Dehydration process adjustment method and system for retaining plant aromatic substances based on simulation

Abstract

The present invention relates to a dehydration process regulation method and system for retaining plant aromatic substances based on analog quantity. The method collects first fragrance information through an olfactory sensor and collects first dehydration image information through an image acquisition unit, constructs a fragrance database and a dehydration image database, establishes a mapping relationship between the fragrance information and the dehydration state, and constructs a dehydration parameter regulation model. In the actual plant dehydration process, the method collects second fragrance information and second dehydration image information, and uses the dehydration parameter regulation model to adjust the dehydration parameters in the actual plant dehydration process, thereby realizing comprehensive monitoring and dataization of the fragrance and dehydration state in the plant dehydration process, more accurately adjusting the dehydration parameters according to the real-time fragrance and dehydration state, better coping with the differences of plant raw materials, and optimizing the dehydration effect. The technical solution is more scientific and efficient, improves the standardization level of plant processing, and ensures the quality of the dehydration process.

Description

Technical Field

[0001] The invention relates to the field of computers, and in particular to a method and system for regulating a dehydration process based on analog quantity to retain plant aromatic substances. Background Art

[0002] The dehydration process of plants refers to the process of using certain methods and steps to make the fresh leaves of plants undergo a certain degree of oxidation and fermentation to form a specific aroma and taste during the plant processing. For example, the traditional Wuyi rock tea primary processing technology mainly relies on the experience of tea-making technicians to achieve it through touching, smelling, and observing the environment. Not only is the production efficiency low, but the quality control is completely dependent on the skills of tea-making technicians and is difficult to effectively guarantee. In addition, the labor cost remains high due to the lack of mature technology. Summary of the invention

[0003] In view of the above problems, the present invention provides a method and system for regulating the dehydration process based on analog quantity retention of plant aromatic substances, which solves the problem that the existing plant dehydration process relies entirely on manual operation.

[0004] To achieve the above object, in a first aspect, the present invention provides a method for regulating a dehydration process for retaining plant aromatic substances based on a simulated amount, comprising:

[0005] Acquire first fragrance information collected by the olfactory sensor in different time periods and different plant categories, the first fragrance information including the concentration of volatiles;

[0006] Building a fragrance database according to the first fragrance information and fragrance tags of different plant categories, the fragrance database including concentration ranges of volatile substances corresponding to different fragrances;

[0007] and, obtaining a first dehydration parameter corresponding to the first fragrance information, the first dehydration parameter including a first flipping frequency, a first temperature information and a first humidity information;

[0008] constructing a first sample parameter set according to the first dehydration parameter and the aroma database;

[0009] and, when the olfactory sensor collects the first fragrance information, synchronously acquiring first dehydrated image information of the plant in different dehydrated states;

[0010] constructing a dehydration image database according to the first dehydration image information and the dehydration state label;

[0011] and, obtaining a second dehydration parameter corresponding to the first dehydration image information, the second dehydration parameter including a second flipping frequency, second temperature information and second humidity information;

[0012] constructing a second sample parameter set according to the second dehydration parameter and the dehydration image database;

[0013] Establishing a mapping relationship between the fragrance database and the dehydrated image database;

[0014] A dehydration parameter adjustment model is constructed according to the mapping relationship, the fragrance database, and the dehydration image database, and the output result of the dehydration parameter adjustment model is the plant dehydration parameter;

[0015] The first sample parameter set and the second sample parameter set are used as training data for the dehydration parameter adjustment model, and the dehydration parameter adjustment model is trained until an output result of the dehydration parameter adjustment model is within a range of a preset accuracy threshold;

[0016] Acquire the second aroma information and the second dehydration image information in the actual dehydration process in real time, and input the second aroma information and the second dehydration image information into the trained dehydration parameter adjustment model to obtain the actual dehydration parameters;

[0017] Adjust the parameter values ​​in the current actual dehydration process according to the actual dehydration parameters.

[0018] In some embodiments, obtaining first fragrance information collected by the olfactory sensor in different time periods and different plant categories, wherein the first fragrance information includes the concentration of volatile substances, includes:

[0019] Get the category label of the plant, which includes any one of Dahongpao, Tieguanyin, Baijiguan, Narcissus, Osmanthus tea and Huangjingui;

[0020] collecting first aroma information of the plants of the current category in the dehydration stage at a first preset frequency, wherein the first aroma information is the concentration of volatiles generated by the plants of the current category in the dehydration stage, and the volatiles include at least one of ester compounds, alcohol compounds, aldehyde compounds, terpene compounds and phenolic compounds;

[0021] Filling the first fragrance information collected at the first preset frequency into the fragrance data set corresponding to the current category label in chronological order to obtain the first fragrance information data set corresponding to the current category label;

[0022] Repeat the above steps until the first fragrance information data set corresponding to all category labels is obtained;

[0023] A fragrance database is constructed according to the first fragrance information and fragrance tags of different plant categories. The fragrance database includes concentration ranges of volatile substances corresponding to different fragrances, including:

[0024] The following steps are performed for each first aroma information data set:

[0025] Draw a volatile matter change curve of the current category label according to the first fragrance information data set;

[0026] and, obtaining a fragrance label corresponding to the current category label, the fragrance label comprising at least one of floral fragrance, fruity fragrance, honey fragrance, baking fragrance, medicinal fragrance, mineral fragrance, and smoky fragrance;

[0027] Selecting a corresponding dehydration time period according to the aroma tag in the volatile matter variation curve, and using a plurality of volatile matter concentration variation ranges within the dehydration time period as aroma threshold information of the current aroma tag;

[0028] Mapping and storing aroma threshold information and category labels;

[0029] Repeat the above steps until the aroma threshold information of all category labels is generated and stored to obtain the aroma database.

[0030] In some embodiments, when the olfactory sensor collects the first fragrance information, synchronously acquiring the first dehydrated image information of the plant in different dehydrated states includes:

[0031] Collecting first dehydrated image information of plants of the current category at the dehydration stage at a second preset frequency and arranging the information in chronological order to obtain a first dehydrated image dataset of plants of the current category, wherein the first dehydrated image information includes a leaf dehydrated image of the current plant;

[0032] Preprocessing the first dehydrated image data set, the preprocessing including noise removal processing and image enhancement processing, to obtain a first processed image data set, the first processed image data set including a plurality of first processed images;

[0033] Extracting features from the first processed images one by one to obtain first image features corresponding to each first processed image, where the first image features include first color features, first texture features, and first shape features;

[0034] splicing the multiple first image features to obtain a first dehydrated image feature set corresponding to the current category label;

[0035] Repeat the above steps until the first dehydrated image feature set corresponding to all category labels is obtained;

[0036] Constructing a dehydration image database according to the first dehydration image information and the dehydration state label includes:

[0037] Obtain a dehydration status tag, where the dehydration status tag is status information of different dehydration degrees;

[0038] Draw a color feature change curve, a texture feature change curve, and a shape feature change curve corresponding to the dehydration state label one by one according to the first dehydration image feature set;

[0039] The color feature change curve, the texture feature change curve and the shape feature change curve of the same category label are divided into a group to obtain a first feature change curve group;

[0040] Repeat the above steps until the first characteristic change curve group corresponding to all category labels is generated to obtain a dehydrated image database.

[0041] In some embodiments, establishing a mapping relationship between the fragrance database and the dehydrated image database includes:

[0042] The aroma database is expressed by formula (1), which is as follows:

[0043] ;

[0044] In formula (1), For the fragrance database, Indicates Category label Aroma threshold information corresponding to each aroma tag;

[0045] The dehydrated image database is expressed by formula (2), which is as follows:

[0046] ;

[0047] In formula (2), For the dehydrated image database, Indicates Category label A first characteristic change curve group corresponding to a dehydration state label;

[0048] No. Category label The first characteristic change curve group corresponding to the dehydration state label is expressed by formula (3), which is as follows:

[0049] ;

[0050] In formula (3), For the Category label The first color feature of the dehydration status label, For the Category label The first texture feature of the dehydration state label, For the Category label a first shape feature of a dehydrated state label;

[0051] The dehydrated image database is taken as the independent variable and the fragrance database is taken as the dependent variable. The mapping function between the dehydrated image database and the fragrance database is constructed and expressed by formula (4). Formula (4) is as follows:

[0052] ;

[0053] In formula (4), is a mapping function, and the mapping function is a multinomial regression function, is the intercept term, yes The regression coefficient of yes of Power, is the random error term, is the order of the polynomial.

[0054] In some embodiments, a dehydration parameter adjustment model is constructed according to the mapping relationship and the fragrance database and the dehydration image database, and the output result of the dehydration parameter adjustment model is the plant dehydration parameter, and the following further includes:

[0055] When collecting the first dehydrated image information, synchronously obtaining first spectrum information corresponding to the first dehydrated image information;

[0056] Extracting features from the first spectrum information to obtain first spectrum features and constructing a spectrum feature database;

[0057] A dehydration parameter adjustment model is constructed based on the mapping relationship, the fragrance database, and the dehydration image database. The output result of the dehydration parameter adjustment model is the plant dehydration parameters, which also include:

[0058] Based on the spectral feature database, mapping relationship, aroma database and dehydrated image database as multimodal features, a multimodal learning model is constructed, which is expressed by formula (5) and formula (6). Formula (5) is as follows:

[0059] ;

[0060] Formula (6) is as follows:

[0061] ;

[0062] In formula (5) and formula (6), is the output of the multimodal learning model, It is a multimodal learning model based on neural network. is a multimodal feature, are the weight parameters and bias parameters of each neural layer, For the fragrance database, For the dehydrated image database, is the spectral feature database;

[0063] The multimodal features are input into the LSTM model to obtain the prediction results corresponding to the multimodal features, which are expressed by formula (7). Formula (7) is as follows:

[0064] ;

[0065] In formula (7), To predict the results, is the function representation of the LSTM model;

[0066] The output result of the multimodal learning model is fused with the prediction result of the LSTM model to obtain the fusion result, which is expressed by formula (8). Formula (8) is as follows:

[0067] ;

[0068] In formula (8), is the fusion result;

[0069] The fusion result is input into the fully connected layer to obtain the plant dehydration parameter, which is expressed by formula (9). Formula (9) is as follows:

[0070] ;

[0071] In formula (9), is the plant dehydration parameter, is the function representation of the fully connected layer, are the bias parameters and weight parameters of the fully connected layer.

[0072] In some embodiments, inputting the fusion result into a fully connected layer to obtain the plant dehydration parameter further comprises:

[0073] The humidity change factor, temperature change factor and flipping frequency change factor are used as feedback information, and the feedback information and the fusion result are input into the fully connected layer to obtain the plant dehydration parameter, which is expressed by formula (10). Formula (10) is as follows:

[0074] ;

[0075] In formula (10), is the humidity variation factor, is the temperature variation factor, is the flip frequency change factor;

[0076] The humidity change factor is expressed by formula (11), which is as follows:

[0077] ;

[0078] In formula (11), For humidity The dehydration rate of plants under is the influence coefficient of humidity on dehydration rate, is the base humidity, is the baseline dehydration rate;

[0079] The temperature change factor is expressed by formula (12), which is as follows:

[0080] ;

[0081] In formula (12), For the temperature The dehydration rate of plants under is the base dehydration rate, is the activation energy, is the gas constant, is the reference temperature;

[0082] The flip frequency change factor is expressed by formula (13), which is as follows:

[0083] ;

[0084] In formula (13), For the flip frequency The dehydration rate of plants under is the base dehydration rate, is the influence coefficient of the turning frequency on the dehydration rate, is the attenuation coefficient of the flipping frequency;

[0085] In formula (11) to formula (13), .

[0086] In some embodiments, using the first sample parameter set and the second sample parameter set as training data for the dehydration parameter adjustment model, and training the dehydration parameter adjustment model until the output result of the dehydration parameter adjustment model is within a preset accurate threshold range includes:

[0087] Dividing the first sample parameter set into a first sample training set and a first sample test set, and dividing the second sample parameter set into a second sample training set and a second sample test set;

[0088] The first sample training set and the second sample training set are sequentially input into the dehydration parameter adjustment model for training, and a loss function is generated during the training process to calculate the output error in the current dehydration parameter adjustment model, and the parameters of the dehydration parameter adjustment model are adjusted according to the output error until the output error is within a preset error range;

[0089] Inputting the first sample test set and the second sample test set into the dehydration parameter adjustment model in sequence, obtaining a first model output result of the first sample test set and a second model output result of the second sample test set;

[0090] Calculating the accuracy of the dehydration parameter adjustment model according to the output results of the first model and the output results of the second model;

[0091] Determine whether the accuracy is within a preset accuracy threshold;

[0092] If yes, it means that the dehydration parameter adjustment model training is completed.

[0093] In a second aspect, the present invention further provides a dehydration process regulation system for retaining plant aromatic substances based on analog quantity, which is applicable to the method described in the first aspect. The system comprises an outer cylinder, an inner cylinder, a first driving unit, a hot air blower, a shell, a humidifier, a sensor assembly and a control unit. The outer cylinder is provided with a plurality of first air holes and a first monitoring window; the inner cylinder is arranged inside the outer cylinder, a first chamber is provided between the inner cylinder and the outer cylinder, the first chamber is used for placing plants, a second chamber is provided inside the inner cylinder, and a plurality of second air holes are provided on the inner cylinder; the first driving unit is transmission-connected to the outer cylinder, and the first driving unit is used to drive the outer cylinder to rotate relative to the inner cylinder; the hot air blower is connected to the second chamber;

[0094] The shell cover is arranged on the outside of the outer cylinder; the humidifier is arranged on the shell; the sensor assembly is arranged on the shell, and the sensor assembly is arranged in a first preset area. When the outer cylinder is rotated to a preset angle, the first preset area coincides with the first monitoring window. The sensor assembly includes an olfactory sensor, an image acquisition unit, a temperature sensor and a humidity sensor. The olfactory sensor is used to collect the concentration of volatiles of plants in the first chamber, and the image acquisition unit is used to collect dehydration image information of plants in the first chamber; the control unit is electrically connected to the sensor assembly, the humidifier, the hot air blower and the first driving unit respectively, and the control unit is used to execute the method described in the first aspect.

[0095] Different from the prior art, in the above technical solution, the first fragrance information is collected by the olfactory sensor and the first dehydration image information is collected by the image acquisition unit, a fragrance database and a dehydration image database are constructed, a mapping relationship between the fragrance information and the dehydration state is established, and a dehydration parameter adjustment model is constructed. In the actual plant dehydration process, the second fragrance information and the second dehydration image information are collected, and the dehydration parameters in the actual plant dehydration process are adjusted with the help of the dehydration parameter adjustment model, so as to realize the comprehensive monitoring and dataization of the fragrance and dehydration state in the plant dehydration process, and more accurately adjust the dehydration parameters according to the real-time fragrance and dehydration state, better cope with the differences of plant raw materials, and optimize the dehydration effect. Compared with the traditional experience adjustment, the method shown in this technical solution is more scientific and efficient, improves the standardization level of plant processing, and ensures the stability of product quality.

[0096] The above-mentioned records related to the invention content are only an overview of the technical solution of the present application. In order to enable ordinary technicians in the field to more clearly understand the technical solution of the present application, and then implement it according to the text of the specification and the contents recorded in the drawings, and to make the above-mentioned purpose and other purposes, features and advantages of the present application easier to understand, the following is an explanation in combination with the specific implementation mode and drawings of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] The drawings are only used to illustrate the principles, implementation methods, applications, characteristics and effects of the specific embodiments of the present invention and other related contents, and shall not be considered as limitations of the present application.

[0098] In the drawings of the specification:

[0099] Figure 1 It is a schematic diagram of steps S101 to S111 of the dehydration process adjustment method based on the simulation amount to retain plant aromatic substances according to the specific implementation method;

[0100] Figure 2 It is a schematic diagram of steps S201 to S204 of the dehydration process adjustment method based on the simulation amount to retain the plant aromatic substances described in the specific implementation method;

[0101] Figure 3 It is a schematic diagram of steps S301 to S304 of the dehydration process adjustment method based on the simulation amount to retain the plant aromatic substances described in the specific implementation method;

[0102] Figure 4 It is a schematic diagram of steps S401 to S406 of the dehydration process adjustment method based on the simulation amount to retain the plant aromatic substances described in the specific implementation method;

[0103] Figure 5It is a schematic diagram of a dehydration process regulating system for retaining plant aromatic substances based on simulation quantity as described in a specific implementation method.

[0104] Reference numerals:

[0105] 1. Inner tube;

[0106] 2. Outer cylinder;

[0107] 3. Shell;

[0108] 4. Sensor components;

[0109] 5. Hot air blower;

[0110] 6. A first driving unit;

[0111] 7. Humidifier. DETAILED DESCRIPTION

[0112] In order to explain in detail the possible application scenarios, technical principles, specific schemes that can be implemented, and the purposes and effects that can be achieved, the following is a detailed description of the specific embodiments listed in conjunction with the accompanying drawings. The embodiments described herein are only used to more clearly illustrate the technical solutions of the present application, and are therefore only used as examples, and cannot be used to limit the scope of protection of the present application.

[0113] Reference to "embodiment" herein means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present application. The term "embodiment" appearing in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or association with other embodiments. In principle, in the present application, as long as there is no technical contradiction or conflict, the various technical features mentioned in the embodiments can be combined in any way to form a corresponding implementable technical solution.

[0114] Unless otherwise defined, the technical terms used in this document have the same meanings as those generally understood by those skilled in the art to which this application belongs; the use of relevant terms in this document is only for describing specific embodiments and is not intended to limit this application.

[0115] In the description of this application, the term "and / or" is an expression used to describe the logical relationship between objects, indicating that three relationships may exist, for example, A and / or B, which means: A exists, B exists, and A and B exist at the same time. In addition, the character " / " in this article generally indicates that the objects before and after are in an "or" logical relationship.

[0116] In the present application, terms such as “first” and “second” are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship of quantity, priority or sequence between these entities or operations.

[0117] Without further limitations, in this application, the words "include", "comprises", "has" or other similar open-ended expressions used in sentences are intended to cover non-exclusive inclusion. These expressions do not exclude the presence of additional elements in the process, method or product including the elements, so that the process, method or product including a series of elements may include not only those limited elements, but also other elements not explicitly listed, or also include elements inherent to such process, method or product.

[0118] See also Figure 1 In a first aspect, this embodiment provides a method for regulating a dehydration process for retaining plant aromatic substances based on a simulated amount, comprising:

[0119] S101, obtaining first fragrance information collected by the olfactory sensor in different time periods and different plant categories, where the first fragrance information includes the concentration of volatiles;

[0120] S102, constructing a fragrance database according to the first fragrance information and fragrance tags of different plant categories, the fragrance database including concentration ranges of volatile substances corresponding to different fragrances; and obtaining a first dehydration parameter corresponding to the first fragrance information, the first dehydration parameter including a first flipping frequency, first temperature information and first humidity information;

[0121] S103, constructing a first sample parameter set according to the first dehydration parameter and the aroma database;

[0122] S104, when the olfactory sensor collects the first fragrance information, synchronously obtaining first dehydrated image information of the plant in different dehydrated states;

[0123] S105, constructing a dehydration image database according to the first dehydration image information and the dehydration state label; and obtaining a second dehydration parameter corresponding to the first dehydration image information, the second dehydration parameter including a second flipping frequency, second temperature information and second humidity information;

[0124] S106, constructing a second sample parameter set according to the second dehydration parameter and the dehydration image database;

[0125] S107, establishing a mapping relationship between the fragrance database and the dehydrated image database;

[0126] S108, constructing a dehydration parameter adjustment model according to the mapping relationship, the fragrance database, and the dehydration image database, wherein the output result of the dehydration parameter adjustment model is the plant dehydration parameter;

[0127] S109, using the first sample parameter set and the second sample parameter set as training data for a dehydration parameter adjustment model, and training the dehydration parameter adjustment model until an output result of the dehydration parameter adjustment model is within a range of a preset accuracy threshold;

[0128] S110, acquiring the second aroma information and the second dehydration image information in the actual dehydration process in real time, and inputting the second aroma information and the second dehydration image information into the trained dehydration parameter adjustment model to obtain the actual dehydration parameters;

[0129] S111. Adjust parameter values ​​in the current actual dehydration process according to actual dehydration parameters.

[0130] In step S101, the olfactory sensor is arranged inside the plant dehydration process regulating system, that is, inside the shell described later, and the olfactory sensor can regularly collect the concentration of volatiles generated by the plant during the dehydration process. The first fragrance information is collected in different time periods and different plant categories, that is, in the process of constructing the fragrance database, it is necessary to collect the concentration change information of volatiles of different plant categories in different time periods during the dehydration process one by one, and distinguish these first fragrance information according to the plant category and fragrance label. It should be noted that the concentration of volatiles is the concentration of alcohols, esters, aldehydes, phenols and other compounds emitted by the plant itself after kneading and dehydration during the dehydration process. It can be understood that by setting different olfactory sensors, the value of the first fragrance information can be collected more comprehensively. Preferably, multiple olfactory sensors can be further set during the collection process to reduce the collection error of the first fragrance information. Furthermore, in this embodiment, the aromatic substances of the plants vary according to the specific plant categories. Plants include vegetables, fruits, tea leaves, substitute teas, and other plant categories that can be used in the dehydration process. The specific aroma characteristics and category labels will also vary depending on the plant category.

[0131] In step S102, after obtaining a plurality of first fragrance information, it is associated with the fragrance tags corresponding to different plant categories, so as to construct a fragrance database. Further, the first fragrance information is quantified to obtain the concentration range of volatiles corresponding to the fragrance tags in different plant categories. In addition, the plants of each plant category need to be properly turned over during the dehydration process to avoid uneven heating of local plants due to the stacking of multiple plants, which affects the overall dehydration effect. In the process of the olfactory sensor collecting the first fragrance information, the turning frequency, the temperature during the turning, and the humidity during the turning are the main factors affecting the fragrance, and are also the essence of the dehydration process. On this basis, while the olfactory sensor collects the first fragrance information, the temperature, humidity and turning frequency corresponding to the first fragrance information are synchronously recorded and recorded as the first dehydration parameter. The first dehydration parameter includes the first turning frequency, the first temperature information and the first humidity information.

[0132] In step S103, a first sample parameter set is constructed based on the first dehydration parameter and the fragrance database. The specific process may be: grouping the first dehydration parameters according to the plant category, and on this basis, obtaining the fragrance tag corresponding to the current plant category, and in the process of mapping the fragrance tag with the first fragrance information, synchronously storing the data of the first dehydration parameter corresponding to the fragrance tag to form a first dehydration parameter set, forming a group of multiple first dehydration parameter sets according to the group of plant category, and then forming the first dehydration parameter sets of multiple plant categories into a first sample parameter set. The construction process of the second sample parameter set in the following text can also be understood with reference to the first sample parameter set.

[0133] In step S104, when the olfactory sensor collects the first fragrance information, the first dehydrated image information of the plant in different dehydrated states is collected synchronously. The dehydration process belongs to the production process of Wuyi rock tea. In essence, it is to volatilize the organic matter and water in the plant through temperature, humidity and turning operations, so that the tea fragrance of the plant can be retained in the plant body, and the influence of external fungi on the plant is reduced by the discharge of water, making it easier to store. The image information of the plant collected in the dehydration process can also be understood as the image of the plant during the dehydration process. Based on this, the dehydration state also represents the basis for the change of the current plant in different time periods during the dehydration stage. It can be understood that the dehydration state can be quantified according to actual needs. For example, it is quantified by the water loss rate of the plant. The dehydration state can be reflected as the water loss rate of the plant. From 0% water loss rate to 100% water loss rate, every 1% increase here can be used as a representation of a dehydration state. Through the timed acquisition of the image acquisition unit, the first dehydration image information of the current plant category in different dehydration states can be obtained, and multiple first dehydration image information can be collected for plants of each plant category during the dehydration process, so as to improve the comprehensiveness of the dehydration image acquisition data.

[0134] In step S105, the dehydration state label is as described above, and the dehydration rate can be 0% to 100% as specific label information. In other embodiments, the dehydration rate can be roughly divided into ranges and different dehydration state descriptions can be assigned to each range to form the dehydration state label, such as setting the dehydration state of 10% to 30% as a light dehydration label, setting the dehydration state of 40% to 60% as a semi-dehydration label, and so on. A dehydration image database is constructed based on the first dehydration image information and the dehydration state label. It can be understood that during the construction process, the first dehydration image information uses plant categories as the basis for group division to better distinguish the differences in dehydration parameters of subdivided plant categories in rock tea, and it is also more in line with the actual adjustment requirements of the dehydration process of multiple rock tea categories.

[0135] Furthermore, this process also involves the adjustment of dehydration parameters corresponding to different dehydration states. To facilitate the distinction from the aforementioned first dehydration parameter, the dehydration parameter synchronously acquired during the first dehydration image information acquisition process is recorded as the second dehydration parameter. It should be noted that the first dehydration parameter is acquired synchronously when the first fragrance information is acquired, and the second dehydration parameter is acquired synchronously when the first dehydration image information is acquired. When the frequency of acquiring the first fragrance information is consistent with the frequency of acquiring the first dehydration image information, the data of the first dehydration parameter and the second dehydration parameter are the same.

[0136] In step S106, a second sample parameter set is constructed according to the second dehydration parameter and the dehydration image database. It can be understood that the second dehydration parameters are grouped according to plant categories to obtain a second dehydration parameter set for each plant category under different dehydration states. On this basis, multiple second dehydration parameter sets are sorted and merged to obtain a second sample parameter set. Different from the first sample parameter set, the dehydration parameters in the second sample parameter set are associated with the dehydration state.

[0137] In step S107, the mapping relationship can be understood as reflecting the association between the first fragrance information in the fragrance database and the first dehydrated image information in the dehydrated image database through a mapping relationship. More specifically, the connection between the fragrance database and the dehydrated image database can be reflected through a functional relationship. It can be understood that the mapping relationship here is based on the mapping under the same plant category, and mapping in different plant categories is not of practical significance. In actual application, the mapping relationship of each plant category tends to be unified for the dehydration process of the same rock tea, and this mapping relationship can be uniformly represented by the functional relationship described later.

[0138] In step S108, a dehydration parameter adjustment model is constructed. It can be understood that the dehydration parameter adjustment model shown in this embodiment is based on a neural network model. After the fragrance database and the dehydrated image database are associated through a mapping relationship, the dehydration parameter adjustment model is constructed in the form of input features. In step S109, the constructed dehydration parameter adjustment model is trained with the first sample parameter set and the second sample parameter set as training data until the output result of the dehydration parameter adjustment model is within the range of a preset accurate threshold. It can be understood that the output result of the dehydration parameter adjustment model is the plant dehydration parameter in a future period of time that is adapted to the current plant dehydration state.

[0139] In step S110, after the dehydration parameter adjustment model is trained, the second aroma information and the second dehydration image information in the actual plant dehydration process can be obtained. It should be noted that the second aroma information and the second dehydration image information are collected in real time during the plant dehydration process. By inputting the second aroma information and the second dehydration image information into the dehydration parameter adjustment model, the actual dehydration parameters in the next time period can be obtained; in step S111, the corresponding equipment in the current plant dehydration process adjustment system is adjusted based on the actual dehydration parameters, and the parameter values ​​include temperature value, humidity value and flipping frequency.

[0140] Specifically, the difficulty of the plant dehydration process is that the tea maker needs to adjust various parameters in real time according to the state of the plant in the dehydration stage to ensure the dynamic control of the parameters during the plant dehydration process. The dehydration parameter adjustment model shown in this embodiment is used to simulate the real-time tracking state of the tea maker, and the dehydration process is divided into multiple different time segments. The second fragrance information, the second dehydration image information and the mapping relationship between the two generated in the actual dehydration process of the currently collected plant are input into the dehydration parameter adjustment model. The dehydration parameter adjustment model will output the dehydration parameter adjustment value in the next time segment based on this data, thereby realizing the dynamic adjustment of the parameters in the dehydration process. On the basis of what is shown in this embodiment, the number of time segment divisions can be further specified to perform more precise parameter adjustment during the dehydration process, thereby ensuring the quality of the plant dehydration process.

[0141] This embodiment collects the first fragrance information through the olfactory sensor and the first dehydration image information through the image acquisition unit, constructs a fragrance database and a dehydration image database, establishes a mapping relationship between the fragrance information and the dehydration state, and constructs a dehydration parameter adjustment model. In the actual plant dehydration process, the second fragrance information and the second dehydration image information are collected, and the dehydration parameters in the actual plant dehydration process are adjusted with the help of the dehydration parameter adjustment model, so as to achieve comprehensive monitoring and dataization of the fragrance and dehydration state in the plant dehydration process, more accurately adjust the dehydration parameters according to the real-time fragrance and dehydration state, better cope with the differences of plant raw materials, and optimize the dehydration effect. Compared with traditional empirical adjustment, the method shown in this embodiment is more scientific and efficient, improves the standardization level of plant processing, and ensures the stability of product quality.

[0142] See also Figure 2 and Figure 3 In some embodiments, obtaining first fragrance information collected by the olfactory sensor in different time periods and different plant categories, the first fragrance information including the concentration of volatile substances includes:

[0143] S201, obtaining a plant category label;

[0144] S202, collecting first aroma information of the plants of the current category in the dehydration stage according to a first preset frequency, the first aroma information being the concentration of volatiles generated by the plants of the current category in the dehydration stage, the volatiles comprising at least one of ester compounds, alcohol compounds, aldehyde compounds, terpene compounds and phenolic compounds;

[0145] S203, filling the first fragrance information collected at the first preset frequency into the fragrance data set corresponding to the current category label in chronological order, to obtain the first fragrance information data set corresponding to the current category label;

[0146] S204, repeat the above steps until the first fragrance information data set corresponding to all category labels is obtained;

[0147] A fragrance database is constructed according to the first fragrance information and fragrance tags of different plant categories. The fragrance database includes concentration ranges of volatile substances corresponding to different fragrances, including:

[0148] The following steps are performed for each first aroma information data set:

[0149] S301, plotting a volatile matter change curve of the current category label according to the first aroma information data set; and obtaining an aroma label corresponding to the current category label, the aroma label including at least one of floral aroma, fruity aroma, honey aroma, baking aroma, medicinal aroma, mineral aroma, and smoky aroma;

[0150] S302, selecting a corresponding dehydration time period according to the aroma tag in the volatile matter variation curve, and using a plurality of volatile matter concentration variation ranges within the dehydration time period as aroma threshold information of the current aroma tag;

[0151] S303, mapping and storing the aroma threshold information and the category label;

[0152] S304, repeat the above steps until the aroma threshold information of all category labels is generated and stored to obtain an aroma database.

[0153] In step S201, the category label of the plant can be understood as a further subdivision of the plant. Specifically, the major category of the plant is associated with the subdivision. For example, when the plant is Wuyi Rock Tea, the category labels include Dahongpao, Tieguanyin, Baijiguan, Narcissus, Osmanthus Tea and Huangjingui. Further, the category label can be defined according to the plant category names of various manufacturers, for example, cinnamon, narcissus, white lotus, etc. The division and definition of specific category labels can be modified according to actual needs, and this embodiment does not limit this. For another example, when the plant is a mushroom, the category label can be shiitake mushroom, morel, cordyceps flower, velvet mushroom, tea tree mushroom and other categories; when the plant is a scented tea, the category label can be rose tea, jasmine tea, chrysanthemum tea, etc. For another example, when the plant is a dried vegetable product, the category label corresponds to a specific category of vegetables, such as cabbage, cauliflower, Chinese cabbage, lettuce, sweet potato, purple sweet potato, etc.; for another example, when the plant is a dried fruit product, the category label corresponds to a specific category of fruit, and so on. The category labels shown in this embodiment are associated with the plant categories.

[0154] In step S202, after obtaining the category label of the plant, the first aroma information of the plant in the dehydration stage under the current category is collected according to the first preset frequency. Different volatiles will form different aroma information after mixing. The first preset frequency can be set according to actual needs, or it can be the same as the second preset frequency described later to reduce the computing power of data processing. In this embodiment, an olfactory sensor with a composite collection function can be used to simultaneously obtain the concentrations of different volatiles, or a plurality of olfactory sensors with a single collection function can be used to collect the concentrations of different volatiles. Furthermore, alcohol compounds such as ethanol, isopentanol, etc. can bring some sweet tastes; aldehyde compounds such as geranyl aldehyde have the characteristics of floral or fruity aromas; ester compounds such as ethyl acetate will bring the aroma of fruits; terpene compounds such as limonene, α-pinene, etc., bring the aroma of herbal plants; phenolic compounds such as phenolic hydroxyl compounds, etc., phenolic compounds will also produce some aroma under some suitable environmental conditions.

[0155] In step S203, the first fragrance information collected at the first preset frequency is sorted in time order to form a fragrance data set, which is recorded as the first fragrance information data set. It can be understood that each plant category has a first fragrance information data set, and during the collection process of the first fragrance information data set, the total time of the dehydration process is consistent. The dehydration time of the plant category with the longest dehydration time is used as the total time of the current first fragrance information collection. After being sorted into the first fragrance information data set, the dynamic change data of the concentration of volatiles of different plant categories in the same dehydration time can be known.

[0156] In step S204, the aforementioned first aroma information collection step is repeated until all category labels correspond to a set of first aroma information data sets.

[0157] On this basis, the construction process of the fragrance database is as follows Figure 3 shown.

[0158] In step S301, time is used as an independent variable, and the concentration change value of each volatile in the first aroma information data set is used as a dependent variable to draw a volatile change curve. Then, the first aroma information data set can draw at least one volatile change curve. When there are multiple volatile change curves, multiple volatiles can exist simultaneously in the same time period. The concentration changes of multiple volatiles are different, which creates the difference in aroma in different time periods in the dehydration process. At the same time, the aroma label corresponding to the current category label is obtained. It should be noted that different plant categories correspond to different aroma labels, but they are not unique. For example, taking Dahongpao as an example, there will be two stages of flower fragrance and fruit fragrance in the dehydration process. The flower fragrance is used as a fragrance label, and the fruit fragrance is used as another fragrance label. The flower fragrance is mainly dominated by terpenoid compounds, and the fruit fragrance is mainly dominated by ester compounds. When the concentration value of terpenoid compounds is high, Dahongpao mainly presents flower fragrance, and when the concentration value of ester compounds is high, Dahongpao mainly presents fruit fragrance.

[0159] In step S302 and step S303, according to the definition of the fragrance tag, the corresponding dehydration time period is selected in the volatile change curve, and further, the upper limit and lower limit of the volatile in this dehydration time period are sorted into the threshold range of this volatile. When there are multiple volatiles, the same dehydration time period corresponds to multiple threshold ranges of volatiles. This data is stored with the fragrance tag to form the fragrance threshold information under the current plant category. This process can also be understood as further extracting the characteristics of the fragrance analog quantity after quantifying the concentration of the volatiles to obtain the fragrance characteristics, and storing the fragrance characteristics in the category tag of the current plant category in the form of fragrance threshold information.

[0160] In step S304, the above steps are repeated until the aroma threshold information corresponding to all category tags is generated. It is understandable that there may be multiple aroma threshold information in the same category tag, and each aroma threshold information corresponds to a different dehydration time period. The multiple aroma threshold information is sorted to form an aroma database.

[0161] This embodiment systematically collects and analyzes the concentration of volatiles of different plant categories during the dehydration stage, and draws corresponding concentration change curves to quantify the concentration changes of different volatiles within a specific time period. This process allows the characteristics of plant fragrance to be clearly extracted and defined, which facilitates the construction of the subsequent dehydration parameter adjustment model and the generation of mapping relationships.

[0162] In some embodiments, when the olfactory sensor collects the first fragrance information, synchronously acquiring the first dehydrated image information of the plant in different dehydrated states includes:

[0163] Collecting first dehydrated image information of plants of the current category at the dehydration stage at a second preset frequency and arranging the information in chronological order to obtain a first dehydrated image dataset of plants of the current category, wherein the first dehydrated image information includes a leaf dehydrated image of the current plant;

[0164] Preprocessing the first dehydrated image data set, the preprocessing including noise removal processing and image enhancement processing, to obtain a first processed image data set, the first processed image data set including a plurality of first processed images;

[0165] Extracting features from the first processed images one by one to obtain first image features corresponding to each first processed image, where the first image features include first color features, first texture features, and first shape features;

[0166] splicing the multiple first image features to obtain a first dehydrated image feature set corresponding to the current category label;

[0167] Repeat the above steps until the first dehydrated image feature set corresponding to all category labels is obtained;

[0168] Constructing a dehydration image database according to the first dehydration image information and the dehydration state label includes:

[0169] Obtain a dehydration status tag, where the dehydration status tag is status information of different dehydration degrees;

[0170] Draw a color feature change curve, a texture feature change curve, and a shape feature change curve corresponding to the dehydration state label one by one according to the first dehydration image feature set;

[0171] The color feature change curve, the texture feature change curve and the shape feature change curve of the same category label are divided into a group to obtain a first feature change curve group;

[0172] Repeat the above steps until the first characteristic change curve group corresponding to all category labels is generated to obtain a dehydrated image database.

[0173] In this embodiment, the second preset frequency may be the same as or different from the first preset frequency, and the second preset frequency may be greater than the first preset frequency, so as to collect plant dehydration image data more precisely. In this embodiment, the first dehydration image information collected according to the second preset frequency is arranged in chronological order to obtain a first dehydration image data set. Specifically, the first dehydration image information includes a leaf dehydration image of the current plant, which may be a leaf dehydration image of multiple leaves in a flat state, or a leaf dehydration image of a single leaf. It is set according to actual needs. For example, if the relationship between plant dehydration and plant fragrance information is studied separately, the leaf dehydration image of a single leaf can be used as the first dehydration image information. If it is from the perspective of batch production, it is preferred to use the leaf dehydration image of multiple leaves as the first dehydration image information. This embodiment does not limit this.

[0174] After obtaining the first dehydrated image data set under the current plant category, the first dehydrated image data set is preprocessed to remove image noise in the first dehydrated image information and highlight image features in the first dehydrated image information, so as to facilitate subsequent feature extraction. In this embodiment, the preprocessing includes noise removal and image enhancement processing. Noise removal can be implemented using a Gaussian filter algorithm, and image enhancement processing can be implemented using an FFT transform or a wavelet transform to highlight image features. For ease of distinction, the processed first dehydrated image data set is recorded as a first processed image data set, and the processed first dehydrated image information is recorded as a first processed image.

[0175] The first processed images are subjected to feature extraction one by one. The specific feature extraction process may be: calculating the first-order, second-order, and third-order statistical moments of the color of the first processed image to obtain the first color feature of the current first processed image, calculating the gray-level co-occurrence matrix of the first processed image to obtain the first texture feature corresponding to the first processed image, and performing Zernike moment transformation on the first processed image to obtain the first shape feature corresponding to the first processed image. For ease of description, the first color feature, the first texture feature, and the first shape feature corresponding to each first processed image are organized into a feature vector, which is recorded as the first image feature.

[0176] The multiple first image features are spliced ​​in chronological order to obtain a first dehydration image feature set corresponding to the current category label, which records the dynamic change amount of dehydration displayed by the image of the plant of the current category label.

[0177] Similar to the first aroma information dataset, the above steps are repeated until each category label corresponds to a first dehydrated image feature set.

[0178] Furthermore, a dehydration status label is obtained. As mentioned above, the dehydration status label can be quantified into a water loss rate percentage, or a dehydration status label can be artificially defined, such as "semi-dehydration, slight dehydration, complete dehydration", "mild dehydration, moderate dehydration, severe dehydration", "shallow dehydration, medium dehydration, deep dehydration", etc. It is preferred to use the water loss rate percentage to define the dehydration status label to facilitate the accurate construction of the dehydration image database.

[0179] Specifically, according to the first dehydration image feature set, a color feature change curve, a texture feature change curve and a shape feature change curve are drawn in sequence. It should be noted that each plant category corresponds to a first dehydration image feature set, and each first dehydration image feature set can draw a color feature change curve, a texture feature change curve and a shape feature change curve.

[0180] The color feature change curve, texture feature change curve and shape feature change curve of the same category label are divided into a group to obtain a first feature change curve group. In the first feature change curve group, time is used as an independent variable, and the dehydration state label corresponds to the color feature change interval, texture feature change interval and shape feature change interval of a certain time period. It should be noted that the generation process of the first feature change curve group can be understood as the extraction process of the dehydration image features of each plant category.

[0181] By repeating the above steps, a plurality of first characteristic change curve groups can be obtained, and the plurality of first characteristic change curve groups are mapped and stored with the category labels to obtain a dehydration image database.

[0182] This embodiment systematically collects the first dehydration image information of multiple categories of plants in the dehydration stage, and constructs a complete first dehydration image data set in chronological order. By preprocessing the first dehydration image data set, including noise removal and image enhancement, the image quality is improved, which is conducive to more accurate extraction of image features. The features of each first dehydration image information are extracted from the three dimensions of color, texture and shape, and detailed first image features are obtained. The first feature change curve group reflects the evolution law of the image features of plants in the dehydration state, constructs a complete dehydration image database, integrates the comprehensive image feature information of plants in the dehydration stage, and provides an important reference for subsequent intelligent analysis and parameter control.

[0183] In some embodiments, establishing a mapping relationship between the fragrance database and the dehydrated image database includes:

[0184] The aroma database is expressed by formula (1), which is as follows:

[0185] ;

[0186] In formula (1), For the fragrance database, Indicates Category label Aroma threshold information corresponding to each aroma tag;

[0187] The dehydrated image database is expressed by formula (2), which is as follows:

[0188] ;

[0189] In formula (2), For the dehydrated image database, Indicates Category label A first characteristic change curve group corresponding to a dehydration state label;

[0190] No. Category label The first characteristic change curve group corresponding to the dehydration state label is expressed by formula (3), which is as follows:

[0191] ;

[0192] In formula (3), For the Category label The first color feature of the dehydration status label, For the Category label The first texture feature of the dehydration state label, For the Category label a first shape feature of a dehydrated state label;

[0193] The dehydrated image database is taken as the independent variable and the fragrance database is taken as the dependent variable. The mapping function between the dehydrated image database and the fragrance database is constructed and expressed by formula (4). Formula (4) is as follows:

[0194] ;

[0195] In formula (4), is a mapping function, and the mapping function is a multinomial regression function, is the intercept term, yes The regression coefficient of yes of Power, is the random error term, is the order of the polynomial.

[0196] In this embodiment, the mapping relationship is expressed in the form of a function. For ease of understanding, the fragrance database and the dehydrated image database are respectively expressed in the form of matrices, wherein in the matrix representation of the fragrance database, the behavior category label is listed as a fragrance label, and in the matrix representation of the dehydrated image database, the behavior category label is listed as a dehydration state label.

[0197] In this embodiment, the dehydrated image database is used as an independent variable, that is, the dehydrated image features in the dehydrated image database are used as input features, and the fragrance database is used as a dependent variable, that is, the fragrance features in the fragrance database are used as output features, so as to construct a mapping relationship between the two, that is, a mapping function. In this embodiment, the fragrance feature is a feature quantified by a complex analog quantity, which is specifically reflected in that the fragrance threshold information may contain multiple concentration thresholds of volatile substances, and a single plant category may contain multiple fragrance threshold information. Based on this, this embodiment uses a polynomial regression function as a mapping function, as shown in formula (4), which is used to fully reflect the mapping relationship between the fragrance database and the dehydrated image database.

[0198] This embodiment establishes a mapping relationship between the fragrance database and the dehydration image database, and uses a polynomial regression function to effectively use the dehydration image features as independent variables and the fragrance threshold information as the dependent variable, accurately reflecting the complex relationship between the fragrance characteristics of the plant and its dehydration state. By representing the two databases in matrix form and mapping them using a polynomial regression model, the changes in the fragrance characteristics of different types of plants under different dehydration states can be captured comprehensively and systematically, thereby providing a scientific basis for the optimization and adjustment of the plant dehydration process, and improving the intelligence level of plant processing and the consistency of product quality.

[0199] In some embodiments, a dehydration parameter adjustment model is constructed according to the mapping relationship and the fragrance database and the dehydration image database, and the output result of the dehydration parameter adjustment model is the plant dehydration parameter, and the following further includes:

[0200] When collecting the first dehydrated image information, synchronously obtaining first spectrum information corresponding to the first dehydrated image information;

[0201] Extracting features from the first spectrum information to obtain first spectrum features and constructing a spectrum feature database;

[0202] A dehydration parameter adjustment model is constructed based on the mapping relationship, the fragrance database, and the dehydration image database. The output result of the dehydration parameter adjustment model is the plant dehydration parameters, which also include:

[0203] Based on the spectral feature database, mapping relationship, aroma database and dehydrated image database as multimodal features, a multimodal learning model is constructed, which is expressed by formula (5) and formula (6). Formula (5) is as follows:

[0204] ;

[0205] Formula (6) is as follows:

[0206] ;

[0207] In formula (5) and formula (6), is the output of the multimodal learning model, It is a multimodal learning model based on neural network. is a multimodal feature, are the weight parameters and bias parameters of each neural layer, For the fragrance database, For the dehydrated image database, is the spectral feature database;

[0208] The multimodal features are input into the LSTM model to obtain the prediction results corresponding to the multimodal features, which are expressed by formula (7). Formula (7) is as follows:

[0209] ;

[0210] In formula (7), To predict the results, is the function representation of the LSTM model;

[0211] The output result of the multimodal learning model is fused with the prediction result of the LSTM model to obtain the fusion result, which is expressed by formula (8). Formula (8) is as follows:

[0212] ;

[0213] In formula (8), is the fusion result;

[0214] The fusion result is input into the fully connected layer to obtain the plant dehydration parameter, which is expressed by formula (9). Formula (9) is as follows:

[0215] ;

[0216] In formula (9), is the plant dehydration parameter, is the function representation of the fully connected layer, are the bias parameters and weight parameters of the fully connected layer.

[0217] In this embodiment, the collection of spectral information of plants is introduced, and the image collection unit can be changed into a hyperspectral image collection device to realize the simultaneous collection of the first spectral information during the first dehydration image information collection process.

[0218] Specifically, after the first spectral information is collected, the first spectral information can be processed by denoising, calibration, normalization, etc. to reduce the noise in the first spectral information and highlight the characteristic data in the first spectral information. In this embodiment, the first spectral information is set to better monitor the change in the water loss rate of the plant during the dehydration process. The change in water loss rate will cause the change in the wavelength of the spectrum reflected by the outer surface of the plant. On this basis, the processed first spectral information is feature extracted to obtain the first spectral feature. Specifically, the first spectral feature can be understood as the extracted information of the spectral feature of a specific wavelength, for example, it can be the extracted information of the change amount of the visible light band and the near-infrared band. In the same category label, by extracting features from multiple first spectral information, the spectral feature data set corresponding to the current category label is obtained, and the multiple spectral feature data sets are organized to form a spectral feature database.

[0219] Furthermore, in the process of constructing the dehydration parameter adjustment model, this embodiment uses a multimodal learning model, introduces the encoding of the long short-term memory network into the multimodal learning model, and fuses the two through a fully connected layer to form a dehydration parameter adjustment model.

[0220] Specifically, the data in the spectral feature database, mapping relationship, fragrance database and dehydrated image database are used to construct multimodal features according to plant categories, that is, multiple features under each plant category are integrated and represented as a multimodal feature. In this embodiment, the multimodal learning model is constructed using the framework of a deep neural network model, as shown in formula (5) and formula (6), where the multimodal feature is It should be noted that the multimodal learning model can be represented as a stack of multiple fully connected layers. On this basis, convolutional layers and pooling layers can be introduced to construct a more complex multimodal learning model to improve the accuracy of the output results of the multimodal learning model.

[0221] In this embodiment, the multimodal features are input into the LSTM model, and the LSTM model encodes and decodes the multimodal features containing time series information to obtain the final hidden state information, that is, the prediction result. On this basis, the prediction result is concatenated and fused with the output result of the multimodal learning model to obtain a fusion result. Further, the fusion result is input into the fully connected layer to obtain the plant dehydration parameter. It should be noted that the fully connected layer in formula (9) can be understood as a concatenation of a linear transformation and a nonlinear activation function. Specifically, the parameter information in the linear transformation is , Parameter optimization can be achieved through gradient descent, Adam, etc. It can be understood as a representation of a nonlinear activation function. The nonlinear activation function may be selected such as ReLU, Sigmoid or Tanh, etc., and this embodiment does not limit this.

[0222] This embodiment incorporates the spectral feature database into the construction of the multimodal learning model, so that the model can more comprehensively capture the changing patterns of various physical and chemical properties of plants during the dehydration process. It adopts a multimodal learning framework that combines a deep neural network and an LSTM model to effectively integrate different types of feature information, and performs feature fusion on the output results of the multimodal learning model and the prediction results of the LSTM model, further enriching the final plant dehydration parameter output and improving the accuracy of dehydration parameter prediction.

[0223] In some embodiments, inputting the fusion result into a fully connected layer to obtain the plant dehydration parameter further comprises:

[0224] The humidity change factor, temperature change factor and flipping frequency change factor are used as feedback information, and the feedback information and the fusion result are input into the fully connected layer to obtain the plant dehydration parameter, which is expressed by formula (10). Formula (10) is as follows:

[0225] ;

[0226] In formula (10), is the humidity variation factor, is the temperature variation factor, is the flip frequency change factor;

[0227] The humidity change factor is expressed by formula (11), which is as follows:

[0228] ;

[0229] In formula (11), For humidity The dehydration rate of plants under is the influence coefficient of humidity on dehydration rate, is the base humidity, is the baseline dehydration rate;

[0230] The temperature change factor is expressed by formula (12), which is as follows:

[0231] ;

[0232] In formula (12), For the temperature The dehydration rate of plants under is the base dehydration rate, is the activation energy, is the gas constant, is the reference temperature;

[0233] The flip frequency change factor is expressed by formula (13), which is as follows:

[0234] ;

[0235] In formula (13), For the flip frequency The dehydration rate of plants under is the base dehydration rate, is the influence coefficient of the turning frequency on the dehydration rate, is the attenuation coefficient of the flipping frequency;

[0236] In formula (11) to formula (13), .

[0237] In this embodiment, the influence of temperature, humidity and flipping frequency is reflected in formula (9) in the form of a variation factor, and an updated formula (9), that is, formula (10), is obtained. The temperature variation factor, humidity variation factor and flipping frequency variation factor are introduced into the fully connected layer, so that the prediction result of the entire dehydration parameter adjustment model can be closer to the influence of the parameter information of the actual dehydration process.

[0238] In this embodiment, the humidity change factor is expressed in the form of a linear function, the temperature change factor is expressed in the form of an Arrhenius equation, and the flipping frequency change factor is expressed in a nonlinear form, which is consistent with the influence of three different parameters on the dehydration rate in the plant dehydration process. It can be understood that the specific values ​​corresponding to the constants or parameters in the above formulas can be obtained by collecting and measuring data on the plant dehydration process.

[0239] This embodiment feeds back the real-time environmental parameter changes to the dehydration parameter adjustment model, so that the plant dehydration parameters finally outputted can better adapt to the current actual process conditions, making the prediction of the specific values ​​of the dehydration parameters more accurate and reliable.

[0240] See also Figure 4 In some embodiments, the first sample parameter set and the second sample parameter set are used as training data for the dehydration parameter adjustment model, and the dehydration parameter adjustment model is trained until the output result of the dehydration parameter adjustment model is within a preset accurate threshold range, including:

[0241] S401, dividing the first sample parameter set into a first sample training set and a first sample test set, and dividing the second sample parameter set into a second sample training set and a second sample test set;

[0242] S402, inputting the first sample training set and the second sample training set into the dehydration parameter adjustment model in sequence for training, generating a loss function to calculate the output error in the current dehydration parameter adjustment model during the training process, and adjusting the parameters of the dehydration parameter adjustment model according to the output error until the output error is within a preset error range;

[0243] S403, inputting the first sample test set and the second sample test set into the dehydration parameter adjustment model in sequence, obtaining a first model output result of the first sample test set and a second model output result of the second sample test set;

[0244] S404, calculating the accuracy of the dehydration parameter adjustment model according to the output results of the first model and the output results of the second model;

[0245] S405, determining whether the accuracy is within a preset accuracy threshold;

[0246] S406: If yes, it means that the dehydration parameter adjustment model training is completed.

[0247] In step S401, the first sample parameter set is divided into a first sample training set and a first sample data set. Similarly, the second sample parameter set is divided into a second sample training set and a second sample data set to facilitate subsequent training of a dehydration parameter adjustment model.

[0248] In step S402, the first sample training set and the second sample training set are input into the dehydration parameter adjustment model for training. Furthermore, the loss function in the training process can be calculated to track the output error of the dehydration parameter adjustment model, so as to gradually modify the weight value and bias value of each connection layer in the dehydration parameter adjustment model during the training process until the output error meets the preset error range, that is, the fitting state of the current dehydration parameter adjustment model is good.

[0249] In step S403, the first sample test set and the second sample test set are sequentially input into the dehydration parameter adjustment model to verify the training effect of the current dehydration parameter adjustment model. For ease of description, the output result of the first sample test set in the dehydration parameter adjustment model is recorded as the first model output result, and the output result of the second sample test set in the dehydration parameter adjustment model is recorded as the second model output result.

[0250] In step S404, the accuracy is calculated based on the output results of the first model and the actual dehydration parameter values ​​in the first sample test set, and the accuracy is calculated based on the output results of the second model and the actual dehydration parameter values ​​in the second sample test set, so as to obtain the accuracy of the dehydration parameter adjustment model. It can be understood that this accuracy can be a range, a specific value, or a specific array.

[0251] In step S405, it is determined whether the accuracy is within the range of a preset accuracy threshold. It should be noted that the preset accuracy threshold can be set manually, which reflects the user's requirements for the prediction accuracy of the current dehydration parameter adjustment model. For example, the preset accuracy threshold can be set to 90%.

[0252] In step S406, if the accuracy is within the range of the preset accuracy threshold, it indicates that the current dehydration parameter adjustment model training is completed and the actual dehydration parameter prediction can be performed. If not, it is necessary to repeat the above steps to modify and adjust the bias parameters and weight parameters in the dehydration parameter adjustment model until the accuracy is within the range of the preset accuracy threshold.

[0253] This embodiment uses a training set and a test set to train and verify the model, which can effectively avoid the problem of overfitting and ensure good generalization performance. Through cyclic iterative training, it is ensured that the model eventually achieves a good fitting effect and improves the accuracy of dehydration parameter prediction. This model based on a double-sample training set and a test set makes full use of the available dehydration parameter data, improves the reliability and generalization ability of model training through cross-validation, and constructs a high-quality dehydration parameter prediction model.

[0254] See also Figure 5 In the second aspect, the present embodiment further provides a dehydration process regulation system for retaining plant aromatic substances based on analog quantity, which is applicable to the method described in the first aspect. The system comprises an outer cylinder 2, an inner cylinder 1, a first driving unit 6, a hot air blower 5, a shell 3, a humidifier 7, a sensor assembly 4 and a control unit. The outer cylinder 2 is provided with a plurality of first air holes and a first monitoring window; the inner cylinder 1 is arranged inside the outer cylinder 2, and a first chamber is provided between the inner cylinder 1 and the outer cylinder 2, and the first chamber is used to place plants, and a second chamber is provided inside the inner cylinder 1, and a plurality of second air holes are provided on the inner cylinder 1; the first driving unit 6 is transmission-connected to the outer cylinder 2, and the first driving unit 6 is used to drive the outer cylinder 2 to rotate relative to the inner cylinder 1; the hot air blower 5 is connected to the second chamber;

[0255] The shell 3 is covered on the outside of the outer cylinder 2; the humidifier 7 is arranged on the shell 3; the sensor assembly 4 is arranged on the shell 3, and the sensor assembly 4 is arranged in a first preset area. When the outer cylinder 2 rotates to a preset angle, the first preset area coincides with the first monitoring window. The sensor assembly 4 includes an olfactory sensor, an image acquisition unit, a temperature sensor and a humidity sensor. The olfactory sensor is used to collect the concentration of volatiles of plants in the first chamber, and the image acquisition unit is used to collect dehydration image information of plants in the first chamber; the control unit is electrically connected to the sensor assembly 4, the humidifier 7, the hot air blower 5 and the first driving unit 6 respectively, and the control unit is used to execute the method described in the first aspect.

[0256] In this embodiment, the plant dehydration process adjustment system is embodied by a drum-type dehydration device. For details, please refer to Figure 5 As shown, the system includes an inner cylinder 1, an outer cylinder 2, a first drive unit 6, a sensor assembly 4, a shell 3, a humidifier 7, a hot air blower 5 and a control unit. Among them, the outer cylinder 2 is sleeved on the outside of the inner cylinder 1, and the outer cylinder 2 can roll relative to the inner cylinder 1. There is a first chamber between the outer cylinder 2 and the inner cylinder 1, and the first chamber is filled with plants of a preset mass. By rolling the outer cylinder 2, the plants can be turned over. The second chamber in the inner cylinder 1 is connected to the hot air blower 5, and the hot air flow generated by the hot air blower 5 can enter the first chamber through the second air vent. Similarly, the humidifier 7 is arranged on the shell 3, and the shell 3 is covered on the outside of the outer cylinder 2. The moisture generated by the humidifier 7 can enter the first chamber through the first air vent, thereby providing the plants with the humidity required in the dehydration process. In this embodiment, the first driving unit 6 can be a servo motor or an engine, etc. The sensor assembly 4 includes an olfactory sensor, an image acquisition unit, a humidity sensor and a temperature sensor. It can be understood that the number of sensor assemblies 4 can be multiple, and they are arranged in different areas of the shell 3, the outer tube 2 and the inner tube 1 to comprehensively monitor the environmental data inside the shell 3.

[0257] In this embodiment, the control unit is also electrically connected to the sensor assembly 4, the humidifier 7, the hot air blower 5, the first drive unit 6 and the image acquisition unit, and the image acquisition unit can be an industrial camera or a camera with a spectral information acquisition function. The control unit can be a microcomputer chip. By executing the method described in the first aspect on the control unit, the adjustment of the dehydration parameters of the entire system during the plant dehydration process can be achieved. For the specific content of the method described in the first aspect, please refer to the above text, and this embodiment will not be redundantly described here.

[0258] In a third aspect, this embodiment further provides a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described in the first aspect is implemented.

[0259] The computer program involved in this embodiment can be stored in a computer device readable storage medium, which includes but is not limited to a disk, a magnetic tape, a magnetic card, a floppy disk, a flash memory, an optical disk, an optical card, a read-only memory (ROM), a random access memory (RAM), an erasable programmable ROM (EPROM) and an electrically erasable programmable ROM (EEPROM), etc., and also includes other biological, physical or chemical structures that can achieve the same or equivalent functions as the storage media listed above, such as DNA, RNA, protein and other units with information storage capabilities. In a specific embodiment, the storage medium involved can be one of the above-mentioned media types, or a combination of the above-mentioned media types. In different embodiments, the computer program involved in the embodiment can be stored in a single medium in a centralized manner, or it can be stored in multiple media in a distributed manner. The memory containing the computer device readable storage medium can be a non-volatile memory or a random access memory. These computer device readable storage media can be built into the device, or they can be connected to the device involved in the embodiment as an external device or a part of an external device. In some embodiments, a memory with a computer device readable storage medium is deployed locally; in other embodiments, a solution of deploying the memory away from the processor may also be adopted, such as a network attached memory accessed via an RF circuit or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), etc., or a suitable combination thereof, as long as the computer device can access the memory. In addition, the computer program involved in the embodiment may be stored in plaintext / ciphertext form, or may be designed as training data, which may be integrated and reorganized through model training and implicitly stored in the parameter state of a deep neural network or other machine learning model.

[0260] In a fourth aspect, this embodiment further provides an electronic device, comprising a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method described in the first aspect.

[0261] The processor described in this embodiment can be implemented by hardware, firmware, software or a combination thereof, and can use a circuit, a single or multiple application-specific integrated circuits (Application Specific Integrated Circuit, ASIC), a digital signal processor (Digital Signal Processor, DSP), a digital signal processing device (Digital Signal Processing Device, DSPD), a programmable logic device (Programmable Logic Device, PLD), a field programmable gate array (Field Programmable Gate Array, FPGA), a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, at least one of a microprocessor, and also includes other physical, biological or chemical structures that can achieve similar or equivalent functions to the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of the present application, or any combination of the steps mentioned therein.

[0262] Different from the prior art, the above technical solution collects the first fragrance information through the olfactory sensor and the first dehydration image information through the image acquisition unit, constructs a fragrance database and a dehydration image database, establishes a mapping relationship between the fragrance information and the dehydration state, and constructs a dehydration parameter adjustment model. In the actual plant dehydration process, the second fragrance information and the second dehydration image information are collected, and the dehydration parameters in the actual plant dehydration process are adjusted with the help of the dehydration parameter adjustment model, so as to realize the comprehensive monitoring and dataization of the fragrance and dehydration state in the plant dehydration process, more accurately adjust the dehydration parameters according to the real-time fragrance and dehydration state, better cope with the differences of plant raw materials, and optimize the dehydration effect. Compared with the traditional experience adjustment, the method shown in this technical solution is more scientific and efficient, improves the standardization level of plant processing, and ensures the stability of product quality.

[0263] Finally, it should be noted that although the above embodiments have been described in the specification and drawings of this application, this does not limit the scope of patent protection of this application. All technical solutions generated by replacing or modifying equivalent structures or equivalent processes based on the essential concept of this application using the contents recorded in the specification and drawings of this application, and directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, etc., are included in the scope of patent protection of this application.

Claims

1. A method for regulating the dehydration process of retaining plant aromatic substances based on simulated quantity, characterized in that: include: Acquire first fragrance information collected by the olfactory sensor in different time periods and different plant categories, wherein the first fragrance information includes the concentration of volatiles; construct a fragrance database according to the first fragrance information and the fragrance tag, wherein the fragrance database includes the concentration ranges of volatiles corresponding to different fragrances; Acquire a first dehydration parameter corresponding to the first fragrance information, the first dehydration parameter including a first flipping frequency, first temperature information and first humidity information; construct a first sample parameter set according to the first dehydration parameter and the fragrance database; Synchronously acquiring first dehydration image information of the plant under different dehydration states; Building a dehydration image database according to the first dehydration image information and the dehydration state label; acquiring a second dehydration parameter corresponding to the first dehydration image information, wherein the second dehydration parameter includes a second flipping frequency, second temperature information, and second humidity information; constructing a second sample parameter set according to the second dehydration parameter and the dehydration image database; Establishing a mapping relationship between the fragrance database and the dehydrated image database; Constructing a dehydration parameter adjustment model according to the mapping relationship, the fragrance database, and the dehydration image database, wherein the output result of the dehydration parameter adjustment model is the plant dehydration parameter; Using the first sample parameter set and the second sample parameter set as training data for the dehydration parameter adjustment model, and training the dehydration parameter adjustment model until an output result of the dehydration parameter adjustment model is within a range of a preset accuracy threshold; Acquire the second aroma information and the second dehydration image information in the actual dehydration process in real time, and input the second aroma information and the second dehydration image information into the trained dehydration parameter adjustment model to obtain the actual dehydration parameters; Adjusting parameter values ​​in the current actual dehydration process according to the actual dehydration parameters; Establishing a mapping relationship between the fragrance database and the dehydrated image database includes: The aroma database is represented by formula (1), which is as follows: In formula (1), A is the aroma database, α ij Indicates the aroma threshold information corresponding to the jth aroma label of the i-th category label; The dehydrated image database is expressed by formula (2), and the formula (2) is as follows: In formula (2), B is the dehydrated image database, β im The first characteristic change curve group corresponding to the mth dehydration state label of the i-th category label; The first characteristic change curve group corresponding to the mth dehydration state label of the i-th category label is expressed by formula (3), and the formula (3) is as follows: In formula (3), β im ′ is the first color feature of the mth dehydrated state label of the ith category label, β im ″ is the first texture feature of the mth dehydrated state label of the ith category label, β im ″′ is the first shape feature of the mth dehydrated state label of the i-th category label; The dehydrated image database is used as an independent variable and the aroma database is used as a dependent variable to construct a mapping function between the dehydrated image database and the aroma database, which is expressed by formula (4). The formula (4) is as follows: In formula (4), f(B) is a mapping function, and the mapping function is a multinomial regression function, ω0 is the intercept term, ω k It is B k The regression coefficient, B k is B raised to the kth power, ∈ is the random error term, and n is the order of the polynomial.

2. The dehydration process adjustment method for retaining plant aromatic substances based on simulated amount as claimed in claim 1 is characterized in that: Acquiring first fragrance information collected by the olfactory sensor in different time periods and different plant categories, wherein the first fragrance information includes the concentration of volatile substances including: Get the plant's category label; collecting first aroma information of the plants of the current category in the dehydration stage at a first preset frequency, wherein the first aroma information is the concentration of volatiles generated by the plants of the current category in the dehydration stage, wherein the volatiles include at least one of ester compounds, alcohol compounds, aldehyde compounds, terpene compounds and phenolic compounds; Filling the first fragrance information collected at the first preset frequency into the fragrance data set corresponding to the current category label in chronological order to obtain the first fragrance information data set corresponding to the current category label; Repeat the above steps until the first fragrance information data set corresponding to all category labels is obtained; A fragrance database is constructed according to the first fragrance information and fragrance tags of different plant categories, wherein the fragrance database includes concentration ranges of volatile substances corresponding to different fragrances, including: For each of the first aroma information data sets, the following steps are performed: Draw a volatile matter change curve of the current category label according to the first fragrance information data set; Obtaining a fragrance label corresponding to the current category label, wherein the fragrance label includes at least one of floral fragrance, fruity fragrance, honey fragrance, baking fragrance, medicinal fragrance, mineral fragrance, and smoky fragrance; Selecting a corresponding dehydration time period in the volatile matter variation curve according to the aroma tag, and using a plurality of volatile matter concentration variation ranges within the dehydration time period as aroma threshold information of the current aroma tag; Mapping and storing the aroma threshold information and the category label; Repeat the above steps until the aroma threshold information of all category labels is generated and stored to obtain the aroma database.

3. The method for regulating the dehydration process of plant aromatic substances based on simulated amount retention as claimed in claim 2, characterized in that When the olfactory sensor collects the first fragrance information, synchronously acquiring the first dehydrated image information of the plant in different dehydrated states includes: Collecting first dehydrated image information of plants of the current category at a dehydration stage at a second preset frequency and arranging the information in chronological order to obtain a first dehydrated image dataset of plants of the current category, wherein the first dehydrated image information includes a dehydrated image of leaves of the current plant; Preprocessing the first dehydrated image data set, wherein the preprocessing includes noise removal processing and image enhancement processing, to obtain a first processed image data set, wherein the first processed image data set includes a plurality of first processed images; Extracting features from the first processed images one by one to obtain first image features corresponding to each of the first processed images, where the first image features include first color features, first texture features, and first shape features; splicing the plurality of the first image features to obtain a first dehydrated image feature set corresponding to the current category label; Repeat the above steps until the first dehydrated image feature set corresponding to all category labels is obtained; Constructing a dehydration image database according to the first dehydration image information and the dehydration state label includes: Acquire a dehydration status tag, wherein the dehydration status tag is status information of different dehydration degrees; Draw a color feature change curve, a texture feature change curve, and a shape feature change curve corresponding to the dehydration state label one by one according to the first dehydration image feature set; The color feature change curve, the texture feature change curve and the shape feature change curve of the same category label are divided into a group to obtain a first feature change curve group; Repeat the above steps until the first characteristic change curve group corresponding to all category labels is generated to obtain the dehydration image database.

4. The dehydration process adjustment method for retaining plant aromatic substances based on simulated amount as claimed in claim 1 is characterized in that: A dehydration parameter adjustment model is constructed according to the mapping relationship, the fragrance database, and the dehydration image database, wherein the output result of the dehydration parameter adjustment model is the plant dehydration parameter, and the following further includes: When collecting the first dehydrated image information, synchronously acquiring first spectrum information corresponding to the first dehydrated image information; Extracting features from the first spectral information to obtain first spectral features, and constructing a spectral feature database; A dehydration parameter adjustment model is constructed according to the mapping relationship, the fragrance database, and the dehydration image database. The output result of the dehydration parameter adjustment model is the plant dehydration parameter, and further includes: According to the spectral feature database, mapping relationship, fragrance database and dehydrated image database as multimodal features, a multimodal learning model is constructed, which is expressed by formula (5) and formula (6). Formula (5) is as follows: Y=f DNN (X,Θ); The formula (6) is as follows: X = [A, B, C]; In formula (5) and formula (6), Y is the output result of the multimodal learning model, f DNN (X, Θ) is a multimodal learning model based on a neural network, X is the multimodal feature, Θ is the weight parameter and bias parameter of each neural layer, A is the fragrance database, B is the dehydrated image database, and C is the spectral feature database; The multimodal features are input into the LSTM model to obtain the prediction results corresponding to the multimodal features, which are expressed by formula (7). Formula (7) is as follows: H = LSTM(X); In formula (7), H is the prediction result, and LSTM is the function representation of the LSTM model; The output result of the multimodal learning model is feature fused with the prediction result of the LSTM model to obtain a fusion result, which is expressed by formula (8). The formula (8) is as follows: Z = [Y, H]; In formula (8), Z is the fusion result; The fusion result is input into the fully connected layer to obtain the plant dehydration parameter, which is expressed by formula (9). The formula (9) is as follows: Z′=f′(Z,θ′); In formula (9), Z′ is the plant dehydration parameter, f′ is the function representation of the fully connected layer, and θ′ is the bias parameter and weight parameter of the fully connected layer.

5. The dehydration process adjustment method for retaining plant aromatic substances based on simulated amount as claimed in claim 4 is characterized in that: Inputting the fusion result into the fully connected layer to obtain the plant dehydration parameter also includes: The humidity change factor, the temperature change factor and the flipping frequency change factor are used as feedback information, and the feedback information and the fusion result are input into the fully connected layer to obtain the plant dehydration parameter, which is expressed by formula (10). The formula (10) is as follows: Z′=f′(Z,Q,T,L,θ′); In formula (10), Q is the humidity change factor, T is the temperature change factor, and L is the flipping frequency change factor; The humidity change factor is expressed by formula (11), and the formula (11) is as follows: r(Q)=r0·(1+μ·(Q-Q0)); In formula (11), r(Q) is the plant dehydration rate under humidity Q, μ is the coefficient of humidity on dehydration rate, Q0 is the reference humidity, and r0 is the reference dehydration rate; The temperature change factor is expressed by formula (12), and the formula (12) is as follows: In formula (12), p(T) is the plant dehydration rate at temperature T, p0 is the reference dehydration rate, and E a is the activation energy, R is the gas constant, and T0 is the reference temperature; The flip frequency change factor is expressed by formula (13), and the formula (13) is as follows: l(L)=l0·(1+γ·e -δ·L ); In formula (13), l(L) is the plant dehydration rate under the turning frequency of L, l0 is the reference dehydration rate, γ is the influence coefficient of the turning frequency on the dehydration rate, and δ is the attenuation coefficient of the turning frequency; In formula (11) to formula (13), r0=p0=l0.

6. The dehydration process adjustment method for retaining plant aromatic substances based on simulated amount as claimed in claim 1 is characterized in that: Using the first sample parameter set and the second sample parameter set as training data for the dehydration parameter adjustment model, and training the dehydration parameter adjustment model until the output result of the dehydration parameter adjustment model is within a preset accurate threshold range includes: Dividing the first sample parameter set into a first sample training set and a first sample test set, and dividing the second sample parameter set into a second sample training set and a second sample test set; The first sample training set and the second sample training set are sequentially input into the dehydration parameter adjustment model for training, and a loss function is generated during the training process to calculate the output error in the current dehydration parameter adjustment model, and the parameters of the dehydration parameter adjustment model are adjusted according to the output error until the output error is within a preset error range; Inputting the first sample test set and the second sample test set into the dehydration parameter adjustment model in sequence to obtain a first model output result of the first sample test set and a second model output result of the second sample test set; Calculating the accuracy of the dehydration parameter adjustment model according to the output result of the first model and the output result of the second model; Determining whether the accuracy is within the range of the preset accuracy threshold; If yes, it means that the dehydration parameter adjustment model training is completed.

7. A dehydration process regulation system for retaining plant aromatic substances based on analog quantity, characterized in that: The method according to any one of claims 1 to 6, wherein the system comprises: An outer cylinder, wherein the outer cylinder is provided with a plurality of first air holes and a first monitoring window; An inner cylinder is arranged inside the outer cylinder, a first chamber is provided between the inner cylinder and the outer cylinder, the first chamber is used to place plants, a second chamber is provided inside the inner cylinder, and a plurality of second air holes are provided on the inner cylinder; A first driving unit, drivingly connected to the outer cylinder, and configured to drive the outer cylinder to rotate relative to the inner cylinder; a hot air blower, connected to the second chamber; A shell body, which is arranged outside the outer cylinder; A humidifier, disposed on the housing; A sensor assembly is arranged on the housing, the sensor assembly is arranged in a first preset area, when the outer cylinder rotates to a preset angle, the first preset area coincides with the first monitoring window, the sensor assembly includes an olfactory sensor, an image acquisition unit, a temperature sensor and a humidity sensor, the olfactory sensor is used to collect the concentration of volatiles of the plant in the first chamber, and the image acquisition unit is used to collect dehydration image information of the plant in the first chamber; A control unit is electrically connected to the sensor assembly, the humidifier, the hot air blower and the first driving unit respectively, and the control unit is used to execute the method described in any one of claims 1 to 6.

Images