Quality loss monitoring and feedback system and method in processing process of large yellow tea

Abstract

The invention discloses a large yellow tea processing process quality loss monitoring and feedback system and method. The method comprises the following steps: monitoring tea moisture, tea polyphenol and amino acid indexes on line in real time in each workshop section of large yellow tea; constructing a change curve of tea moisture, tea polyphenol and amino acid indexes in the heaping yellowing section, constructing a correlation model between tea polyphenol / amino acid and processing time / tea stacking temperature / moisture variation, and converging the correlation model according to the correspondingly required tea polyphenol / amino acid / moisture variation at the end of the heaping yellowing section; by taking the shortest processing time of the heaping yellowing section as a screening condition, determining the initial moisture of the tea leaves corresponding to the heaping yellowing starting time point, and the processing time and the tea leaf stacking temperature in the heaping yellowing section; determining the transfer loss values of the fixation section and the heaping for yellowing section and the tea fixation moisture at the fixation end point, and regulating and terminating the fixation section based on the tea fixation moisture; a relation is established between the fixation section and the heaping for yellowing section, and the production efficiency of the whole heaping for yellowing section is improved.

Description

Technical Field

[0001] This invention relates to the field of tea processing technology, specifically to a system and method for monitoring and feedback on quality loss during the processing of Huangda tea. Background Technology

[0002] Huangda tea is a type of yellow tea made from relatively coarse and old raw materials through a key "yellowing" process, characterized by "yellow leaves, yellow liquor, and a mellow, caramelized aroma." The processing of Huangda tea includes withering, fixation, yellowing, and roasting stages. To promote the transformation and upgrading of the tea industry from traditional experience-based judgment to a digital and intelligent modern industry, near-infrared spectroscopy can be used to measure key quality indicators such as moisture, tea polyphenols, and amino acids in the tea leaves during the Huangda tea processing.

[0003] Regarding the withering process, although Huangda tea leaves are coarse and old, and the withering degree is relatively light, changes in moisture and internal components are still the foundation for subsequent processes and need to be monitored. Fixing requires quickly stopping enzyme activity; moisture and leaf temperature are key. However, indicators such as tea polyphenols are more for result verification than process control points. The real focus of monitoring is the piling and yellowing process after rolling. The final roasting is the flavor-setting stage; the degree of moisture and heat (reflected in further changes in tea polyphenols / amino acids) determines whether the characteristic aroma can be formed.

[0004] Currently, when using near-infrared spectroscopy to measure key quality indicators such as moisture, tea polyphenols, and amino acids in tea, most methods involve installing near-infrared spectroscopy analyzers at various stages of the tea processing to measure changes in these indicators. However, when initially establishing time-series control of moisture, tea polyphenols, and amino acids in the withering, fixation, yellowing, and baking stages of Huangda tea processing, most methods only monitor quality indicator data independently and online in real time at each stage, without establishing process connections between consecutive stages. Therefore, the independent control method for each stage cannot improve the processing efficiency of Huangda tea. Summary of the Invention

[0005] The purpose of this invention is to provide a system and method for monitoring and feedback on quality loss during the processing of Huangda tea, in order to solve the technical problem that most existing technologies only monitor quality index data independently online in real time at each stage, without establishing process connections between two stages. Therefore, the independent control method of each stage cannot improve the processing efficiency of Huangda tea.

[0006] To solve the above-mentioned technical problems, the present invention specifically provides the following technical solution: A method for monitoring and providing feedback on quality loss during the processing of Huangda tea includes the following steps: Online real-time monitoring of tea moisture, tea polyphenols, and amino acid indicators during the withering, fixation, yellowing, and baking processes of Huangda tea; The change curves of tea moisture, tea polyphenols and amino acids in the yellowing process were constructed. A correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change was constructed. The correlation model was converged by the required change of tea polyphenols / amino acids / moisture at the end of the yellowing process to determine the initial moisture of tea at the start time of yellowing. Using the shortest yellowing process as the screening criterion, the initial moisture content of the tea corresponding to the starting time of yellowing under this screening condition, as well as the processing time and tea stacking temperature in the yellowing process, were determined. The changes in tea moisture content caused by the transfer process between the withering and yellowing stages are collected to determine the transfer loss value between the withering and yellowing stages. The initial moisture content of the tea corresponding to the yellowing start time point is compensated by the transfer loss value to determine the withering moisture content of the tea at the end of the withering stage. Based on the withering moisture content of the tea at the end of the withering stage, the withering stage is terminated.

[0007] As a preferred embodiment of the present invention, a near-infrared spectroscopy analyzer is installed in the withering, fixing, yellowing and baking sections of Huangda tea, and the data of the near-infrared spectroscopy analyzer is transmitted to the industrial control computer in real time via Ethernet or fieldbus. The industrial control computer constructs an online prediction model to convert the raw spectrum of the near-infrared spectrometer into real-time indicators of tea moisture, tea polyphenols, and amino acids. The model is then validated and trained until it can monitor the tea moisture, tea polyphenols, and amino acid indicators online in real time using the near-infrared spectrometer.

[0008] As a preferred embodiment of the present invention, the method for constructing a correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change is as follows: Design experimental conditions: Set different initial moisture gradients for tea leaves in the yellowing stage, and set different temperature and humidity environments for the stacking environment in the yellowing stage. Arbitrarily combine different initial moisture gradients and different temperature and humidity environments for the stacking environment to form multiple sets of experimental conditions. Monitoring multiple parameters: Start the yellowing process and insert temperature and humidity sensors into the surface, middle and core layers of the tea pile to continuously record the tea pile temperature and initial moisture content, forming a multi-variable system of "initial moisture content of tea - tea pile temperature - processing time". Monitor the "tea polyphenols - amino acids" index online in real time at the time point after the yellowing process begins. Construct the association model: [P, A] = f(W, T, t); Where P is the content of tea polyphenols, A is the content of amino acids, W is the initial moisture content of tea corresponding to the start time of the yellowing process, T is the stacking temperature of tea in the yellowing process, t is the processing time of the yellowing process, and f() is the mathematical function relationship between processing time / change in tea stacking temperature / moisture content and tea polyphenols / amino acids.

[0009] As a preferred embodiment of the present invention, the method for constructing the association model is as follows: A machine learning model is constructed, and high-level features corresponding to multiple sets of experimental conditions formed by "initial moisture content of tea leaves - stacking temperature of tea leaves - processing time" are constructed. The high-level features and the multiple sets of experimental conditions formed by "initial moisture content of tea leaves - stacking temperature of tea leaves - processing time" are combined to form an experimental dataset. The training set of this experimental dataset is used as the input data of the machine learning model. Using online real-time monitoring of "tea polyphenols-amino acids" as output data, the study aimed to learn the effects of the interaction of initial moisture content of tea leaves, tea stacking temperature, and processing time on the composition of tea polyphenols / amino acids. The machine learning model was validated using the test set of the experimental dataset. The prediction results of the machine learning model were compared with the real-time monitoring results of tea polyphenols and amino acids under the corresponding experimental conditions. When the root mean square error between the prediction results and the real-time monitoring results was less than a set threshold, the machine learning model was used as a correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change.

[0010] As a preferred embodiment of the present invention, the advanced features include: accumulated temperature of tea leaves, temperature-humidity ratio / water activity, and the rate of change of initial moisture or temperature of tea leaves between adjacent time points; Wherein, the accumulated temperature of tea leaves = Σ(average temperature of each time interval * time interval).

[0011] As a preferred embodiment of the present invention, the correlation model is converged based on the required tea polyphenols / amino acids / moisture at the end of the yellowing process to determine the initial moisture content of the tea corresponding to the yellowing start time, as well as the processing time and tea stacking temperature in the yellowing process.

[0012] As a preferred embodiment of the present invention, the initial moisture content of the tea leaves corresponding to the yellowing start time point obtained by the convergence of the correlation model, as well as the processing time and tea accumulation temperature in the yellowing process, are used to form a dataset. The relationship between processing time and tea stacking temperature in the yellowing stage and the initial moisture content of the tea is constructed, resulting in t=g(W,T); where W is the initial moisture content of the tea, T is the tea stacking temperature in the yellowing stage, t is the processing time in the yellowing stage, and g() is the functional relationship corresponding to the relationship. Using the shortest processing time t in the yellowing stage as the screening condition, the initial moisture content of the tea corresponding to the starting time of yellowing under this screening condition is determined.

[0013] As a preferred embodiment of the present invention, the method for learning the changes in tea moisture content caused by the transfer process between the fixing and yellowing stages using a deep learning model is as follows: Collect the tea's moisture content W1 at the end time of the fixation stage and the initial moisture content W2 at the start time of the yellowing stage. By coupling the tea fixation moisture W1 and the initial tea moisture W2, the transfer loss value of the fixation stage and the yellowing stage is determined. The transfer loss value is used as the compensation value to determine the tea moisture at the end of fixation. The initial tea moisture corresponding to the selected yellowing start time point is combined with the compensation value to reversely determine the tea fixation moisture at the end of fixation.

[0014] In addition, the present invention also provides a system for monitoring and feedback of quality loss during the processing of Huangda tea, comprising: The tea index monitoring module is used to monitor the moisture, tea polyphenols, and amino acid indicators of tea leaves online in real time during the withering, fixation, yellowing, and baking processes of Huangda tea. The correlation model building module is used to build correlation models between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change. The yellowing period control module is used to converge the correlation model based on the required changes in tea polyphenols / amino acids / moisture at the end of the yellowing process, and to determine the initial moisture content of the tea corresponding to the start time of the yellowing process by using the shortest processing time of the yellowing process as the screening condition. The transfer loss calculation module is used to collect the changes in tea moisture index caused by the transfer process in the withering and yellowing stages, so as to determine the transfer loss value in the withering and yellowing stages. The reverse calculation module for fixation moisture content uses the initial moisture content of the tea leaves at the start of the yellowing process, as well as the transfer loss values ​​during the fixation and yellowing stages, to reverse calculate the fixation moisture content of the tea leaves at the end of the fixation process.

[0015] As a preferred embodiment of the present invention, the tea moisture content at the end of the fixation process is reversed based on the fixation moisture reverse deduction module, and the real-time tea moisture content of the fixation process is monitored in real time by the tea index monitoring module. When the real-time tea moisture content is the same as the tea fixation moisture content, the fixation process is terminated. The initial moisture content of the tea leaves corresponding to the start time of the yellowing process is determined based on the yellowing period control module, and the processing time of the yellowing process is obtained by converging the correlation model. Based on the correlation model [P,A] = f(W, T, t) between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change, the tea stacking temperature of the yellowing stage is obtained based on the correlation model, where the initial moisture W, tea polyphenol content P, amino acid content A, and processing time t of the yellowing stage are known. When the real-time temperature of the tea leaves is the same as the tea accumulation temperature in the yellowing stage obtained based on the correlation model, the yellowing stage is terminated.

[0016] Compared with the prior art, the present invention has the following advantages: This invention establishes a link between the fixation and yellowing stages. Based on the optimal processing time and final tea stacking temperature of the yellowing stage, the invention regulates the end of the yellowing stage. Furthermore, it combines the initial moisture content of the tea corresponding to the start time of yellowing with the transfer loss values ​​of the fixation and yellowing stages to reversely determine the fixation moisture content of the tea at the end of fixation. By controlling the tea moisture content in the fixation stage, the processing time of the yellowing stage for Huangda tea can be minimized while ensuring that the requirements for tea polyphenols / amino acids are met, thereby improving the overall production efficiency of the yellowing stage. Attached Figure Description

[0017] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the quality loss monitoring and feedback method according to an embodiment of the present invention. Figure 2 This is a block diagram of the overall structure of the quality loss monitoring and feedback system according to an embodiment of the present invention; The labels in the diagram represent the following: Tea indicator monitoring module 1, correlation model construction module 2, yellowing period control module 3, transfer loss calculation module 4, and fixation moisture reverse deduction module 5. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 As shown, this invention provides a method for monitoring and providing feedback on quality loss during the processing of Huangda tea, comprising the following steps: Online real-time monitoring of tea moisture, tea polyphenols, and amino acid levels is conducted during the withering, fixation, yellowing, and baking processes of Huangda tea.

[0021] The variation curves of tea moisture, tea polyphenols, and amino acids during the yellowing process were constructed. A correlation model was built between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change. The correlation model was converged by the required changes in tea polyphenols / amino acids / moisture at the end of the yellowing process to determine the initial moisture of the tea at the start time of yellowing, as well as the processing time and tea stacking temperature during the yellowing process.

[0022] Using the shortest yellowing process as the screening criterion, the initial moisture content of the tea corresponding to the starting time of yellowing under this screening condition, as well as the processing time and tea stacking temperature in the yellowing process, were determined.

[0023] The changes in tea moisture content caused by the transfer process between the withering and yellowing stages are collected to determine the transfer loss value between the withering and yellowing stages. The initial moisture content of the tea corresponding to the yellowing start time point is compensated by the transfer loss value to determine the withering moisture content of the tea at the end of the withering stage. Based on the withering moisture content of the tea at the end of the withering stage, the withering stage is terminated.

[0024] In the preparation process of Huangda tea, the moisture, tea polyphenols, and amino acid indicators of the tea leaves are monitored online in real time during the withering, fixation, yellowing, and baking stages. The core purpose is to precisely control its unique "fixation", "yellowing", and "baking" processes.

[0025] Furthermore, this implementation method is based on real-time monitoring data of tea moisture, tea polyphenols, amino acids, processing time, and tea stacking temperature during the yellowing process. It constructs a correlation model between tea polyphenols / amino acids and changes in processing time / tea stacking temperature / moisture content, thereby achieving dynamic process modeling of "multivariate input (processing time - tea stacking temperature - moisture change) - multivariate output (tea polyphenols - amino acids)". When the tea polyphenols / amino acids meet the requirements and the processing time of the yellowing process is minimized, the production efficiency of the entire yellowing process is improved. This determines the optimal initial moisture content of the tea entering the yellowing process, as well as the corresponding processing time and tea stacking temperature. When the processing time and tea stacking temperature of the yellowing process reach the set values ​​obtained from the correlation model, the yellowing process is terminated.

[0026] Then, based on the changes in tea moisture content caused by the transfer process between the withering and yellowing stages, the transfer loss values ​​of the withering and yellowing stages are determined to determine the optimal initial moisture content and transfer loss value of the tea entering the yellowing stage. This allows for the determination of the optimal withering moisture content at the end of the withering stage, thereby controlling the termination of the withering stage.

[0027] Therefore, this implementation method establishes a connection between the withering and yellowing stages. By controlling the moisture content of the tea leaves in the withering stage, the processing time in the yellowing stage of Huangda tea can be minimized while ensuring that the requirements for tea polyphenols / amino acids are met, thereby improving the overall production efficiency of the yellowing stage.

[0028] Specifically, near-infrared spectroscopy analyzers are installed in the withering, fixing, yellowing, and baking sections of Huangda tea. The near-infrared spectroscopy analyzers transmit data to the industrial control computer in real time via Ethernet or fieldbus.

[0029] An industrial control computer was used to build an online prediction model to convert the raw spectra of a near-infrared spectroscopy analyzer into real-time indicators of tea moisture, tea polyphenols, and amino acids. The online prediction model was then validated and trained until it was possible to monitor the indicators of tea moisture, tea polyphenols, and amino acids in real time online using a near-infrared spectroscopy analyzer.

[0030] The technical principle of using near-infrared spectroscopy (NIRS) to detect key indicators of tea leaves, such as moisture, tea polyphenols, and amino acids, is as follows: It mainly utilizes the absorption and reflection of near-infrared light (wavelength range typically 780-2500 nm) by substances. When light shines on tea leaves, the hydrogen-containing groups in the tea leaves (such as OH, NH, CH, etc.) will undergo harmonic and combination vibrations, absorbing light of specific wavelengths.

[0031] The strong absorption peaks of moisture and OH bonds are closely related.

[0032] Organic compounds such as tea polyphenols, amino acids, and caffeine have unique absorption characteristics related to CH and NH bonds.

[0033] By detecting the reflected or transmitted spectra, a sample's "chemical fingerprint" can be obtained. By establishing a mathematical model with a standard sample of known chemical composition (determined by traditional chemical methods) (this process is called "calibration" or "modeling"), it is possible to make rapid and non-destructive predictions of the chemical composition of a new sample.

[0034] This implementation method primarily utilizes near-infrared spectroscopy to achieve real-time, continuous monitoring of changes in moisture, tea polyphenols, and amino acids in tea leaves on the production line. Specifically, diffuse reflectance probes are installed on the Huangda tea production line. On the production line, offline sampling was conducted simultaneously at different stages of high, medium, and low moisture content (corresponding to withering, fixation, yellowing, and baking processes, respectively): ①Spectral acquisition: Allow the material to pass under the probe and save its spectrum.

[0035] ② Reference value determination: Immediately remove the corresponding material and accurately determine its moisture content using the standard oven method (this is the "gold standard").

[0036] ③ Model Establishment: The collected online spectra are correlated with the corresponding real moisture values, and algorithms such as PLS are used to build a model. Preprocessing algorithms are particularly important, and it is necessary to focus on eliminating interference from the online environment, such as baseline drift and temperature effects.

[0037] The first stage involves constructing a correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change, which specifically includes: Design experimental conditions: Set different initial moisture gradients for tea leaves in the yellowing stage, and set different temperature and humidity environments for the stacking environment in the yellowing stage. Arbitrarily combine different initial moisture gradients and different temperature and humidity environments for the stacking environment to form multiple sets of experimental conditions. Monitoring multiple parameters: Start the yellowing process and insert temperature and humidity sensors into the surface, middle and core layers of the tea pile to continuously record the tea pile temperature and initial moisture content, forming a multi-variable system of "initial moisture content of tea - tea pile temperature - processing time". Monitor the "tea polyphenols - amino acids" index online in real time at the time point after the yellowing process begins. Construct the association model: [P, A] = f(W, T, t); Where P is the content of tea polyphenols, A is the content of amino acids, W is the initial moisture content of tea corresponding to the start time of the yellowing process, T is the stacking temperature of tea in the yellowing process, t is the processing time of the yellowing process, and f() is the mathematical function relationship between processing time / change in tea stacking temperature / moisture content and tea polyphenols / amino acids.

[0038] Constructing a correlation model between "processing time - tea stacking temperature - moisture change" and "tea polyphenols - amino acid changes" during the yellow tea fermentation process is a crucial step in moving from experience-based control to digitalization and precision in yellow tea processing. This is a typical dynamic process modeling problem involving "multivariate inputs (processing time, tea stacking temperature, moisture change) - multivariate outputs (tea polyphenols, amino acids)".

[0039] Its core logic is to transform the human-perceived process conditions (touching humidity, smelling aroma, and observing leaf color) into quantifiable data through experimental design and intensive sampling, and then use mathematical tools to find the inherent patterns between these data.

[0040] The specific steps for building an association model can be summarized into four stages: “designing experiments, intensive sampling, data analysis, and model validation”.

[0041] The specific steps for process experiment design and data acquisition in the first stage are as follows: 1. Design multi-level process experiments: Controlled variables: In a controllable workshop or experimental chamber, set different initial moisture gradients (e.g., set the moisture content of blanched leaves to 40%, 45%, and 50%) and different temperature and humidity environments for stacking (e.g., low temperature and low humidity, medium temperature and medium humidity, and high temperature and high humidity).

[0042] Experiment: For each set of conditions (e.g., "initial moisture 50% + ambient temperature 40℃"), start the yellowing process.

[0043] 2. Implement intensive monitoring and sampling: Real-time monitoring of environmental parameters: Temperature and humidity sensors are inserted into the surface, middle, and core layers of the tea pile to continuously record temperature and relative humidity (which can be converted into the equilibrium moisture content of the tea leaves). This is the source of "moisture-temperature-time" data.

[0044] Destructive sampling: At key time points after the start of the yellowing process (such as 0h, 2h, 4h, 6h, 8h, 12h, 24h...), representative samples (approximately 50-100 grams) are taken from different locations.

[0045] Rapid sample processing: Immediately after sampling, a. Rapid physical measurement: Determine the immediate moisture content of the sample using a portable moisture meter. b. Termination of transformation: Rapidly inactivate the sample using microwave or high temperature to prevent further changes after sampling. c. Preparation of analytical samples: Grind a portion of the sample into powder for subsequent precise analysis.

[0046] 3. "Gold Standard" Laboratory Testing: For each ground sample, the absolute values ​​of moisture, tea polyphenols, and amino acids were precisely determined using standard chemical methods (oven drying method, Folin-Ciocalteu method, ninhydrin method). This serves as the "true value" benchmark for all models.

[0047] The implementation method for constructing the association model is specifically the second stage, which includes: A machine learning model is constructed, and high-level features corresponding to multiple sets of experimental conditions formed by "initial moisture content of tea leaves - stacking temperature of tea leaves - processing time" are constructed. The high-level features and the multiple sets of experimental conditions formed by "initial moisture content of tea leaves - stacking temperature of tea leaves - processing time" are combined to form an experimental dataset. The training set of this experimental dataset is used as the input data of the machine learning model. The high-level features include: accumulated temperature of tea leaves, temperature-humidity ratio / water activity, and the rate of change of initial moisture content or temperature of tea leaves between adjacent time points. Wherein, the accumulated temperature of tea leaves = Σ(average temperature of each time interval * time interval).

[0048] Using online real-time monitoring of "tea polyphenols-amino acids" as output data, the study aimed to learn the effects of the interaction of initial moisture content of tea leaves, tea stacking temperature, and processing time on the composition of tea polyphenols / amino acids. The machine learning model was validated using the test set of the experimental dataset. The prediction results of the machine learning model were compared with the real-time monitoring results of tea polyphenols and amino acids under the corresponding experimental conditions. When the root mean square error between the prediction results and the real-time monitoring results was less than a set threshold, the machine learning model was considered as the correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change.

[0049] Phase Two: Data Processing and Model Building (Uncovering Patterns from Data), specifically including: Data processing and exploration: The data is organized into tables: each row represents a sample, including its sampling time, temperature and moisture at the location, as well as the corresponding tea polyphenol content and amino acid content.

[0050] Construct a correlation model: Independent variables: time, temperature, moisture (can be instantaneous values, average values, integral values, such as "accumulated temperature"); Dependent variables: tea polyphenol content, amino acid content, or their conversion rate.

[0051] Machine learning models, such as random forests, support vector machines, and neural networks, offer better prediction accuracy when the relationships are highly complex and nonlinear. They automatically capture the influence of complex interactions between water, temperature, and time on composition. The model output is a mathematical function or algorithm that takes a set of inputs (W, T, t) as input and outputs the predicted values ​​for P and A.

[0052] Model validation: Cross-validation is used: modeling is done with a portion of the experimental data, and the prediction accuracy is tested with another portion of data not used in the modeling. Key evaluation metric: Coefficient of determination (R²). 2 ) and the root mean square error of prediction (RMSEP). R 2 The closer to 1, the smaller the RMSEP, and the better the model.

[0053] The correlation model of tea polyphenols / amino acids / moisture required at the end of the yellowing process converges to determine the initial moisture content of the tea at the start time of the yellowing process, as well as the processing time and tea stacking temperature in the yellowing process.

[0054] The initial moisture content of the tea leaves corresponding to the starting time of the yellowing process obtained by the convergence of the correlation model, as well as the processing time and tea stacking temperature in the yellowing process, are used to form a dataset. The relationship between processing time and tea stacking temperature in the yellowing stage and the initial moisture content of the tea is constructed, resulting in t=g(W,T); where W is the initial moisture content of the tea, T is the tea stacking temperature in the yellowing stage, t is the processing time in the yellowing stage, and g() is the functional relationship corresponding to the relationship. Using the shortest processing time t in the yellowing stage as the screening condition, the initial moisture content of the tea corresponding to the starting time of yellowing under this screening condition is determined.

[0055] Regarding the above-constructed correlation model: [P, A] = f(W, T, t), the specific application method is as follows: after inputting the required tea polyphenol content P and amino acid content A in the yellowing process into the correlation model, the output can be obtained as the initial moisture content W of the tea corresponding to the yellowing start time, the tea stacking temperature T of the yellowing process, and the processing time t of the yellowing process.

[0056] The initial moisture content W of the tea leaves, the tea leaf stacking temperature T, and the processing time t can be combined in different ways. By using the shortest processing time t in the yellowing stage as the screening condition, the initial moisture content W of the tea leaves corresponding to the starting time of the yellowing stage and the final tea leaf stacking temperature in the yellowing stage can be determined.

[0057] Most existing technologies set the initial moisture content of the tea leaves at the starting point of the yellowing process as the moisture content of the tea leaves during the fixing process. However, this implementation takes into account the changes in the moisture content of the tea leaves caused by the transfer process between the fixing and yellowing processes.

[0058] Specifically, the method for using a deep learning model to learn the changes in tea moisture content caused by the transfer process between the withering and yellowing stages is as follows: Collect the tea's moisture content W1 at the end of the fixation stage and the initial moisture content W2 at the beginning of the yellowing stage. By coupling the tea fixation moisture W1 and the initial tea moisture W2, the transfer loss value of the fixation stage and the yellowing stage is determined. The transfer loss value is used as the compensation value to determine the tea moisture at the end of fixation. The initial tea moisture corresponding to the selected yellowing start time point is combined with the compensation value to reversely determine the tea fixation moisture at the end of fixation.

[0059] This implementation method, based on the optimal processing time of the yellowing stage and the final tea stacking temperature of the yellowing stage, regulates the end of the yellowing stage. It also combines the initial moisture content of the tea corresponding to the start time of yellowing with the transfer loss values ​​of the fixation and yellowing stages to reverse-determine the fixation moisture content of the tea at the end of fixation. By establishing a connection between the fixation and yellowing stages, and controlling the tea moisture content in the fixation stage, the processing time of the yellowing stage for Huangda tea can be minimized while ensuring that the requirements for tea polyphenols / amino acids are met, thus improving the overall production efficiency of the yellowing stage.

[0060] In addition, such as Figure 2As shown, the present invention also provides a system for monitoring and feedback of quality loss during the processing of Huangda tea, including: a tea index monitoring module 1, a correlation model construction module 2, a yellowing period control module 3, a transfer loss calculation module 4, and a reverse deduction module for fixation moisture 5.

[0061] The tea index monitoring module 1 is used to monitor the moisture, tea polyphenols, and amino acid indicators of tea leaves online in real time during the withering, fixation, yellowing, and baking processes of Huangda tea.

[0062] The correlation model building module 2 is used to build a correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change.

[0063] The yellowing period control module 3 is used to converge the correlation model based on the changes in tea polyphenols / amino acids / moisture at the end of the yellowing process, and to determine the initial moisture content of the tea corresponding to the start time of the yellowing process by using the shortest processing time of the yellowing process as the screening condition.

[0064] The transfer loss calculation module 4 is used to collect the changes in tea moisture index caused by the transfer process in the withering and yellowing stages, in order to determine the transfer loss value in the withering and yellowing stages.

[0065] The reverse deduction module 5 for fixation moisture is based on the initial moisture content of the tea leaves at the start time of the yellowing process, as well as the transfer loss values ​​of the fixation and yellowing processes, and reverse deduces the fixation moisture content of the tea leaves at the end of the fixation process.

[0066] Specifically, based on the reverse deduction module 5 for fixing moisture, the fixing moisture of the tea at the end of fixing is deduced. The real-time tea moisture of the fixing section is monitored in real time by the tea index monitoring module 1. When the real-time tea moisture is the same as the fixing moisture of the tea, the fixing section is terminated.

[0067] The initial moisture content of the tea leaves is determined based on the yellowing period control module 3, and the processing time of the yellowing section is obtained by converging the correlation model.

[0068] Based on the correlation model [P,A] = f(W, T, t) between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change, the tea stacking temperature of the yellowing stage is obtained based on the correlation model, where the initial moisture W, tea polyphenol content P, amino acid content A, and processing time t of the yellowing stage are known. When the real-time temperature of the tea leaves is the same as the tea accumulation temperature in the yellowing stage obtained based on the correlation model, the yellowing stage is terminated.

[0069] This invention establishes a link between the withering and yellowing stages. By controlling the moisture content of the tea leaves in the withering stage, the processing time in the yellowing stage of Huangda tea can be minimized while ensuring that the requirements for tea polyphenols / amino acids are met, thereby improving the overall production efficiency of the yellowing stage.

[0070] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.

Claims

1. A method for monitoring and feedback on quality loss during the processing of Huangda tea, characterized in that, Includes the following steps: Online real-time monitoring of tea moisture, tea polyphenols, and amino acid indicators during the withering, fixation, yellowing, and baking processes of Huangda tea; The change curves of tea moisture, tea polyphenols and amino acids in the yellowing process were constructed. A correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change was constructed. The correlation model was converged by the required change of tea polyphenols / amino acids / moisture at the end of the yellowing process to determine the initial moisture of tea at the start time of yellowing. Using the shortest yellowing process as the screening criterion, the initial moisture content of the tea corresponding to the starting time of yellowing under this screening condition, as well as the processing time and tea stacking temperature in the yellowing process, were determined. The changes in tea moisture content caused by the transfer process between the withering and yellowing stages are collected to determine the transfer loss value between the withering and yellowing stages. The initial moisture content of the tea corresponding to the yellowing start time point is compensated by the transfer loss value to determine the withering moisture content of the tea at the end of the withering stage. Based on the withering moisture content of the tea at the end of the withering stage, the withering stage is terminated.

2. The method for monitoring and feedback on quality loss during the processing of Huangda tea according to claim 1, characterized in that, Near-infrared spectroscopy analyzers are installed in the withering, fixing, yellowing, and baking sections of Huangda tea, and the data from the near-infrared spectroscopy analyzers is transmitted to the industrial control computer in real time via Ethernet or fieldbus. The industrial control computer constructs an online prediction model to convert the raw spectrum of the near-infrared spectrometer into real-time indicators of tea moisture, tea polyphenols, and amino acids. The model is then validated and trained until it can monitor the tea moisture, tea polyphenols, and amino acid indicators online in real time using the near-infrared spectrometer.

3. The method for monitoring and feedback on quality loss during the processing of Huangda tea according to claim 1, characterized in that, The method for constructing a correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change is as follows: Design experimental conditions: Set different initial moisture gradients for tea leaves in the yellowing stage, and set different temperature and humidity environments for the stacking environment in the yellowing stage. Arbitrarily combine different initial moisture gradients and different temperature and humidity environments for the stacking environment to form multiple sets of experimental conditions. Monitoring multiple parameters: Start the yellowing process and insert temperature and humidity sensors into the surface, middle and core layers of the tea pile to continuously record the tea pile temperature and initial moisture content, forming a multi-variable "initial moisture content of tea - tea pile temperature - processing time". Monitor the "tea polyphenols - amino acids" index online in real time at the time point after the yellowing process begins. Construct the association model: [P, A] = f(W, T, t); Where P is the content of tea polyphenols, A is the content of amino acids, W is the initial moisture content of tea corresponding to the start time of the yellowing process, T is the stacking temperature of tea in the yellowing process, t is the processing time of the yellowing process, and f() is the mathematical function relationship between processing time / change in tea stacking temperature / moisture content and tea polyphenols / amino acids.

4. The method for monitoring and feedback on quality loss during the processing of Huangda tea according to claim 3, characterized in that, The implementation method for constructing the association model is as follows: A machine learning model is constructed, and high-level features corresponding to multiple sets of experimental conditions formed by "initial moisture content of tea leaves - stacking temperature of tea leaves - processing time" are constructed. The high-level features and the multiple sets of experimental conditions formed by "initial moisture content of tea leaves - stacking temperature of tea leaves - processing time" are combined to form an experimental dataset. The training set of this experimental dataset is used as the input data of the machine learning model. Using online real-time monitoring of "tea polyphenols-amino acids" as output data, the study aimed to learn the effects of the interaction of initial moisture content of tea leaves, tea stacking temperature, and processing time on the composition of tea polyphenols / amino acids. The machine learning model was validated using the test set of the experimental dataset. The prediction results of the machine learning model were compared with the real-time monitoring results of tea polyphenols and amino acids under the corresponding experimental conditions. When the root mean square error between the prediction results and the real-time monitoring results was less than a set threshold, the machine learning model was used as a correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change.

5. The system and method for monitoring and feedback on quality loss during the processing of Huangda tea according to claim 4, characterized in that, The advanced features include: tea accumulated temperature, temperature-humidity ratio / water activity, and the rate of change of initial moisture or temperature of tea between adjacent time points; Wherein, the accumulated temperature of tea leaves = Σ(average temperature of each time interval * time interval).

6. The method for monitoring and feedback on quality loss during the processing of Huangda tea according to claim 4, characterized in that, The correlation model converges with the required tea polyphenols / amino acids / moisture at the end of the yellowing process to determine the initial moisture content of the tea at the start time of the yellowing process, as well as the processing time and tea stacking temperature in the yellowing process.

7. The method for monitoring and feedback on quality loss during the processing of Huangda tea according to claim 6, characterized in that, The initial moisture content of the tea leaves corresponding to the starting time of the yellowing process obtained by the convergence of the correlation model, as well as the processing time and tea stacking temperature in the yellowing process, are used to form a dataset. The relationship between processing time and tea stacking temperature in the yellowing stage and the initial moisture content of the tea is constructed, resulting in t=g(W,T); where W is the initial moisture content of the tea, T is the tea stacking temperature in the yellowing stage, t is the processing time in the yellowing stage, and g() is the functional relationship corresponding to the relationship. Using the shortest processing time t in the yellowing stage as the screening condition, the initial moisture content of the tea corresponding to the starting time of yellowing under this screening condition is determined.

8. The method for monitoring and feedback on quality loss during the processing of Huangda tea according to claim 1, characterized in that, The method for learning the changes in tea moisture content caused by the transfer process between the withering and yellowing stages using a deep learning model is as follows: Collect the tea's moisture content W1 at the end time of the fixation stage and the initial moisture content W2 at the start time of the yellowing stage. By coupling the tea fixation moisture W1 and the initial tea moisture W2, the transfer loss value of the fixation stage and the yellowing stage is determined. The transfer loss value is used as the compensation value to determine the tea moisture at the end of fixation. The initial tea moisture corresponding to the selected yellowing start time point is combined with the compensation value to reversely determine the tea fixation moisture at the end of fixation.

9. A system for monitoring and feedback of quality loss during the processing of Huangda tea according to any one of claims 1-8, characterized in that, include: The tea index monitoring module (1) is used to monitor the moisture, tea polyphenols and amino acid indicators of tea in real time during the withering, fixation, yellowing and baking processes of Huangda tea. The correlation model construction module (2) is used to construct a correlation model between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change. The yellowing period control module (3) is used to converge the correlation model based on the change in tea polyphenols / amino acids / moisture required at the end of the yellowing process, and to determine the initial moisture content of the tea corresponding to the yellowing start time by using the shortest yellowing process processing time as the screening condition. The transfer loss calculation module (4) is used to collect the changes in tea moisture index caused by the transfer process in the fixing and yellowing stages, so as to determine the transfer loss value in the fixing and yellowing stages. The reverse deduction module for fixation moisture (5) reverse deduces the fixation moisture of tea at the end of fixation based on the initial moisture of tea corresponding to the start time of yellowing, as well as the transfer loss value of fixation and yellowing stages.

10. The system according to claim 9, characterized in that, Based on the reverse deduction module (5) for the deduction of the deduction of the deduction of the tea at the end of the deduction process, the real-time tea moisture of the deduction process is monitored in real time by the tea index monitoring module (1). When the real-time tea moisture is the same as the tea deduction moisture, the deduction process is terminated. Based on the yellowing period control module (3), the initial moisture content of the tea leaves corresponding to the yellowing start time is determined, and the processing time of the yellowing section is obtained by converging the correlation model. Based on the correlation model [P, A] = f(W, T, t) between tea polyphenols / amino acids and processing time / tea stacking temperature / moisture change, the tea stacking temperature of the yellowing stage is obtained based on the correlation model, where the initial moisture W, tea polyphenol content P, amino acid content A, and processing time t of the yellowing stage are known. When the real-time temperature of the tea leaves is the same as the tea accumulation temperature in the yellowing stage obtained based on the correlation model, the yellowing stage is terminated.

Images