Semi-refined tea grading method, media, and equipment based on charcoal roasting parameters

Abstract

This invention discloses a semi-refined tea grading method, medium, and equipment based on charcoal roasting parameters. The method collects charcoal roasting process parameters and basic tea parameters, extracts thermodynamic characteristics and Maillard reaction characteristic spectra, and constructs a variety adaptability model. It then performs multimodal fusion of thermodynamic characteristics, Maillard reaction characteristic spectra, and variety adaptability parameters to generate a charcoal roasting quality prediction vector. A grading decision tree model is used for preliminary grade determination, and secondary-grade teas are re-graded after cascaded roasting optimization. Finally, the grading results are linked to the charcoal roasting parameters and stored on a blockchain. This method achieves dynamic correlation analysis between charcoal roasting process parameters and tea quality, solving the problem of inaccurate evaluation caused by the separation between traditional grading methods and the charcoal roasting process. It significantly improves the scientific validity and reliability of the grading results and provides data support for process optimization.

Description

Technical Field

[0001] This invention relates to the field of tea processing technology, and in particular to a method, medium and equipment for grading semi-refined tea based on charcoal roasting parameters. Background Technology

[0002] Semi-refined tea refers to tea leaves that have undergone basic processing (such as withering, fixation, rolling, and drying) but have not yet completed refined processing (such as sorting, grading, blending, and roasting). Grading semi-refined tea can optimize quality consistency, facilitate subsequent refined processing, enhance the market value of the finished tea, and meet diverse consumer demands. However, because tea grading is often completed before the charcoal roasting process, it fails to fully utilize the technological data generated during subsequent processing. This fragmented evaluation method can easily lead to some raw materials with quality improvement potential being prematurely classified as low-grade. Furthermore, the lack of continuous monitoring and analysis of processing parameters makes it difficult to specifically optimize these parameters to improve product quality. In addition, this static grading method cannot provide effective feedback for subsequent process adjustments, limiting the optimization space of the entire production process. Establishing a dynamic link between the processing process and quality evaluation has become a key issue that needs to be addressed to improve tea grading levels. Summary of the Invention

[0003] In view of this, the purpose of this invention is to propose a semi-refined tea grading method, medium and equipment based on carbon roasting parameters, to solve the problem that the existing grading method is disconnected from the subsequent refining process, resulting in inaccurate quality evaluation and lack of basis for process optimization.

[0004] To achieve the aforementioned technical objectives, in a first aspect, this application provides a method for grading semi-refined tea based on charcoal roasting parameters, comprising:

[0005] Collect parameters of the charcoal roasting process and basic parameters of the tea leaves. The basic parameters of the tea leaves include variety information, moisture content information and initial processing grade.

[0006] Thermodynamic features were extracted from the carbon roasting process parameters, including the carbon roasting uniformity index and the duration of critical temperatures.

[0007] In addition, hyperspectral imaging information of tea leaves in the carbon roasting space is collected, Maillard reaction characteristic spectra are extracted within a preset band range, and the spectral parameters of tea leaves are calculated based on the Maillard reaction characteristic spectra.

[0008] A variety adaptability model is dynamically constructed based on tea spectral parameters and basic tea parameters to obtain variety adaptability parameters;

[0009] Thermodynamic characteristics, Maillard reaction characteristic spectrum and variety adaptability parameters are fused in a multimodal manner to generate a carbon roasting quality prediction vector. The multimodal fusion includes temperature-material conversion co-splicing and spatial non-uniformity compensation.

[0010] The carbon roasting quality prediction vector is input into the hierarchical decision tree model to obtain the preliminary grade judgment result, which includes secondary, superior and special grades. The hierarchical decision tree model is configured to trigger the cascade fire optimization process based on the clustering quality index.

[0011] For tea leaves whose initial grade is classified as secondary, cascade fire optimization is performed to obtain optimized tea leaves. The optimized tea leaves are then re-graded to obtain updated grade results. The optimization strategy of cascade fire optimization is configured to use a material transformation model to combine and adjust the fire duration and temperature.

[0012] Based on the preliminary grading results and the updated grading results, the final grading result of the current tea is generated, and the final grading result is stored in the blockchain in association with the charcoal roasting process parameters, generating a grading report that includes the grading distribution probability, the location of process defects, and energy consumption assessment.

[0013] In some embodiments, the carbon roasting process parameters include three-dimensional temperature field information of the roasting cage and the duration;

[0014] Feature extraction was performed on the carbon roasting process parameters to obtain thermodynamic characteristics, including the carbon roasting uniformity index and the duration of critical temperatures, including:

[0015] The carbon roasting uniformity index is generated by calculating the ratio of the standard deviation to the mean of the temperature values ​​at each temperature measurement point based on the three-dimensional temperature field information of the roasting cage.

[0016] Furthermore, a temperature-time curve is constructed based on the three-dimensional temperature field information and duration of the baking cage, and temperature plateau intervals with durations exceeding a preset time threshold are extracted from it. Each temperature plateau interval corresponds to a temperature fluctuation range. Temperature plateau intervals whose temperature fluctuation ranges meet the preset fluctuation deviation range are recorded as key plateaus, and the duration corresponding to the key plateau is recorded as the key temperature duration.

[0017] Furthermore, temperature gradient distribution information is generated based on the temperature values ​​of adjacent temperature measurement points, and abnormal temperature points in the temperature gradient distribution information that exceed the preset average temperature range are corrected by interpolation compensation based on spatial weighted neighborhood to obtain corrected temperature field information.

[0018] Thermodynamic characteristics are generated based on the carbon roasting uniformity index, the duration of critical temperatures, and the corrected temperature field information.

[0019] In some embodiments, hyperspectral imaging information of tea leaves within a carbon-roasting space is acquired, Maillard reaction characteristic spectra within a preset wavelength range are extracted, and spectral parameters of the tea leaves are calculated based on the Maillard reaction characteristic spectra, including:

[0020] The reflectance spectrum data of the tea surface is collected by a hyperspectral imaging system, and at least two characteristic bands are selected within a preset band range. The characteristic bands include the first reaction characteristic band and the second reaction characteristic band.

[0021] The absorbance of the reflectance spectral data of the first and second reaction characteristic bands is converted to generate the time-absorbance change curves for the corresponding bands.

[0022] The time-absorbance change curve is smoothed to eliminate noise interference, and feature points on the curve are extracted, including absorbance maxima, inflection points, and plateau intervals.

[0023] Based on the location and trend of the feature points, the absorbance of the first reaction feature band and the absorbance of the second reaction feature band are integrated to generate a reaction progress index. The integral ratio is calculated using the trapezoidal numerical integration method.

[0024] Monitor the rate of change of reaction progress indicators over time, and mark the Maillard reaction critical inflection point when the rate of change of reaction progress indicators over time exceeds the set threshold.

[0025] Based on the temperature and duration corresponding to the key inflection point of the Maillard reaction, the cumulative amount of melanoidins and the dynamic trend of the aldehyde-ketone ratio were calculated. The cumulative amount of melanoidins was obtained by summing the changes in absorbance after the key inflection point of the Maillard reaction, and the dynamic trend of the aldehyde-ketone ratio was obtained by establishing a time series ratio curve.

[0026] Tea spectral parameters were generated based on the cumulative amount of melanoidins, the dynamic trend of aldehyde-ketone ratio, and reaction process indicators.

[0027] In some embodiments, a variety adaptability model is dynamically constructed based on tea spectral parameters and basic tea parameters to obtain variety adaptability parameters, including:

[0028] The variety information is matched with the preset variety feature library to obtain the benchmark substance conversion curve corresponding to the current variety information. The benchmark substance conversion curve includes the expected change range of aldehyde-ketone ratio and the threshold of melanoidin accumulation.

[0029] Based on the dynamic trend information of aldehyde-ketone ratio in the spectral parameters of tea, the dynamic time-normalized distance between the real-time aldehyde-ketone ratio curve and the conversion curve of the reference substance is calculated, and the first matching degree index is generated.

[0030] Furthermore, based on the cumulative amount of melanoidins in the spectral parameters of tea leaves, the percentage of relative deviation between the current cumulative amount of melanoidins and the expected value of the conversion curve of the reference substance is calculated, and a second matching index is generated.

[0031] The first matching degree index and the second matching degree index are weighted and fused to obtain the comprehensive matching degree index. The first matching weight corresponding to the dynamic change trend information of aldehyde-ketone ratio increases with the increase of carbon roasting temperature, and the second matching weight of melanoidin accumulation increases with the extension of duration.

[0032] The comprehensive matching index is compared with the preset index threshold range, and temperature adjustment information is generated based on the comparison results.

[0033] In addition, information on the variation of carbon roasting amplitude is generated based on moisture content information;

[0034] Variety adaptability parameters are generated based on temperature regulation information and carbon roasting amplitude variation information.

[0035] In some embodiments, thermodynamic characteristics, Maillard reaction characteristic spectra, and varietal adaptability parameters are fused in a multimodal manner to generate a carbon roasting quality prediction vector. The multimodal fusion includes temperature-material conversion co-splicing and spatial inhomogeneity compensation, including:

[0036] Based on the carbon roasting uniformity index, a temperature field compensation coefficient is generated, and the dynamic change trend information of aldehyde-ketone ratio in the Maillard reaction characteristic spectrum is regionally corrected, which is denoted as the correction trend information.

[0037] The temperature regulation information in the variety adaptability parameters is coupled with the duration of key temperature to obtain a time-weighted factor. The time-weighted factor is used as the weight value of the Maillard reaction characteristic spectrum at different carbon roasting stages and weighted to obtain the weighted reaction characteristic spectrum.

[0038] Furthermore, based on the moisture content information, the temperature field compensation coefficient and time weighting factor are dynamically scaled to generate the moisture content adaptation adjustment coefficient.

[0039] A spatial grid mapping relationship is constructed, and the correction trend information, weighted response characteristic spectrum and moisture content adaptation adjustment coefficient are integrated according to spatial location. Furthermore, the abnormal grid points in the spatial grid mapping relationship are subjected to neighborhood smoothing. The smoothing process preserves the temperature gradient characteristics while eliminating isolated noise points, thus obtaining spatial grid information.

[0040] The spatial grid information is expanded along the time dimension to generate a carbon roasting quality prediction vector that includes temperature-material transformation synergistic features and spatial distribution features.

[0041] In some embodiments, the carbon roasting quality prediction vector is input into a grading decision tree model to obtain preliminary grade determination results. These preliminary grade determination results include secondary, superior, and exceptional grades, including:

[0042] A hierarchical decision tree model is established, which includes a temperature-matter conversion synergistic feature branch, a spatial distribution feature branch, and a Maillard reaction integrity branch.

[0043] Input the temperature-material conversion synergistic feature value in the carbon roasting quality prediction vector into the temperature-material conversion synergistic feature branch. When the temperature-material conversion synergistic feature value is lower than the first feature threshold, it is determined to be secondary.

[0044] The spatial distribution features in the carbon roasting quality prediction vector are input into the spatial distribution feature branch to calculate the quality uniformity index. When the quality uniformity index is lower than the second feature threshold, it is judged as excellent.

[0045] Input the Maillard reaction feature in the carbon roasting quality prediction vector into the Maillard reaction integrity branch. When the product of the slope of the aldehyde-ketone ratio curve and the cumulative amount of melanoidins exceeds the third feature threshold, it is judged as a special grade.

[0046] Based on the variety information in the basic parameters of tea, the corresponding first feature threshold, second feature threshold and third feature threshold are extracted from the preset grading standard library;

[0047] Based on the actual carbon roasting time in the carbon roasting process parameters, time decay compensation is applied to the first characteristic threshold, the second characteristic threshold, and the third characteristic threshold.

[0048] The output includes preliminary grade determination results, including secondary, superior, and special grades;

[0049] The hierarchical decision tree model is configured to trigger a cascaded fire optimization process based on the clustering quality index, including:

[0050] The clustering quality index of the preliminary grade determination results is calculated. The clustering quality index is determined by the variance of the feature space distance between the top-grade samples and the superior samples.

[0051] When the clustering quality index is lower than the preset clustering threshold, the cascade fire optimization process is activated.

[0052] In some embodiments, the optimization strategy for cascade afterfiring optimization is configured to generate a combination of afterfiring duration and temperature using a material conversion model, including:

[0053] A material conversion kinetic model is established. The input features of the material conversion kinetic model include the deviation of the temperature-material conversion synergistic feature and the spatial distribution non-uniformity in the carbon roasting quality prediction vector.

[0054] The optimal combination of afterburning parameters is calculated based on the material transformation kinetics model, and an optimization strategy for cascade afterburning optimization is generated based on the optimal combination of afterburning parameters.

[0055] For tea leaves initially classified as substandard, cascaded re-firing optimization is performed to obtain optimized tea leaves. The optimized tea leaves are then re-graded to obtain updated grading results, including:

[0056] The carbon roasting process parameters and hyperspectral imaging information of tea leaves were re-acquired and optimized, denoted as optimized carbon roasting parameters and optimized imaging information. The carbon roasting quality prediction vector was updated and denoted as optimized quality prediction vector.

[0057] The optimized quality prediction vector is input into the hierarchical decision tree model for secondary judgment, generating an updated level judgment result.

[0058] In some embodiments, the final grading result of the current tea is generated based on the preliminary grading result and the updated grading result, and the final grading result is associated with the charcoal roasting process parameters and stored in the blockchain to generate a grading report containing grade distribution probability, process defect location, and energy consumption assessment, including:

[0059] The weighted average of the proportions of preliminary special grade, preliminary excellent grade, and preliminary secondary grade in the preliminary grade determination results and the proportions of updated special grade, updated excellent grade, and updated secondary grade in the updated grade determination results is calculated to generate the final special grade proportion, final excellent grade proportion, and final secondary grade proportion in the final grade determination results.

[0060] Establish a correlation mapping relationship between the final grading results and the three-dimensional temperature field information of the roasting cage and the duration of key temperatures in the carbon roasting process parameters. The correlation mapping relationship includes the spatial correspondence between temperature field uniformity and quality grade distribution, and the temporal correspondence between the duration of key temperatures and the integrity of Maillard reaction.

[0061] The final grading results, carbon roasting process parameters, and associated mapping relationships are hashed to generate data blocks, which are then written into the blockchain network for distributed storage. The data blocks contain a grade distribution probability matrix, coordinates of process defect locations, and carbon roasting energy consumption statistics.

[0062] Based on data blocks, a visualized hierarchical report is generated, which includes a three-dimensional temperature field distribution and a hierarchical distribution overlaid with a thermal map, suggestions for improving process defects, and an energy efficiency optimization analysis report.

[0063] In a second aspect, the present invention also provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the method described in the first aspect.

[0064] In a third aspect, the present invention also provides an electronic device including a memory and a processor, the memory being used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method described in the first aspect.

[0065] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art: The present invention discloses a semi-refined tea grading method, medium, and equipment based on charcoal roasting parameters. The method collects charcoal roasting process parameters and basic tea parameters, extracts thermodynamic characteristics and Maillard reaction characteristic spectra, and constructs a variety adaptability model; it then performs multimodal fusion of thermodynamic characteristics, Maillard reaction characteristic spectra, and variety adaptability parameters to generate a charcoal roasting quality prediction vector; it performs preliminary grade determination through a grading decision tree model, and performs cascaded supplementary roasting optimization on secondary tea leaves before re-determining; finally, it stores the grading results in association with the charcoal roasting parameters on a blockchain. This method realizes dynamic correlation analysis between charcoal roasting process parameters and tea quality, solves the problem of inaccurate evaluation caused by the separation between traditional grading methods and charcoal roasting processes, significantly improves the scientificity and reliability of grading results, and provides data support for process optimization. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention or 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart illustrating steps S101 to S107 of the method described in the specific implementation embodiment;

[0068] Figure 2 This is a flowchart illustrating steps S201 to S207 of the method described in the specific implementation embodiment;

[0069] Figure 3 This is a schematic diagram of the structure of the electronic device described in the specific embodiment.

[0070] The reference numerals used in the above figures are explained as follows:

[0071] 1. Electronic equipment;

[0072] 11. Memory;

[0073] 12. Processor. Detailed Implementation

[0074] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be particularly noted that the following embodiments are for illustrative purposes only and do not limit the scope of the invention. Similarly, the following embodiments are only some, not all, embodiments of the present invention, and all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] Please see Figure 1 In a first aspect, this embodiment provides a method for grading semi-refined tea based on charcoal roasting parameters, including:

[0076] S101. Collect the carbon roasting process parameters and the basic parameters of the tea leaves. The carbon roasting process parameters include the three-dimensional temperature field distribution curve of the roasting cage and the duration. The basic parameters of the tea leaves include variety information, moisture content information and initial processing grade.

[0077] S102. Extract features from the carbon roasting process parameters to obtain thermodynamic features, including carbon roasting uniformity index and key temperature duration. The carbon roasting uniformity index is obtained by calculating the spatial temperature standard deviation using infrared thermal imaging information.

[0078] In addition, hyperspectral imaging information of tea leaves in the carbon roasting space is collected, Maillard reaction characteristic spectrum within the preset band range of 900-2500nm is extracted, and tea spectral parameters are calculated based on the Maillard reaction characteristic spectrum. The tea spectral parameters include the aldehyde-ketone ratio change rate and the inflection point temperature of melanoidin formation.

[0079] S103. Based on the spectral parameters and basic parameters of tea leaves, a variety adaptability model is dynamically constructed to obtain variety adaptability parameters, which include leaf edge scorch compensation coefficient, aroma retention rate threshold and water runoff gradient parameter.

[0080] S104. Thermodynamic characteristics, Maillard reaction characteristic spectrum and variety adaptability parameters are fused in multiple modes to generate a carbon roasting quality prediction vector. The multimodal fusion includes temperature-material conversion co-splicing and spatial non-uniformity compensation.

[0081] S105. Input the carbon roasting quality prediction vector into the hierarchical decision tree model to obtain the preliminary grade judgment result. The preliminary grade judgment result includes secondary, superior and special grades. The hierarchical decision tree model is configured to trigger the cascade fire optimization process based on the clustering quality index.

[0082] S106. Perform cascade fire optimization on tea leaves with a preliminary grade of secondary to obtain optimized tea leaves. Then, re-grade the optimized tea leaves to obtain updated grade results. The optimization strategy of cascade fire optimization is configured to use a material transformation model to combine and adjust the fire duration and temperature.

[0083] S107. Based on the preliminary grading results and the updated grading results, generate the final grading result of the current tea, and store the final grading result in association with the carbon roasting process parameters in the blockchain to generate a grading report that includes the grading distribution probability, process defect location, and energy consumption assessment.

[0084] In step S101, the parameters of the charcoal roasting process can be collected through a distributed temperature sensor array. The three-dimensional temperature field distribution curve of the roasting cage characterizes the dynamic changes in the spatial thermal field during charcoal roasting, and the duration records the duration of each temperature zone. Among the basic parameters of the tea, variety information can be obtained through a variety identification system; moisture content information can be measured using a near-infrared moisture meter. Preferably, the initial processing grade is determined comprehensively based on sensory evaluation (manual evaluation) combined with instrumental testing results (such as moisture content and color) after initial processing, encompassing both subjective quality evaluation and objective quantitative indicators. These parameters collectively constitute the basic data source for tea quality evaluation.

[0085] In step S102, thermodynamic feature extraction is achieved using an infrared thermal imaging system. The carbon roasting uniformity index is obtained by calculating the standard deviation of the temperature in each region of the roasting cage, reflecting the stability of the thermal field distribution. The duration of key temperatures is calculated as the cumulative exposure time within a specific temperature range. A hyperspectral imaging system acquires Maillard reaction characteristic spectra in the 900-2500 nm band, where the aldehyde-ketone ratio change rate characterizes the glycosamine reaction process, and the inflection point temperature for melanoidin formation indicates the critical state of the browning reaction. These characteristic parameters collectively reflect the physicochemical changes of tea leaves during the carbon roasting process.

[0086] In step S103, the varietal adaptability model is constructed using a machine learning algorithm. The leaf edge scorch compensation coefficient is calculated based on the cuticle thickness characteristics of the varietal, the aroma retention rate threshold is set according to the volatile component characteristics of the varietal, and the water loss gradient parameter reflects the differences in water loss rates among different varieties. The model dynamically adjusts these parameters to adapt to the charcoal roasting characteristics of different tea varieties.

[0087] In step S104, preferably, the multimodal fusion employs a feature-level fusion strategy. Temperature-material transformation co-splitting aligns thermodynamic and spectral parameters according to a time series, thereby establishing a precise correspondence between material transformation and temperature changes during processing. Spatial non-uniformity compensation corrects parameter differences between the edge and center regions of the roasting cage using a weighted algorithm, thereby eliminating the impact of temperature distribution differences within the roasting cage on tea quality and ensuring overall processing uniformity. This process generates a carbon-roasted quality prediction vector containing multidimensional features.

[0088] In step S105, the hierarchical decision tree model uses a pre-trained classifier. Preferably, the clustering quality index is calculated using the ratio of intra-class distance to inter-class distance. When this index falls below a set threshold, a cascaded fire optimization process is triggered. The model outputs preliminary judgment results for three categories: secondary, superior, and exceptional.

[0089] In step S106, preferably, the cascade re-firing optimization is achieved through a material transformation model. This model establishes the temperature-time-material transformation relationship based on the Arrhenius equation, generating an optimized parameter combination for the secondary tea leaves. The tea leaves after re-firing need to undergo a complete grading and evaluation process again.

[0090] In step S107, the blockchain storage can employ distributed ledger technology, the grade distribution probability is obtained through historical data statistics, the process defect location is based on parameter deviation analysis, and the energy consumption assessment is calculated by combining temperature field data and duration. The aforementioned grade distribution probability, process defect location, and energy consumption assessment together constitute a traceable graded quality report.

[0091] This embodiment achieves closed-loop control of the charcoal roasting process and quality grading through three core steps: multi-source data acquisition, feature fusion, and dynamic optimization. Thermodynamic characteristics reflect the processing environment, spectral parameters characterize the degree of material transformation, and a variety adaptability model ensures personalized evaluation standards, ultimately forming a scientific and objective grading decision-making system. A cascaded fire-replenishing mechanism provides a secondary optimization opportunity for quality improvement, while blockchain technology ensures data authenticity and traceability. This embodiment organically integrates three-dimensional temperature field distribution, critical states of material transformation, and variety characteristic parameters to form a closed-loop control system. This ensures both the uniformity and controllability of the charcoal roasting process and the traceability of grading results through blockchain storage, significantly improving the scientific rigor and reliability of semi-refined tea grading.

[0092] In some embodiments, the carbon roasting process parameters include three-dimensional temperature field information of the roasting cage and the duration;

[0093] Feature extraction was performed on the carbon roasting process parameters to obtain thermodynamic characteristics, including the carbon roasting uniformity index and the duration of critical temperatures, including:

[0094] The carbon roasting uniformity index is generated by calculating the ratio of the standard deviation to the mean of the temperature values ​​at each temperature measurement point based on the three-dimensional temperature field information of the roasting cage.

[0095] Furthermore, a temperature-time curve is constructed based on the three-dimensional temperature field information and duration of the baking cage, and temperature plateau intervals with durations exceeding a preset time threshold are extracted from it. Each temperature plateau interval corresponds to a temperature fluctuation range. Temperature plateau intervals whose temperature fluctuation ranges meet the preset fluctuation deviation range are recorded as key plateaus, and the duration corresponding to the key plateau is recorded as the key temperature duration.

[0096] Furthermore, temperature gradient distribution information is generated based on the temperature values ​​of adjacent temperature measurement points, and abnormal temperature points in the temperature gradient distribution information that exceed the preset average temperature range are corrected by interpolation compensation based on spatial weighted neighborhood to obtain corrected temperature field information.

[0097] Thermodynamic characteristics are generated based on the carbon roasting uniformity index, the duration of critical temperatures, and the corrected temperature field information.

[0098] In this embodiment, the three-dimensional temperature field information of the roasting cage refers to the data set reflecting the temperature distribution inside the roasting cage, which is collected in real time by a distributed temperature sensor array. The three-dimensional reconstruction of the spatial temperature field is achieved through a gridded arrangement of temperature measurement points. The duration characterizes the complete process cycle from the start to the end of carbon roasting, and its combination with the temporal sequence of the temperature field information forms the basis for constructing the temperature-time curve.

[0099] The carbon roasting uniformity index is calculated by the ratio of the standard deviation to the mean of the temperature values ​​at each temperature measurement point. This index is used to quantify the uniformity of the temperature distribution inside the roasting cage. The lower the ratio, the better the spatial consistency of the temperature field.

[0100] The temperature-time curve is a continuous curve formed by correlating time-series temperature data with a duration coordinate system. A preset time threshold is established based on the optimal charcoal roasting process requirements for each tea variety, used to screen temperature plateaus of technological significance. A temperature plateau interval refers to a continuous period within the curve where temperature fluctuations remain within a specific range; its identification is achieved by calculating the slope change between adjacent sampling points. A critical plateau refers to a plateau interval that meets both the duration threshold requirements and the temperature fluctuation range is controlled within a preset deviation range; its corresponding duration reflects the stability of the core process stage.

[0101] Temperature gradient distribution information is vector field data generated by analyzing the temperature difference between adjacent temperature measurement points, used to characterize the spatial trend of temperature variation. The preset average temperature range is set based on historical process data. Anomaly point identification is achieved by comparing the statistical characteristics of that point with its surrounding temperature range. Spatial weighted neighborhood interpolation compensation correction calculates a replacement value based on the spatial distribution weights of valid temperature measurement points around the anomaly point; the weight allocation is inversely proportional to the spatial distance between the temperature measurement points. Corrected temperature field information is an optimized temperature distribution model obtained after eliminating interference from abnormal temperature measurement data. This model, along with the carbon roasting uniformity index and the duration of critical temperatures, constitutes a thermodynamic characteristic system reflecting the quality of the carbon roasting process.

[0102] This embodiment extracts multi-dimensional features to transform raw temperature data into quantifiable thermodynamic characteristics for evaluating process quality. The spatial uniformity, temporal stability, and data reliability dimensions corroborate each other, providing data support for the standardized control of the carbon roasting process.

[0103] Please see Figure 2 In some embodiments, hyperspectral imaging information of tea leaves within the carbon roasting space is acquired, Maillard reaction characteristic spectra within a preset wavelength range are extracted, and spectral parameters of the tea leaves are calculated based on the Maillard reaction characteristic spectra, including:

[0104] S201. Collect the reflectance spectrum data of the tea surface through a hyperspectral imaging system, and select at least two characteristic bands within a preset band range. The characteristic bands include the first reaction characteristic band and the second reaction characteristic band.

[0105] S202. Perform absorbance conversion on the reflectance spectral data of the first and second reaction characteristic bands to generate time-absorbance change curves for the corresponding bands.

[0106] S203. Smooth the time-absorbance change curve to eliminate noise interference and extract feature points on the curve, including absorbance maxima, inflection points and plateau intervals.

[0107] S204. Based on the location and trend of the feature points, calculate the integral ratio of the absorbance of the first reaction feature band to the absorbance of the second reaction feature band to generate a reaction progress index. The integral ratio is calculated using the trapezoidal numerical integration method.

[0108] S205. Monitor the rate of change of reaction progress indicators over time. When the rate of change of reaction progress indicators over time exceeds the set threshold, mark it as a critical inflection point of the Maillard reaction.

[0109] S206. Based on the temperature and duration corresponding to the key inflection point of the Maillard reaction, calculate the cumulative amount of melanoidins and the dynamic trend information of the aldehyde-ketone ratio. The cumulative amount of melanoidins is obtained by summing the absorbance changes after the key inflection point of the Maillard reaction, and the dynamic trend information of the aldehyde-ketone ratio is obtained by establishing a time series ratio curve.

[0110] S207. Generate tea spectral parameters based on the cumulative amount of melanoidins, the dynamic trend of aldehyde-ketone ratio, and reaction process indicators.

[0111] In step S201, the preset band range of the hyperspectral imaging system is determined by the characteristics of the Maillard reaction. Reflectance spectral data refers to the optical information of the tea surface acquired by the hyperspectral imaging system. The first and second reaction characteristic bands correspond to the characteristic absorption peaks of products at different stages of the Maillard reaction, respectively. The selection of these characteristic bands is based on the spectral characteristics of the characteristic compounds produced during the Maillard reaction.

[0112] In step S202, absorbance conversion is a mathematical process that converts reflectance data into absorbance. The time-absorbance change curve is used to characterize the change law of absorbance over time in a specific wavelength band, which can intuitively reflect the reaction process.

[0113] In step S203, the smoothing process eliminates spectral noise using a digital filtering algorithm, ensuring data reliability. The absorbance maxima characterize the peak moment of the reaction rate, the inflection point reflects the transition characteristics of the reaction mechanism, and the plateau interval indicates the stage where the reaction tends to stabilize. Accurate extraction of these feature points is crucial for subsequent analysis.

[0114] In step S204, the integral ratio refers to the ratio of the areas under the absorbance curves of the two characteristic bands. The trapezoidal numerical integration method is used to improve the calculation accuracy.

[0115] In step S205, the threshold is a critical judgment value determined based on the kinetic characteristics of the Maillard reaction, used to accurately identify the key inflection point of the Maillard reaction.

[0116] In step S206, the cumulative amount of melanoidins refers to the total amount of reaction products obtained by accumulating the changes in absorbance, and the dynamic trend information of the aldehyde-ketone ratio refers to the reaction characteristics reflected by establishing a time series ratio curve.

[0117] In step S207, the spectral parameters of tea leaves refer to a combination of quantitative indicators that comprehensively reflect the Maillard reaction process, providing data support for subsequent analysis.

[0118] This embodiment utilizes hyperspectral imaging technology to dynamically monitor the Maillard reaction during the charcoal roasting process of tea, employing spectral characteristic changes in characteristic bands to characterize the reaction progress. By selecting the first and second characteristic bands of the reaction, a time-absorbance change curve is established, and key feature points such as absorbance maxima, inflection points, and plateau intervals are extracted, enabling precise tracking of the Maillard reaction process. The reaction progress index is calculated using integral ratios, and key inflection points are accurately identified by setting thresholds, thereby calculating the cumulative amount of melanoidins and the dynamic trends of the aldehyde-ketone ratio. This scheme achieves real-time, quantitative monitoring of the Maillard reaction during charcoal roasting through non-contact optical detection, providing objective and reliable spectral parameter indicators for the control of the tea charcoal roasting process. It effectively solves the subjectivity problem of traditional manual experience-based judgment, significantly improving the scientific nature and accuracy of charcoal roasting process control. This embodiment, through spectral analysis of characteristic bands, achieves dynamic monitoring and quantitative characterization of the Maillard reaction during the tea charcoal roasting process, providing a reliable optical detection method for charcoal roasting process control.

[0119] In some embodiments, a variety adaptability model is dynamically constructed based on tea spectral parameters and basic tea parameters to obtain variety adaptability parameters, including:

[0120] The variety information is matched with the preset variety feature library to obtain the benchmark substance conversion curve corresponding to the current variety information. The benchmark substance conversion curve includes the expected change range of aldehyde-ketone ratio and the threshold of melanoidin accumulation.

[0121] Based on the dynamic trend information of aldehyde-ketone ratio in the spectral parameters of tea, the dynamic time-normalized distance between the real-time aldehyde-ketone ratio curve and the conversion curve of the reference substance is calculated, and the first matching degree index is generated.

[0122] Furthermore, based on the cumulative amount of melanoidins in the spectral parameters of tea leaves, the percentage of relative deviation between the current cumulative amount of melanoidins and the expected value of the conversion curve of the reference substance is calculated, and a second matching index is generated.

[0123] The first matching degree index and the second matching degree index are weighted and fused to obtain the comprehensive matching degree index. The first matching weight corresponding to the dynamic change trend information of aldehyde-ketone ratio increases with the increase of carbon roasting temperature, and the second matching weight of melanoidin accumulation increases with the extension of duration.

[0124] The comprehensive matching index is compared with the preset index threshold range, and temperature adjustment information is generated based on the comparison result. When the comprehensive matching index is lower than the lower limit of the preset index threshold range, a heating acceleration command is generated, and when the comprehensive matching index is higher than the upper limit of the preset index threshold range, a constant temperature maintenance command is generated.

[0125] Additionally, based on the moisture content information, information on the variation of carbon roasting amplitude is generated, with the adjustment amplitude increased when the moisture content is high and decreased when the moisture content is low.

[0126] Variety adaptability parameters are generated based on temperature regulation information and carbon roasting amplitude variation information.

[0127] In this embodiment, the variety adaptability model is a dynamic parameter adjustment system based on the characteristics of tea varieties. It optimizes the process by matching real-time monitoring data with baseline parameters. The preset variety feature library contains material transformation characteristic data of different tea varieties under standard charcoal roasting conditions. Among them, the baseline material transformation curve characterizes the aldehyde-ketone ratio change trajectory and melanoidin accumulation trend of a specific variety under ideal process conditions. The expected change range of the aldehyde-ketone ratio is obtained through statistical analysis of historical process data, and the melanoidin accumulation threshold is preset according to the quality requirements of the variety.

[0128] The real-time aldehyde-ketone ratio curve is obtained from actual reaction progress data through hyperspectral monitoring. The dynamic time warping distance quantifies the morphological difference between the actual curve and the baseline curve. This distance calculation is achieved through a dynamic time warping algorithm, which eliminates the influence of nonlinear deformation on the time axis. The first matching degree index reflects the degree of process compliance of the aldehyde-ketone ratio change; the smaller the value, the closer the actual reaction process is to the ideal state. The current cumulative amount of melanoidins refers to the real-time measurement value obtained through spectral analysis. The relative deviation percentage is obtained by calculating the difference between the measured value and the baseline expected value. The generated second matching degree index is used to assess whether the melanoidin production progress meets the target.

[0129] Weighted fusion is a process of dynamically weighting and summing two matching indices based on the characteristics of the process stage. The first matching weight is positively correlated with the carbon roasting temperature, reflecting the dominance of the aldehyde-ketone reaction in the high-temperature stage; the second matching weight is positively correlated with the duration, reflecting the time-delay characteristics of the melanoidin-like cumulative effect.

[0130] Preferably, the comprehensive matching degree index is obtained by linear weighted calculation, and its preset index threshold range is determined according to the variety characteristics and quality requirements. The lower threshold triggers the heating acceleration command to promote the reaction rate, and the upper threshold triggers the isothermal maintenance command to stabilize the reaction state.

[0131] The information on the variation in charcoal roasting amplitude refers to the correction amount of process parameters dynamically adjusted based on the initial moisture content of the tea leaves. The adjustment logic follows the negative correlation between moisture content and heat transfer efficiency, and the adjustment amplitude is quantitatively controlled through a proportional coefficient. Temperature regulation information and the information on the variation in charcoal roasting amplitude together constitute the varietal adaptability parameter, which is dynamically adjusted in real time through a feedback control system.

[0132] This embodiment establishes a dual-matching evaluation mechanism to dynamically compare the spectral parameters of tea leaves with the conversion curves of reference substances corresponding to the current variety information. Combined with a moisture content compensation mechanism, it achieves customized control of the charcoal roasting process for specific varieties. Specifically, the dynamic time warping algorithm solves the problem of temporal nonlinearity in the reaction process, and the dual-weighting mechanism ensures the evaluation focus at different process stages. The resulting variety adaptability parameters provide a precise basis for the regulation of the charcoal roasting process.

[0133] In some embodiments, thermodynamic characteristics, Maillard reaction characteristic spectra, and varietal adaptability parameters are fused in a multimodal manner to generate a carbon roasting quality prediction vector. The multimodal fusion includes temperature-material conversion co-splicing and spatial inhomogeneity compensation, including:

[0134] Based on the carbon roasting uniformity index, a temperature field compensation coefficient is generated, and the dynamic change trend information of aldehyde-ketone ratio in the Maillard reaction characteristic spectrum is regionally corrected, which is denoted as the correction trend information.

[0135] The temperature regulation information in the variety adaptability parameters is coupled with the duration of key temperature to obtain a time-weighted factor. The time-weighted factor is used as the weight value of the Maillard reaction characteristic spectrum at different carbon roasting stages and weighted to obtain the weighted reaction characteristic spectrum.

[0136] Furthermore, based on the moisture content information, the temperature field compensation coefficient and time weighting factor are dynamically scaled to generate the moisture content adaptation adjustment coefficient.

[0137] A spatial grid mapping relationship is constructed, and the correction trend information, weighted response characteristic spectrum and moisture content adaptation adjustment coefficient are integrated according to spatial location. Furthermore, the abnormal grid points in the spatial grid mapping relationship are subjected to neighborhood smoothing. The smoothing process preserves the temperature gradient characteristics while eliminating isolated noise points, thus obtaining spatial grid information.

[0138] The spatial grid information is expanded along the time dimension to generate a carbon roasting quality prediction vector that includes temperature-material transformation synergistic features and spatial distribution features.

[0139] In this embodiment, the temperature field compensation coefficient is a correction factor obtained by converting the carbon roasting uniformity index. Its value is negatively correlated with the uniformity index and is used to perform region-specific correction on the dynamic trend information of the aldehyde-ketone ratio, thereby eliminating the influence of uneven temperature distribution on the reaction process. That is, the lower the carbon roasting uniformity index, the more uneven the temperature distribution, and the larger the temperature field compensation coefficient, thus enhancing the regional correction strength for the dynamic trend of the aldehyde-ketone ratio. The generated correction trend information characterizes the spatially compensated reaction process data.

[0140] The time-weighted factor is a dynamic weight value obtained by nonlinearly coupling temperature regulation information with the duration of key temperatures. Its calculation process considers the synergistic effect of temperature regulation amplitude and duration. Furthermore, the calculation process of the time-weighted factor includes:

[0141] The temperature regulation information (heating acceleration or constant temperature maintenance command) is normalized and converted into a regulation intensity coefficient;

[0142] The product of the regulation intensity coefficient and the duration of the key temperature is used to reflect the cumulative effect of temperature regulation over time.

[0143] By mapping to the [0,1] interval using an S-shaped function, the weight values ​​are ensured to conform to the probability distribution characteristics, and higher weights will be obtained during the long-term high-temperature phase.

[0144] The time-weighted factor, as the weighting coefficient of the Maillard reaction characteristic spectrum, enables the reaction characteristics of different carbon roasting stages to be characterized differently. The final weighted reaction characteristic spectrum reflects the material transformation state after process regulation.

[0145] The moisture content adaptation adjustment coefficient is a scaling factor that dynamically adjusts the temperature field compensation coefficient and time weighting factor based on moisture content information. Its adjustment direction is positively correlated with moisture content, ensuring that process parameters adapt to differences in the initial state of the tea leaves. Preferably, the dynamic adjustment process of the moisture content adaptation adjustment coefficient is as follows: when the initial moisture content of the tea leaves is high, the temperature field compensation coefficient (strengthening temperature correction) and time weighting factor (extending the weight of key temperatures) are proportionally amplified to compensate for the hysteresis effect of heat conduction caused by high moisture content; conversely, the coefficient is appropriately reduced when the initial moisture content is low. The adjustment range is based on a preset moisture content-adjustment coefficient mapping table, which is established using historical process data to ensure that tea leaves with different moisture contents can obtain matching thermodynamic parameter configurations.

[0146] The spatial grid mapping relationship is an established three-dimensional spatial coordinate system, whose grid division is consistent with the distribution of temperature field measurement points in the roasting cage. Preferably, when integrating correction trend information, weighted reaction characteristic spectrum, and moisture content adaptation adjustment coefficient according to spatial location, a gridded interpolation algorithm is used to ensure data spatial alignment. Further, the specific process of the gridded interpolation algorithm is as follows:

[0147] The carbon-roasted region is divided into uniform two-dimensional grid cells, and each grid node stores parameters such as correction trend, weighted characteristic spectrum and moisture content coefficient.

[0148] The bicubic spline interpolation method is used to calculate the integrated value at any position inside the grid based on the parameter values ​​of adjacent nodes, ensuring a smooth spatial transition between different parameters;

[0149] Gradient detection eliminates potential edge abrupt changes caused by interpolation, ensuring that temperature compensation parameters are continuously distributed in a three-dimensional space, including planar position and process time.

[0150] The identification of abnormal grid points is based on local statistical features. Preferably, it is achieved by calculating the standard deviation and mean deviation of each grid point from its 8 neighbors. When the data of a certain point deviates from the mean of its neighbors by more than 3 times the standard deviation, it is marked as abnormal.

[0151] Preferably, the neighborhood smoothing process employs an anisotropic filtering algorithm. During smoothing, the anisotropic filter enhances the filtering intensity along the temperature gradient direction (suppressing outliers), while maintaining isotropic smoothing in uniform regions. The filter kernel size is dynamically adjusted with the local coefficient of variation, eliminating isolated noise points while preserving the temperature gradient characteristics. The generated spatial grid information contains complete spatial distribution characteristics.

[0152] The carbon roasting quality prediction vector is a high-dimensional feature vector generated by unfolding spatial grid information along the time dimension. Its temperature-mass conversion synergy characteristics reflect the correlation between thermodynamic parameters and reaction progress, while its spatial distribution characteristics reflect the three-dimensional distribution of process parameters. The generation process of the carbon roasting quality prediction vector is achieved through tensor expansion operations, and the resulting multi-dimensional data structure provides a comprehensive feature representation for quality prediction.

[0153] Preferably, the generation of the carbon-roasted quality prediction vector employs high-order tensor expansion. The three-dimensional tensor (spatial location, time, parameter type) composed of multi-source process parameters such as temperature field, moisture content, and time series is modally expanded into multiple two-dimensional matrices. Principal component features under each mode are extracted using alternating least squares, and finally fused into a prediction vector containing spatial distribution patterns, time-varying characteristics, and parameter coupling relationships. The carbon-roasted quality prediction vector is decoded using a convolutional neural network, outputting the probability distributions of quality indicators such as moisture content uniformity and caramelization degree.

[0154] This embodiment establishes a multi-source data fusion framework to achieve deep coupling of thermodynamic environmental parameters, material transformation characteristics, and variety adaptability. Specifically, a temperature field compensation mechanism addresses spatial inhomogeneity, a dynamic weighting strategy reflects the differences in process stages, moisture content adaptation ensures parameter generalization ability, and spatial grid processing guarantees spatial consistency of features. The resulting carbon roasting quality prediction vector possesses three-dimensional features encompassing time, space, and material transformation, providing high-precision feature input for subsequent quality grading. The fusion process in this embodiment strictly follows the laws of physicochemical change and achieves quantitative characterization of the carbon roasting process through a data-driven approach.

[0155] In some embodiments, the carbon roasting quality prediction vector is input into a grading decision tree model to obtain preliminary grade determination results. These preliminary grade determination results include secondary, superior, and exceptional grades, including:

[0156] A hierarchical decision tree model is established, which includes a temperature-matter conversion synergistic feature branch, a spatial distribution feature branch, and a Maillard reaction integrity branch.

[0157] Input the temperature-material conversion synergistic feature value in the carbon roasting quality prediction vector into the temperature-material conversion synergistic feature branch. When the temperature-material conversion synergistic feature value is lower than the first feature threshold, it is determined to be secondary.

[0158] The spatial distribution features in the carbon roasting quality prediction vector are input into the spatial distribution feature branch to calculate the quality uniformity index. When the quality uniformity index is lower than the second feature threshold, it is judged as excellent.

[0159] Input the Maillard reaction feature in the carbon roasting quality prediction vector into the Maillard reaction integrity branch. When the product of the slope of the aldehyde-ketone ratio curve and the cumulative amount of melanoidins exceeds the third feature threshold, it is judged as a special grade.

[0160] Based on the variety information in the basic parameters of tea, the corresponding first feature threshold, second feature threshold and third feature threshold are extracted from the preset grading standard library;

[0161] Based on the actual carbon roasting time in the carbon roasting process parameters, time decay compensation is applied to the first characteristic threshold, the second characteristic threshold, and the third characteristic threshold.

[0162] The output includes preliminary grade determination results of secondary, superior and special grades, and the preliminary grade determination results are accompanied by feature threshold comparison data of each branch;

[0163] The hierarchical decision tree model is configured to trigger a cascaded fire optimization process based on the clustering quality index, including:

[0164] The clustering quality index of the preliminary grade determination results is calculated. The clustering quality index is determined by the variance of the feature space distance between the top-grade samples and the superior samples.

[0165] When the clustering quality index is lower than the preset clustering threshold, the cascade fire optimization process is activated.

[0166] In this embodiment, the hierarchical decision tree model is a tea quality grading algorithm framework based on multi-branch condition judgment. Its temperature-material transformation synergistic feature branch is used to evaluate the degree of matching between thermodynamic parameters and material transformation, the spatial distribution feature branch is used to detect the spatial uniformity of quality indicators, and the Maillard reaction integrity branch is used to quantify the completion of key reactions.

[0167] Temperature-material transformation synergistic feature value is a scalar parameter extracted from the carbon-roasted quality prediction vector. Its value reflects the coupling strength between the temperature field and the material transformation process. The first feature threshold is a classification boundary value determined based on the statistics of historical high-quality tea samples.

[0168] The quality uniformity index is a comprehensive evaluation parameter calculated based on spatial distribution characteristics. It is quantified by analyzing the spatial variation coefficients of parameters such as moisture content and caramelization degree. The second characteristic threshold characterizes the lower limit of permissible quality fluctuation.

[0169] The slope of the aldehyde-ketone ratio curve in the Maillard reaction integrity branch refers to the instantaneous rate of change in the concentration of aldehydes and ketones during the reaction process. The cumulative amount of melanoidins is an indicator of pigment production extracted through near-infrared spectroscopy. The product of the two reflects the reaction kinetics. The third characteristic threshold corresponds to the reaction completion benchmark required for premium-grade tea.

[0170] The aforementioned graded decision tree model employs a three-level progressive judgment structure to achieve strict quality grading. When the temperature-matter conversion synergistic characteristic value is below the first characteristic threshold, the model directly classifies it as secondary and terminates subsequent branch evaluations, indicating a severe mismatch between the thermodynamic environment and the matter conversion process. Only when the temperature-matter conversion synergistic characteristic value meets the standard will the model enter the spatial distribution characteristic branch evaluation, judging whether the superior grade standard is met through the quality uniformity index; if this index is below the second characteristic threshold, the superior grade determination is terminated, reflecting that although the thermodynamic conditions are qualified, there is a spatial non-uniformity problem. Finally, only samples that pass the first two levels of judgment will enter the Maillard reaction integrity branch, determining the special grade qualification by multiplying the slope of the aldehyde-ketone ratio curve with the cumulative amount of melanoidins. This mechanism ensures that special grade tea must simultaneously meet the three core requirements of thermodynamic matching, spatial uniformity, and reaction integrity. Through the cascaded judgment logic and the establishment of a strict admission mechanism, precise differentiation of quality grades is achieved.

[0171] The preset grading standard library refers to a database that stores the corresponding grading parameters for different tea varieties. Its threshold data can be obtained through variety-specific process tests.

[0172] Time decay compensation refers to the dynamic adjustment of the feature threshold based on the actual carbon roasting time. The compensation coefficient is negatively correlated with the carbon roasting time and is used to eliminate the influence of feature signal decay caused by long-term carbon roasting.

[0173] The feature threshold comparison data includes the percentage deviation between the actual feature value and the threshold for each branch, which is used to assist in the interpretability analysis of the grade determination.

[0174] The clustering quality index is a model discrimination performance indicator calculated through the dispersion of the feature space distribution. It is quantified by calculating the variance of the Mahalanobis distance between top-grade and superior-grade samples in a three-dimensional feature space encompassing temperature-material conversion synergistic features, spatial distribution features, and Maillard reaction features. The preset clustering threshold is a stability boundary value determined based on performance verification during the model training phase. Values ​​below the preset threshold indicate fuzzy boundaries in the grading results. The cascaded calcination optimization process is a process adjustment mechanism triggered by batches with insufficient clustering quality, re-optimizing the carbon roasting parameters through feedback control.

[0175] This embodiment achieves refined grading through a three-level progressive judgment structure. Temperature-material transformation synergy characteristics serve as the basic screening condition, spatial uniformity as the criterion for intermediate quality, and Maillard reaction integrity as the determining factor for top-level quality. The grading process incorporates a variety-adaptive dynamic threshold mechanism and time decay compensation to ensure that the judgment criteria match the actual process. A clustering quality monitoring mechanism guarantees the reliability of the grading results and provides data support for subsequent process optimization.

[0176] In some embodiments, the optimization strategy for cascade afterfiring optimization is configured to generate a combination of afterfiring duration and temperature using a material conversion model, including:

[0177] A material conversion kinetic model is established. The input features of the material conversion kinetic model include the deviation of the temperature-material conversion synergistic feature and the spatial distribution non-uniformity in the carbon roasting quality prediction vector.

[0178] The optimal combination of afterburning parameters is calculated based on the material transformation kinetics model. An optimization strategy for cascade afterburning optimization is then generated based on this optimal combination of parameters. The optimal combination of afterburning parameters calculated based on the material transformation kinetics model includes:

[0179] For regions where the temperature-material conversion synergistic characteristic value is below the threshold, the amount of increase in ignition temperature and the amount of extension of action time are generated.

[0180] For regions with uneven spatial distribution characteristics, calculate the turning frequency adjustment coefficient and heat energy redistribution scheme;

[0181] For samples with incomplete Maillard reactions, determine the optimal heating rate curve for the afterburning stage;

[0182] For tea leaves initially classified as substandard, cascaded re-firing optimization is performed to obtain optimized tea leaves. The optimized tea leaves are then re-graded to obtain updated grading results, including:

[0183] The carbon roasting process parameters and hyperspectral imaging information of tea leaves were re-acquired and optimized, denoted as optimized carbon roasting parameters and optimized imaging information. The carbon roasting quality prediction vector was updated and denoted as optimized quality prediction vector.

[0184] The optimized quality prediction vector is input into the hierarchical decision tree model for secondary judgment, generating an updated grade judgment result.

[0185] When the updated level determination result is still secondary and the number of fire boosting attempts has not reached the upper limit, the cascaded fire boosting optimization process is executed iteratively.

[0186] Record the parameter adjustments and corresponding quality changes for each ignition operation, and dynamically update the parameter weights of the material transformation kinetic model.

[0187] In this embodiment, the optimization strategy for cascaded fire optimization is generated through a material conversion kinetics model. This model uses the temperature-material conversion synergistic characteristic deviation and spatial distribution characteristic non-uniformity as core input features. The temperature-material conversion synergistic characteristic deviation refers to the difference between the actual characteristic value and the standard characteristic value, and the spatial distribution characteristic non-uniformity is calculated as the percentage deviation of the quality uniformity index from the standard value.

[0188] The optimal combination of afterburning parameters is a process adjustment scheme calculated for different types of quality defects. Its generation process adopts the principle of differentiated treatment: for areas where the temperature-material conversion synergistic characteristic value is lower than the threshold, the material conversion is enhanced by increasing the afterburning temperature and extending the action time; for areas with uneven spatial distribution, the material movement trajectory is changed by adjusting the turning frequency coefficient, and the temperature field distribution is optimized in conjunction with the heat energy redistribution scheme; for samples with incomplete Maillard reaction, the reaction kinetics process is controlled by the optimal heating rate curve.

[0189] Optimized tea refers to improved products that have undergone cascaded re-firing optimization treatment. Its quality assessment is achieved through newly acquired optimized charcoal roasting parameters and hyperspectral imaging information. Optimized charcoal roasting parameters include newly added temperature-time curves and turning operation records from the re-firing stage, while optimized imaging information is used to extract updated material distribution characteristics. The optimized quality prediction vector is a secondary evaluation feature set constructed by fusing newly acquired data. The updated grade determination result generated after inputting this vector into a hierarchical decision tree model is used to verify the optimization effect.

[0190] The upper limit for the number of re-firing operations is a safety boundary value set based on the heat resistance characteristics of tea leaves. The parameter adjustment record includes the temperature correction, duration adjustment, and changes to the roasting scheme for each re-firing operation. Quality change data is obtained by comparing the differences in feature values ​​before and after optimization, and is used for dynamic updating of parameter weights in the material transformation kinetic model. This update mechanism achieves adaptive optimization of the model through machine learning algorithms.

[0191] This embodiment establishes a mapping relationship between defect characteristics and process parameters to achieve precise generation of targeted afterburning strategies. An iterative optimization mechanism ensures quality improvement, and dynamic updates to model parameters maintain the adaptability of the optimization strategy. This transforms traditional experience-based afterburning operations into a quantitative control process based on material transformation laws, providing a scientific basis for the refined control of the carbon roasting process.

[0192] In some embodiments, the final grading result of the current tea is generated based on the preliminary grading result and the updated grading result, and the final grading result is associated with the charcoal roasting process parameters and stored in the blockchain to generate a grading report containing grade distribution probability, process defect location, and energy consumption assessment, including:

[0193] The weighted average of the proportions of preliminary special grade, preliminary excellent grade, and preliminary secondary grade in the preliminary grade determination results and the proportions of updated special grade, updated excellent grade, and updated secondary grade in the updated grade determination results is calculated to generate the final special grade proportion, final excellent grade proportion, and final secondary grade proportion in the final grade determination results.

[0194] Establish a correlation mapping relationship between the final grading results and the three-dimensional temperature field information of the roasting cage and the duration of key temperatures in the carbon roasting process parameters. The correlation mapping relationship includes the spatial correspondence between temperature field uniformity and quality grade distribution, and the temporal correspondence between the duration of key temperatures and the integrity of Maillard reaction.

[0195] The final grading results, carbon roasting process parameters, and associated mapping relationships are hashed to generate data blocks, which are then written into the blockchain network for distributed storage. The data blocks contain a grade distribution probability matrix, coordinates of process defect locations, and carbon roasting energy consumption statistics.

[0196] Based on data blocks, a visualized hierarchical report is generated, which includes a three-dimensional temperature field distribution and a hierarchical distribution overlaid with a thermal map, suggestions for improving process defects, and an energy efficiency optimization analysis report.

[0197] In this embodiment, the final grading result refers to the final evaluation of tea quality generated by integrating the preliminary judgment and the optimized judgment data. The proportions of preliminary special grade, preliminary superior grade, and preliminary secondary grade in the preliminary grade judgment result reflect the quality distribution under the original charcoal roasting state. The proportions of updated special grade, updated superior grade, and updated secondary grade in the updated grade judgment result represent the quality improvement effect after the supplementary roasting optimization.

[0198] The correlation mapping relationship refers to establishing a traceability relationship between grading results and carbon roasting process parameters. Specifically, it includes the analysis of the regional matching degree between the spatial uniformity index of the three-dimensional temperature field and the distribution of tea grades, as well as the correlation verification between the temperature persistence during key process periods and the completion degree of the Maillard reaction. This correlation mapping relationship provides a basis for the process-related analysis of the grading results.

[0199] The final grading results, carbon roasting process parameters, and associated mapping relationships are hashed to generate immutable data blocks, which include the grade distribution probability matrix, coordinates of process defect locations, and carbon roasting energy consumption statistics. These data are stored in a distributed manner to ensure the traceability of the grading process.

[0200] In the generation of visualized hierarchical reports, the preferred method is to use spatial interpolation technology to render a bivariate heat map by overlaying a three-dimensional temperature field distribution and a hierarchical distribution; the process defect improvement suggestions are automatically generated based on defect location and parameter correlation; and the energy efficiency optimization analysis report provides energy-saving suggestions by analyzing energy consumption data and quality-output ratio.

[0201] This embodiment constructs a verification and traceability system for grading results. Blockchain technology ensures the authenticity of grading data, and visual analysis tools enable process traceability of grading conclusions. This embodiment generates accurate final grading results through a weighted calculation that integrates preliminary and updated grading results. It also establishes a mapping relationship between these final grading results and the three-dimensional temperature field information and key temperature durations of the carbon roasting process parameters, enabling traceability analysis of quality grades and process parameters. Blockchain storage of grading results, process parameters, and mapping relationships ensures data immutability and traceability. The visualized grading report generated based on data blocks, combined with overlaid heatmaps of temperature field and grade distribution, process defect improvement suggestions, and energy efficiency optimization analysis, forms a complete closed loop of quality assessment and process optimization, significantly improving the reliability and guiding value of the grading results.

[0202] In a second aspect, this embodiment also provides a computer-readable storage medium storing computer program instructions thereon, which, when executed by a processor, implement the method described in the first aspect.

[0203] The computer program involved in this embodiment can be stored in a computer device readable storage medium, which includes, but is not limited to, disks, magnetic tapes, magnetic cards, floppy disks, flash memory, optical disks, optical cards, read-only memory (ROM), random access memory (RAM), erasable programmable ROM (EPROM), and electrically erasable programmable ROM (EEPROM), etc. It also includes other biological, physical, or chemical structures capable of performing similar or equivalent functions to the storage media listed above, such as DNA, RNA, proteins, and other units with information storage capabilities. In specific embodiments, the storage medium involved can be one of the above-mentioned media types or a combination of the above media types. In different embodiments, the computer program involved in the embodiment can be centrally stored in a single medium or distributed across multiple media. The memory containing the computer device readable storage medium can be non-volatile memory or random access memory. These computer device readable storage media can be built into the device or connected to the device involved in the embodiment as an external device or part of an external device. In some embodiments, the memory having a computer device readable storage medium is deployed locally; in other embodiments, the memory may be deployed remotely from the processor, for example, as a network-attached memory accessed via RF circuitry or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), or a suitable combination thereof, as long as computer device access to the memory is enabled. Furthermore, the computer program involved in the embodiments may be stored in plaintext / ciphertext form, or it may be designed as training data, integrated and recombined through model training and implicitly stored in the parameter states of a deep neural network or other machine learning model.

[0204] Please see Figure 3 In a third aspect, this embodiment also provides an electronic device 1, including a memory 11 and a processor 12, wherein the memory 11 is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor 12 to implement the method described in the first aspect.

[0205] The processor described in this embodiment can be implemented by hardware, firmware, software, or a combination thereof. It can be a circuit, one or more of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field-programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, or a microprocessor. It also includes other physical, biological, or chemical structures that can implement the same or equivalent functions as the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of this application, or any combination of the steps mentioned therein.

[0206] Unlike existing technologies, the above technical solution has the following beneficial effects:

[0207] This invention establishes a complete grading system for semi-refined tea by integrating carbon roasting process parameters, tea spectral parameters, and varietal adaptability parameters. The system acquires the three-dimensional temperature field distribution curve of the roasting cage using an infrared thermal imaging system, extracts Maillard reaction characteristic spectra using a hyperspectral imaging system, and generates a carbon roasting quality prediction vector through multimodal fusion. Based on a grading decision tree model, the system analyzes the carbon roasting quality prediction vector, outputting preliminary grade judgments including premium, superior, and secondary grades. The secondary tea is then further improved through a cascaded fire optimization process. Finally, the grading results are linked to the carbon roasting process parameters and stored on a blockchain, generating a visualized grading report containing a superimposed heatmap of the three-dimensional temperature field distribution and grade distribution. This technical solution achieves a precise correlation between carbon roasting process parameters and tea quality grading, replacing traditional experience-based judgment with a data-driven approach, ensuring the objectivity and traceability of the grading results, and providing reliable technical support for the quality control of semi-refined tea.

[0208] The above description is only a part of the embodiments of the present invention and does not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made based on the content of the present invention specification and drawings, or direct or indirect application in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for grading semi-refined tea based on charcoal roasting parameters, characterized in that, include: Collect charcoal roasting process parameters and basic tea parameters. The basic tea parameters include variety information, moisture content information and initial processing grade. The charcoal roasting process parameters include three-dimensional temperature field information of the roasting cage and duration. Feature extraction is performed on the carbon roasting process parameters to obtain thermodynamic features, including the carbon roasting uniformity index and the duration of critical temperatures, including: The carbon roasting uniformity index is generated by calculating the ratio of the standard deviation to the mean of the temperature values ​​at each temperature measurement point based on the three-dimensional temperature field information of the roasting cage. Furthermore, a temperature-time curve is constructed based on the three-dimensional temperature field information of the baking cage and the duration, and temperature plateau intervals with durations exceeding a preset time threshold are extracted from it. Each temperature plateau interval corresponds to a temperature fluctuation range. Temperature plateau intervals whose temperature fluctuation ranges meet the preset fluctuation deviation range are recorded as key plateaus, and the duration corresponding to the key plateau is recorded as the key temperature duration. Furthermore, temperature gradient distribution information is generated based on the temperature values ​​of adjacent temperature measurement points, and abnormal temperature points in the temperature gradient distribution information that exceed the preset average temperature range are corrected by interpolation compensation based on spatial weighted neighborhood to obtain corrected temperature field information. The thermodynamic characteristics are generated based on the carbon baking uniformity index, the duration of critical temperatures, and the corrected temperature field information. In addition, hyperspectral imaging information of tea leaves in the carbon roasting space is collected, Maillard reaction characteristic spectra are extracted within a preset band range, and tea spectral parameters are calculated based on the Maillard reaction characteristic spectra. Based on the tea spectral parameters and basic tea parameters, a variety adaptability model is dynamically constructed to obtain the variety adaptability parameters; The thermodynamic characteristics, Maillard reaction characteristic spectrum and variety adaptability parameters are fused in a multimodal manner to generate a carbon roasting quality prediction vector. The multimodal fusion includes temperature-material conversion co-splicing and spatial inhomogeneity compensation. The carbon roasting quality prediction vector is input into the hierarchical decision tree model to obtain the preliminary grade judgment result, which includes secondary, superior and special grades. The hierarchical decision tree model is configured to trigger the cascade fire optimization process based on the clustering quality index. For tea leaves whose initial grade is classified as secondary, cascaded supplementary fire optimization is performed to obtain optimized tea leaves. The optimized tea leaves are then re-graded to obtain updated grade results. The optimization strategy of the cascaded supplementary fire optimization is configured to use a material transformation model to combine and adjust the supplementary fire duration and temperature. Based on the preliminary grading results and the updated grading results, the final grading result of the current tea is generated, and the final grading result is stored in the blockchain in association with the charcoal roasting process parameters, generating a grading report that includes the grading distribution probability, the location of process defects, and energy consumption assessment.

2. The semi-refined tea grading method based on charcoal roasting parameters according to claim 1, characterized in that, Hyperspectral imaging information of tea leaves within a carbon-roasting space is acquired, Maillard reaction characteristic spectra within a preset wavelength range are extracted, and spectral parameters of the tea leaves are calculated based on the Maillard reaction characteristic spectra, including: The reflectance spectrum data of the tea surface is collected by a hyperspectral imaging system, and at least two characteristic bands are selected within a preset band range. The characteristic bands include a first reaction characteristic band and a second reaction characteristic band. The absorbance of the reflectance spectral data of the first and second reaction characteristic bands is converted to generate the time-absorbance change curves of the corresponding bands. The time-absorbance change curve is smoothed to eliminate noise interference, and feature points on the curve are extracted, including absorbance maxima, inflection points, and plateau intervals. Based on the location and trend of the feature points, the absorbance of the first reaction feature band and the absorbance of the second reaction feature band are integrated to generate a reaction progress index. The integrated ratio is calculated using the trapezoidal numerical integration method. Monitor the rate of change of the reaction process index over time, and mark the Maillard reaction critical inflection point when the rate of change of the reaction process index over time exceeds a set threshold. Based on the temperature and duration corresponding to the key inflection point of the Maillard reaction, the cumulative amount of melanoidins and the dynamic trend information of the aldehyde-ketone ratio are calculated. The cumulative amount of melanoidins is obtained by summing the absorbance changes after the key inflection point of the Maillard reaction, and the dynamic trend information of the aldehyde-ketone ratio is obtained by establishing a time series ratio curve. The spectral parameters of the tea leaves are generated based on the information on the cumulative amount of melanoidins, the dynamic trend of the aldehyde-ketone ratio, and the reaction process indicators.

3. The semi-refined tea grading method based on charcoal roasting parameters according to claim 1, characterized in that, Based on the tea spectral parameters and basic tea parameters, a variety adaptability model is dynamically constructed to obtain variety adaptability parameters, including: The variety information is matched with a preset variety feature library to obtain the benchmark substance conversion curve corresponding to the current variety information. The benchmark substance conversion curve includes the expected change range of aldehyde-ketone ratio and the threshold of melanoidin accumulation. Based on the dynamic trend information of the aldehyde-ketone ratio in the spectral parameters of the tea, the dynamic time-normalized distance between the real-time aldehyde-ketone ratio curve and the conversion curve of the reference substance is calculated, and the first matching degree index is generated. Furthermore, based on the cumulative amount of melanoidins in the spectral parameters of the tea leaves, the relative percentage deviation between the current cumulative amount of melanoidins and the expected value of the conversion curve of the reference substance is calculated, and a second matching index is generated. The first matching degree index and the second matching degree index are weighted and fused to obtain a comprehensive matching degree index. The first matching weight corresponding to the dynamic change trend information of the aldehyde-ketone ratio increases with the increase of carbon roasting temperature, and the second matching weight of the cumulative amount of melanoidins increases with the extension of duration. The comprehensive matching index is compared with the preset index threshold range, and temperature adjustment information is generated based on the comparison results. And, based on the moisture content information, information on the variation of carbon roasting amplitude is generated; Variety adaptability parameters are generated based on the temperature regulation information and the carbon roasting amplitude change information.

4. The semi-refined tea grading method based on charcoal roasting parameters according to claim 1, characterized in that, The thermodynamic characteristics, Maillard reaction characteristic spectrum, and variety adaptability parameters are fused in a multimodal manner to generate a carbon roasting quality prediction vector. This multimodal fusion includes temperature-material transformation co-splicing and spatial inhomogeneity compensation, comprising: Based on the carbon roasting uniformity index, a temperature field compensation coefficient is generated, and the dynamic change trend information of the aldehyde-ketone ratio in the Maillard reaction characteristic spectrum is regionally corrected, which is denoted as the correction trend information. The temperature regulation information in the variety adaptability parameters is coupled with the duration of key temperature to obtain a time weighting factor. The time weighting factor is used as the weight value of Maillard reaction characteristic spectra at different carbon roasting stages and weighted to obtain a weighted reaction characteristic spectrum. Furthermore, based on the moisture content information, the temperature field compensation coefficient and time weighting factor are dynamically scaled to generate a moisture content adaptation adjustment coefficient. A spatial grid mapping relationship is constructed, and the correction trend information, weighted response characteristic spectrum and moisture content adaptation adjustment coefficient are integrated according to spatial location. Furthermore, the abnormal grid points in the spatial grid mapping relationship are subjected to neighborhood smoothing processing. The smoothing processing retains the temperature gradient characteristics while eliminating isolated noise points, thereby obtaining spatial grid information. The spatial grid information is expanded along the time dimension to generate a carbon roasting quality prediction vector that includes temperature-material transformation synergistic features and spatial distribution features.

5. The semi-refined tea grading method based on charcoal roasting parameters according to claim 1, characterized in that, The carbon roasting quality prediction vector is input into the grading decision tree model to obtain preliminary grade determination results. The preliminary grade determination results include secondary, excellent, and special grades, including: A hierarchical decision tree model is established, which includes a temperature-matter conversion synergistic feature branch, a spatial distribution feature branch, and a Maillard reaction integrity branch. The temperature-material conversion synergistic feature value in the carbon roasting quality prediction vector is input into the temperature-material conversion synergistic feature branch. When the temperature-material conversion synergistic feature value is lower than the first feature threshold, it is determined to be secondary. The spatial distribution features in the carbon roasting quality prediction vector are input into the spatial distribution feature branch to calculate the quality uniformity index. When the quality uniformity index is lower than the second feature threshold, it is judged as excellent. The Maillard reaction feature in the carbon roasting quality prediction vector is input into the Maillard reaction integrity branch. When the product of the slope of the aldehyde-ketone ratio curve and the cumulative amount of melanoidins exceeds the third feature threshold, it is determined to be of the special grade. Based on the variety information in the basic parameters of tea, the corresponding first feature threshold, second feature threshold and third feature threshold are extracted from the preset grading standard library; Based on the actual carbon roasting time in the carbon roasting process parameters, time decay compensation is applied to the first feature threshold, the second feature threshold, and the third feature threshold. The output includes preliminary grade determination results, including secondary, superior, and special grades; The hierarchical decision tree model is configured to trigger a cascaded fire optimization process based on the clustering quality index, including: Calculate the clustering quality index of the preliminary grade determination results, wherein the clustering quality index is determined by the variance of the feature space distance between the top-grade samples and the superior samples; When the clustering quality index is lower than the preset clustering threshold, the cascade fire optimization process is activated.

6. The semi-refined tea grading method based on charcoal roasting parameters according to claim 1, characterized in that, The optimization strategy for the cascaded afterfiring is configured to combine and adjust the afterfiring time and temperature using a material conversion model, including: A material conversion kinetic model is established, wherein the input features of the material conversion kinetic model include the temperature-material conversion synergistic feature deviation and the spatial distribution feature non-uniformity in the carbon roasting quality prediction vector; The optimal combination of afterburning parameters is calculated based on the material transformation kinetics model, and the optimization strategy for cascade afterburning optimization is generated based on the optimal combination of afterburning parameters. For tea leaves initially classified as substandard, cascaded re-firing optimization is performed to obtain optimized tea leaves. The optimized tea leaves are then re-graded to obtain updated grading results, including: The carbon roasting process parameters and hyperspectral imaging information of the optimized tea leaves are re-acquired and denoted as optimized carbon roasting parameters and optimized imaging information. The carbon roasting quality prediction vector is then updated and denoted as optimized quality prediction vector. The optimized quality prediction vector is input into the hierarchical decision tree model for secondary judgment, generating the updated level judgment result.

7. The semi-refined tea grading method based on charcoal roasting parameters according to claim 1, characterized in that, Based on the preliminary and updated grading results, the final grading result for the tea is generated. This final grading result is then linked to the charcoal roasting process parameters and stored on the blockchain. A grading report is generated, including grading probability distribution, process defect location, and energy consumption assessment. The weighted average of the preliminary special grade proportion, preliminary excellent grade proportion, and preliminary secondary grade proportion in the preliminary grade determination results and the updated special grade proportion, updated excellent grade proportion, and updated secondary grade proportion in the updated grade determination results is calculated to generate the final special grade proportion, final excellent grade proportion, and final secondary grade proportion in the final grade determination results. Establish a correlation mapping relationship between the final grading results and the three-dimensional temperature field information of the roasting cage and the duration of key temperatures in the carbon roasting process parameters. The correlation mapping relationship includes the spatial correspondence between temperature field uniformity and quality grade distribution, and the temporal correspondence between the duration of key temperatures and the integrity of Maillard reaction. The final grading result, the carbon roasting process parameters, and the associated mapping relationship are used to generate a data block through hash operation, and the data block is written into the blockchain network for distributed storage. The data block includes a grade distribution probability matrix, process defect location coordinates, and carbon roasting energy consumption statistics. Based on the data blocks, a visualized hierarchical report is generated, which includes a three-dimensional temperature field distribution and a hierarchical distribution superimposed thermal map, a process defect improvement plan, and an energy efficiency optimization analysis report.

8. A computer-readable storage medium storing computer program instructions thereon, characterized in that, The computer program instructions, when executed by a processor, implement the method as described in any one of claims 1 to 7.

9. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in any one of claims 1 to 7.

Images