Preparation method and device of fermented agilawood tea

Abstract

The invention relates to the technical field of tea leaf processing, and discloses a preparation method and device of fermented agilawood tea, and the method comprises the following steps: putting agilawood leaves and tea leaves into a fermentation stirring tank, collecting total mass data, temperature data and humidity data, and determining a time interval of first intervention during initial fermentation; after each first intervention, a plurality of pH values in the fermentation stirring tank are collected, and whether the fermentation stage is changed or not is judged; when the transformation is judged, initial fermentation characteristics are collected to be compared with a historical fermentation data set, and when similarities are all smaller than a similarity threshold value, the time interval of second intervention and the number of second intervention times in low-temperature delayed fermentation are determined based on a convolutional neural network model; collecting carbon dioxide concentration data in the fermentation stirring tank after the last second intervention to judge whether the second intervention is prolonged or not. By introducing the data-driven control method, the defect that the traditional fermentation process depends on artificial experience is overcome, and the accuracy and consistency of the agilawood tea fermentation process are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tea processing, and in particular to a method and device for preparing fermented agarwood tea. Background Art

[0002] Agarwood tea, a new health drink, is gaining increasing attention in the tea market for its unique flavor and aroma. The fermentation process of combining agarwood leaves with tea leaves involves not only changes in the tea leaves themselves but also the penetration and integration of agarwood components, making the fermentation process even more complex. Therefore, precisely controlling the fermentation process to ensure optimal flavor and aroma has become a key issue in current fermented agarwood tea preparation technology.

[0003] Currently, traditional fermentation techniques rely on manual control of fermentation time, temperature, and humidity. However, due to variability in environmental factors and fluctuations in microbial activity, ensuring consistent fermentation results is often difficult, leading to varying tea quality. While timely intervention (such as stirring and ventilation) is necessary to adapt to tea changes during the fermentation process, traditional methods often rely on empirical experience and lack the ability to adjust based on real-time data during the fermentation process. This can lead to missed optimal intervention times, impacting the tea's final flavor.

[0004] Therefore, it is necessary to design a preparation method and device for fermented agarwood tea to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes a preparation method and device for fermented agarwood tea, aiming to solve the problems of inaccurate control of the fermentation process of fermented agarwood tea and lack of feedback regulation mechanism.

[0006] In one aspect, the present invention provides a method for preparing fermented agarwood tea, comprising: placing withered and rolled agarwood leaves and tea leaves in a ratio of 1:4 into a fermentation stirring tank, collecting total mass data of the fermented material in the fermentation stirring tank, and collecting temperature data and humidity data in the fermentation stirring tank, and determining a time interval for a first intervention during initial fermentation based on the total mass data, temperature data, and humidity data, wherein the first intervention includes stirring and passing 25°C air; After each of the first interventions, collecting a plurality of pH values in the fermentation stirring tank to construct a pH value data set, and determining whether to change the fermentation stage according to the pH value data set; When it is determined that the fermentation stage is to be changed, initial fermentation characteristics are collected, and the initial fermentation characteristics are compared with a historical fermentation data set. When the similarity between the historical fermentation data set and the initial fermentation characteristics is less than a similarity threshold, a time interval and a number of second interventions in the low-temperature delayed fermentation are determined based on a convolutional neural network model, wherein the initial fermentation characteristics include the initial fermentation time, the initial fermentation temperature change rate, and the initial fermentation carbon dioxide concentration change rate. The historical fermentation data set includes historical fermentation characteristics, a historical second intervention time interval, and a historical second intervention number, and each of the historical fermentation characteristics corresponds to a historical second intervention time interval and a historical second intervention number. When determining the time interval and number of second interventions in low-temperature delayed fermentation based on the convolutional neural network model, the carbon dioxide concentration data of the preset time period in the fermentation stirring tank after the last second intervention is collected to establish a concentration data set, and whether to extend the second intervention is determined based on the concentration data set, and the final number of second interventions is determined, and the time interval between the initial fermentation characteristics and the second intervention and the final number of second interventions are stored.

[0007] Furthermore, the collecting of total mass data of the fermented material in the fermentation stirring tank, and the collecting of temperature data and humidity data in the fermentation stirring tank, and determining the time interval of the first intervention during the initial fermentation according to the total mass data, temperature data, and humidity data, includes:

[0008] Among them, T1 represents the time interval of the first intervention during initial fermentation, G represents total mass data, W represents temperature data, S represents humidity data, Gz represents standard mass data, Wz represents standard temperature data, Sz represents standard humidity data, T0 represents the interval reference duration, and the value range of T0 is [0.5h-0.75h].

[0009] Furthermore, when determining whether to change the fermentation stage according to the pH value data set, the method includes: When all pH values in the pH value data set are less than or equal to 5.5, it is determined that the fermentation stage will not be changed; When there are pH values greater than 5.5 and there are pH values less than or equal to 5.5 in the pH value data set, determining a mean pH value of the pH value data set, and listing data in the pH value data set with a pH value greater than the mean pH value in a first set, and listing data in the pH value data set with a pH value less than the mean pH value in a second set, obtaining a comprehensive pH value based on the first set, the second set, and the mean pH value, and determining whether to change the fermentation stage based on the comprehensive pH value; When all pH values in the pH value data set are greater than 5.5, it is determined that the fermentation stage has been changed.

[0010] Furthermore, obtaining a comprehensive pH value based on the first set, the second set, and the average pH value, and determining whether to change the fermentation stage based on the comprehensive pH value includes:

[0011] Where E represents the comprehensive pH value, E0 represents the mean pH value, N represents the number of pH values in the first set, M represents the number of pH values in the second set, Ei represents the i-th pH value in the first set, and Ej represents the j-th pH value in the second set; Comparing the pH comprehensive value with the pH comprehensive threshold, and determining whether to change the fermentation stage according to the comparison result; When the pH comprehensive value is greater than the pH comprehensive threshold, it is determined that the fermentation stage is changed; when the pH comprehensive value is less than or equal to the pH comprehensive threshold, it is determined that the fermentation stage is not changed.

[0012] Furthermore, when comparing the initial fermentation characteristics with the historical fermentation data set, it includes: When data having a similarity with the initial fermentation characteristics greater than or equal to a similarity threshold exists in the historical fermentation dataset, determining a time interval and a number of second interventions in the low-temperature delayed fermentation according to the historical fermentation dataset; When the similarity between the historical fermentation data set and the initial fermentation characteristics is less than a similarity threshold, the time interval and the number of second interventions in the low-temperature delayed fermentation are determined based on a convolutional neural network model.

[0013] Furthermore, determining the time interval and number of second interventions in the low-temperature delayed fermentation according to the historical fermentation data set includes: establishing data whose similarity between the historical fermentation data set and the initial fermentation characteristics is greater than or equal to a similarity threshold as a similarity set, and determining the time interval and number of second interventions in the low-temperature delayed fermentation according to the similarity set; When the data in the same type set is unique, the historical second intervention time interval and the historical second intervention number corresponding to the historical fermentation feature are used as the second intervention time interval and the second intervention number; When the data in the same type set is not unique, the maximum historical second intervention time interval and the maximum historical second intervention number in the same type set are selected as the second intervention time interval and the second intervention number.

[0014] Furthermore, when determining the time interval and the number of second interventions in low-temperature delayed fermentation based on the convolutional neural network model, the convolutional neural network model is obtained by: The historical fermentation data set is sampled according to a ratio of 4:1 to obtain a training subset and a test subset; Iteratively training the neural network model according to the training subset, evaluating the iteratively trained neural network model according to the test subset, and determining whether to stop the iterative training according to the evaluation value; If the evaluation value of the neural network model after the current iterative training is less than the evaluation value of the neural network model after the previous iterative training, the amplitude of the change of the neural network model in the gradient direction is reduced, and the iterative training is continued until the preset number of iterations is reached; if the evaluation value of the neural network model after the current iterative training is greater than or equal to the evaluation value of the neural network model after the previous iterative training, the iterative training is stopped to obtain the convolutional neural network model.

[0015] Furthermore, determining whether to extend the second intervention time according to the concentration data set includes: Arrange the concentration data sets according to the acquisition time, and obtain the concentration data change rates in sequence; When the concentration data change rate decreases successively and finally stabilizes to 0, it is determined that the second intervention will not be extended; otherwise, it is determined that the second intervention will be extended, and the number of interventions is increased according to the concentration data change rate to determine the final number of the second interventions.

[0016] Furthermore, determining to increase the number of interventions according to the concentration data change rate and determining the final number of the second interventions includes: Obtaining an average concentration change rate based on all concentration data change rates, comparing the average concentration change rate with a first preset concentration change rate and a second preset concentration change rate, respectively, and determining the increase in the number of interventions based on the comparison results, the first preset concentration change rate being less than the second preset concentration change rate; When the average concentration change rate is less than or equal to the first preset concentration change rate, the number of interventions is determined to be increased as the first preset number; when the average concentration change rate is greater than the first preset concentration change rate and less than or equal to the second preset concentration change rate, the number of interventions is determined to be increased as the second preset number; when the average concentration change rate is greater than the second preset concentration change rate, the number of interventions is determined to be increased as the third preset number; the first preset number is less than the second preset number, and the second preset number is less than the third preset number.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: by introducing a data-driven intelligent control method, the drawbacks of relying on manual experience in the traditional fermentation process are solved, and the accuracy and consistency of the agarwood tea fermentation process are improved. By collecting the total mass, temperature, humidity and other parameters in the fermentation stirring tank, the time interval of the first intervention is dynamically determined based on real-time data, ensuring the timeliness of intervention measures such as stirring and ventilation, and avoiding missing the fermentation opportunity. By constructing and analyzing the pH value data set, the transition of the fermentation stage is judged, and the fermentation process is further optimized. When the fermentation stage changes, the convolutional neural network model is used to predict the second intervention time and number in low-temperature delayed fermentation based on historical fermentation data and initial fermentation characteristics, thereby effectively controlling the progress of the fermentation process. By real-time monitoring of changes in carbon dioxide concentration, it is dynamically determined whether the intervention needs to be extended to ensure complete fermentation, thereby ensuring the fusion of the flavor and aroma of the agarwood tea.

[0018] On the other hand, the present application also provides a device for preparing fermented agarwood tea, which is used to apply the above-mentioned method for preparing fermented agarwood tea, comprising: A fermentation stirring tank, including a tank body, a stirring motor and a stirring rod; a temperature-controlled gas reservoir, connected to the tank body, and used to fill the fermentation stirring tank with gas; A control module connected to the stirred fermentation tank and the temperature-controlled gas reservoir, the control module including a collection unit, a judgment unit, a processing unit, and an adjustment unit; The collecting unit is configured to collect total mass data of the fermented material in the fermentation stirring tank, and collect temperature data and humidity data in the fermentation stirring tank, and determine a time interval of a first intervention during initial fermentation based on the total mass data, temperature data, and humidity data, wherein the first intervention includes stirring and passing air at 25° C.; The judgment unit is configured to collect a plurality of pH values in the fermentation stirring tank after each first intervention, construct a pH value data set, and judge whether to change the fermentation stage according to the pH value data set; The processing unit is configured to, when determining to transition to the fermentation stage, collect initial fermentation characteristics, compare the initial fermentation characteristics with a historical fermentation dataset, and determine, based on a convolutional neural network model, a time interval and a number of second interventions for the low-temperature delayed fermentation when a similarity between the historical fermentation dataset and the initial fermentation characteristics is less than a similarity threshold, wherein the initial fermentation characteristics include an initial fermentation duration, a rate of change of the initial fermentation temperature, and a rate of change of the initial fermentation carbon dioxide concentration; the historical fermentation dataset includes historical fermentation characteristics, a historical time interval and a historical number of second interventions, and each of the historical fermentation characteristics corresponds to a historical time interval and a historical number of second interventions; The adjustment unit is configured to, when determining the time interval and the number of second interventions in low-temperature delayed fermentation according to the convolutional neural network model, collect carbon dioxide concentration data in the fermentation stirring tank for a preset period of time after the last second intervention to establish a concentration data set, determine whether to extend the second intervention according to the concentration data set, determine the final number of second interventions, and store the time interval between the initial fermentation characteristics and the second intervention and the final number of second interventions.

[0019] It is understandable that the above-mentioned preparation method and device of fermented agarwood tea have the same beneficial effects and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings: Figure 1 A flow chart of a method for preparing fermented agarwood tea provided in an embodiment of the present invention; Figure 2 A schematic structural diagram of a device for preparing fermented agarwood tea provided in an embodiment of the present invention; Figure 3 This is a functional block diagram of the control module in the device for preparing fermented agarwood tea provided by an embodiment of the present invention.

[0021] Among them, 110 is the tank body; 120 is the stirring motor; 130 is the stirring rod; 210 is the temperature-controlled gas storage device; 310 is the control module; and 410 is the pressure relief valve. DETAILED DESCRIPTION

[0022] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, unless there is a conflict, the embodiments of the present disclosure and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0023] In some embodiments of the present application, see Figure 1 As shown, a method for preparing fermented agarwood tea comprises: S100: Withered and rolled agarwood leaves and tea leaves are placed in a fermentation stirring tank in a ratio of 1:4, the total mass data of the fermented material in the fermentation stirring tank is collected, and the temperature data and humidity data in the fermentation stirring tank are collected. The time interval of the first intervention during the initial fermentation is determined based on the total mass data, temperature data, and humidity data. The first intervention includes stirring and passing 25°C air.

[0024] S200: After each first intervention, a plurality of pH values in the fermentation stirring tank are collected to construct a pH value data set, and whether to change the fermentation stage is determined according to the pH value data set.

[0025] S300: When determining to transition to a fermentation stage, collect initial fermentation characteristics, and compare the initial fermentation characteristics with the historical fermentation data set. When the similarity between the historical fermentation data set and the initial fermentation characteristics is less than the similarity threshold, determine the time interval and number of second interventions in low-temperature delayed fermentation based on the convolutional neural network model. The initial fermentation characteristics include the initial fermentation time, the initial fermentation temperature change rate, and the initial fermentation carbon dioxide concentration change rate. The historical fermentation data set includes historical fermentation characteristics, the time interval of the historical second intervention, and the number of historical second interventions. Each historical fermentation characteristic corresponds to a historical second intervention time interval and a historical second intervention number.

[0026] S400: When determining the time interval and number of second interventions in low-temperature delayed fermentation based on the convolutional neural network model, collect the carbon dioxide concentration data of the preset time period in the fermentation stirring tank after the last second intervention to establish a concentration data set, determine whether to extend the second intervention based on the concentration data set, determine the final number of second interventions, and store the initial fermentation characteristics, the time interval of the second intervention, and the final number of second interventions.

[0027] Specifically, in S100, withered and rolled agarwood leaves and tea leaves are placed in a fermentation mixing tank at a ratio of 1:4. A mixed enzyme solution containing acetic acid bacteria, penicillium, lactic acid bacteria, and active yeast is placed in the fermentation mixing tank. By collecting total mass, temperature, and humidity data within the fermentation mixing tank, preliminary environmental information about the fermentation process is obtained. This data is used to calculate the time interval for the first intervention during initial fermentation. Temperature and humidity are important factors affecting fermentation rate and microbial activity, particularly the metabolic rate of microorganisms and the changes in fermentation progress at different temperatures. Based on this environmental data, a precise initial intervention interval can be set. The first intervention includes stirring and aeration with 25°C air, which helps evenly distribute the temperature and humidity in the fermentation environment, promotes microbial activity, and avoids localized overheating or overhumidification, thereby improving fermentation efficiency. In S200, after each first intervention, multiple pH values are collected within the fermentation mixing tank. The pH value reflects changes in acidic substances during fermentation and effectively indicates the progress of the fermentation stage. By collecting these pH value data, a pH value data set is established, and the data set is analyzed to determine whether the fermentation process has entered a new stage. For example, the range of pH value changes can help determine the degree of microbial activity, and then decide whether to conduct phased intervention. In S300, it is determined that the fermentation stage has changed, and the initial fermentation characteristics are collected, including the initial fermentation time, the initial fermentation temperature change rate, and the initial fermentation carbon dioxide concentration change rate. These characteristics can reflect the characteristics of the fermentation process. These characteristics are compared with the historical fermentation data set. If the similarity with the fermentation characteristics in the historical fermentation data set is less than the set threshold, it indicates that there are differences between the fermentation process and the historical data. Based on the new data features, the convolutional neural network model is used to learn the initial fermentation characteristics and historical fermentation data. The convolutional neural network can automatically extract the laws and patterns in the fermentation process through the learning of multiple layers of neurons, and then accurately predict the time interval and number of interventions for the second intervention in the low-temperature delayed fermentation stage. At the beginning of fermentation, the fermentation temperature is 25°C and the humidity is 50-75%. During stirring and aeration, the gas temperature is maintained at 25°C. During low-temperature extended fermentation, the aeration temperature is 20°C, maintaining the tank temperature at around 20°C. The introduction of a convolutional neural network model can handle complex nonlinear relationships, ensuring that every intervention adjustment during the fermentation process maximizes the flavor and quality of the tea. During the low-temperature extended fermentation phase, the S400 continues to collect carbon dioxide concentration data within the fermentation stirring tank. This data effectively reflects microbial metabolic activity and fermentation progress. Based on changes in carbon dioxide concentration, it is determined whether the second intervention time needs to be extended. This concentration dataset provides important information on whether fermentation is progressing adequately, further determining whether the intervention frequency needs to be adjusted. Concentration datasets are established and analyzed to determine the fermentation intensity during the fermentation process.If the concentration reaches a plateau, fermentation is nearing completion. If it fluctuates, the second intervention is extended to ensure full fermentation. This dynamic adjustment ensures optimal fermentation for each batch of tea. After fermentation is complete, the fermented agarwood tea is baked to finalize its shape.

[0028] It is understandable that precise control of the fermentation process of agarwood tea is achieved through real-time data monitoring and optimization of the convolutional neural network model. The time interval for initial intervention can be determined based on real-time data such as temperature, humidity, and total mass, ensuring the timeliness and accuracy of each intervention, avoiding the limitations of traditional experience-based methods. pH monitoring and data set analysis provide a basis for judging the transition between fermentation stages, avoiding stage-by-stage intervention errors caused by lack of experience. The introduction of a convolutional neural network model to adjust the low-temperature delayed fermentation automatically learns and optimizes based on historical data and initial fermentation characteristics, so that every detail of the fermentation process is adjusted and optimized. Real-time monitoring of carbon dioxide concentration further increases sensitivity to the fermentation process, ensuring the quality and flavor of the final product. The production efficiency and product consistency of fermented agarwood tea are improved.

[0029] In some embodiments of the present application, collecting total mass data of the fermented material in the fermentation stirring tank, and collecting temperature data and humidity data in the fermentation stirring tank, and determining the time interval of the first intervention during the initial fermentation based on the total mass data, temperature data, and humidity data, includes:

[0030] Among them, T1 represents the time interval of the first intervention during initial fermentation, G represents total mass data, W represents temperature data, S represents humidity data, Gz represents standard mass data, Wz represents standard temperature data, Sz represents standard humidity data, T0 represents the interval reference duration, and the value range of T0 is [0.5h-0.75h].

[0031] Comparing key fermentation parameters (such as total mass, temperature, and humidity) with standard reference data to calculate the time interval for the first intervention breaks through the limitations of traditional fermentation processes that rely on manual experience and fixed time rules. Dynamic adjustment methods based on real-time data enable more accurate control of every step of the fermentation process, ensuring that each intervention is optimized for the current fermentation environment, thereby improving the consistency of tea quality and flavor stability. Standard mass data typically ranges from 5-10kg, standard temperature data is typically 25°C, and standard humidity data is typically 50%.

[0032] In some embodiments of the present application, when determining whether to change the fermentation stage based on a pH value dataset, the method includes: when all pH values in the pH value dataset are less than or equal to 5.5, determining not to change the fermentation stage. When there are pH values greater than 5.5 and there are pH values less than or equal to 5.5 in the pH value dataset, determining the pH mean of the pH value dataset, and listing the data in the pH value dataset that are greater than the pH mean in the first set, and listing the data in the pH value dataset that are less than the pH mean in the second set, obtaining a pH comprehensive value based on the first set, the second set, and the pH mean, and determining whether to change the fermentation stage based on the pH comprehensive value. When all pH values in the pH value dataset are greater than 5.5, determining to change the fermentation stage.

[0033] In some embodiments of the present application, obtaining a comprehensive pH value based on the first set, the second set, and the average pH value, and determining whether to change the fermentation stage based on the comprehensive pH value includes:

[0034] Wherein, E represents the comprehensive pH value, E0 represents the mean pH value, N represents the number of pH values in the first set, M represents the number of pH values in the second set, Ei represents the i-th pH value in the first set, and Ej represents the j-th pH value in the second set.

[0035] The pH value is compared with the pH threshold, and the fermentation stage is determined based on the comparison result. If the pH value is greater than the threshold, the fermentation stage is determined to be changed. If the pH value is less than or equal to the threshold, the fermentation stage is determined not to be changed.

[0036] In some embodiments of the present application, comparing the initial fermentation characteristics with a historical fermentation dataset includes: when data in the historical fermentation dataset exists whose similarity to the initial fermentation characteristics is greater than or equal to a similarity threshold, determining the time interval and number of second interventions in the low-temperature delayed fermentation based on the historical fermentation dataset. When the similarity between the historical fermentation dataset and the initial fermentation characteristics is less than the similarity threshold, determining the time interval and number of second interventions in the low-temperature delayed fermentation based on a convolutional neural network model.

[0037] In some embodiments of the present application, when determining the time interval and the number of second interventions in low-temperature delayed fermentation based on a historical fermentation data set, it includes: establishing data whose similarity between the historical fermentation data set and the initial fermentation characteristics is greater than or equal to a similarity threshold as a similarity set, and determining the time interval and the number of second interventions in low-temperature delayed fermentation based on the similarity set.

[0038] Specifically, when the data in a similar set is unique, the time interval and number of historical second interventions corresponding to the historical fermentation feature are used as the time interval and number of second interventions. When the data in a similar set is not unique, the maximum time interval and number of historical second interventions in the similar set are selected as the time interval and number of second interventions.

[0039] Specifically, pH is an indicator of tea acidity during fermentation. As fermentation progresses, changes in pH reflect the accumulation of microbial metabolites. In the early stages of fermentation, the pH is low, ranging from 4.5 to 5.5, or even lower. During this stage, the original acidic substances (such as tea polyphenols and amino acids) and microbial metabolites (such as lactic acid and acetic acid) in the tea leaves and agarwood begin to manifest, causing the pH to drop. As fermentation progresses, the pH gradually stabilizes and then rises slightly. During this period, organic acids (such as lactic acid and acetic acid) in the tea leaves and agarwood begin to be converted by microbial metabolism, particularly into volatile organic compounds. The pH during this stage is typically between 5.5 and 6.0. If all pH values in the pH dataset are less than or equal to 5.5, fermentation is still in its early stages and has not yet entered a significant fermentation acceleration phase. Therefore, the fermentation phase is considered to be unchanged. If the pH dataset contains both pH values greater than 5.5 and pH values less than or equal to 5.5, the mean (E0) of the pH dataset is calculated. Based on whether the pH value is greater than the mean, the data is divided into a first set (pH values greater than the mean) and a second set (pH values less than the mean). A comprehensive pH value (E) is calculated and compared with a pre-set pH threshold. If E is greater than the threshold, fermentation has entered the next stage and a transition to the next stage is considered. If E is less than or equal to the threshold, fermentation has not yet entered the next stage and a transition to the next stage is considered. A pH collector is used to monitor pH in real time. The averaged comprehensive value allows for dynamic assessment of the fermentation stage, making stage determination more accurate and avoiding errors due to human experience. To determine if a transition has occurred, initial fermentation characteristics are collected, such as initial fermentation duration, initial fermentation temperature change rate, and initial fermentation carbon dioxide concentration change rate, reflecting the initial fermentation environment and microbial activity. A historical fermentation dataset contains characteristic data and intervention records from previous fermentations. Comparing the current fermentation characteristics with the historical dataset determines whether to adopt historical intervention experience or perform new optimizations. Similarity calculation methods (such as Euclidean distance and cosine similarity) are used to compare the initial fermentation characteristics with the historical dataset. If the similarity is greater than or equal to the preset similarity threshold, the time interval and frequency of the second intervention are determined based on the historical fermentation data. The second intervention includes stirring and the introduction of 20°C air. If the data in the same set is unique, the historical second intervention time interval and the historical second intervention frequency corresponding to the historical fermentation feature are directly used as the time and frequency of the current intervention. If the data in the same set is not unique, the maximum historical second intervention time interval and the maximum historical second intervention frequency in the set are selected to optimize the intervention strategy. If the historical data does not match the current fermentation feature, the optimal second intervention plan is predicted based on the current initial fermentation feature.A convolutional neural network was used to analyze the relationship between historical fermentation data and current fermentation characteristics, thereby predicting the time interval and number of interventions during low-temperature delayed fermentation. The convolutional neural network model automatically learns complex nonlinear relationships from the data. When the historical fermentation data showed a low degree of similarity to the initial characteristics, the convolutional neural network model inferred the time interval and number of the second intervention.

[0040] It's clear that the fermentation process of agarwood tea has been optimized through the application of pH analysis, historical fermentation data comparison, and convolutional neural networks. Through meticulous analysis and calculation of pH datasets, it's possible to dynamically determine whether fermentation has entered a new stage, avoiding the stage-by-stage misjudgments often associated with traditional methods that rely on empirical experience. The comparison of historical data and the introduction of a convolutional neural network model further refine the timing and frequency of secondary interventions during the fermentation process, enabling personalized adjustments based on the specific fermentation conditions. Compared to traditional empirical methods, this data-driven intelligent decision-making model improves the controllability and consistency of the fermentation process.

[0041] In some embodiments of the present application, when determining the time interval and the number of second interventions in low-temperature delayed fermentation based on a convolutional neural network model, the convolutional neural network model is obtained by: The historical fermentation dataset was sampled at a ratio of 4:1 to obtain training and test subsets.

[0042] The neural network model is iteratively trained according to the training subset, the neural network model after iterative training is evaluated according to the test subset, and whether to stop the iterative training is determined according to the evaluation value.

[0043] If the evaluation value of the neural network model after the current iterative training is less than the evaluation value of the neural network model after the previous iterative training, the amplitude of the change in the gradient direction of the neural network model is reduced, and iterative training is continued until the preset number of iterations is reached. If the evaluation value of the neural network model after the current iterative training is greater than or equal to the evaluation value of the neural network model after the previous iterative training, the iterative training is stopped to obtain a convolutional neural network model.

[0044] Specifically, the training data is derived from various fermentation characteristics accumulated during historical fermentation processes, including initial fermentation duration, temperature fluctuations, and CO2 concentration changes. The historical fermentation dataset is partitioned into a training subset and a test subset in a 4:1 ratio. The training subset is used to train the neural network, while the test subset is used to evaluate the performance of the neural network model. This ensures data diversity during neural network training while also assessing the network's generalization ability through the test set to avoid overfitting. The convolutional neural network model is iteratively trained using the training subset. In each iteration, the neural network adjusts model parameters based on the input data (such as fermentation characteristics) and output labels (such as the time interval and number of second interventions) to minimize the loss function (error or mean loss). After each training iteration, the test subset is used to evaluate the performance of the current model. The model's error (such as mean squared error or accuracy) on the test set is calculated to generate an evaluation value, which is then compared with the evaluation value from the previous iteration. If the evaluation value from the current iteration is lower than the previous one, it indicates that the model has not converged and requires further training. At this point, the gradient update amplitude (that is, the learning rate) is reduced, and iterative training continues to fine-tune the model parameters and avoid overtraining or oscillation. If the evaluation value of the current iteration is greater than or equal to the previous evaluation value, it indicates that the model performance has reached or improved, and training is terminated, ultimately resulting in a trained convolutional neural network model. By gradually reducing the training step size, overfitting or overly rapid convergence that may occur during training is avoided. The gradient descent algorithm is an optimization method for convolutional neural network training, and the gradient update determines the amplitude of the model parameter adjustments. To avoid excessive or insufficient steps during model training, dynamic adjustment of the learning rate (for example, by reducing the learning rate to slow down the updates) can help more stably converge to a good optimal solution. The maximum number of training iterations is preset, typically determined through methods such as cross-validation. Setting an appropriate number of iterations helps prevent overtraining, and early stopping (stopping training after a few consecutive iterations without significant improvement in the model evaluation value) can further optimize the training process. The trained and optimized convolutional neural network model can ultimately predict the time interval and number of secondary interventions during low-temperature delayed fermentation based on input initial fermentation characteristics (such as initial fermentation duration, temperature changes, and CO2 concentration changes). Not only can it derive a reasonable intervention plan based on historical data, but it can also gradually improve its predictive capabilities as more data is input.

[0045] It's clear that through iterative training on fermentation data, the convolutional neural network can continuously optimize the prediction model, improve the accuracy of fermentation interventions, and avoid biases caused by human judgment. The model can adaptively adjust based on real-time fermentation data, flexibly adjusting the timing and frequency of interventions to suit the varying fermentation conditions of each batch, thereby ensuring consistent flavor and quality across each batch of agarwood tea. By adjusting the learning rate and the preset number of iterations, combined with cross-validation, the model effectively prevents overfitting and improves its generalization, enabling it to adapt to diverse fermentation environments and conditions.

[0046] In some embodiments of the present application, determining whether to extend the second intervention according to the concentration data set includes: arranging the concentration data set according to acquisition time, and sequentially obtaining concentration data change rates.

[0047] Specifically, when the concentration data change rate decreases successively and finally stabilizes to 0, it is determined that the second intervention will not be extended. Otherwise, it is determined that the second intervention will be extended, and the number of interventions is increased according to the concentration data change rate to determine the final number of the second interventions.

[0048] In some embodiments of the present application, the number of interventions to be increased is determined based on the concentration data change rate, and when determining the second final number of interventions, it includes: obtaining an average concentration change rate based on all concentration data change rates, comparing the average concentration change rate with a pre-set first preset concentration change rate and a second preset concentration change rate, respectively, and determining the number of interventions to be increased based on the comparison results, the first preset concentration change rate is less than the second preset concentration change rate.

[0049] Specifically, when the average concentration change rate is less than or equal to a first preset concentration change rate, the number of interventions is determined to be increased to the first preset number. When the average concentration change rate is greater than the first preset concentration change rate and less than or equal to a second preset concentration change rate, the number of interventions is determined to be increased to the second preset number. When the average concentration change rate is greater than the second preset concentration change rate, the number of interventions is determined to be increased to the third preset number. The first preset number is less than the second preset number, and the second preset number is less than the third preset number.

[0050] As can be understood, the rate of change in concentration provides real-time insight into the fermentation process, determining whether interventions need to be extended or increased, ensuring each fermentation stage is fully completed. By tracking the concentration rate's changing trends, the system can monitor the fermentation progress and automatically adjust the number of interventions based on the data, minimizing instances of insufficient intervention. This data-driven approach reduces the impact of human factors on the fermentation process, ensuring more consistent fermentation conditions for each batch of agarwood tea and ensuring consistent quality and flavor.

[0051] In the above embodiment, by introducing a data-driven intelligent control method, the drawbacks of relying on manual experience in the traditional fermentation process are solved, and the accuracy and consistency of the agarwood tea fermentation process are improved. By collecting the total mass, temperature, humidity and other parameters in the fermentation stirring tank, the time interval of the first intervention is dynamically determined based on real-time data, ensuring the timeliness of intervention measures such as stirring and ventilation, and avoiding missing the fermentation opportunity. By constructing and analyzing the pH value data set, the transition of the fermentation stage is judged, and the fermentation process is further optimized. When the fermentation stage changes, the convolutional neural network model is used to predict the second intervention time and number in low-temperature delayed fermentation based on historical fermentation data and initial fermentation characteristics, thereby effectively controlling the progress of the fermentation process. By real-time monitoring of changes in carbon dioxide concentration, it is dynamically determined whether the intervention needs to be extended to ensure complete fermentation, thereby ensuring the fusion of the flavor and aroma of the agarwood tea.

[0052] In another preferred embodiment based on the above embodiment, refer to Figure 2-3 As shown, this embodiment provides a device for preparing fermented agarwood tea, which is used to apply the above-mentioned method for preparing fermented agarwood tea, including: The fermentation stirring tank includes a tank body 110 , a stirring motor 120 and a stirring rod 130 .

[0053] The temperature-controlled gas storage device 210 is in communication with the tank body 110 and is used to charge the fermentation stirring tank with gas.

[0054] The control module 310 is connected to the stirring fermentation tank and the temperature-controlled gas storage 210. The control module includes a collection unit, a judgment unit, a processing unit and an adjustment unit.

[0055] The collection unit is configured to collect the total mass data of the fermented material in the fermentation stirring tank, and collect the temperature data and humidity data in the fermentation stirring tank, and determine the time interval of the first intervention during the initial fermentation based on the total mass data, temperature data and humidity data. The first intervention includes stirring and passing 25°C air.

[0056] The judgment unit is configured to collect a plurality of pH values in the fermentation stirring tank after each first intervention, construct a pH value data set, and judge whether to change the fermentation stage according to the pH value data set.

[0057] The processing unit is configured to collect initial fermentation characteristics when determining the transition to a fermentation stage, compare the initial fermentation characteristics with a historical fermentation data set, and when the similarity between the historical fermentation data set and the initial fermentation characteristics is less than a similarity threshold, determine the time interval and number of second interventions in low-temperature delayed fermentation based on a convolutional neural network model. The initial fermentation characteristics include the initial fermentation time, the initial fermentation temperature change rate, and the initial fermentation carbon dioxide concentration change rate. The historical fermentation data set includes historical fermentation characteristics, the time interval of the historical second intervention, and the number of historical second interventions, and each historical fermentation characteristic corresponds to a historical second intervention time interval and a historical second intervention number.

[0058] The adjustment unit is configured to, when determining the time interval and number of second interventions in low-temperature delayed fermentation according to a convolutional neural network model, collect carbon dioxide concentration data for a preset period of time in the fermentation stirring tank after the last second intervention to establish a concentration data set, determine whether to extend the second intervention according to the concentration data set, determine the final number of second interventions, and store the initial fermentation characteristics and the time interval of the second intervention and the final number of second interventions.

[0059] Specifically, a mass sensor can be used to collect total mass data. The mass sensor can be installed at the bottom of the tank. A temperature sensor, humidity sensor, pH sensor, and carbon dioxide concentration sensor are integrated into the tank to collect data from the fermentation mixing tank. A pressure relief valve 410 is also installed at the top of the tank to balance the pressure inside the tank when gas is added.

[0060] It is understandable that by introducing data-driven intelligent control methods, the drawbacks of relying on manual experience in the traditional fermentation process have been solved, and the accuracy and consistency of the agarwood tea fermentation process have been improved. By collecting parameters such as the total mass, temperature, and humidity in the fermentation stirring tank, the time interval for the first intervention is dynamically determined based on real-time data, ensuring the timeliness of intervention measures such as stirring and ventilation, and avoiding missing the fermentation opportunity. By constructing and analyzing the pH value data set, the transition of the fermentation stage is judged and the fermentation process is further optimized. When the fermentation stage changes, the convolutional neural network model is used to predict the timing and number of the second intervention in low-temperature delayed fermentation based on historical fermentation data and initial fermentation characteristics, thereby effectively controlling the progress of the fermentation process. By monitoring the changes in carbon dioxide concentration in real time, it is dynamically determined whether the intervention needs to be extended to ensure complete fermentation, thereby ensuring the fusion of the flavor and aroma of the agarwood tea.

[0061] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0062] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0063] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0064] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A method for preparing fermented agarwood tea, characterized in that: include: placing withered and rolled agarwood leaves and tea leaves in a ratio of 1:4 into a fermentation stirring tank, collecting total mass data of the fermented material in the fermentation stirring tank, and collecting temperature data and humidity data in the fermentation stirring tank, and determining a time interval for a first intervention during initial fermentation based on the total mass data, temperature data, and humidity data, wherein the first intervention includes stirring and passing 25°C air; After each of the first interventions, collecting a plurality of pH values in the fermentation stirring tank to construct a pH value data set, and determining whether to change the fermentation stage according to the pH value data set; When it is determined that the fermentation stage is to be changed, initial fermentation characteristics are collected, and the initial fermentation characteristics are compared with a historical fermentation data set. When the similarity between the historical fermentation data set and the initial fermentation characteristics is less than a similarity threshold, a time interval and a number of second interventions in the low-temperature delayed fermentation are determined based on a convolutional neural network model, wherein the initial fermentation characteristics include the initial fermentation time, the initial fermentation temperature change rate, and the initial fermentation carbon dioxide concentration change rate. The historical fermentation data set includes historical fermentation characteristics, a historical second intervention time interval, and a historical second intervention number, and each of the historical fermentation characteristics corresponds to a historical second intervention time interval and a historical second intervention number. When determining the time interval and number of second interventions in low-temperature delayed fermentation based on the convolutional neural network model, the carbon dioxide concentration data of the preset time period in the fermentation stirring tank after the last second intervention is collected to establish a concentration data set, and whether to extend the second intervention is determined based on the concentration data set, and the final number of second interventions is determined, and the time interval between the initial fermentation characteristics and the second intervention and the final number of second interventions are stored.

2. The method for preparing fermented agarwood tea according to claim 1, wherein: The collecting of the total mass data of the fermented material in the fermentation stirring tank, and the collecting of the temperature data and the humidity data in the fermentation stirring tank, and determining the time interval of the first intervention during the initial fermentation according to the total mass data, the temperature data, and the humidity data, includes: ； Among them, T1 represents the time interval of the first intervention during initial fermentation, G represents total mass data, W represents temperature data, S represents humidity data, Gz represents standard mass data, Wz represents standard temperature data, Sz represents standard humidity data, T0 represents the interval reference duration, and the value range of T0 is [0.5h-0.75h].

3. The method for preparing fermented agarwood tea according to claim 1, wherein: When determining whether to change the fermentation stage according to the pH value data set, the method includes: When all pH values in the pH value data set are less than or equal to 5.5, it is determined that the fermentation stage will not be changed; When there are pH values greater than 5.5 and there are pH values less than or equal to 5.5 in the pH value data set, determining a mean pH value of the pH value data set, and listing data in the pH value data set with a pH value greater than the mean pH value in a first set, and listing data in the pH value data set with a pH value less than the mean pH value in a second set, obtaining a comprehensive pH value based on the first set, the second set, and the mean pH value, and determining whether to change the fermentation stage based on the comprehensive pH value; When all pH values in the pH value data set are greater than 5.5, it is determined that the fermentation stage has been changed.

4. The method for preparing fermented agarwood tea according to claim 3, wherein: Obtaining a comprehensive pH value based on the first set, the second set, and the average pH value, and determining whether to change the fermentation stage based on the comprehensive pH value includes: ； Where E represents the comprehensive pH value, E0 represents the mean pH value, N represents the number of pH values in the first set, M represents the number of pH values in the second set, Ei represents the i-th pH value in the first set, and Ej represents the j-th pH value in the second set; Comparing the pH comprehensive value with the pH comprehensive threshold, and determining whether to change the fermentation stage according to the comparison result; When the pH comprehensive value is greater than the pH comprehensive threshold, it is determined that the fermentation stage is changed; when the pH comprehensive value is less than or equal to the pH comprehensive threshold, it is determined that the fermentation stage is not changed.

5. The method for preparing fermented agarwood tea according to claim 4, wherein: When comparing the initial fermentation characteristics with historical fermentation datasets, including: When data having a similarity with the initial fermentation characteristics greater than or equal to a similarity threshold exists in the historical fermentation dataset, determining a time interval and a number of second interventions in the low-temperature delayed fermentation according to the historical fermentation dataset; When the similarity between the historical fermentation data set and the initial fermentation characteristics is less than a similarity threshold, the time interval and the number of second interventions in the low-temperature delayed fermentation are determined based on a convolutional neural network model.

6. The method for preparing fermented agarwood tea according to claim 5, characterized in that: Determining the time interval and number of second interventions in low-temperature delayed fermentation according to the historical fermentation data set includes: establishing data whose similarity between the historical fermentation data set and the initial fermentation characteristics is greater than or equal to a similarity threshold as a similarity set, and determining the time interval and number of second interventions in the low-temperature delayed fermentation according to the similarity set; When the data in the same type set is unique, the historical second intervention time interval and the historical second intervention number corresponding to the historical fermentation feature are used as the second intervention time interval and the second intervention number; When the data in the same type set is not unique, the maximum historical second intervention time interval and the maximum historical second intervention number in the same type set are selected as the second intervention time interval and the second intervention number.

7. The method for preparing fermented agarwood tea according to claim 6, characterized in that: When determining the time interval and number of second interventions in low-temperature delayed fermentation based on a convolutional neural network model, the convolutional neural network model is obtained by: The historical fermentation data set is sampled according to a ratio of 4:1 to obtain a training subset and a test subset; Iteratively training the neural network model according to the training subset, evaluating the iteratively trained neural network model according to the test subset, and determining whether to stop the iterative training according to the evaluation value; If the evaluation value of the neural network model after the current iterative training is less than the evaluation value of the neural network model after the previous iterative training, the amplitude of the change of the neural network model in the gradient direction is reduced, and the iterative training is continued until the preset number of iterations is reached; If the evaluation value of the neural network model after the current iterative training is greater than or equal to the evaluation value of the neural network model after the previous iterative training, the iterative training is stopped to obtain the convolutional neural network model.

8. The method for preparing fermented agarwood tea according to claim 7, characterized in that: Determining whether to extend the second intervention according to the concentration dataset includes: Arrange the concentration data sets according to the acquisition time, and obtain the concentration data change rates in sequence; When the concentration data change rate decreases successively and finally stabilizes to 0, it is determined that the second intervention will not be extended; otherwise, it is determined that the second intervention will be extended, and the number of interventions is increased according to the concentration data change rate to determine the final number of the second interventions.

9. The method for preparing fermented agarwood tea according to claim 8, characterized in that: Determining an increase in the number of interventions according to the concentration data change rate, and determining a final number of the second interventions, includes: Obtaining an average concentration change rate based on all concentration data change rates, comparing the average concentration change rate with a first preset concentration change rate and a second preset concentration change rate, respectively, and determining the increase in the number of interventions based on the comparison results, the first preset concentration change rate being less than the second preset concentration change rate; When the average concentration change rate is less than or equal to the first preset concentration change rate, the number of interventions is determined to be increased as the first preset number; when the average concentration change rate is greater than the first preset concentration change rate and less than or equal to the second preset concentration change rate, the number of interventions is determined to be increased as the second preset number; when the average concentration change rate is greater than the second preset concentration change rate, the number of interventions is determined to be increased as the third preset number; the first preset number is less than the second preset number, and the second preset number is less than the third preset number.

10. A device for preparing fermented agarwood tea, used for applying the method for preparing fermented agarwood tea according to any one of claims 1 to 9, characterized in that: include: A fermentation stirring tank, including a tank body, a stirring motor and a stirring rod; a temperature-controlled gas reservoir, connected to the tank body, and used to fill the fermentation stirring tank with gas; A control module connected to the stirred fermentation tank and the temperature-controlled gas reservoir, the control module including a collection unit, a judgment unit, a processing unit, and an adjustment unit; The collecting unit is configured to collect total mass data of the fermented material in the fermentation stirring tank, and collect temperature data and humidity data in the fermentation stirring tank, and determine a time interval of a first intervention during initial fermentation based on the total mass data, temperature data, and humidity data, wherein the first intervention includes stirring and passing air at 25° C.; The judgment unit is configured to collect a plurality of pH values in the fermentation stirring tank after each first intervention, construct a pH value data set, and judge whether to change the fermentation stage according to the pH value data set; The processing unit is configured to, when determining to transition to the fermentation stage, collect initial fermentation characteristics, compare the initial fermentation characteristics with a historical fermentation dataset, and determine, based on a convolutional neural network model, a time interval and a number of second interventions for the low-temperature delayed fermentation when a similarity between the historical fermentation dataset and the initial fermentation characteristics is less than a similarity threshold, wherein the initial fermentation characteristics include an initial fermentation duration, a rate of change of the initial fermentation temperature, and a rate of change of the initial fermentation carbon dioxide concentration; the historical fermentation dataset includes historical fermentation characteristics, a historical time interval and a historical number of second interventions, and each of the historical fermentation characteristics corresponds to a historical time interval and a historical number of second interventions; The adjustment unit is configured to, when determining the time interval and the number of second interventions in low-temperature delayed fermentation according to the convolutional neural network model, collect carbon dioxide concentration data in the fermentation stirring tank for a preset period of time after the last second intervention to establish a concentration data set, determine whether to extend the second intervention according to the concentration data set, determine the final number of second interventions, and store the time interval between the initial fermentation characteristics and the second intervention and the final number of second interventions.

Images