Tobacco leaf appearance quality control method based on microenvironment intelligent maintenance and related equipment

Abstract

The invention discloses a tobacco appearance quality control method and related equipment based on microenvironment intelligent maintenance, a tobacco appearance quality control model is constructed, an appearance feature extraction network is utilized, feature extraction is performed on appearance time sequence data of tobacco based on aligned timestamps, and features reflecting appearance changes of the tobacco are obtained; and performing feature extraction on the time sequence data of the environment where the tobacco leaves are located based on the aligned timestamps through an environment feature extraction network to obtain features of environment change. The appearance change characteristics and the environment change characteristics are mapped into a control network for appearance quality change analysis, the influence of the appearance change characteristics and the environment change characteristics on appearance quality change is found through the change relation between the appearance change characteristics and the environment change characteristics after alignment and time stamping, and the environment parameters are adjusted based on the appearance quality change analysis. And the tobacco appearance quality change speed is accurately regulated and controlled, so that the tobacco is in the best appearance state during use.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial data processing, and in particular to a tobacco leaf appearance quality control method based on micro-environment intelligent maintenance and related equipment. Background Art

[0002] In the tobacco industry, tobacco leaves, as a core raw material, play a decisive role in the quality of the final tobacco product, and appearance is a key factor in measuring tobacco leaf quality. Traditional tobacco leaf curing methods rely heavily on manual experience, making it difficult to achieve precise control and fully meet the individual needs of tobacco leaves at different growth stages and during storage.

[0003] In recent years, intelligent microenvironmental curing technology has gradually emerged and is being applied to tobacco leaf lifecycle management. Leveraging various sensors, controllers, and intelligent systems, this technology enables real-time monitoring, precise regulation, and intelligent management of the microenvironment in which tobacco leaves grow and store. This significantly improves curing efficiency and quality, enabling personalized curing and environmental management throughout the entire in-store curing lifecycle. However, despite this, several pressing challenges remain in controlling tobacco leaf appearance quality.

[0004] Tobacco leaf appearance quality is determined by a combination of indicators, including color, maturity, leaf structure, oil content, and chroma. These indicators are highly susceptible to a variety of factors, including environmental parameters in the warehouse, such as temperature, humidity, light, ventilation, and stacking methods. While some research has been conducted on regulating these environmental parameters, most focus on general adjustments to the overall tobacco growth or storage environment, and do not adequately address the core objective of tobacco leaf appearance quality. A systematic strategy, oriented towards precise control of tobacco leaf appearance quality, is lacking, capable of dynamically and precisely adjusting warehouse environmental parameters and optimizing stacking methods based on different tobacco leaf varieties, grades, growth stages, or storage times. This makes it difficult to appropriately manage the rate of change in tobacco leaf appearance quality based on tobacco leaf usage plans and aging schedule requirements, preventing the leaves from reaching peak appearance quality levels during use. Even if low-oxygen curing is subsequently implemented to slow the decline in appearance quality, the effectiveness is limited by inaccurate initial appearance quality control.

[0005] To sum up, in the existing technical background of applying micro-environment intelligent maintenance to tobacco leaf appearance quality control, the existing micro-environment intelligent maintenance technology lacks environmental parameter regulation and tobacco stack management strategies oriented towards precise control of tobacco leaf appearance quality, which makes it difficult to precisely control the speed of change of tobacco leaf appearance quality according to demand, and cannot fully ensure that the tobacco leaves are in the best appearance state when used. Summary of the Invention

[0006] Based on the problems raised by the above background technology, the purpose of the present invention is to provide a tobacco leaf appearance quality control method and related equipment based on micro-environment intelligent maintenance, which solves the problem that the existing micro-environment intelligent maintenance technology lacks environmental parameter regulation and tobacco stack management strategies guided by precise control of tobacco leaf appearance quality, making it difficult to accurately control the speed of change of tobacco leaf appearance quality according to demand, and unable to fully ensure that the tobacco leaves are in the best appearance state when used.

[0007] The present invention is achieved through the following technical solutions:

[0008] The first aspect of the present invention provides a method for controlling the appearance quality of tobacco leaves based on microenvironment intelligent maintenance, comprising the following steps:

[0009] Obtain appearance data and environmental data of tobacco leaves during intelligent curing;

[0010] Performing time series alignment on the appearance data and the environment data to generate appearance time series data and environment time series data with alignment timestamps;

[0011] Constructing a tobacco leaf appearance quality control model, wherein the tobacco leaf appearance quality control model includes an appearance feature extraction network, an environmental feature extraction network, and a control network;

[0012] The appearance feature extraction network extracts features from the appearance time series data based on the aligned timestamps to obtain appearance change features;

[0013] The environmental feature extraction network extracts features from the environmental time series data based on the aligned timestamps to obtain environmental change features;

[0014] The appearance change characteristics and the environment change characteristics are mapped into the control network to perform appearance quality change analysis, and microenvironment regulation is performed according to the results of the appearance quality change analysis.

[0015] In the above technical solution, the appearance data and environmental data of tobacco leaves during intelligent curing are collected, and the acquired appearance data and environmental data are time-aligned to ensure that they have the same timestamp reference, so that the data are consistent and comparable in the time dimension, which facilitates subsequent comprehensive analysis.

[0016] A tobacco leaf appearance quality control model is constructed, which consists of three parts: appearance feature extraction network, environmental feature extraction network and control network. Among them, the appearance feature extraction network is used to extract features of the tobacco leaf appearance time series data based on the aligned timestamps, and obtain features reflecting the changes in tobacco leaf appearance, such as the change characteristics of indicators such as color and maturity; the environmental feature extraction network is used to extract features of the tobacco leaf environment time series data based on the aligned timestamps, and obtain features of environmental changes, such as the change characteristics of environmental parameters such as temperature and humidity.

[0017] The appearance change characteristics and environmental change characteristics are mapped to the control network for appearance quality change analysis. The change relationship between the appearance change characteristics and environmental change characteristics after aligned timestamps is used to find their impact on appearance quality changes. Based on the appearance quality change analysis, the environmental parameters are adjusted to accurately control the speed of tobacco leaf appearance quality changes, so that the tobacco leaves are in the best appearance state when used.

[0018] In an optional embodiment, the appearance feature extraction network includes: a geometric feature extraction channel, a color feature extraction channel, and a texture feature extraction channel;

[0019] Extracting geometric features of the appearance time series data in time series using the geometric feature extraction channel, calculating geometric variations of the geometric features in time series, and integrating the geometric features and the geometric variations to generate geometric variation features;

[0020] Extracting color features of the appearance time series data in time series using the color feature extraction channel, calculating a color change amount of the color features in time series, and integrating the color features and the color change amount to generate a color change feature;

[0021] The texture feature extraction channel is used to extract the texture features of the appearance time series data in time series, and the texture variation of the texture features in time series is calculated. The texture features and the texture variation are integrated to generate texture variation features.

[0022] In an optional embodiment, extracting the color features of the appearance time series data in time series using the color feature extraction channel and calculating the color change of the color features in time series includes the following steps:

[0023] Extract tobacco leaf image frames from the appearance time series data in sequence, perform color space conversion on the tobacco leaf image frames, and perform histogram statistics on the tobacco leaf image frames after color space conversion to obtain a color histogram set C = (c1, c2, ..., c i ,...,c t );

[0024] The color change amount of the color histogram set is calculated to obtain a color change amount set ΔC=(Δc 1，2 ,Δc 2，3 ,...,Δc i,i+1 ,...,Δc t-k,t ); the calculation formula for the color change is as follows:

[0025] Δc i,i+k =c i+k -ci

[0026] In the above formula, c i is the color histogram of the tobacco leaf image frame corresponding to the i-th timestamp, Δc i,i+k is the color histogram difference between the tobacco leaf image frame corresponding to the i-th timestamp and the tobacco leaf image frame corresponding to the i+k-th timestamp, and t is the total time sequence.

[0027] In an optional embodiment, mapping the appearance change feature and the environment change feature to the control network includes:

[0028] Obtain appearance evaluation data;

[0029] Mapping the environmental features in the environmental change features to the bottom layer of the control network, mapping the appearance features in the appearance change features to the middle layer of the control network, and mapping the appearance evaluation data to the top layer of the control network;

[0030] performing correlation calculation on the change amount of the environmental change feature and the change amount of the appearance change feature to obtain first correlation data, and mapping the first correlation data into directed edges between the bottom layer and the middle layer;

[0031] A correlation calculation is performed on the change amount of the appearance change feature and the appearance evaluation data to obtain second correlation data, and the second correlation data is mapped into a directed edge between the middle layer and the top layer.

[0032] In an optional embodiment, performing correlation calculation on the change amount of the environmental change feature and the change amount of the appearance change feature includes the following steps:

[0033] Obtaining initial impact parameters of the environmental data on the appearance data;

[0034] Calculating a difference between a change amount of the environmental change feature and a change amount of the appearance change feature at the aligned timestamps;

[0035] Performing abnormal mutation analysis on the variation difference to obtain abnormal mutation data points, and marking the variation of the environmental change feature and the variation of the appearance change feature corresponding to the abnormal mutation data points as abnormal variation;

[0036] Performing an abnormality analysis on the abnormal variation, and if an abnormality exists, deleting the variation difference corresponding to the abnormal variation to generate a non-abnormal variation difference;

[0037] The initial impact parameter and the difference between the variables without abnormalities are comprehensively calculated to obtain the impact correlation value.

[0038] In an optional embodiment, performing a comprehensive calculation on the initial influencing parameter and the non-abnormal variable difference includes:

[0039]

[0040] In the above formula, θ is the difference of the non-abnormal variable, r is the initial influencing parameter, is the average value of the change ΔA of the appearance change feature, is the average value of the change ΔE of the environmental change feature, ΔG is the change of the geometric feature, ΔC is the change of the color feature, and ΔO is the change of the texture feature.

[0041] In an optional embodiment, microenvironment regulation is performed according to the results of the appearance quality change analysis, including the following steps:

[0042] sorting the appearance data according to the results of the appearance quality change analysis to obtain an appearance change sequence;

[0043] Traversing the appearance change sequence, and associating each appearance data in the appearance change sequence with the environment data according to the result of the appearance quality change analysis, to obtain an appearance-environment association sequence; wherein the environment data in the appearance-environment association sequence is a subsequence of the corresponding associated appearance data;

[0044] Performing intersection and union calculation on all subsequences in the appearance-environment association sequence to obtain an intersection sequence and a non-intersection sequence;

[0045] The environmental data in the intersection sequence is used as the main control data, and the environmental data in the non-intersection sequence is used as the auxiliary control data. The main control data and the auxiliary control data are used to perform microenvironment control.

[0046] The second aspect of the present invention provides a tobacco leaf appearance quality control system based on micro-environment intelligent maintenance, comprising:

[0047] A data acquisition module is used to obtain the appearance data and environmental data of tobacco leaves during the intelligent curing process;

[0048] A timing processing module, configured to perform timing alignment on the appearance data and the environment data, and generate appearance timing data and environment timing data with alignment timestamps;

[0049] A model building module is used to build a tobacco leaf appearance quality control model, wherein the tobacco leaf appearance quality control model includes an appearance feature extraction network, an environmental feature extraction network, and a control network;

[0050] An appearance feature module, configured for the appearance feature extraction network to extract features from the appearance time series data based on the aligned timestamps to obtain appearance change features;

[0051] An environmental feature module, configured for the environmental feature extraction network to extract features from the environmental time series data based on the aligned timestamps to obtain environmental change features;

[0052] The change analysis module is used to map the appearance change characteristics and the environment change characteristics into the control network to perform appearance quality change analysis, and to perform microenvironment regulation according to the results of the appearance quality change analysis.

[0053] The third aspect of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a method for controlling the appearance quality of tobacco leaves based on intelligent microenvironment maintenance is implemented.

[0054] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for controlling the appearance quality of tobacco leaves based on micro-environment intelligent maintenance.

[0055] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0056] Based on the analysis of changes in appearance quality, environmental parameters are adjusted to accurately control the speed of change in tobacco leaf appearance quality, so that the tobacco leaves are in the best appearance state when used. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the examples. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be considered as limiting the scope. A person of ordinary skill in the art can also derive other relevant drawings based on these drawings without inventive effort. In the drawings:

[0058] Figure 1 A schematic flow chart of a method for controlling tobacco leaf appearance quality based on microenvironment intelligent curing provided in Example 1 of the present invention;

[0059] Figure 2 This is a schematic diagram of the structure of a tobacco leaf appearance quality control system based on micro-environment intelligent maintenance provided in Example 2 of the present invention;

[0060] Figure 3 This is a structural diagram of an electronic device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0061] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0062] Example 1

[0063] Figure 1 The flowchart of the tobacco leaf appearance quality control method based on microenvironment intelligent maintenance provided in Example 1 of the present invention is as follows: Figure 1 As shown, the tobacco leaf appearance quality control method based on micro-environment intelligent maintenance includes the following steps:

[0064] Obtain appearance data and environmental data of tobacco leaves during intelligent curing;

[0065] Performing time series alignment on the appearance data and the environment data to generate appearance time series data and environment time series data with alignment timestamps;

[0066] Constructing a tobacco leaf appearance quality control model, wherein the tobacco leaf appearance quality control model includes an appearance feature extraction network, an environmental feature extraction network, and a control network;

[0067] The appearance feature extraction network extracts features from the appearance time series data based on the aligned timestamps to obtain appearance change features;

[0068] The environmental feature extraction network extracts features from the environmental time series data based on the aligned timestamps to obtain environmental change features;

[0069] The appearance change characteristics and the environment change characteristics are mapped into the control network to perform appearance quality change analysis, and microenvironment regulation is performed according to the results of the appearance quality change analysis.

[0070] It should be noted that the appearance data and environmental data of tobacco leaves during intelligent maintenance are collected, and the acquired appearance data and environmental data are time-aligned to ensure that they have the same timestamp reference, so that the data are consistent and comparable in the time dimension, which is convenient for subsequent comprehensive analysis.

[0071] A tobacco leaf appearance quality control model is constructed, which consists of three parts: appearance feature extraction network, environmental feature extraction network and control network. Among them, the appearance feature extraction network is used to extract features of the tobacco leaf appearance time series data based on the aligned timestamps, and obtain features reflecting the changes in tobacco leaf appearance, such as the change characteristics of indicators such as color and maturity; the environmental feature extraction network is used to extract features of the tobacco leaf environment time series data based on the aligned timestamps, and obtain features of environmental changes, such as the change characteristics of environmental parameters such as temperature and humidity.

[0072] The appearance change characteristics and environmental change characteristics are mapped to the control network for appearance quality change analysis. The change relationship between the appearance change characteristics and environmental change characteristics after aligned timestamps is used to find their impact on appearance quality changes. Based on the appearance quality change analysis, the environmental parameters are adjusted to accurately control the speed of tobacco leaf appearance quality changes, so that the tobacco leaves are in the best appearance state when used.

[0073] In an optional embodiment, the appearance feature extraction network includes: a geometric feature extraction channel, a color feature extraction channel, and a texture feature extraction channel;

[0074] Extracting geometric features of the appearance time series data in time series using the geometric feature extraction channel, calculating geometric variations of the geometric features in time series, and integrating the geometric features and the geometric variations to generate geometric variation features;

[0075] Extracting color features of the appearance time series data in time series using the color feature extraction channel, calculating a color change amount of the color features in time series, and integrating the color features and the color change amount to generate a color change feature;

[0076] The texture feature extraction channel is used to extract the texture features of the appearance time series data in time series, and the texture variation of the texture features in time series is calculated. The texture features and the texture variation are integrated to generate texture variation features.

[0077] It should be noted that the appearance feature extraction includes but is not limited to the geometric features, color features and texture features proposed in this embodiment, as well as multiple index features such as maturity, leaf structure, oil content, chroma, etc. that are concretized thereunder.

[0078] In an optional embodiment, extracting the color features of the appearance time series data in time series using the color feature extraction channel and calculating the color change of the color features in time series includes the following steps:

[0079] Extract tobacco leaf image frames from the appearance time series data in sequence, perform color space conversion on the tobacco leaf image frames, and perform histogram statistics on the tobacco leaf image frames after color space conversion to obtain a color histogram set C = (c1, c2, ..., c i ,...,c t );

[0080] The color change amount of the color histogram set is calculated to obtain a color change amount set ΔC=(Δc 1，2 ,Δc 2，3 ,...,Δc i,i+1 ,...,Δct-k,t ); the calculation formula for the color change is as follows:

[0081] Δc i,i+k =c i+k -c i

[0082] In the above formula, c i is the color histogram of the tobacco leaf image frame corresponding to the i-th timestamp, Δc i,i+k is the color histogram difference between the tobacco leaf image frame corresponding to the i-th timestamp and the tobacco leaf image frame corresponding to the i+k-th timestamp, and t is the total time sequence.

[0083] It should be noted that the color feature extraction channel extracts the temporal color features and color change features of the appearance time series data. The appearance time series data consists of a sequence of multiple tobacco leaf images with timestamps, arranged in chronological order. These image frames are extracted one by one for subsequent color feature analysis. The extracted tobacco leaf image frames undergo color space conversion. Typically, the color information of tobacco leaf images is represented in RGB (red, green, blue) format. However, the RGB color space is not always suitable for color feature extraction. Therefore, it is necessary to convert it to a more suitable color space for analysis, such as HSV (hue, saturation, brightness) or LAB color space. Color space conversion can help extract color features more effectively and reduce the impact of varying lighting conditions on color analysis. Histogram statistics are performed on the tobacco leaf image frames after color space conversion to obtain a color histogram set. Histogram statistics are a method for calculating the pixel distribution of different color channels in an image. Color histograms provide an overall characteristic of the color distribution in an image, reflecting the color composition and relative proportions within the image. Color change calculations are performed on the color histogram set to obtain a color change value set. Color change is calculated by comparing the differences between the color histograms of adjacent image frames. There are various methods for calculating color change, such as Euclidean distance and chi-square distance. By calculating the color change between adjacent frames, a color change set can be obtained, which describes the temporal changes in tobacco leaf color.

[0084] The difference k between the timestamps is determined by the specific application and is not further limited in this embodiment.

[0085] Furthermore, the geometric feature extraction channel is used to extract the geometric features of the appearance time series data in time series, and the geometric variation of the geometric features in time series is calculated. The geometric features and the geometric variation are integrated to generate a geometric variation feature, including:

[0086] Extract tobacco leaf image frames from the appearance time series data in sequence according to the time series, perform edge recognition on the tobacco leaf image frames, and obtain a curl degree set G = (g1, g2, ..., g i ,...,g t );

[0087] The curl degree set is subjected to curl degree variation calculation to obtain a curl degree variation set ΔG=(Δg 1，2 ,Δg 2，3 ,...,Δg i,i+1 ,...,Δg t-k,t ); wherein, the calculation formula of the curl change is as follows:

[0088] Δg i,i+k =g i+k -g i

[0089] In the above formula, g i is the curl of the tobacco leaf image frame corresponding to the i-th timestamp, Δg i,i+k is the curl difference between the tobacco leaf image frame corresponding to the i-th timestamp and the tobacco leaf image frame corresponding to the i+k-th timestamp, and t is the total time sequence.

[0090] Furthermore, the texture feature extraction channel is used to extract the texture features of the appearance time series data in time series, and the texture variation of the texture features in time series is calculated, and the texture features and the texture variation are integrated to generate a texture variation feature, including:

[0091] Extract tobacco leaf image frames from the appearance time series data in sequence according to the time series, perform texture recognition on the tobacco leaf image frames, and obtain a texture set O=(o1, o2, ..., o i ,...,o t );

[0092] The texture variation is calculated for the texture set to obtain a texture variation set ΔO=(Δo 1，2 ,Δo 2，3 ,...,Δo i,i+1 ,...,Δo t-k,t ); the calculation formula of texture change is as follows:

[0093] Δo i,i+k =o i+k -o i

[0094] In the above formula, o i is the texture parameter of the tobacco leaf image frame corresponding to the i-th timestamp, Δo i,i+kis the texture parameter difference between the tobacco leaf image frame corresponding to the i-th timestamp and the tobacco leaf image frame corresponding to the i+k-th timestamp, and t is the total time sequence.

[0095] It should be emphasized that the color variation, curl variation and texture variation are the basis for controlling the network to perform variation analysis.

[0096] In an optional embodiment, mapping the appearance change feature and the environment change feature to the control network includes:

[0097] Obtain appearance evaluation data;

[0098] Mapping the environmental features in the environmental change features to the bottom layer of the control network, mapping the appearance features in the appearance change features to the middle layer of the control network, and mapping the appearance evaluation data to the top layer of the control network;

[0099] performing correlation calculation on the change amount of the environmental change feature and the change amount of the appearance change feature to obtain first correlation data, and mapping the first correlation data into directed edges between the bottom layer and the middle layer;

[0100] A correlation calculation is performed on the change amount of the appearance change feature and the appearance evaluation data to obtain second correlation data, and the second correlation data is mapped into a directed edge between the middle layer and the top layer.

[0101] It should be noted that the appearance evaluation data is objective evaluation information on the appearance quality of tobacco leaves, such as the appearance quality evaluation method described in the authorized patent CN113759079B. This embodiment only uses its correlation relationship with the appearance of tobacco leaves according to each dimension. On this basis, the comprehensive score of tobacco leaf appearance quality is mapped to the top layer of the control network; the appearance features in the appearance change features are mapped to the middle layer of the control network; and the environmental features in the environmental change features are mapped to the bottom layer of the control network. The correlation between the environment and appearance, and between appearance and the comprehensive score of tobacco leaf appearance quality are respectively explored; at the same time, the comprehensive score of tobacco leaf appearance quality is traced to find the environmental data that affects the comprehensive score of tobacco leaf appearance quality, and environmental regulation is carried out accordingly.

[0102] The correlation between the change in environmental change characteristics and the change in appearance change characteristics is calculated to obtain first correlation data. This first correlation data is mapped as a directed edge between the bottom and middle layers, reflecting how environmental changes affect tobacco leaf appearance. The correlation between the change in appearance change characteristics and the appearance evaluation data is calculated to obtain second correlation data. This second correlation data is mapped as a directed edge between the middle and top layers, revealing the impact of appearance changes on the comprehensive evaluation of appearance quality.

[0103] The second related data may adopt the existing weight.

[0104] This step constructs a multi-layered control network to organically integrate environmental change characteristics, appearance change characteristics, and appearance evaluation data. The bottom-layer environmental characteristics serve as the foundation, and changes in appearance characteristics at the middle layer ultimately influence the appearance evaluation at the top layer. Through correlation calculation and directed edge mapping, the causal relationships and influence levels between each layer are clarified, providing a basis for precise control of tobacco leaf appearance quality.

[0105] In an optional embodiment, performing correlation calculation on the change amount of the environmental change feature and the change amount of the appearance change feature includes the following steps:

[0106] Obtaining initial impact parameters of the environmental data on the appearance data;

[0107] Calculating a difference between a change amount of the environmental change feature and a change amount of the appearance change feature at the aligned timestamps;

[0108] Performing abnormal mutation analysis on the variation difference to obtain abnormal mutation data points, and marking the variation of the environmental change feature and the variation of the appearance change feature corresponding to the abnormal mutation data points as abnormal variation;

[0109] Performing an abnormality analysis on the abnormal variation, and if an abnormality exists, deleting the variation difference corresponding to the abnormal variation to generate a non-abnormal variation difference;

[0110] The initial impact parameter and the difference between the variables without abnormalities are comprehensively calculated to obtain the impact correlation value.

[0111] It should be noted that the initial impact parameter of environmental data on tobacco leaf appearance data is obtained through historical data and experience. This parameter reflects the initial impact of environmental factors on appearance. The initial impact parameter here is the initial weight of environmental data for appearance data, which is set by technical personnel in this field according to actual projects.

[0112] Under aligned timestamps, calculate the difference in the amount of change between environmental and appearance change features. Aligning timestamps ensures temporal consistency in analysis and allows for direct comparison and analysis of the change relationship between the two. Perform an abnormal mutation analysis on the difference in amount of change, identifying the mutation points as abnormal mutation data points. Label the corresponding environmental and appearance changes as abnormal, facilitating subsequent targeted processing.

[0113] In this embodiment, the abnormal mutation data point is an abnormal mutation data point between the environment data and the appearance data. The abnormal analysis is performed based on the abnormal mutation data point, that is, the change amount of the environment change feature and the change amount of the appearance change feature corresponding to the abnormal mutation data point are marked as abnormal change amounts, and the change amount of the appearance change feature corresponding to the abnormal change amount is calculated with the change amount of other environment change features. If the change amount of the appearance change feature is also an abnormal change amount in the change amount of other environment change features, it is considered that there is an abnormal situation, and the change amount of the appearance change feature corresponding to the abnormal change amount is deleted, and all data associated with it are deleted, that is, all environment change features under the timestamp.

[0114] Similarly, the abnormal change amount is processed in the same way as the change amount of the environmental change feature. If the change amount of the environmental change feature and the change amount of other appearance change features are also abnormal, they are also deleted in the same way.

[0115] In an optional embodiment, performing a comprehensive calculation on the initial influencing parameter and the non-abnormal variable difference includes:

[0116]

[0117] In the above formula, θ is the difference of the non-abnormal variable, r is the initial influencing parameter, is the average value of the change ΔA of the appearance change feature, is the average value of the change ΔE of the environmental change feature, ΔG is the change of the geometric feature, ΔC is the change of the color feature, and ΔO is the change of the texture feature.

[0118] In an optional embodiment, microenvironment regulation is performed according to the results of the appearance quality change analysis, including the following steps:

[0119] sorting the appearance data according to the results of the appearance quality change analysis to obtain an appearance change sequence;

[0120] Traversing the appearance change sequence, and associating each appearance data in the appearance change sequence with the environment data according to the result of the appearance quality change analysis, to obtain an appearance-environment association sequence; wherein the environment data in the appearance-environment association sequence is a subsequence of the corresponding associated appearance data;

[0121] Performing intersection and union calculation on all subsequences in the appearance-environment association sequence to obtain an intersection sequence and a non-intersection sequence;

[0122] The environmental data in the intersection sequence is used as the main control data, and the environmental data in the non-intersection sequence is used as the auxiliary control data. The main control data and the auxiliary control data are used to perform microenvironment control.

[0123] It should be noted that, using the results of the appearance quality change analysis, all appearance data are sorted by degree of change to generate an appearance change sequence, clearly illustrating the overall evolution of tobacco leaf appearance quality over time. Each appearance data point in the appearance change sequence is then individually linked to environmental data based on temporal correspondence, constructing an appearance-environment correlation sequence. Within this sequence, each environmental data point serves as a subsequence of the corresponding appearance data point, clarifying the correspondence between the two and facilitating subsequent environmental parameter adjustment. Intersection sequences contain environmental data points that appear in multiple appearance data subsequences; non-intersection sequences contain environmental data points that appear only in individual appearance data subsequences. Environmental data in intersection sequences have a multi-factor impact on appearance quality. Adjusting environmental data in an intersection sequence will result in changes to multiple appearance data points, and therefore serves as the primary control data. Environmental data in non-intersection sequences, on the other hand, affects only one appearance data point and serves as auxiliary control data.

[0124] Example 2

[0125] Figure 2 This is a schematic diagram of the structure of the tobacco leaf appearance quality control system based on micro-environment intelligent maintenance provided in Example 2 of the present invention, as shown in FIG. Figure 2 As shown, the tobacco leaf appearance quality control system based on micro-environment intelligent maintenance includes:

[0126] A data acquisition module is used to obtain the appearance data and environmental data of tobacco leaves during the intelligent curing process;

[0127] A timing processing module, configured to perform timing alignment on the appearance data and the environment data, and generate appearance timing data and environment timing data with alignment timestamps;

[0128] A model building module is used to build a tobacco leaf appearance quality control model, wherein the tobacco leaf appearance quality control model includes an appearance feature extraction network, an environmental feature extraction network, and a control network;

[0129] An appearance feature module, configured for the appearance feature extraction network to extract features from the appearance time series data based on the aligned timestamps to obtain appearance change features;

[0130] An environmental feature module, configured for the environmental feature extraction network to extract features from the environmental time series data based on the aligned timestamps to obtain environmental change features;

[0131] The change analysis module is used to map the appearance change characteristics and the environment change characteristics into the control network to perform appearance quality change analysis, and to perform microenvironment regulation according to the results of the appearance quality change analysis.

[0132] Example 3

[0133] Figure 3 This is a schematic diagram of the structure of an electronic device provided in Example 3 of the present invention, such as Figure 3 As shown, the electronic device includes a processor 21, a memory 22, an input device 23 and an output device 24; the number of processors 21 in the computer device can be one or more. Figure 3 In the figure, a processor 21 is taken as an example; the processor 21, memory 22, input device 23 and output device 24 in the electronic device can be connected by a bus or other means. Figure 3 The bus connection is taken as an example.

[0134] Memory 22, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules. Processor 21 executes the software programs, instructions, and modules stored in memory 22 to perform various electronic device functions and data processing, thereby implementing the tobacco leaf appearance quality control method based on microenvironment intelligent curing of Example 1.

[0135] The memory 22 may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data created based on the use of the terminal, etc. Furthermore, the memory 22 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state memory device. In some instances, the memory 22 may further include a memory remotely located relative to the processor 21, and these remote memories may be connected to the electronic device via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0136] The input device 23 can be used to receive the ID and password input by the user, etc. The output device 24 is used to output the network configuration page.

[0137] Example 4

[0138] Example 4 of the present invention also provides a computer-readable storage medium, and the computer-executable instructions, when executed by a computer processor, are used to implement the tobacco leaf appearance quality control method based on micro-environment intelligent maintenance as provided in Example 1.

[0139] An embodiment of the present invention provides a storage medium containing computer-executable instructions, and its computer-executable instructions are not limited to the method operations provided in Example 1, but can also execute related operations in the tobacco leaf appearance quality control method based on micro-environment intelligent maintenance provided by any embodiment of the present invention.

[0140] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for controlling tobacco leaf appearance quality based on microenvironment intelligent maintenance, characterized in that: The steps include: Obtain appearance data and environmental data of tobacco leaves during intelligent curing; Performing time series alignment on the appearance data and the environment data to generate appearance time series data and environment time series data with alignment timestamps; Constructing a tobacco leaf appearance quality control model, wherein the tobacco leaf appearance quality control model includes an appearance feature extraction network, an environmental feature extraction network, and a control network; The appearance feature extraction network extracts features from the appearance time series data based on the aligned timestamps to obtain appearance change features; The environmental feature extraction network extracts features from the environmental time series data based on the aligned timestamps to obtain environmental change features; The appearance change characteristics and the environment change characteristics are mapped into the control network to perform appearance quality change analysis, and microenvironment regulation is performed according to the results of the appearance quality change analysis.

2. The method for controlling tobacco leaf appearance quality based on microenvironment intelligent maintenance according to claim 1, characterized in that: The appearance feature extraction network includes: a geometric feature extraction channel, a color feature extraction channel and a texture feature extraction channel; Extracting geometric features of the appearance time series data in time series using the geometric feature extraction channel, calculating geometric variations of the geometric features in time series, and integrating the geometric features and the geometric variations to generate geometric variation features; Extracting color features of the appearance time series data in time series using the color feature extraction channel, calculating a color change amount of the color features in time series, and integrating the color features and the color change amount to generate a color change feature; The texture feature extraction channel is used to extract the texture features of the appearance time series data in time series, and the texture variation of the texture features in time series is calculated. The texture features and the texture variation are integrated to generate texture variation features.

3. The method for controlling tobacco leaf appearance quality based on microenvironment intelligent maintenance according to claim 2, characterized in that: Extracting the color features of the appearance time series data in time series by using the color feature extraction channel and calculating the color change amount of the color features in time series includes the following steps: Extract tobacco leaf image frames from the appearance time series data in sequence, perform color space conversion on the tobacco leaf image frames, and perform histogram statistics on the tobacco leaf image frames after color space conversion to obtain a color histogram set C = (c1, c2, ..., c i ,...,c t ); The color change amount of the color histogram set is calculated to obtain a color change amount set ΔC=(Δc 1，2 ,Δc 2，3 ,...,Δc i,i+1 ,...,Δc t-k,t ); the calculation formula for the color change is as follows: Δc i,i+k =c i+k -c i In the above formula, c i is the color histogram of the tobacco leaf image frame corresponding to the i-th timestamp, Δc i,i+k is the color histogram difference between the tobacco leaf image frame corresponding to the i-th timestamp and the tobacco leaf image frame corresponding to the i+k-th timestamp, and t is the total time sequence.

4. The tobacco leaf appearance quality control method based on microenvironment intelligent maintenance according to claim 1, characterized in that: Mapping the appearance change feature and the environment change feature to the control network includes: Obtain appearance evaluation data; Mapping the environmental features in the environmental change features to the bottom layer of the control network, mapping the appearance features in the appearance change features to the middle layer of the control network, and mapping the appearance evaluation data to the top layer of the control network; performing correlation calculation on the change amount of the environmental change feature and the change amount of the appearance change feature to obtain first correlation data, and mapping the first correlation data into directed edges between the bottom layer and the middle layer; A correlation calculation is performed on the change amount of the appearance change feature and the appearance evaluation data to obtain second correlation data, and the second correlation data is mapped into a directed edge between the middle layer and the top layer.

5. The method for controlling tobacco leaf appearance quality based on microenvironment intelligent maintenance according to claim 4, characterized in that: Calculating the correlation between the change amount of the environmental change feature and the change amount of the appearance change feature includes the following steps: Obtaining initial impact parameters of the environmental data on the appearance data; Calculating a difference between a change amount of the environmental change feature and a change amount of the appearance change feature at the aligned timestamps; Performing abnormal mutation analysis on the variation difference to obtain abnormal mutation data points, and marking the variation of the environmental change feature and the variation of the appearance change feature corresponding to the abnormal mutation data points as abnormal variation; Performing an abnormality analysis on the abnormal variation, and if an abnormality exists, deleting the variation difference corresponding to the abnormal variation to generate a non-abnormal variation difference; The initial impact parameter and the difference between the variables without abnormalities are comprehensively calculated to obtain the impact correlation value.

6. The method for controlling tobacco leaf appearance quality based on microenvironment intelligent maintenance according to claim 5, characterized in that: Performing a comprehensive calculation on the initial influencing parameter and the difference between the variables without abnormalities, including: ΔA={ΔG,ΔC,ΔO} φ=r -θ In the above formula, θ is the difference of the non-abnormal variable, r is the initial influencing parameter, is the average value of the change ΔA of the appearance change feature, is the average value of the change ΔE of the environmental change feature, ΔG is the change of the geometric feature, ΔC is the change of the color feature, and ΔO is the change of the texture feature.

7. The method for controlling tobacco leaf appearance quality based on microenvironment intelligent maintenance according to claim 1, characterized in that: Microenvironment regulation is performed based on the results of the appearance quality change analysis, including the following steps: sorting the appearance data according to the results of the appearance quality change analysis to obtain an appearance change sequence; Traversing the appearance change sequence, and associating each appearance data in the appearance change sequence with the environment data according to the result of the appearance quality change analysis, to obtain an appearance-environment association sequence; wherein the environment data in the appearance-environment association sequence is a subsequence of the corresponding associated appearance data; Performing intersection and union calculation on all subsequences in the appearance-environment association sequence to obtain an intersection sequence and a non-intersection sequence; The environmental data in the intersection sequence is used as the main control data, and the environmental data in the non-intersection sequence is used as the auxiliary control data. The main control data and the auxiliary control data are used to perform microenvironment control.

8. A tobacco leaf appearance quality control system based on micro-environment intelligent maintenance, characterized in that: include: A data acquisition module is used to obtain the appearance data and environmental data of tobacco leaves during the intelligent curing process; A timing processing module, configured to perform timing alignment on the appearance data and the environment data, and generate appearance timing data and environment timing data with alignment timestamps; A model building module is used to build a tobacco leaf appearance quality control model, wherein the tobacco leaf appearance quality control model includes an appearance feature extraction network, an environmental feature extraction network, and a control network; An appearance feature module, configured for the appearance feature extraction network to extract features from the appearance time series data based on the aligned timestamps to obtain appearance change features; An environmental feature module, configured for the environmental feature extraction network to extract features from the environmental time series data based on the aligned timestamps to obtain environmental change features; The change analysis module is used to map the appearance change characteristics and the environment change characteristics into the control network to perform appearance quality change analysis, and to perform microenvironment regulation according to the results of the appearance quality change analysis.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for controlling the appearance quality of tobacco leaves based on microenvironment intelligent maintenance as described in any one of claims 1 to 7 is implemented.

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 method for controlling the appearance quality of tobacco leaves based on microenvironment intelligent maintenance as described in any one of claims 1 to 7 is implemented.

Images