Residual cigarette processing flow optimization system

Abstract

The invention discloses a residual cigarette processing flow optimization system, which comprises a trademark unit used for storing residual cigarettes according to trademarks; the weighing unit is used for acquiring weighing data of the residual cigarettes and the cut tobaccos with selected brands; the hyperspectral unit is used for collecting a hyperspectral image of the residual cigarette with the selected trademark, obtaining hyperspectral data of the residual cigarette through a hyperspectral detection model I, separating cigarette paper, a filter stick and tobacco shreds when the hyperspectral data of the residual cigarette meets the requirement, collecting a hyperspectral image of the tobacco shreds, and outputting the hyperspectral image of the tobacco shreds; hyperspectral data of the tobacco shreds are obtained through a hyperspectral detection model II; and the early warning unit is used for judging whether sundries, oil smoke and paper sheets exist in the defective cigarettes and the tobacco shreds or not according to the obtained hyperspectral data and a preset early warning range, and when the result is' yes', early warning is carried out and the defective cigarettes and the tobacco shreds are removed. According to the method, the components of the residual cigarettes are accurately and rapidly judged, the screening efficiency and accuracy of sundries, oil smoke and paper scraps are improved, the flow time of the optimized residual cigarette treatment system is shortened, anti-machine early warning is added, and the working efficiency and accuracy are improved.

Description

Technical Field

[0001] The invention relates to the field of residual smoke treatment in cigarette manufacturing, and in particular to a residual smoke treatment process optimization system. Background Art

[0002] Residual tobacco treatment is the process of recycling and treating a large number of defective cigarettes produced during cigarette production. The rolling and packaging workshop collects the residual cigarettes by brand and transports them to the residual tobacco treatment area. In order to avoid tobacco waste, the recycling process must be received, weighed, and recorded by statisticians (about 2 hours). The residual tobacco treatment personnel test the received residual tobacco by brand according to the process requirements, process the tobacco without problems, and weigh and record the tobacco and mouthpieces according to categories (about 3 hours). The process personnel then re-test the tobacco without problems (the oil smoke test uses the oven method, which takes 2 hours) and record it; finally, a series of operations such as packing and weighing are carried out.

[0003] That is, the existing residual smoke treatment process has problems such as lengthy time and low labor efficiency. Summary of the invention

[0004] In view of the above shortcomings, the present invention provides a residual smoke processing process optimization system. The present invention automatically identifies residual smoke, automatically detects and determines the residual smoke through the optimization system, transforms the residual smoke processing from manual to automatic processing, greatly improves the detection efficiency, and optimizes the processing process. The technical solution adopted by the present invention to solve the technical problem is: Residual smoke treatment process optimization system, including Brand unit, used to store remaining cigarettes by brand; Weighing unit, used to obtain the weighing data of the remaining cigarettes and tobacco of the selected brand; The detection unit is used to collect the hyperspectral image of the remaining cigarette of the selected brand, and obtain the hyperspectral data of the remaining cigarette through the hyperspectral detection model I, and when the hyperspectral data of the remaining cigarette meets the requirements, separate the cigarette paper, the filter rod and the tobacco, collect the hyperspectral image of the tobacco, and obtain the hyperspectral data of the tobacco through the hyperspectral detection model II; The early warning unit is used to determine whether there are debris, oil smoke and paper pieces in the remaining cigarettes and tobacco according to the acquired hyperspectral data and the preset early warning range. If the result is "yes", the early warning is initiated and the remaining cigarettes and tobacco corresponding to the image section are removed; The self-learning unit is used to enable the hyperspectral detection model I and the hyperspectral detection model II to perform self-learning through data accumulation, thereby continuously improving the model accuracy.

[0005] Furthermore, the hyperspectral detection model I and the hyperspectral detection model II collect hyperspectral images of residual cigarettes and tobacco shreds through hyperspectral imaging, correct the hyperspectral images, extract the regions of interest, and obtain hyperspectral data.

[0006] Furthermore, the hyperspectral detection model II is also used to detect the moisture content of cut tobacco.

[0007] Furthermore, the system also includes a box storage unit, which is used to weigh, mark, and automatically box the tobacco when the judgment result of the hyperspectral data obtained by the hyperspectral detection model II is "no".

[0008] The beneficial effects brought by the present invention are: The present invention aims to optimize the residual cigarette processing process, convert the weighing, detection and other tasks from manual to automatic weighing, and use a hyperspectral detection model to perform image recognition of residual cigarettes and tobacco, thereby achieving accurate and rapid determination of the components of residual cigarettes and improving the screening efficiency and accuracy of debris, oil smoke and paper scraps.

[0009] The optimized residual smoke treatment system has shortened the process time and added machine-protected early warning to improve work efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0011] Figure 1 is a system block diagram of the present invention; Figure 2 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0012] The following will be combined with the embodiments of the present invention and the accompanying drawings to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] This embodiment relates to a residual smoke processing process optimization system, referring to Figure 1 , Figure 2 , this system includes Brand unit, used to store remaining cigarettes by brand; Weighing unit, used to obtain the weighing data of the remaining cigarettes and tobacco of the selected brand, and automatically upload it to the system; Afterwards, the remaining cigarettes of the selected brand are fed into the conveyor belt and subjected to hyperspectral detection by the hyperspectral unit: The hyperspectral detection unit is used to collect the hyperspectral image of the remaining cigarettes of the selected brand, and obtain the hyperspectral data of the remaining cigarettes through the hyperspectral detection model I, and when the hyperspectral data of the remaining cigarettes meet the requirements, separate the cigarette paper, the filter rod and the tobacco, collect the hyperspectral image of the tobacco, and obtain the hyperspectral data of the tobacco through the hyperspectral detection model II; The early warning unit is used to determine whether there are debris, oil smoke and paper pieces in the remaining cigarettes and tobacco according to the acquired hyperspectral data and the preset early warning range. If the result is "yes", the early warning is initiated and the remaining cigarettes and tobacco corresponding to the image section are removed; During the execution process, the residual cigarette is detected and judged for oil smoke and debris based on the hyperspectral detection model I. If its hyperspectral data does not meet the preset warning range, it indicates that the residual cigarette corresponding to the image sampling section contains paper pieces, debris, oil smoke, etc., then the system enters the warning and removes the residual cigarette; if its hyperspectral data meets the preset warning range, the next step of processing is carried out, that is, the separation of cigarette paper, filter rod and tobacco, and the tobacco is transported for secondary hyperspectral detection. The moisture content and whether the tobacco still contains paper pieces, debris, oil smoke, etc. are detected based on the hyperspectral detection model II, and the hyperspectral data is combined to make a judgment. If there is unqualified tobacco, the system enters the warning and removes the tobacco corresponding to the image sampling section. If there is no problem, the next step of packaging and storage is carried out.

[0014] This embodiment specifically describes the principle of residual smoke warning determination in detail in conjunction with the following experimental examples.

[0015] 1. Purpose of the experiment Hyperspectral imaging technology is used to detect mildew in tobacco leaves. The hyperspectral data is output through the detection model and combined with the early warning range to determine whether the tobacco leaves have mildew defects.

[0016] 2. Experimental Materials and Methods Experimental materials: A certain number of tobacco leaf samples were selected, some of which were known to be moldy.

[0017] Experimental equipment: Hyperspectral imaging system, including hyperspectral camera, light source, image acquisition card, etc.

[0018] Experimental methods: Place the tobacco leaf sample on the detection platform of the hyperspectral imaging system to ensure that the sample surface is flat and the light is uniform.

[0019] Adjust the parameters of the hyperspectral camera, such as spectral range, resolution, exposure time, etc., to meet the detection requirements of tobacco leaf samples.

[0020] Start the hyperspectral imaging system, scan the tobacco leaf samples, and obtain their hyperspectral image data.

[0021] The acquired hyperspectral image data is preprocessed, including denoising and correction, to improve data quality.

[0022] 3. Hyperspectral Data Acquisition Process Spectral data collection: Pixel points of tobacco leaf samples are selected in a specific band (such as 400~1000nm).

[0023] For each selected pixel, record its spectral reflectance or absorbance data at different wavelengths.

[0024] Data organization and storage: The collected spectral data are sorted according to pixels to form a hyperspectral data cube.

[0025] Each pixel in the data cube corresponds to a complete spectral curve, which contains the spectral information of the pixel at different wavelengths.

[0026] Save the organized hyperspectral data in an appropriate file format (such as ENVI, HDF, etc.) for subsequent analysis and processing.

[0027] Hyperspectral detection model output and judgment: The established hyperspectral detection models I and II are used to process and analyze the hyperspectral data of tobacco leaf sample images pixel by pixel.

[0028] The model does not directly output the judgment result, but outputs the processed hyperspectral data (that is, data processed by the model algorithm but maintaining the spectral characteristics).

[0029] Combined with the preset warning range (such as the spectral characteristic threshold of moldy tobacco leaves), the output hyperspectral data is judged.

[0030] If the spectral feature of a certain pixel point matches the warning range, it is determined that the tobacco leaf area corresponding to the pixel point has a mold defect.

[0031] 4. Experimental Results and Analysis By collecting, sorting and analyzing the hyperspectral data of tobacco leaf samples, we can clearly see the difference in spectral characteristics between the moldy area and the normal area. Experiments have shown that the mold defects in tobacco leaves can be accurately identified by using the hyperspectral data output by the hyperspectral detection model and combining it with the early warning range for judgment.

[0032] As an optimization of the above embodiment, the system further includes a self-learning unit, which is used to enable the hyperspectral detection model I and the hyperspectral detection model II to perform self-learning through data accumulation, thereby continuously improving the model accuracy.

[0033] As an optimization of the above embodiment, the system further includes a box storage unit, which is used to weigh, mark, and automatically box the tobacco when the judgment result of the hyperspectral data obtained by the hyperspectral detection model II is "no".

[0034] As an optional implementation of this embodiment, the hyperspectral detection model I and the hyperspectral detection model II collect hyperspectral images of residual cigarettes and tobacco shreds through hyperspectral imaging, correct the hyperspectral images, extract the region of interest, and obtain hyperspectral data.

[0035] The detection method of the hyperspectral detection model I and II includes 1. Experimental Materials and Methods Experimental materials: Select a certain number of residual cigarettes and tobacco samples to ensure that the samples are representative and contain different types of impurities and defects.

[0036] Experimental equipment: hyperspectral imaging system (including hyperspectral camera, light source, image acquisition card, etc.), computer workstation, image processing software.

[0037] Experimental methods: A hyperspectral imaging system is used to scan residual cigarettes and tobacco samples to obtain hyperspectral images.

[0038] The acquired hyperspectral images are preprocessed, including correction, denoising and other steps.

[0039] Extract the region of interest (ROI), which is the area containing impurities or defects.

[0040] The extracted ROI is further processed using hyperspectral detection models I and II to output hyperspectral data.

[0041] 2. Image Recognition and Processing Hyperspectral image acquisition Steps: Place the remaining cigarettes and tobacco samples on the detection platform of the hyperspectral imaging system, adjust the light source and camera parameters to ensure clear image quality and complete spectral information. Start the system, scan the samples, and obtain hyperspectral images.

[0042] Image Correction Purpose: To eliminate image distortion, shadows, and uneven lighting, and improve image quality.

[0043] Steps: Use image processing software to perform geometric correction, radiation correction and spectral correction on the hyperspectral image. Geometric correction is used to eliminate image distortion; radiation correction is used to adjust the image brightness to make it more uniform; spectral correction is used to ensure the accuracy of spectral information.

[0044] Region of Interest (ROI) Extraction Purpose: To extract regions containing impurities or defects from hyperspectral images for subsequent analysis.

[0045] Steps: Use manual or automatic segmentation tools in image processing software to extract ROI based on color, texture and other features. Manual segmentation requires manual operation and is suitable for complex or irregular shaped ROIs; automatic segmentation is based on a preset algorithm and can quickly and accurately extract ROIs.

[0046] Hyperspectral data output Purpose: To obtain hyperspectral data from the extracted ROI for subsequent analysis and judgment.

[0047] Steps: Use hyperspectral detection models I and II to process the extracted ROI. Model I may be used for primary data preprocessing and feature extraction, such as removing background noise and enhancing spectral features; Model II is used for further data analysis and judgment, such as identifying impurity types and calculating impurity content based on ROI shape and area. At the same time, the model outputs processed hyperspectral data, including spectral reflectance, absorptivity and other parameters.

[0048] 3. Experimental Results and Analysis By collecting, correcting, extracting ROI and outputting data from the hyperspectral images of residual cigarettes and tobacco samples, we obtained hyperspectral data containing impurities and defects. These data can be used for subsequent analysis and judgment, and a detection and classification model can be established using machine learning algorithms to achieve automatic identification and classification of impurities and defects.

[0049] By applying this embodiment, residual smoke of different brands is processed, and the process time and processing results are statistically shown in Table 1 and Table 2 respectively.

[0050] Table 1 Record of warning status of residual smoke treatment for different brands

[0051] It can be seen that the system of the present invention can realize accurate and rapid determination of the components of residual cigarettes, and improve the efficiency and accuracy of screening of debris, oil smoke and paper pieces.

[0052] Table 2 Time record of the processing process of different brands of residual cigarettes

[0053] It can be seen that the system of the present invention can effectively shorten the residual smoke treatment process time and improve work efficiency.

[0054] As an optimization of the above embodiment, the hyperspectral detection model II is also used to detect the moisture content of cut tobacco. Referring to Table 3, it can be seen that the moisture content of recycled cut tobacco detected by this system has a high accuracy.

[0055] Table 3 Comparison of moisture content detected in tobacco by this system and moisture content detected in oven

[0056] The method for detecting moisture content of cut tobacco by using the hyperspectral detection model II in this embodiment includes: Spectral data acquisition: Use a hyperspectral imaging system to scan target objects (such as soil, plants, food, etc.) and obtain their hyperspectral image data. These data contain the reflectivity or absorption rate information of the target objects at different wavelengths.

[0057] Spectral data processing: The collected spectral data is preprocessed, including denoising, correction and other steps to improve data quality. Then, a specific algorithm or model is used to extract features from the spectral data to extract spectral features related to the moisture content of the target object.

[0058] Establishing a quantitative relationship between moisture content and spectral characteristics: Establish a quantitative relationship model between the moisture content of the target object and its spectral characteristics through regression analysis, machine learning, etc. This model can be linear or nonlinear, depending on the characteristics of the target object and the distribution of the spectral data.

[0059] Moisture content calculation: The hyperspectral data of the target object to be measured is input into the established model, and the moisture content value is calculated by the model. This value can be an absolute value (such as the percentage of moisture content) or a relative value (such as the difference or ratio compared with a certain standard value).

[0060] It should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. Residual smoke processing process optimization system, characterized by: include Brand unit, used to store remaining cigarettes by brand; Weighing unit, used to obtain the weighing data of the remaining cigarettes and tobacco of the selected brand; The detection unit is used to collect the hyperspectral image of the remaining cigarette of the selected brand, and obtain the hyperspectral data of the remaining cigarette through the hyperspectral detection model I, and when the hyperspectral data of the remaining cigarette meets the requirements, separate the cigarette paper, the filter rod and the tobacco, collect the hyperspectral image of the tobacco, and obtain the hyperspectral data of the tobacco through the hyperspectral detection model II; The early warning unit is used to determine whether there are debris, oil smoke and paper pieces in the remaining cigarettes and tobacco according to the acquired hyperspectral data and the preset early warning range. When the result is "yes", it enters the early warning and removes the remaining cigarettes and tobacco corresponding to the image section.

2. The residual smoke processing process optimization system according to claim 1, characterized in that: The system also includes a self-learning unit, which is used to enable the hyperspectral detection model I and the hyperspectral detection model II to perform self-learning through data accumulation, thereby continuously improving the model accuracy.

3. The residual smoke processing process optimization system according to claim 1, characterized in that: The hyperspectral detection model I and the hyperspectral detection model II collect hyperspectral images of residual cigarettes and tobacco shreds through hyperspectral imaging, correct the hyperspectral images, extract the regions of interest, and obtain hyperspectral data.

4. The residual smoke processing process optimization system according to claim 1, characterized in that: The hyperspectral detection model II is also used to detect the moisture content of cut tobacco.

5. The residual smoke processing process optimization system according to claim 1, characterized in that: The system also includes a box storage unit, which is used to weigh, mark, and automatically box the tobacco when the judgment result of the hyperspectral data obtained by the hyperspectral detection model II is "no".

Images