of the pixel point and the flow rate of the cut tobacco according to formulas (9) to (13).

[0038]

[0039] A·V = b, (12)

[0040]

[0041] wherein, is the brightness gradient of the pixels in the neighborhood of the pixel point in the x direction, is the brightness gradient of the pixels in the neighborhood of the pixel point in the y direction, is the rate of change of the pixels in the neighborhood of the pixel point over time, N is the number of pixel points, A is the brightness gradient matrix, b is the brightness rate of change matrix over time, V is the flow rate matrix, v(x, y, t) is the component of the flow rate vector in the y direction, u(x, y, t) is the component of the flow rate vector in the x direction, and |V(x, y, t)| is the flow rate of the pixels in the neighborhood of the pixel point.

[0042] Optionally, obtain the infrared image of the cut tobacco through an infrared sensor to obtain the temperature of the cut tobacco, including:

[0043] Determine the temperature of the cut tobacco according to formula (14).

[0044]

[0045] wherein, I(λ, T) is the radiation intensity at wavelength λ, T is the temperature, h is Planck's constant, c is the speed of light, and k is Boltzmann's constant.

[0046] Optionally, input the flow rate, temperature, and humidity into a multi-dimensional parameter flavor generation model to calculate the generation rate of aromatic compounds, including:

[0047] Calculate the generation rate of aromatic compounds according to formula (15).

[0048]

[0049] wherein, g(T(x,y,t), H(x,y,t), V(x,y,t)) is a non-linear function of temperature, humidity, and flow rate, T(x,y,t) is a non-linear function of temperature, H(x,y,t) is a non-linear function of humidity, V(x,y,t) is a non-linear function of flow rate, E a is the activation energy for the generation of aromatic compounds, R is the gas constant, T is the temperature, R f (t) is the generation rate of aromatic compounds.

[0050] Optionally, a flavor compound distribution map is constructed according to the production rates of the aromatic compounds and volatile components to optimize the cut tobacco drying conditions, including:

[0051] Adjust the heating power according to formula (16),

[0052]

[0053] Adjust the fan air velocity according to formula (17),

[0054] P fan-adjusted (t)=V target (t)·(1+α H ·ΔH(x,y,t)), (17)

[0055] Adjust the flow rate according to formula (18),

[0056] v opt (t)=V target (t)·(1+β v ·|V(x,y,t)|), (18)

[0057] Wherein, is the temperature gradient, ΔH(x,y,t) is the humidity deviation, α T and α H are the temperature and humidity adjustment coefficients respectively, P target is the target heating power, V target is the target flow rate, P heat-adjusted (t) is the adjusted heating power, P fan-adjusted (t) is the adjusted air velocity power, β v is the flow rate correction coefficient, v opt (t) is the adjusted flow rate.

[0058] Optionally, the flow rate, temperature, and humidity are input into a multi-dimensional parameter flavor generation model to calculate the production rate of aromatic compounds, including:

[0059] Construct a multi-dimensional parameter flavor generation model;

[0060] Input the flow rate, temperature, and humidity into the multi-dimensional parameter flavor generation model for training to update the weights and parameters.

[0061] On the other hand, the present invention provides a tobacco product flavor optimization system based on a multi-spectral sensing system, and the system includes:

[0062] An acquisition module, configured to acquire the flow rate, temperature, and humidity of cut tobacco during the cut tobacco drying process;

[0063] A processing module, connected to the acquisition module, configured to execute any one of the above optimization methods;

[0064] A display module, connected to the processing module, for displaying the flavor compound distribution map;

[0065] A cloud module, connected to the acquisition module, for receiving the flow rate, temperature, and humidity and real-time monitoring the state of the cut tobacco.

[0066] Through the above technical solution, the present invention provides a method and system for optimizing the flavor of tobacco products based on a multispectral sensing system. By dynamically monitoring and modeling key parameters such as the flow rate, temperature, and humidity of the cut tobacco, the generation and release of aromatic substances can be more precisely controlled, thereby effectively improving the flavor and taste of tobacco products. Moreover, the flavor compound distribution map is used to show the generation of aromatic substances in each region under different temperature, humidity, and flow rate conditions. Operators can adjust the operating parameters of the cut tobacco drying equipment according to the feedback of the distribution map, optimize and adjust the key parameters in the cut tobacco drying process, improve the flavor and taste of tobacco products, and ensure the uniformity and consistency of the generation of flavor compounds throughout the production process.

[0067] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent specific implementation part. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementation, they are used to explain the embodiments of the present invention, but do not constitute a limitation to the embodiments of the present invention. In the drawings:

[0069] Figure 1 is a flowchart of an optimization method according to an embodiment of the present invention;

[0070] Figure 2 is a flowchart of obtaining the flow rate, temperature, and humidity according to an embodiment of the present invention;

[0071] Figure 3 is a flowchart of image preprocessing according to an embodiment of the present invention;

[0072] Figure 4 is a flowchart of adjusting the humidity value according to an embodiment of the present invention;

[0073] Figure 5 is a flowchart of obtaining the flow rate of the cut tobacco according to an embodiment of the present invention;

[0074] Figure 6 is a flowchart of adjusting the cut tobacco drying conditions according to an embodiment of the present invention;

[0075] Figure 7It is a flowchart for training a multi-dimensional parameter flavor generation model according to an embodiment of the present invention;

[0076] Figure 8 It is a tabular graph of experimental data according to an embodiment of the present invention. Detailed Embodiments

[0077] The following will describe in detail the specific embodiments of the embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the embodiments of the present invention, and are not used to limit the embodiments of the present invention.

[0078] It should be noted that the acquisition, transmission, storage, use, processing, etc. of data in the technical solution of this application all comply with the relevant regulations of national laws and regulations. In the embodiments of this application, certain industry-existing solutions such as software, components, models, etc. may be mentioned. They should be regarded as exemplary. The purpose is only to illustrate the feasibility in the implementation of the technical solution of this application, but it does not mean that the applicant has already or necessarily used this solution.

[0079] Figure 1 It is a flowchart of an optimization method according to an embodiment of the present invention. In this figure, the optimization method includes:

[0080] In step S1, the flow rate, temperature, and humidity of cut tobacco during the cut tobacco drying process are acquired. By real-time monitoring the flow rate, temperature, and humidity data of cut tobacco, the dynamic changes of cut tobacco during the drying process can be accurately captured. Compared with traditional single-temperature or single-humidity monitoring, the present invention can more comprehensively reflect the physical state of cut tobacco during the drying process through real-time acquisition and analysis of multi-dimensional data.

[0081] In step S2, the flow rate, temperature, and humidity are input into the multi-dimensional parameter flavor generation model to calculate the generation rate of aromatic compounds. The present invention constructs a flavor generation model based on chemical reaction kinetics. By comprehensively analyzing the non-linear relationship among temperature, humidity, and flow rate, the generation rates of aromatic compounds and volatile components are calculated. This model ensures the stable generation of flavor substances under different working conditions, thereby optimizing the flavor and taste of tobacco products.

[0082] In step S3, a flavor compound distribution map is constructed according to the generation rates of aromatic compounds and volatile components to optimize the cut tobacco drying conditions. The flavor compound distribution map is used to show the generation situation of aromatic compounds in different regions. Combining with temperature and humidity data, the distribution map helps operators optimize the cut tobacco drying conditions to ensure uniform flavor throughout the drying process.

[0083] By collecting the flow rate, temperature, and humidity data of cut tobacco in real time, combining with a multi-dimensional parameter flavor generation model to calculate the production rate of aromatic compounds, and using a flavor compound distribution map to show the generation of aromatic substances in each region under different temperature, humidity, and flow rate conditions, operators can adjust the operating parameters of the cut tobacco drying equipment according to the feedback of the distribution map, optimize and adjust the key parameters in the cut tobacco drying process, improve the flavor and taste of tobacco products, and ensure the uniformity and consistency of the generation of flavor compounds throughout the production process.

[0084] In this embodiment, there are various ways to collect the flow rate, temperature, and humidity of cut tobacco, which are known to those skilled in the art. In one example of the present invention, these parameters are obtained through a multispectral sensor. Specifically, as Figure 2 shown, it includes:

[0085] In step S11, a visible light image of the cut tobacco is collected through a visible light sensor, the brightness of each pixel point is extracted using an image algorithm, and a flow algorithm is used for consecutive visible light images to obtain the flow rate of the cut tobacco.

[0086] In step S12, an infrared image of the cut tobacco is obtained through an infrared sensor to obtain the temperature of the cut tobacco.

[0087] For the convenience of subsequent analysis, it is necessary to perform preliminary preprocessing on the infrared light image and the visible light image. The specific steps are as Figure 3 shown, and it includes:

[0088] In step 121, image alignment and registration are performed on the visible light image and the infrared image. Due to the different optical characteristics of the visible light and infrared sensors, differences in the field of view and resolution may occur. Through image registration technology, the visible light image is geometrically aligned with the infrared image to ensure that the two sets of data can be accurately corresponding to the same spatial position. The visible light sensor usually uses a camera in the 400 - 700nm band to capture the flow pattern and movement trajectory of the cut tobacco. Infrared sensor: The infrared camera uses a dual-band of 3 - 5μm and 8 - 14μm, corresponding to high temperatures within 550°C and low-temperature environments such as 20°C to 100°C respectively. These bands can provide sufficient sensitivity for monitoring the cut tobacco and the surface of the cavity within a wide temperature range. To ensure that the visible light and infrared light images are captured at the same moment, the system should be equipped with a synchronization trigger to control the two sensors to collect images simultaneously. According to the movement speed of the cut tobacco, the frame rates of the visible light sensor and the infrared sensor need to be dynamically adjusted. Generally speaking, the visible light sensor requires a higher frame rate (such as 120 frames per second) to capture the details of the fast-moving cut tobacco. While the frame rate of the infrared sensor can be lower (30 - 60 frames per second) for temperature monitoring. The timestamp technology is used to ensure that each pair of visible light and infrared images has the same timestamp for subsequent image fusion and data processing.

[0089] In step 122, feature points of the visible light image and the infrared image are extracted and matched. Multiple feature points such as the edges or key nodes of cut tobacco are selected from the visible light and infrared images, and registration is performed through image registration algorithms such as SIFT, ORB, etc. to ensure the consistency of the spatial positions of the images.

[0090] In step 123, the infrared image is filtered. The collected infrared image is filtered to reduce noise and enhance the contrast of the image.

[0091] In step 124, the infrared image is subjected to pseudo-color processing and non-uniformity correction. By performing pseudo-color processing on the infrared image, the temperature information is converted into a visual image, and at the same time, the image quality is improved through non-uniformity correction NUC, and the error of temperature measurement is reduced.

[0092] In step S13, a polynomial fitting model of brightness-temperature-humidity is constructed according to formula (1), H(I, T) = a 0 +a 1 I + a 2 T + a 3 IT + a 4 I 2 + a 5 T 2 , (1)

[0093] where a 0 , a 1 , a 2 , a 3 , a 4 , a 5 are fitting coefficients obtained from experimental data, I is the brightness, T is the temperature, and H(I, T) is the humidity value.

[0094] In step S14, the brightness and temperature are input into the polynomial fitting model of brightness-temperature-humidity to obtain the humidity value of the cut tobacco.

[0095] In step S15, the humidity value is dynamically adjusted according to environmental factors and flow rate.

[0096] In step S16, the fitting coefficients of the polynomial fitting model of brightness-temperature-humidity are dynamically adjusted according to formula (2). The present invention dynamically optimizes the polynomial fitting model of brightness-temperature-humidity through an adaptive parameter adjustment mechanism. The adaptive mechanism dynamically adjusts the fitting coefficients in the humidity model according to the temperature, brightness, and flow rate data monitored in real time. For the adjustment method, there are various methods known to those skilled in the art. In one example of the present invention, it can be adjusted according to formula (2). Specifically, it includes:

[0097] Dynamically adjust the fitting coefficients of the polynomial fitting model for brightness-temperature-humidity according to formula (2), a i (t) = a i base ·(1 + ρ·(T(t) - T reference )), (2)

[0098] where a i (t) is the adjusted coefficient, a i base is the basic fitting coefficient, ρ is the adaptive adjustment coefficient, T(t) is the real-time temperature, and T reference is the reference temperature.

[0099] In order to improve the dynamic response ability and estimation accuracy of humidity estimation, in this embodiment, the humidity value is dynamically adjusted according to environmental factors and flow rate. Specifically, as Figure 4 shown, it includes:

[0100] In step S151, the humidity value is dynamically adjusted according to formulas (3) to (4),

[0101] k H = 1 + α·(T ambient - T), (3)

[0102] H adjusted (I, T) = k H ·H(I, T), (4)

[0103] where k H is the humidity response coefficient, α is the humidity response adjustment coefficient, T ambient is the environmental temperature, and H adjusted (I, T) is the corrected humidity value.

[0104] In step S152, the corrected humidity value is dynamically adjusted according to formula (5),

[0105] H final (I, T, V) = H adjusted (I, T)·(1 - k V ·|V(x, y, t)|), (5)

[0106] where H final (I, T, V) is the final humidity value, k V is the flow rate correction coefficient, and V(x, y, t) is the flow rate of the cut tobacco. Since the flow rate of the cut tobacco in the cavity affects the change of humidity, the humidity estimation value is dynamically corrected by the flow rate, so that the humidity estimation value can dynamically adapt to the change of the flow rate and further improve the accuracy.

[0107] To enhance the robustness and self - adaptability of the neural network - optimized brightness - temperature - humidity model, in this embodiment, the present invention further trains the historical humidity, brightness, and temperature data through the neural network - optimized brightness - temperature - humidity model to fit complex non - linear relationships. Specifically, it includes the following steps:

[0108] In step S17, a neural network - optimized brightness - temperature - humidity model is constructed.

[0109] In step S18, according to formula (6), the loss function of the neural network - optimized brightness - temperature - humidity model is determined.

[0110]

[0111] where MSE is the loss function. is the actual humidity. is the predicted humidity.

[0112] In step S19, brightness and temperature are input into the neural network - optimized brightness - temperature - humidity model to obtain the humidity value. Through training, the model can effectively fit the complex relationship between brightness, temperature, and humidity, further improving the accuracy of humidity estimation.

[0113] In this embodiment, considering that it is necessary to extract brightness information from visible - light images, to improve the accuracy and stability of brightness data, the visible - light image needs to be processed. Specifically, as Figure 5 shown, it includes:

[0114] In step 111, visible - light images of the cut tobacco in the cut - tobacco drying cavity are regularly collected by a visible - light sensor. Each collection generates a two - dimensional grayscale image, and the brightness information is stored in each pixel of the image. Specifically, the visible - light sensor first regularly collects each frame of the image through an internal or external clock, generating an image sequence. Each frame of the image can be represented as a matrix I(x, y, t), where x and y are the pixel coordinates of the image, and t is the time dimension. Then the collected image is usually converted into a grayscale image. For a color image, the pixel values of the RGB three channels can be converted into grayscale values I(x, y), and the conversion formula is usually:

[0115] I(x, y, t)=0.299·R(x, y, t)+0.857·G(x, y, t)+0.114·B(x, y, t), (18)

[0116] where R, G, and B respectively represent the pixel values of the red, green, and blue channels, and I(x, y, t) represents the brightness value of the pixel point in the grayscale image.

[0117] In step 112, denoising, enhancement, and correction processing are performed on the visible light image. Due to the complex internal environment of the device, the image may be affected by noise. By using image filtering techniques such as Gaussian filtering or median filtering, the interference of noise on the luminance data can be reduced. Image enhancement techniques such as histogram equalization are used as needed to improve the contrast of the image, making the luminance information more obvious. The luminance deviation caused by the sensor position or illumination change can be adjusted through the automatic white balance and automatic brightness adjustment functions of the sensor to ensure the accuracy of the image luminance.

[0118] In step 113, the luminance values are extracted from the visible light image. The preprocessed grayscale image contains the luminance information I(x, y, t) of each pixel. The extraction of the luminance value includes two processes: reading the pixel value and regional averaging of the luminance value. For the image I(x, y, t) at each moment t, the luminance value of each pixel is read. These luminance values represent the light reflection intensity of the cut tobacco at that position and have a certain correlation with the density of the cut tobacco. To reduce the influence of local noise, the luminance mean value of the local area can be calculated for each frame of the image. The luminance within a small area, such as 3×3 pixels (i.e., the neighborhood of the pixel point), is averaged to obtain a smoother luminance distribution map.

[0119] After processing the visible light image, the optical flow method is further used to analyze the visible light image. Specifically, it includes the following steps:

[0120] In step S114, the luminance of a single pixel point is obtained according to formulas (7) to (8).

[0121] I(x, y, t) = I(x + Δx, y + Δy, t + Δt), (7)

[0122]

[0123] where I(x, y, t) is the luminance of the pixel at the position (x, y, t), and Δx, Δy, and Δt are the movement amounts of the pixel between individual image frames. is the luminance gradient of the pixel point in the x direction. is the luminance gradient of the pixel point in the y direction. is the rate of change of pixel points over time, u is the velocity component of pixel points in the x direction, and v is the velocity component of pixel points in the y direction. The optical flow method estimates the motion speed of an object in an image by analyzing the brightness changes generated by the object's motion in the image. Let the pixel brightness value of the image at time t be I(x, y, t), where (x, y) are pixel coordinates. Assume that the object has a displacement at time t + Δt and moves to a new position (x + Δx, y + Δy), then the brightness value I(x + Δx, y + Δy, t + Δt) also changes accordingly. According to the basic assumption of optical flow, the brightness of the object remains unchanged in a short period of time, that is, formula (7). By performing a Taylor expansion on this formula and ignoring the high-order terms, the optical flow constraint equation can be obtained as formula (8). Formula (8) contains two unknown velocity components u and v, but there is only one equation. Therefore, additional constraint conditions need to be introduced to solve it.

[0124] In step S115, the brightness matrix of the neighborhood of the pixel points and the flow velocity of the cut tobacco are obtained according to formulas (9) to (13).

[0125]

[0126] A·V = b, (12)

[0127]

[0128] Wherein, is the brightness gradient of the pixels in the neighborhood of the pixel point in the x direction, is the brightness gradient of the pixels in the neighborhood of the pixel point in the y direction, is the rate of change of the pixels in the neighborhood of the pixel point over time, N is the number of pixel points, A is the brightness gradient matrix, b is the brightness change rate matrix over time, V is the flow velocity matrix, v(x, y, t) is the component of the flow velocity vector in the y direction, u(x, y, t) is the component of the flow velocity vector in the x direction, and |V(x, y, t)| is the flow velocity of the pixels in the neighborhood of the pixel point. Assume that within a local window Ω, the motion of all pixel points is the same. Therefore, an optical flow constraint equation can be generated for each pixel point. For the N pixel points within the window, we have N equations. Based on this assumption, the optical flow can be solved using multiple pairs of equations, as shown in formulas (9) to (13).

[0129] In this embodiment, for the method of obtaining the temperature of the cut tobacco, there are various methods known to those skilled in the art. In an example of the present invention, it can be obtained according to formula (14). Specifically, it includes:

[0130] Determine the temperature of the cut tobacco according to formula (14).

[0131]

[0132] Among them, I(λ, T) is the radiation intensity at wavelength λ, T is the temperature, h is Planck's constant, c is the speed of light, and k is Boltzmann's constant. The intensity of infrared radiation energy is directly related to the object temperature. According to Planck's radiation law and combined with the band of the infrared sensor, the temperature of the cut tobacco surface can be calculated.

[0133] In this embodiment, for the constructed multi-dimensional parameter generation model, under the premise of satisfying the calculation of the generation rate of aromatic compounds, there can be various types known to those skilled in the art. In a preferred example of the invention, the generation rate of aromatic compounds can be calculated based on the chemical reaction kinetics and the coupling relationship of temperature, humidity, and flow rate. Specifically, it includes:

[0134] Calculate the generation rate of aromatic compounds according to formula (15),

[0135]

[0136] where g(T(x, y, t), H(x, y, t), V(x, y, t)) is a non-linear function of temperature, humidity, and flow rate, T(x, y, t) is a non-linear function of temperature, H(x, y, t) is a non-linear function of humidity, V(x, y, t) is a non-linear function of flow rate, E a is the activation energy for the generation of aromatic compounds, R is the gas constant, T is the temperature, R f (t) is the generation rate of aromatic compounds. This model calculates the generation rate of aromatic substances automatically based on the principles of chemical reaction kinetics and the generation law of tobacco aroma substances, and predicts the generation of aromatic substances in different regions during the cut tobacco drying process in combination with the dynamic state of the cut tobacco.

[0137] In this embodiment, the present invention calculates and generates a flavor compound distribution map through a flavor generation model to show the generation and distribution of aromatic substances during the cut tobacco drying process. The distribution map shows the generation of aromatic substances in each region under different temperature, humidity, and flow rate conditions. Operators can optimize the operating parameters of the cut tobacco drying equipment, i.e., the cut tobacco drying conditions, according to the feedback of the distribution map to ensure the uniformity and consistency of the generation of flavor compounds throughout the production process. Specifically, as Figure 6 shown, it includes the following steps:

[0138] In step S32, adjust the heating power according to formula (16),

[0139]

[0140] In step S31, adjust the fan air velocity according to formula (17),

[0141] P fan-adjusted (t) = V target(t)·(1 + α H ·ΔH(x, y, t)), (17)

[0142] In step S33, the flow rate is adjusted according to formula (18),

[0143] v opt (t) = V target (t)·(1 + β v ·|V(x, y, t)|), (18)

[0144] wherein, is the temperature gradient, ΔH(x, y, t) is the humidity deviation, α T and α H are the temperature and humidity adjustment coefficients respectively, P target is the target heating power, V target is the target flow rate, P heat-adjusted (t) is the adjusted heating power, P fan-adjusted (t) is the adjusted wind speed power, β v is the flow rate correction coefficient, v opt (t) is the adjusted flow rate. When the temperature gradient is too large, the heating power is automatically reduced to avoid local overheating and keep the tobacco shreds evenly heated, ensuring the stable generation of aromatic substances. And according to the real-time humidity change, the wind speed of the fan is dynamically adjusted to ensure that the humidity of the tobacco shreds during the drying process is kept within the optimal range to avoid the tobacco shreds being too dry or too wet, which affects the taste. And the fan speed is dynamically optimized according to the flow rate deviation to ensure that the tobacco shreds flow evenly during the drying process, avoiding uneven heating of local tobacco shreds, which in turn affects the flavor consistency. By adjusting the heating power, the fan wind speed and the flow rate, the present invention can improve the flavor and taste while effectively controlling the energy consumption. Considering the production efficiency, flavor generation, taste optimization and energy consumption requirements comprehensively, the best balance between flavor quality and energy efficiency is achieved during the production process, significantly improving the production efficiency and energy-saving effect.

[0145] To improve the adaptability and accuracy of the multi-dimensional parameter flavor generation model, the present invention trains the historical production data through machine learning algorithms to automatically optimize the flavor generation model. With the accumulation of more data, the system can adaptively adjust the model parameters to adapt to the production requirements under different tobacco varieties and environmental conditions, ensuring the stable output of product flavor and taste. Specifically, as Figure 7 shown, it includes the following steps:

[0146] In step S21, a multi-dimensional parameter flavor generation model is constructed.

[0147] In step S22, the flow rate, temperature and humidity are input into the multi-dimensional parameter flavor generation model for training to update the weights and parameters.

[0148] On the other hand, the present invention provides a tobacco product flavor optimization system based on a multispectral sensing system, which includes an acquisition module, a processing module, a display module, and a cloud module. The acquisition module is used to obtain the flow rate, temperature, and humidity of tobacco leaves during the drying process. The processing module is connected to the acquisition module and is used to execute any of the above optimization methods. The display module is connected to the processing module and is used to display the flavor compound distribution map. The cloud module is connected to the acquisition module and is used to receive the flow rate, temperature, and humidity and monitor the state of tobacco leaves in real time. Operators can adjust the drying process parameters at any time through the remote platform to ensure the continuous optimization and intelligent management of the production process.

[0149] To verify the effectiveness of the present invention, data such as the flow rate, temperature, and humidity of tobacco leaves during the drying process are collected to evaluate the performance of the aromatic compound generation rate, flavor uniformity, and energy consumption. The experiment mainly collects relevant data in real time through a multispectral sensing system and dynamically adjusts the process parameters to observe the flavor optimization effect under different working conditions. This experiment runs under three different working conditions, and different flow rate, temperature, and humidity parameters are adjusted under each working condition. The aromatic compound generation rate, flavor uniformity, and energy consumption of tobacco leaves are monitored emphatically. The experimental data will be recorded at multiple time points, as Figure 8 shown. Among them, the specific meanings of each parameter are as follows:

[0150] 1. Flow rate V(x, y, t): The flow rate of each working condition changes slightly during the experiment to ensure the uniform flow of tobacco leaves in the drying chamber and avoid local overheating or over-drying.

[0151] 2. Temperature T(x, y, t): The temperature is different under different working conditions. The aromatic compound generation rate is faster at high temperatures, but the temperature needs to be strictly controlled to maintain flavor consistency.

[0152] 3. Humidity H(x, y, t): Under different humidity conditions, the water content of tobacco leaves affects the final flavor, and the system ensures the best taste through dynamic humidity control.

[0153] 4. Aromatic compound generation rate Rf(t): This value represents the generation rate of aromatic substances and is comprehensively affected by temperature, humidity, and flow rate. Too high a temperature or too low a humidity will inhibit the generation of aromatic compounds.

[0154] 5. Flavor uniformity: By estimating the distribution of flavor in the drying chamber, it is ensured that tobacco leaves in different regions can obtain a consistent flavor.

[0155] 6. Energy consumption E(t): The change in energy consumption under different working conditions reflects the effectiveness of energy consumption control on the premise of ensuring flavor optimization.

[0156] Through the above technical solution, the present invention provides a method and system for optimizing the flavor of tobacco products based on a multispectral sensing system. By dynamically monitoring and modeling key parameters such as the flow rate, temperature, and humidity of tobacco shreds, it is possible to more precisely control the generation and release of aroma substances, thereby effectively improving the flavor and taste of tobacco products. Moreover, the flavor compound distribution map shows the generation of aroma substances in each region under different temperature, humidity, and flow rate conditions. Operators can adjust the operating parameters of the tobacco drying equipment according to the feedback of the distribution map, optimize and adjust the key parameters during the tobacco drying process, improve the flavor and taste of tobacco products, and ensure the uniformity and consistency of the generation of flavor compounds throughout the production process.

[0157] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, commodity or device including the element.

[0158] The above are only examples of the present application and are not used to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A tobacco product flavor optimization method based on a multispectral sensing system, characterized in that: The optimization method comprises: Obtain the flow rate, temperature and humidity of tobacco during the drying process; inputting the flow rate, temperature and humidity into a multidimensional parameter flavor generation model to calculate the generation rate of aroma compounds; A flavor compound distribution map is constructed according to the generation rate of the aroma compounds to optimize the drying conditions.

2. The optimization method according to claim 1, characterized in that: Real-time acquisition of flow rate, temperature and humidity of tobacco during drying, including: The visible light image of the tobacco is collected by a visible light sensor, the brightness of each pixel is extracted by an image algorithm, and the flow rate of the tobacco is obtained by using a streamer algorithm on the continuous visible light image; The infrared image of the tobacco is obtained by an infrared sensor to obtain the temperature of the tobacco; According to formula (1), a polynomial fitting model of brightness-temperature-humidity is constructed. H(I,T)=0+1I+a2T+a3IT+a4I 2 +5T 2 ,(1) Among them, a0, a1, a2, a3, a4, and a5 are fitting coefficients obtained from experimental data, I is brightness, T is temperature, and H(I,T) is humidity value; Inputting the brightness and temperature into the polynomial fitting model of brightness-temperature-humidity to obtain the humidity value of the tobacco; Dynamically adjusting the humidity value according to environmental factors and flow rate; According to formula (2), the fitting coefficient of the brightness-temperature-humidity polynomial fitting model is dynamically adjusted. a i (t)= i base ·(1+ρ·(T(t)- reference )),(2) Among them, a i (t) is to adjust you and the coefficient, a i base is the basic fitting coefficient, ρ is the adaptive adjustment coefficient, T(t) is the real-time temperature, T reference is the reference temperature.

3. The optimization method according to claim 2, characterized in that: The humidity value is dynamically adjusted based on environmental factors and flow rate, including: The humidity value is dynamically adjusted according to formulas (3) to (4), k H =1+α·(T ambient -),(3) H adjusted (I,T)= H ·(I,T),(4) in, H is the humidity response coefficient, α is the humidity response adjustment coefficient, T ambient is the ambient temperature, H adjusted (I,T) is the corrected humidity value; According to formula (5), the corrected humidity value is dynamically adjusted. H final (I,T,V)= adjusted (I,T)·(1- V ·|V(x,y,t)|),(5) Among them, H final (I, T, V) is the final humidity value, k V is the flow rate correction coefficient, and V(x,y,t) is the flow rate of the tobacco.

4. The optimization method according to claim 2, characterized in that: Real-time acquisition of flow rate, temperature and humidity of tobacco during drying, including: Construct a neural network to optimize the brightness-temperature-humidity model; The loss function of the neural network optimization brightness-temperature-humidity model is determined according to formula (6): Among them, MSE is the loss function, is the actual humidity, To predict humidity; The brightness and temperature are input into the neural network optimized brightness-temperature-humidity model to obtain the humidity value.

5. The optimization method according to claim 2, characterized in that: The visible light image of the tobacco is collected by a visible light sensor, the brightness of each pixel is extracted by an image algorithm, and the flow rate of the tobacco is obtained by using a streamer algorithm on the continuous visible light image, including: According to formulas (7) to (8), the brightness of a single pixel is obtained. I(x,y,t)=(x+Δx,y+Δy,t+Δt), (7) Where I(x,y,t) is the brightness of the pixel at position (x,,), Δx, Δy, Δt are the movement of the pixel between image frames, is the brightness gradient of the pixel in the x direction, is the brightness gradient of the pixel in the y direction, is the rate of change of the pixel point over time, u is the velocity component of the pixel point in the x direction, and v is the velocity component of the pixel point in the y direction; According to formulas (9) to (13), the brightness matrix of the neighborhood of the pixel point and the flow rate of the tobacco are obtained. A·=b, (12) in, …、 is the brightness gradient of the pixels in the neighborhood of the pixel point in the x direction, …、 is the brightness gradient of the pixels in the neighborhood of the pixel point in the y direction, …、 is the rate of change of pixels in the neighborhood of the pixel over time, N is the number of pixels, A is the brightness gradient matrix, n is the brightness change rate matrix over time, V is the flow velocity matrix, v(x,y,t) is the component of the flow velocity vector in the y direction, u(x,y,t) is the component of the flow velocity vector in the x direction, and |(x,y,t)| is the flow velocity of the pixels in the neighborhood of the pixel.

6. The optimization method according to claim 2, characterized in that: The infrared image of the tobacco is obtained by an infrared sensor to obtain the temperature of the tobacco, including: Determine the temperature of the tobacco according to formula (14): Where I(λ,T) is the radiation intensity at wavelength λ, T is the temperature, h is Planck's constant, c is the speed of light, and k is the Boltzmann constant.

7. The optimization method according to claim 1, characterized in that: The flow rate, temperature and humidity are input into a multidimensional parameter flavor generation model to calculate the generation rate of aroma compounds, including: The aromatic compound generation rate is calculated according to formula (14): Among them, g(T(x,y,t),(x,y,t),(x,,)) is a nonlinear function of temperature, humidity and flow rate, T(x,y,t) is a nonlinear function of temperature, H(x,y,t) is a nonlinear function of humidity, V(x,y,t) is a nonlinear function of flow rate, E a is the activation energy of aromatic compound formation, R is the gas constant, T is the temperature, R f (t) is the rate of aromatic compound formation.

8. The optimization method according to claim 7, characterized in that: Constructing a flavor compound distribution map according to the generation rates of the aroma compounds and volatile components to optimize the drying conditions, including: Adjust the heating power according to formula (16): According to formula (17), the fan speed is adjusted. P fan-adjusted (t)= target ()·(1+α H ·ΔH(x,y,t)),(17) The flow rate is adjusted according to formula (18): v opt (t)= target ()·(1+β v ·|V(x,y,t)|),(18) in, is the temperature gradient, ΔH(x,,) is the humidity deviation, α T and α H are the temperature and humidity adjustment coefficients, P target is the target heating power, V target is the target flow rate, P heat-adjusted (t) is the adjusted heating power, P fan-adjusted (t) is the adjusted wind speed power, β v is the velocity correction factor, v opt (t) is the adjusted flow rate.

9. The optimization method according to claim 1, characterized in that: The flow rate, temperature and humidity are input into a multidimensional parameter flavor generation model to calculate the generation rate of aroma compounds, including: Construct a multi-dimensional parameter flavor generation model; The flow rate, temperature and humidity are input into the multi-dimensional parameter flavor generation model for training to update weights and parameters.

10. A tobacco product flavor optimization system based on a multi-spectral sensing system, characterized in that: The system comprises: A collection module is used to obtain the flow rate, temperature and humidity of the tobacco in the drying process; A processing module, connected to the acquisition module, and configured to execute the optimization method according to any one of claims 1 to 9; A display module, connected to the processing module, for displaying a flavor compound distribution map; The cloud module is connected to the acquisition module and is used to receive the flow rate, temperature and humidity and monitor the status of the tobacco in real time.

Images