Online estimation method and device for humidity of cut tobacco in cut tobacco dryer and storage medium
Through multi-spectral sensing technology and data processing algorithms, a brightness-temperature-humidity mapping model was established, which solved the problem of imprecise and inaccurate humidity monitoring in traditional tobacco drying wires, real-time and accurate estimation of tobacco humidity was achieved, and the quality and production efficiency of tobacco products were improved.
Patent Information
- Application Number
- CN202411719122.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-28
- Publication Date
- 2025-02-28
AI Technical Summary
In the traditional tobacco wire drying process, humidity monitoring has problems such as inefficiency and unreal-time, which is difficult to ensure the accuracy and consistency of humidity, affecting the quality and production efficiency of tobacco wire.
Multispectral sensing technology and data processing algorithms are used to obtain visible light and infrared images, extract the brightness and temperature information of tobacco wire, and combine the humidity response coefficient and optical flow algorithm to establish a brightness-temperature-humidity mapping model to achieve real-time and accurate estimation of tobacco wire humidity.
It improves the accuracy and real-time estimation of tobacco humidity, enhances the flexibility and adaptability of humidity control, ensures the quality and taste of tobacco products, and improves production efficiency and consistency.
Smart Images
Figure CN119523147A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an online estimation method, device and storage medium for the humidity of cut tobacco in a tobacco cut drying machine, and belongs to the technical field of tobacco detection. Background Art
[0002] Tobacco drying is an extremely critical process in tobacco processing. The quality requirements of the tobacco drying process have a significant impact on the final product characteristics of the tobacco, such as aroma, combustibility and taste. During the tobacco drying process, precise control of tobacco humidity is a core challenge in tobacco processing. It not only affects the final quality of the tobacco, but is also directly related to production efficiency and quality consistency. If the humidity is too high, the tobacco will not be dried enough, affecting subsequent processing; if the humidity is too low, it may cause the tobacco to be over-dried, increase leaf breakage, and affect its physical and chemical properties. Traditional tobacco drying process control methods have many shortcomings, especially in humidity monitoring, which often rely on fixed equipment parameters and periodic offline detection. This method is not only inefficient, but also difficult to ensure the real-time and accuracy of humidity.
[0003] With the advancement of technology, the introduction of automated sensors and intelligent control technology has provided new possibilities for the intelligentization and precision of the tobacco drying process. At present, although there are some humidity control technologies, most of them still rely on a single data source (such as temperature sensors). This single parameter monitoring method is powerless when faced with the complex nonlinear relationship between tobacco humidity and multiple factors (such as brightness, temperature, flow rate, etc.) during the tobacco drying process. Therefore, traditional humidity estimation methods are often not accurate enough to meet the needs of modern tobacco processing for high-quality tobacco.
[0004] With the development of multi-spectral sensing technology, sensors based on visible light and infrared light can realize real-time monitoring of tobacco brightness and temperature. This provides a new technical approach for building an online humidity estimation model based on multi-parameter fusion. However, how to use these sensor data to build a model that can accurately estimate tobacco humidity in real time is still a difficult point in current research. Especially in the process of drying tobacco, the dynamic fluidity and temperature changes of tobacco have complex and diverse effects on humidity. Therefore, it is urgent to provide an intelligent humidity estimation method that can integrate parameters such as brightness, temperature and flow rate and has adaptive adjustment capabilities to solve the relevant technical problems faced by existing technologies. Summary of the invention
[0005] The purpose of the present invention is to provide a method, device and storage medium for online estimation of tobacco moisture in a tobacco drying machine, aiming to achieve real-time monitoring and accurate estimation of tobacco moisture during the drying process through multi-spectral sensing technology and data processing algorithm.
[0006] In order to achieve the above objectives / solve the above technical problems, the present invention is implemented by adopting the following technical solutions.
[0007] In one aspect, the present invention provides an online method for estimating the humidity of cut tobacco in a tobacco cut drying machine, comprising: Acquire visible light images and infrared images collected by the multispectral sensor system; Extracting brightness information from the visible light image to obtain a brightness value of the tobacco, and calculating the surface temperature of the tobacco from the infrared image to obtain a temperature value of the tobacco; Inputting the brightness value and temperature value of the tobacco into a brightness-temperature-humidity mapping model to obtain an initial first humidity of the tobacco; Performing a humidity correction on the first tobacco shreds humidity in combination with the humidity response coefficient to obtain a corresponding second tobacco shreds humidity; Based on the optical flow algorithm, the movement state of the tobacco is calculated for the visible light image, and the humidity of the second tobacco is corrected for a second time in combination with the corresponding tobacco flow rate value, so as to obtain the final target tobacco humidity.
[0008] The invention obtains the target tobacco humidity through image acquisition, information extraction, initial humidity estimation, primary humidity correction and secondary humidity correction. The method not only improves the accuracy and real-time performance of tobacco humidity estimation, but also provides reliable data support for humidity control in tobacco processing. By real-time monitoring and adjustment of tobacco humidity, the quality and taste of tobacco products can be ensured to be optimal.
[0009] Optionally, extracting brightness information from the visible light image to obtain a brightness value of the tobacco shreds includes: Perform grayscale conversion on the collected visible light image to obtain a corresponding grayscale image; By manual selection or automatic segmentation method, the region containing the tobacco in the grayscale image is determined as the region of interest; The pixel values in the region of interest are counted to obtain the brightness value of the tobacco.
[0010] Optionally, the collected visible light image is converted into grayscale to obtain a corresponding grayscale image, which is expressed as: ; In the formula, , and Represent the pixel values of the red, green and blue channels respectively, Represents the brightness value of a pixel in a grayscale image.
[0011] Optionally, according to Planck's radiation law, the surface temperature of the tobacco is calculated for the infrared image in combination with the wavelength band of the infrared sensor to obtain a tobacco temperature value, wherein the relationship between the tobacco temperature value and the infrared radiation intensity is as follows: ; In the formula, Indicates wavelength The corresponding infrared radiation intensity is: Indicates the tobacco temperature value. is Planck's constant, represents the speed of light, represents the Boltzmann constant.
[0012] Optionally, based on an optical flow algorithm, the movement state of the tobacco is calculated for the visible light image to obtain a tobacco flow rate value, including: Perform grayscale conversion on the collected visible light image sequence to obtain a corresponding grayscale image; Extract key points from the grayscale image using a feature point detection algorithm; Based on the Lucas-Kanade optical flow algorithm, the extracted key points are tracked by optical flow, and the displacement vectors of the feature points between adjacent frames are calculated; By iterative solution, the optical flow velocity vector corresponding to each key point is obtained; The tobacco flow velocity value on the image plane is calculated based on the optical flow velocity vector.
[0013] Optionally, the tobacco flow velocity value on the image plane is calculated according to the optical flow velocity vector, which is expressed as: ; In the formula, Indicates the tobacco flow rate value, Indicates that the pixel is The speed of movement on the axis, Indicates that the pixel is The speed of movement on the axis.
[0014] Optionally, the visible light image and the infrared image are simultaneously captured through a synchronization trigger, so that each pair of visible light images and infrared images has the same time stamp.
[0015] Optionally, the tobacco brightness value and the tobacco temperature value are input into a brightness-temperature-humidity mapping model to obtain an initial first tobacco humidity, which is expressed as: ; In the formula, Indicates the initial first tobacco humidity, , ,…, represents the fitting coefficient obtained by fitting the experimental data, Represents the real-time tobacco brightness value obtained from the visible light sensor.
[0016] Optionally, the model adjusts the fitting coefficients through an adaptive parameter adjustment mechanism to ensure that the model can adapt to different tofu drying conditions. The adjustment formula is as follows: ; In the formula, express The moment The fitting coefficients, represents the basic fitting coefficient, represents the adaptive adjustment coefficient, Indicates the real-time temperature. Indicates the preset reference temperature.
[0017] Optionally, a humidity correction is performed on the first shredded tobacco humidity in combination with the humidity response coefficient to obtain a corresponding second shredded tobacco humidity, which is expressed as: ; In the formula, It indicates the humidity of the second tobacco cut after the first humidity correction. Represents the humidity response coefficient.
[0018] Optionally, the humidity response coefficient is calculated by the following formula: ; In the formula, represents the humidity response adjustment coefficient, Indicates the natural ambient temperature, Indicates the current tobacco temperature.
[0019] Optionally, the brightness-temperature-humidity mapping model is optimized by the following loss function: ; In the formula, Indicates the actual humidity. represents the humidity predicted by the model, Indicates the predicted humidity amount within a certain period of time.
[0020] Optionally, a secondary humidity correction is performed on the second tobacco shreds in combination with the tobacco shreds flow rate value to obtain a final target tobacco shreds humidity, which is expressed as: ; In the formula, Indicates the final target tobacco humidity. Indicates the flow rate correction factor.
[0021] In a second aspect, the present invention further provides an online estimation device for the humidity of cut tobacco in a tobacco cut drying machine, comprising: An image acquisition module is configured to acquire visible light images and infrared images collected by the multispectral sensor system; The brightness and temperature data determination module is configured to extract brightness information from the visible light image to obtain a tobacco brightness value, and calculate the tobacco surface temperature from the infrared image to obtain a tobacco temperature value; A first cut tobacco humidity determination module is configured to input the cut tobacco brightness value and the cut tobacco temperature value into a brightness-temperature-humidity mapping model to obtain an initial first cut tobacco humidity; A second cut tobacco humidity determination module is configured to perform a humidity correction on the first cut tobacco humidity in combination with a humidity response coefficient to obtain a corresponding second cut tobacco humidity; And, the target tobacco humidity determination module is configured to calculate the tobacco movement state of the visible light image based on the optical flow algorithm, and perform a secondary humidity correction on the second tobacco humidity in combination with the corresponding tobacco flow rate value obtained to obtain the final target tobacco humidity.
[0022] In a third aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for online estimation of tobacco moisture in a tobacco drying machine as described in the first aspect.
[0023] Compared with the prior art, the present invention has the following beneficial effects: The present invention collects brightness, temperature, and flow rate data to establish a multi-dimensional humidity estimation model, comprehensively considers the influence of brightness temperature, and corrects the flow rate to enhance the accuracy, which is significantly better than traditional temperature detection and reduces the influence of environmental process fluctuations. Among them, the flow rate data corrects the humidity estimation, solves the deviation caused by the change of flow velocity, maintains the accuracy under high-speed and low-speed conditions, and enhances the flexibility of humidity control; secondly, the present invention also introduces a humidity response coefficient to dynamically adjust the humidity estimation, adapt to different temperature and humidity conditions, improve the system response speed, and reduce the risk of failure under extreme working conditions; in addition, the present invention also combines machine learning algorithms, trains neural network models, fits complex nonlinear humidity changes with high precision, continuously optimizes and improves robustness accuracy, is suitable for variable tobacco characteristics and process conditions, and improves long-term stability, wherein, by automatically adjusting the model coefficient, dynamically updating with the working conditions, feedback control optimizes humidity monitoring, reduces manual intervention, and improves automated production efficiency. In short, the present invention not only improves the stability and consistency of the drying process and reduces quality fluctuations by accurately monitoring and estimating humidity online, but also further provides an efficient and flexible solution strategy for tobacco processing humidity control, which is highly practical. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 Shown is a schematic flow chart of the online estimation method of tobacco cut moisture in a tobacco cut drying machine of the present invention. DETAILED DESCRIPTION
[0025] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.
[0026] The term "and / or" is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " generally indicates that the associated objects are in an "or" relationship.
[0027] Example 1 refer to Figure 1 This embodiment introduces an online estimation method for the humidity of cut tobacco in a tobacco drying machine, which includes: Acquire visible light images and infrared images collected by the multispectral sensor system; Extracting brightness information from the visible light image to obtain a brightness value of the tobacco, and calculating the surface temperature of the tobacco from the infrared image to obtain a temperature value of the tobacco; Inputting the brightness value and temperature value of the tobacco into a brightness-temperature-humidity mapping model to obtain an initial first humidity of the tobacco; Performing a humidity correction on the first tobacco shreds humidity in combination with the humidity response coefficient to obtain a corresponding second tobacco shreds humidity; Based on the optical flow algorithm, the movement state of the tobacco is calculated for the visible light image, and the humidity of the second tobacco is corrected for a second time in combination with the corresponding tobacco flow rate value, so as to obtain the final target tobacco humidity.
[0028] In practical application, this embodiment first collects visible light images and infrared images in the tobacco drying machine in real time through an advanced multi-spectral sensor system. Then, the brightness information of the collected visible light image is extracted to obtain the brightness value of the tobacco; at the same time, the surface temperature of the tobacco is calculated for the infrared image to obtain the temperature value of the tobacco. The above two parameters are important indicators reflecting the state of the tobacco, and can provide key data for subsequent humidity estimation. Then, the extracted brightness value and temperature value of the tobacco are input into the pre-established brightness-temperature-humidity mapping model, which can accurately estimate the initial first tobacco humidity based on the brightness and temperature information.
[0029] However, due to the complex and changeable environmental conditions in the tobacco drying machine, the directly estimated humidity value may have a certain error. Therefore, this embodiment also combines the humidity response coefficient to perform a humidity correction on the first tobacco humidity to obtain a more accurate second tobacco humidity. Finally, in order to further improve the accuracy of humidity estimation, this embodiment also calculates the motion state of tobacco in the visible light image based on the optical flow algorithm, and performs a secondary humidity correction on the second tobacco humidity in combination with the corresponding tobacco flow rate value obtained. Through the above process, the target tobacco humidity is finally obtained, which provides reliable data support for humidity control in the tobacco processing process. In general, the online estimation method provided in this embodiment not only improves the accuracy and real-time performance of tobacco humidity estimation, but also helps to optimize the humidity control strategy in the tobacco processing process, thereby improving the quality and production efficiency of tobacco products.
[0030] Example 2 On the basis of Example 1, this example comprehensively considers the influence of multiple parameters such as brightness, temperature, flow rate, etc. on the estimation of tobacco humidity to achieve real-time and accurate estimation of tobacco humidity. The specific implementation process is described as follows.
[0031] 1. Acquisition and preprocessing of visible light and infrared images The visible light image and infrared image are obtained by the visible light sensor and infrared sensor in the multispectral sensor system respectively. Among them, the visible light sensor is used to collect the flow form and movement trajectory information of the tobacco, and the infrared sensor is used to measure the temperature information of the tobacco and the cavity.
[0032] The visible light sensor usually uses a camera with a wavelength of 400-700nm to regularly collect images of tobacco in the drying chamber. The visible light sensor collects each frame of image regularly through an internal or external clock to generate an image sequence. Each frame of image can be represented as a matrix ,in, and represents the pixel coordinates of the image, Represents the time dimension.
[0033] The infrared camera corresponding to the infrared sensor uses dual bands of 3-5μm and 8-14μm for acquisition, which are used to correspond to the two temperature regions of high temperature environment (such as within 550℃) and low temperature environment (such as 20℃ to 100℃). These corresponding bands provide sufficient sensitivity to monitor tobacco and cavity surfaces with a wide temperature range.
[0034] In addition, according to the speed of tobacco movement, the frame rate of visible light sensor and infrared sensor also needs to be adjusted dynamically. Generally speaking, the visible light sensor needs a higher frame rate (such as 120 frames per second) to capture the details of high-speed moving tobacco; while the frame rate of infrared sensor can be lower (30-60 frames per second) for temperature monitoring.
[0035] It is worth noting that in order to ensure that the visible light image and the infrared light image are captured at the same time, this embodiment is also equipped with a synchronization trigger in the actual application process to control the two sensors to collect images at the same time. By using the timestamp technology, it is ensured that each pair of visible light images and infrared images can have the same timestamp, so as to facilitate the subsequent fusion and data processing of related images.
[0036] Furthermore, this embodiment also performs a series of preprocessing operations on the acquired image, as described in detail below.
[0037] Image alignment and registration: Due to the different optical properties of visible light sensors and infrared sensors, there may be differences in field of view and resolution. Through image registration technology, visible light images and infrared images can be geometrically aligned to ensure that the two data can accurately correspond to the same spatial position.
[0038] Extraction and matching of feature points: Select multiple feature points (such as the edges or key nodes of tobacco) in visible light images and infrared images, and align them using image registration algorithms (such as SIFT, ORB, etc.) to ensure that the spatial positions of the images are consistent.
[0039] Image filtering and enhancement: The collected images are filtered to reduce the impact of noise and enhance the image contrast. Specifically, for visible light images, due to the complex internal environment of the device, the image may be affected by noise. Image filtering techniques (such as Gaussian filtering or median filtering) can be used to reduce the interference of noise on brightness data to achieve denoising. Image enhancement techniques (such as histogram equalization) can be used to improve the image contrast to make the brightness information more obvious, or edge detection can be used to enhance the motion contour of tobacco to achieve image enhancement. To correct the brightness deviation caused by the sensor position or illumination changes, the sensor's automatic white balance and automatic brightness adjustment functions can be used to ensure the accuracy of the image brightness and achieve image correction. Pseudo-color processing is performed on infrared images to convert temperature information into visual images. At the same time, non-uniformity correction (NUC) is used to improve the quality of infrared images to effectively reduce the error of temperature measurement.
[0040] 2. Acquisition of brightness and temperature data and initial determination of tobacco humidity value First of all, it is worth explaining that the collected visible light images are usually converted into grayscale images, that is, each collected image will be processed accordingly to generate a two-dimensional grayscale image, and the brightness information is stored in each pixel of the image accordingly. The pixel value range of the grayscale image is usually 0-255, indicating the grayscale level from black to white. In actual application, the grayscale image can be corrected according to actual needs through linear transformation, nonlinear transformation and other methods to ensure the accuracy of the grayscale value.
[0041] For color images, this can be done by Convert the pixel values of the three channels to grayscale values , the conversion formula is as follows: ; in, , and Represent the pixel values of the red, green and blue channels respectively, Represents the brightness value of a pixel in a grayscale image.
[0042] The preprocessed grayscale image contains the brightness information of each pixel. The process of extracting the brightness value of tobacco includes the following steps: Pixel value reading: For each moment of the image, read the brightness value of each pixel. These brightness values represent the light reflection intensity of the tobacco at that position, which is related to the humidity of the tobacco.
[0043] Regional average of brightness value: In order to reduce the impact of local noise, it is necessary to calculate the brightness average of the local area for each frame of the image. Specifically, the brightness in a small area (for example, 3×3 pixels) can be averaged to obtain a smoother brightness distribution map.
[0044] Visible light image brightness data :Get the image of tobacco from the visible light sensor, and extract the brightness information of each pixel through the image processing algorithm. Specifically, in the grayscale image, use manual selection, automatic segmentation and other methods to select the area containing tobacco as the region of interest (ROI); count the pixel values in the ROI and calculate the brightness value, which can be the average, median, maximum or minimum value of the pixel values in the ROI; if necessary, the calculated brightness value can be corrected using calibration curves, calibration tables and other methods to eliminate the influence of ambient light, sensor characteristics and other factors on the brightness value. Tobacco brightness data is the basis for tobacco moisture estimation.
[0045] Furthermore, the image obtained by the infrared sensor measures the temperature of the tobacco and the cavity through the law of spectral radiation. The intensity of infrared radiation energy is directly related to the temperature of the object. According to Planck's radiation law, combined with the wavelength of the infrared sensor, the surface temperature distribution of the tobacco corresponding to each position in the cavity can be calculated to provide temperature field data. Surface temperature of tobacco Infrared radiation intensity measured by infrared sensor The relationship can be expressed by the following formula: ; In the formula, Indicates wavelength The radiation intensity under Indicates the surface temperature of tobacco. is Planck's constant, represents the speed of light, Represents the Boltzmann constant. Through this formula, the influence of temperature change is taken into account to ensure the accuracy of the estimation of tobacco moisture under different temperature conditions.
[0046] Furthermore, after the brightness data and temperature data are obtained, this embodiment will adopt a brightness-temperature-humidity mapping model, combining brightness and temperature To estimate the initial tobacco humidity value The brightness-temperature-humidity mapping model is the core of humidity estimation, and the model can be fitted by experimental data.
[0047] The basic brightness-temperature-humidity model is a polynomial fitting model with the following formula: ; In the formula, represents the first tobacco humidity obtained by initial calculation, , ,…, represents the coefficient obtained by fitting the experimental data, brightness and temperature They represent the real-time data obtained from the visible light sensor and infrared sensor respectively.
[0048] In actual application, this embodiment also dynamically optimizes the above model through an adaptive parameter adjustment mechanism to dynamically adjust the fitting coefficient in the humidity model. The adjustment formula is: ; In the formula, express The moment The fitting coefficients, Take 0 to 5, represents the basic fitting coefficient, represents the adaptive adjustment coefficient, Indicates the real-time temperature. Represents the preset reference temperature. This mechanism ensures that the model maintains a high humidity estimation accuracy under different fiber drying conditions.
[0049] 3. A correction of tobacco humidity value In order to improve the dynamic response capability of humidity estimation, this embodiment also introduces a humidity response coefficient , The introduction of is dynamically adjusted according to the physical properties of tobacco, and its specific form is: ; In the formula, represents the humidity response adjustment coefficient, Indicates the natural ambient temperature, Indicates the current temperature of the tobacco, and is used to correct the humidity estimation value according to environmental changes. The humidity estimation formula combined with the humidity response coefficient is: ; In the formula, It indicates the humidity of the second tobacco after the first humidity correction process.
[0050] 4. Acquisition of flow rate data and secondary correction of tobacco humidity value First, a sequence of visible light images should be obtained. These images should contain the motion information of the tobacco. The images are grayed to simplify the calculation process. In the preprocessed images, feature point detection algorithms (such as Shi-Tomasi corner detector, etc.) are used to extract key points for optical flow tracking. These key points are usually points with large changes in grayscale values in the image, such as edges, corners, etc. Select a suitable optical flow algorithm, perform optical flow tracking on the extracted feature points, and calculate the displacement vector of the feature points between adjacent frames. Through iterative solution, the optical flow velocity vector of each feature point is obtained. . The flow velocity of the tobacco on the image plane is calculated based on the optical flow velocity vector of the feature point. This usually requires averaging or weighted averaging the flow velocities of multiple feature points to obtain a more accurate flow velocity value. It should be noted that since the tobacco in the image may present different shapes and motion states, it is necessary to calculate and analyze the tobacco in different areas separately.
[0051] Specifically, the calculation method of the tobacco flow rate in this embodiment is mainly based on the optical flow method (Optical Flow), which is a calculation method for estimating the movement speed of an object by analyzing the pixel movement in continuous image frames. The relevant mathematical model and specific implementation steps are described as follows.
[0052] The optical flow method estimates the speed of an object in an image by analyzing the brightness changes caused by the object's movement in the image. The pixel brightness value is ,in, Represents pixel coordinates, assuming that the object is at time A displacement has occurred, moving to a new location , then the brightness value It also changes accordingly.
[0053] According to the basic assumption of optical flow, if the brightness of an object remains unchanged in a short period of time, the corresponding expression can be shown as follows: ; Furthermore, by performing Taylor expansion on the above expression and ignoring the higher-order terms, the following optical flow constraint equation can be obtained: ; In the formula, , indicating that the pixel is The velocity component in the direction, , indicating that the pixel is The velocity component in the direction, and Respectively represent the images in and Directional brightness gradient, Indicates the rate of change of image brightness over time.
[0054] It can be noted that the above equation contains two unknowns, namely the velocity components and , but there is only one equation, so additional constraints need to be introduced to solve it.
[0055] Therefore, this embodiment uses the Lucas-Kanade optical flow algorithm to solve the corresponding optical flow field. The proposed method assumes that the movement of objects is consistent within a small local window (such as a 3×3 pixel area). Based on this assumption, the system can use multiple pairs of equations to solve the optical flow.
[0056] Specifically, assuming that within a local window Ω, the motion of all pixels is the same, an optical flow constraint equation can be generated for each pixel. N pixels, we have N Equations: ; In the formula, for The matrix represents the image brightness gradient, and the specific form is: ; is the optical flow vector to be solved, expressed in and The component of velocity in the direction.
[0057] yes The column vector represents the rate of change of brightness over time, and the specific form is: ; because In the equation usually , so the optical flow vector can be solved by using the least squares method . Minimize the error of the following equation: ; Through the above equation, the velocity component of the tobacco in the local window can be obtained: and .
[0058] Furthermore, according to the calculated flow velocity vector, the flow velocity of each pixel is calculated using the following formula: ; In the formula, Indicates the flow velocity of each pixel, corresponding to the tobacco flow velocity value. and Respectively represent the flow rate in and Directional velocity component.
[0059] This embodiment uses an optical flow algorithm on continuous visible light image frames to calculate the velocity vector field of the tobacco shreds and obtain the velocity magnitude of each point. Since the velocity of the tobacco shreds affects the measurement of humidity, this embodiment performs a secondary correction on the initially estimated humidity of the tobacco shreds based on the velocity data. The humidity-velocity correction model formula is as follows: ; In the formula, Indicates the final target tobacco humidity value, Represents the flow rate correction factor. Through this correction formula, the humidity estimation value can dynamically adapt to the change of flow rate and further improve the accuracy. This correction formula is used to compensate for the influence of flow rate on humidity estimation to ensure the accuracy of humidity estimation.
[0060] It is worth noting that in order to enhance the robustness and adaptability of the humidity estimation model, this embodiment also optimizes the brightness-temperature-humidity model through a neural network. The neural network model is trained with historical humidity, brightness and temperature data to fit complex nonlinear relationships. Its structure includes: Input layer: brightness and temperature ; Hidden layer: Use 1 to 2 hidden layers of activation functions for learning complex nonlinear relationships; Output layer: Output humidity .
[0061] Through training, the model can effectively fit the complex relationship between brightness, temperature and humidity, and further improve the accuracy of humidity estimation. Among them, the loss function of the target prediction model is the mean square error (MSE), and its optimization goal is to minimize the error between the model predicted humidity and the actual humidity. The expression of the loss function is as follows: ; In the formula, Indicates the actual humidity. represents the humidity predicted by the model, Indicates the predicted humidity amount within a certain period of time.
[0062] In addition, this embodiment also combines historical data and real-time feedback information, and uses a cloud-based machine learning algorithm to continuously optimize the model. Specifically, the model parameters are dynamically updated through the gradient descent method or the Adam optimizer to ensure the long-term accuracy and stability of humidity estimation.
[0063] In actual application, based on the estimated humidity data, the relevant system corresponding to this embodiment can generate a humidity distribution diagram in real time and control the operating parameters of the tofu drying equipment through a feedback mechanism. The control strategy at least includes: Wind speed adjustment: If the humidity is detected to be too high or too low, the system will adjust the fan speed to speed up or slow down the change in humidity; Heating power adjustment: According to the estimated humidity value, the heating power is automatically adjusted to ensure that the temperature distribution in the cavity meets the needs of humidity control.
[0064] At the same time, this embodiment can also support uploading humidity data to the cloud through the Internet of Things (IoT) technology. The cloud system can optimize the humidity estimation model in the long term through big data analysis and machine learning models. At the same time, the cloud system can also generate historical humidity analysis reports to help operators better optimize the drying process.
[0065] In addition, in order to further verify the effectiveness of the provided online estimation method for the humidity of cut tobacco in the tobacco drying machine, this embodiment also collects the flow rate, temperature, brightness and humidity data in the actual drying process, and compares the model-estimated humidity value with the actual humidity value to make an intuitive evaluation of the system's humidity estimation accuracy and the dynamic response capability of the model.
[0066] Experimental preparation stage: Experimental equipment: Tobacco drying machine with multi-spectral sensing system.
[0067] Sensor configuration: used to collect brightness Visible light sensor, and temperature sensor Infrared sensor.
[0068] Flow rate monitoring system: Extract the flow rate of tobacco using optical flow algorithm .
[0069] Humidity sensor: used to collect the actual humidity of tobacco , for comparison with the estimated value.
[0070] Experimental application stage:
[0071] Set different experimental conditions (temperature, wind speed, humidity, etc.) to simulate the changing conditions in actual production. The multi-spectral sensing system collects brightness, temperature and flow rate data in real time and inputs them into the humidity estimation model to calculate the humidity value. Calculate. At the same time, use the humidity sensor to measure the actual humidity . Compare the estimated humidity with the actual humidity, calculate the error and analyze the estimation accuracy of the model.
[0072] Experimental data: The following are experimental data under three different working conditions, which record the brightness, temperature, flow rate, estimated humidity and actual humidity at each moment in the drying process.
[0073]
[0074] Experimental analysis: Humidity estimation accuracy: Under all working conditions, the error between the estimated humidity and the actual humidity is kept within 2%, which proves that the humidity estimation model of this embodiment has high accuracy. Under different flow rates and temperature conditions, the humidity estimation model can dynamically adjust the estimated value and keep it relatively consistent with the actual measured value.
[0075] Flow rate correction effect: In the case of large changes in flow rate (such as the third moment of working condition 1 and the first moment of working condition 3), the model with flow rate correction can effectively adjust the humidity estimation value and avoid the increase of estimation error. The flow rate correction mechanism enhances the flexibility of humidity estimation and enables it to adapt to the dynamic flow of tobacco in the tobacco drying machine.
[0076] Adaptive adjustment and temperature response: Under different temperature conditions, the model can dynamically correct the estimation coefficients through an adaptive parameter adjustment mechanism to adapt to the impact of temperature fluctuations on humidity estimation. For example, the temperature fluctuation in condition 2 is large, but the humidity estimation value can still maintain a high accuracy.
[0077] In general, this embodiment adopts the above technical solution, introduces a multi-spectral sensing system and an innovative brightness-temperature-humidity mapping model, and realizes real-time and accurate estimation of tobacco humidity. Compared with the traditional humidity monitoring method, the technical effects of this embodiment can be comprehensively reflected in the following aspects: Multi-parameter fusion improves accuracy: A multi-dimensional humidity estimation model is established by simultaneously collecting the brightness, temperature and flow rate data of the tobacco through a multi-spectral sensing system. The model combines the effects of brightness and temperature on humidity and enhances the accuracy of humidity estimation through flow rate correction. Compared with traditional single temperature detection, this multi-parameter fusion method significantly improves the accuracy of humidity estimation and reduces the impact of environmental fluctuations or changes in process conditions on humidity estimation.
[0078] Dynamically adapt to different working conditions: By introducing the humidity response coefficient, this embodiment can dynamically adjust the humidity estimation value to adapt it to different temperature and environmental conditions. The adaptive adjustment mechanism ensures the robustness of the model under various tofu drying conditions, which not only improves the response speed of the system, but also reduces the possibility of failure of traditional humidity control methods under extreme conditions.
[0079] Adaptive adjustment and feedback mechanism: The adaptive parameter adjustment mechanism in the present invention can automatically adjust the model coefficients according to the real-time temperature and brightness data, so that the humidity estimation model can be dynamically updated as the working conditions change. This mechanism further optimizes the humidity monitoring process through feedback control, reduces the need for manual intervention, and improves the automation and production efficiency of the system.
[0080] Flow rate correction improves humidity control flexibility: By using flow rate data to correct humidity estimation, the present invention effectively solves the problem of humidity estimation deviation caused by changes in tobacco flow rate during the drying process. The flow rate correction model ensures that humidity estimation can still maintain accuracy under high or low flow conditions, enhancing the flexibility of humidity control.
[0081] Improve production efficiency and quality consistency: By realizing accurate online monitoring and estimation of tobacco cut humidity, the present invention significantly improves the stability and consistency of the tobacco cut drying process and reduces quality fluctuations caused by improper humidity. In addition, intelligent humidity monitoring and adaptive adjustment greatly reduce energy consumption and manual intervention, improve production efficiency, and ensure the continuous output of high-quality tobacco cuts.
[0082] Machine learning optimization enhances robustness: This embodiment combines machine learning algorithms to train a neural network model with historical data to achieve high-precision fitting of complex nonlinear humidity changes. With the input of more real-time data, the model can be continuously optimized to improve the robustness and accuracy of humidity estimation. This feature is particularly suitable for the ever-changing tobacco characteristics and process conditions in the production environment, which greatly improves the long-term stability of the system.
[0083] In summary, the method proposed in this embodiment not only improves the accuracy and stability of tobacco cut moisture estimation, but also improves the automation level of the tobacco cut drying process through intelligent and adaptive control mechanisms. This provides an efficient and flexible solution for humidity control in modern tobacco processing and has broad application prospects.
[0084] Example 3 Based on the same inventive concept as that of Embodiment 1 or 2, this embodiment introduces an online estimation device for the humidity of cut tobacco in a tobacco cut drying machine, which comprises: An image acquisition module is configured to acquire visible light images and infrared images collected by the multispectral sensor system; The brightness and temperature data determination module is configured to extract brightness information from the visible light image to obtain a tobacco brightness value, and calculate the tobacco surface temperature from the infrared image to obtain a tobacco temperature value; A first cut tobacco humidity determination module is configured to input the cut tobacco brightness value and the cut tobacco temperature value into a brightness-temperature-humidity mapping model to obtain an initial first cut tobacco humidity; A second cut tobacco humidity determination module is configured to perform a humidity correction on the first cut tobacco humidity in combination with a humidity response coefficient to obtain a corresponding second cut tobacco humidity; And, the target tobacco humidity determination module is configured to calculate the tobacco movement state of the visible light image based on the optical flow algorithm, and perform a secondary humidity correction on the second tobacco humidity in combination with the corresponding tobacco flow rate value obtained to obtain the final target tobacco humidity.
[0085] refer to Figure 1 The online estimation device for the humidity of shredded tobacco in a tobacco drying machine can execute the online estimation method for the humidity of shredded tobacco in a tobacco drying machine introduced in the above-mentioned embodiment 1 or 2, and the specific functional implementation of each functional module is not repeated here.
[0086] Example 4 This embodiment introduces a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the online estimation method of the humidity of cut tobacco in a tobacco drying machine as described in Embodiment 1 or 2 are implemented.
[0087] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0088] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0089] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0090] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0091] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. An online estimation method for the humidity of cut tobacco in a tobacco drying machine, characterized in that: include: Acquire visible light images and infrared images collected by the multispectral sensor system; Extracting brightness information from the visible light image to obtain a brightness value of the tobacco, and calculating the surface temperature of the tobacco from the infrared image to obtain a temperature value of the tobacco; Inputting the brightness value and temperature value of the tobacco into a brightness-temperature-humidity mapping model to obtain an initial first humidity of the tobacco; Performing a humidity correction on the first tobacco shreds humidity in combination with the humidity response coefficient to obtain a corresponding second tobacco shreds humidity; Based on the optical flow algorithm, the movement state of the tobacco is calculated for the visible light image, and the humidity of the second tobacco is corrected for a second time in combination with the corresponding tobacco flow rate value, so as to obtain the final target tobacco humidity.
2. The method for online estimation of tobacco cut moisture in a tobacco cut drying machine according to claim 1, characterized in that: Extracting brightness information from the visible light image to obtain a brightness value of the tobacco, including: Perform grayscale conversion on the collected visible light image to obtain a corresponding grayscale image; By manual selection or automatic segmentation method, the region containing the tobacco in the grayscale image is determined as the region of interest; Counting the pixel values in the region of interest to obtain the brightness value of the tobacco; Among them, the collected visible light image is converted into grayscale to obtain the corresponding grayscale image, which is expressed as: ； In the formula, , and Represent the pixel values of the red, green and blue channels respectively, Represents the brightness value of a pixel in a grayscale image.
3. The method for online estimation of tobacco cut moisture in a tobacco cut drying machine according to claim 2, characterized in that: According to Planck's radiation law, the surface temperature of the tobacco is calculated for the infrared image in combination with the wavelength of the infrared sensor to obtain the tobacco temperature value, wherein the relationship between the tobacco temperature value and the infrared radiation intensity is as follows: ； In the formula, Indicates wavelength The corresponding infrared radiation intensity is: Indicates the tobacco temperature value. is Planck's constant, represents the speed of light, represents the Boltzmann constant.
4. The method for online estimation of tobacco cut moisture in a tobacco cut drying machine according to claim 3, characterized in that: Based on the optical flow algorithm, the movement state of the tobacco is calculated for the visible light image to obtain the tobacco flow rate value, including: Perform grayscale conversion on the collected visible light image sequence to obtain a corresponding grayscale image; Extract key points from the grayscale image using a feature point detection algorithm; Based on the Lucas-Kanade optical flow algorithm, the extracted key points are tracked by optical flow, and the displacement vectors of the feature points between adjacent frames are calculated; By iterative solution, the optical flow velocity vector corresponding to each key point is obtained; According to the optical flow velocity vector, the tobacco flow velocity value on the image plane is calculated and expressed as: ； In the formula, Indicates the tobacco flow rate value, Indicates that the pixel is The speed of movement on the axis, Indicates that the pixel is The speed of movement on the axis; The visible light image and the infrared image are collected simultaneously through a synchronization trigger, so that each pair of the visible light image and the infrared image has the same time stamp.
5. The method for online estimation of tobacco cut moisture in a tobacco cut drying machine according to claim 4, characterized in that: The brightness value and temperature value of the tobacco are input into the brightness-temperature-humidity mapping model to obtain the initial first humidity of the tobacco, which is expressed as: ； In the formula, Indicates the initial first tobacco humidity, , ,…, represents the fitting coefficient obtained by fitting the experimental data, Indicates the brightness value of tobacco; The model adjusts the fitting coefficient through an adaptive parameter adjustment mechanism to ensure that the model can adapt to different drying conditions. The adjustment formula is as follows: ； In the formula, express The moment The fitting coefficients, represents the basic fitting coefficient, represents the adaptive adjustment coefficient, Indicates the real-time temperature. Indicates the preset reference temperature.
6. The method for online estimation of tobacco cut moisture in a tobacco cut drying machine according to claim 5, characterized in that: The humidity of the first tobacco shreds is corrected once in combination with the humidity response coefficient to obtain the corresponding humidity of the second tobacco shreds, which is expressed as: ； In the formula, It indicates the humidity of the second tobacco cut after the first humidity correction. represents the humidity response coefficient; Wherein, the humidity response coefficient is calculated by the following formula: ； In the formula, represents the humidity response adjustment coefficient, Indicates the natural ambient temperature, Indicates the current tobacco temperature.
7. The method for online estimation of tobacco cut moisture in a tobacco cut drying machine according to claim 6, characterized in that: The brightness-temperature-humidity mapping model is optimized by the following loss function: ； In the formula, Indicates the actual humidity. represents the humidity predicted by the model, Indicates the predicted humidity amount within a certain period of time.
8. The method for online estimation of tobacco cut moisture in a tobacco cut drying machine according to claim 6 or 7, characterized in that: The second tobacco shreds are subjected to secondary humidity correction in combination with the tobacco shreds flow rate value to obtain the final target tobacco shreds humidity, which is expressed as: ； In the formula, Indicates the final target tobacco humidity. Indicates the flow rate correction factor.
9. An online estimation device for the humidity of cut tobacco in a tobacco drying machine, characterized in that: include: An image acquisition module is configured to acquire visible light images and infrared images collected by the multispectral sensor system; The brightness and temperature data determination module is configured to extract brightness information from the visible light image to obtain a brightness value of the tobacco, and to calculate the surface temperature of the tobacco from the infrared image to obtain a temperature value of the tobacco; A first cut tobacco humidity determination module is configured to input the cut tobacco brightness value and the cut tobacco temperature value into a brightness-temperature-humidity mapping model to obtain an initial first cut tobacco humidity; A second cut tobacco humidity determination module is configured to perform a humidity correction on the first cut tobacco humidity in combination with a humidity response coefficient to obtain a corresponding second cut tobacco humidity; And, the target tobacco humidity determination module is configured to calculate the tobacco movement state of the visible light image based on the optical flow algorithm, and perform a secondary humidity correction on the second tobacco humidity in combination with the corresponding tobacco flow rate value obtained to obtain the final target tobacco humidity.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for online estimation of the humidity of cut tobacco in a tobacco drying machine according to any one of claims 1 to 8 are implemented.