A method, medium, and system for optimizing coffee deep processing parameters

Abstract

The application provides a coffee deep processing process parameter optimization method, medium and system, and belongs to the technical field of coffee deep processing. The method comprises the following steps: firstly, collecting historical coffee processing data, including process parameters of each process and corresponding product quality indexes. Then, the correlation between the process parameters and the quality indexes is analyzed by using a Bayesian network and a mutual information algorithm, and a preliminary parameter-quality prediction model is established. Based on the prediction model, a set of candidate process parameter schemes is generated by using data augmentation and optimal transmission theory. Next, small batch experiments are performed on these parameter schemes to obtain corresponding actual quality index data. Finally, according to the experimental data, the optimal process parameters are selected as the target parameter output. The method combines machine learning, optimization algorithms and small batch experiment verification techniques, and solves the technical problem that the existing coffee deep processing process optimization cannot balance the optimization of multiple quality factors.

Description

Technical Field

[0001] This invention belongs to the field of coffee deep processing technology, and specifically relates to a method, medium and system for optimizing coffee deep processing process parameters. Background Technology

[0002] Traditional coffee processing methods rely heavily on baristas' experience, adjusting parameters at each stage—such as roasting temperature and time, extraction temperature and pressure—through repeated trials to achieve high-quality coffee that meets consumer demands. However, this experience-based approach has several drawbacks. First, determining optimal processing parameters requires extensive experimentation and evaluation, resulting in long optimization cycles and high costs. Second, even if the current optimal parameter combination is found, it's difficult to maintain this combination in the long term due to changes in raw materials, environment, and other factors. Furthermore, different consumer groups have varying preferences for coffee quality, making it difficult for a fixed set of processing parameters to cater to diverse needs. In other words, current coffee processing optimization faces the technical challenge of achieving a balanced optimization of multiple quality factors. Summary of the Invention

[0003] In view of this, the present invention provides a method, medium and system for optimizing coffee deep processing parameters, which can solve the technical problem that existing coffee deep processing optimization has difficulty in achieving balanced optimization of multiple quality factors.

[0004] This invention is implemented as follows:

[0005] The first aspect of the present invention provides a method for optimizing process parameters in coffee deep processing, comprising the following steps:

[0006] S10. Obtain the process parameters of multiple steps in the historical coffee deep processing process and the corresponding quality indicators of the finished coffee product.

[0007] S20. Using Bayesian networks and mutual information algorithms, calculate the correlation between each process parameter and each quality indicator.

[0008] S30. Based on the calculated correlation degree, a multi-objective prediction model for coffee quality is constructed. The random forest algorithm is used to establish the mapping relationship between parameters and quality indicators, which is denoted as the preliminary mapping.

[0009] S40. Based on the preliminary mapping, augmented data is obtained by using a data augmentation method according to the process parameters and their corresponding quality indicators of the finished coffee product.

[0010] S50. Apply optimal transmission theory to the augmented data to obtain a set of candidate process parameters;

[0011] S60. Cluster the candidate process array to obtain multiple cluster centers, and record the quality index data of the finished coffee product obtained by small-batch experiments on the process parameters corresponding to each cluster center as experimental data.

[0012] S70. Select the process parameters corresponding to the optimal experimental data as the optimized target process parameters and output them.

[0013] Based on the above technical solution, the method for optimizing coffee deep processing parameters of the present invention can be further improved as follows:

[0014] The process parameters include baking process parameters, extraction process parameters, and post-processing process parameters.

[0015] Furthermore, the baking process parameters include baking temperature, baking time, baking machine speed, feed rate, and hot air speed.

[0016] Furthermore, the extraction process parameters include extraction temperature, extraction pressure, extraction time, water-to-coffee ratio, and grinding particle size.

[0017] Furthermore, the post-processing parameters include cooling temperature, drying time, packaging gas composition, storage temperature, and storage humidity.

[0018] Furthermore, the quality indicators include acidity, bitterness, aroma concentration, body shape, and cleanliness.

[0019] Furthermore, the application of optimal transmission theory to the augmented data specifically involves: using the parameter space generated from all process parameters in the augmented data as the source distribution, and the index space generated from all quality indices in the augmented data as the target distribution; constructing a transmission cost function using Euclidean distance and parameter weights to reflect the relationship between process parameter adjustment and quality index improvement; solving for the optimal transmission plan to obtain the adjustment path from the current parameters to the target quality; selecting adjustment paths that meet the quality requirements and corresponding process parameter combinations based on the preset target quality index, conducting a feasibility assessment based on the preliminary mapping relationship, and finally obtaining a candidate process array.

[0020] Furthermore, the candidate process array is specifically a combination of multiple sets of process parameters that meet the quality objectives and are within the process feasibility range.

[0021] Specifically, step S10 involves obtaining the process parameters and corresponding quality indicators of the finished coffee product from multiple historical sets of coffee deep-processing processes. First, historical data needs to be collected and organized. Assume there are n sets of historical data, each containing m process parameters and k quality indicators. This data can be represented in matrix form: a process parameter matrix P and a quality indicator matrix Q. The elements p of the process parameter matrix P... ijq represents the j-th process parameter of the i-th data set, and is an element of the quality index matrix Q. ij Let represent the j-th quality index of the i-th data set. This historical data provides the foundation for subsequent correlation analysis between process parameters and quality indices.

[0022] Step S20 involves using a Bayesian network and mutual information algorithm to calculate the correlation between each process parameter and each quality indicator. First, a Bayesian network needs to be constructed. Assume there are m process parameter nodes X1, X2, ..., X... m and k quality index nodes Y1, Y2, ..., Y k The structure of a Bayesian network can be determined using structure learning algorithms (such as the K2 algorithm or a greedy search algorithm). For each pair of process parameters X... i and quality index Y j Calculate the mutual information I(X) between them. i ;Y j Calculating mutual information requires estimating the joint probability distribution p(x,y) and the marginal probability distributions p(x) and p(y), respectively. Here, kernel density estimation is used. The final result is a correlation matrix R, where r... ij =I(X) i ;Y j The correlation matrix () represents the correlation between the i-th process parameter and the j-th quality indicator. This correlation matrix reflects the strength of the correlation between each process parameter and the quality indicator, providing a foundation for the subsequent construction of a multi-objective prediction model.

[0023] The specific steps of step S30 are as follows: Based on the calculated correlation degree, construct a multi-objective prediction model for coffee quality, and use the random forest algorithm to establish a mapping relationship between parameters and quality indicators, denoted as the initial mapping. The random forest algorithm is used to establish the mapping relationship between process parameters and quality indicators. For each quality indicator Y... j Construct a Random Forest model (RF) j Each Random Forest model (RF) j It consists of T decision trees, and the prediction result of each decision tree is: The final prediction result is the average of all decision tree predictions: The construction process of each decision tree includes: drawing n samples with replacement from the original samples to form the bootstrap sample set, and randomly selecting m samples. try The process involves identifying key features, finding the optimal split point that maximizes the purity of the child nodes after the split, and repeating this process until a stopping condition is met. The splitting criterion for a decision tree can be the Gini index or information gain. The Random Forest (RF) model is then established through these steps. j This constitutes the process parameters to quality index Y. j The initial mapping relationship.

[0024] The specific steps of step S40 are as follows: Based on the preliminary mapping, augmented data is obtained by using data augmentation methods according to the process parameters and their corresponding quality indicators of the finished coffee product. To expand the original dataset, the following data augmentation methods can be used: 1) Adding Gaussian noise, that is, for each original sample (x... i ,y i ), generate new sample (x) i ′,y i '),in Linear interpolation, that is, for any two original samples (x,y) i ,y i ) and (x j ,y j ), generate new sample (x) new ,y new ), where x new =λx i +(1-λ)x j ,y new =λy i +(1-λ)y j ,λ∈[0,1];3)SMOTE algorithm, that is, for minority class samples, find its k nearest neighbors, and then randomly select a nearest neighbor x j Generate a new sample x new =x i +α(x j -x i ), α∈[0,1]. Using the above method, the original dataset can be expanded to N samples, forming an augmented dataset D. aug .

[0025] The specific steps of step S50 are as follows: Apply optimal transport theory to the augmented data to obtain a set of candidate process parameters. First, define the source distribution μ (process parameter space) and the target distribution v (quality index space) as empirical distributions. Then, define the transport cost function. Where w i It represents the weight of the i-th parameter. The optimal transmission problem can be expressed as: min γ∈Π(μ,v) ∫ X×Y c(x,y)dγ(x,y). To solve this problem, the Sinkhorn algorithm can be used. The iterative steps of the algorithm are: 1) Initialize u (0) =v (0) =1; 2) Repeat u for t=1,2,… (t) =μ. / K(v (t-1) ),v (t) =v. / (K) T u (t)Until convergence, where K = e -C / ∈ is the Gibbs kernel, and ∈ is the regularization parameter. The final transport plan is P. ij =u i K ij v j Based on the transmission plan, the adjustment path from the current parameters to the target quality can be obtained. The path that meets the preset quality target is selected, and a feasibility assessment is conducted based on the preliminary mapping relationship, ultimately yielding a set of candidate process parameters.

[0026] The specific steps of step S60 are as follows: Cluster the candidate process array to obtain multiple cluster centers, and obtain the quality index data of the finished coffee product obtained from small-batch experiments using the process parameters corresponding to each cluster center, recording these as experimental data. First, the K-means algorithm is used to cluster the candidate process parameter group. The objective function of the K-means algorithm is... Where K is the cluster number, It is the i-th sample belonging to the j-th cluster, μ j It is the center of the j-th cluster. The algorithm steps include: 1) randomly initializing K cluster centers; 2) repeating a) assigning each sample to the nearest cluster center, b) updating each cluster center until convergence. To determine the optimal number of clusters K, the silhouette coefficient can be used. Choose the K value that maximizes S as the optimal number of clusters. For each cluster center μ j Small-batch experiments were conducted to obtain experimental data E. j =(e j1 ,e j2 ,…,e jk ).

[0027] Step S70 involves selecting the optimal process parameters corresponding to the experimental data as the optimized target process parameters and outputting them. Through the preceding steps, multiple candidate process parameter groups and their corresponding experimental data have been obtained. Now, the optimal process parameters need to be selected as the final output. Candidate solutions can be evaluated based on the following criteria: 1) Whether all quality indicators of the finished coffee product meet the expected targets; 2) Whether the experimental values ​​of each quality indicator are relatively balanced, avoiding situations where some indicators are too high while others are too low; 3) Whether the process parameters are within a feasible range, avoiding problems such as equipment limitations or excessive costs. Taking all factors into consideration, the optimal process parameters corresponding to the experimental data are selected as the final optimized output.

[0028] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, perform the aforementioned method for optimizing coffee deep processing parameters.

[0029] A third aspect of the present invention provides a coffee deep processing parameter optimization system, wherein the system includes the aforementioned computer-readable storage medium.

[0030] Compared with existing technologies, the beneficial effects of the coffee deep processing parameter optimization method, medium, and system provided by this invention are:

[0031] 1. Improved accuracy and robustness of parameter-quality relationship modeling. Traditional linear or simple nonlinear models struggle to capture the complex parameter interactions in actual production processes. This invention, employing Bayesian networks and random forest algorithms, can better uncover these nonlinear relationships, thereby significantly improving prediction accuracy.

[0032] 2. This invention achieves synergistic optimization of multiple quality factors. Many existing methods optimize only a single quality indicator, failing to achieve a balanced improvement in multiple quality attributes such as acidity, bitterness, and aroma. Based on optimal transport theory, this invention integrates the optimization objectives of various quality indicators into a single framework, providing optimal process parameters that satisfy multiple quality goals.

[0033] 3. The optimization results are highly operable. This invention not only provides ideal target values ​​for quality indicators, but also, through cluster analysis and small-batch experiments, obtains specific and feasible process parameter schemes, providing direct technical support for production practice. Compared with empirical adjustments, this data analysis-based method is more scientific, systematic, and replicable.

[0034] In summary, this invention solves the technical problem of existing coffee deep processing technology optimization, which struggles to achieve a balanced optimization of multiple quality factors. Attached Figure Description

[0035] Figure 1 A flowchart of the method provided by the present invention;

[0036] Figure 2 A parameter-quality correlation matrix diagram;

[0037] Figure 3 This is a graph showing the results of K-means clustering. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0039] like Figure 1 The diagram shown is a flowchart of a method for optimizing coffee deep processing parameters provided by the present invention. This method includes the following steps:

[0040] S10. Obtain the process parameters of multiple steps in the historical coffee deep processing process and the corresponding quality indicators of the finished coffee product.

[0041] S20. Using Bayesian networks and mutual information algorithms, calculate the correlation between each process parameter and each quality indicator.

[0042] S30. Based on the calculated correlation degree, a multi-objective prediction model for coffee quality is constructed. The random forest algorithm is used to establish the mapping relationship between parameters and quality indicators, which is denoted as the preliminary mapping.

[0043] S40. Based on the initial mapping, augmented data is obtained by using data augmentation methods according to the process parameters and their corresponding quality indicators of finished coffee products.

[0044] S50. Apply optimal transmission theory to augmented data to obtain a set of candidate process parameters;

[0045] S60. Cluster the candidate process array to obtain multiple cluster centers. Obtain the process parameters corresponding to each cluster center and conduct small-batch experiments. Record the quality index data of the finished coffee product as experimental data.

[0046] S70. Select the process parameters corresponding to the optimal experimental data as the optimized target process parameters and output them.

[0047] The specific implementation methods of the above steps are described in detail below:

[0048] The specific implementation method of step S10 is as follows:

[0049] First, historical data needs to be collected and organized. Assume there are n sets of historical data, each set containing m process parameters and k quality indicators. This data can be represented in matrix form:

[0050] Process parameter matrix P:

[0051]

[0052] Among them, element p ij This represents the j-th process parameter of the i-th data set.

[0053] Quality Indicator Matrix Q:

[0054]

[0055] Among them, element q ij This represents the j-th quality index of the i-th data set.

[0056] These historical data provide a foundation for subsequent correlation analysis between process parameters and quality indicators.

[0057] The specific implementation method of step S20 is as follows:

[0058] First, a Bayesian network needs to be constructed. Assume there are m process parameter nodes X1, X2, ..., X... m and k quality index nodes Y1, Y2, ..., Y k The structure of a Bayesian network can be determined using structure learning algorithms (such as the K2 algorithm or the greedy search algorithm).

[0059] For each pair of process parameters X i and quality index Y j Calculate the mutual information I(X) between them. i ;Y j ):

[0060]

[0061] Where p(x,y) is X i and Y j The joint probability distributions of X, p(x) and p(y) are respectively... i and Y j The marginal probability distribution.

[0062] To calculate these probability distributions, kernel density estimation methods can be used:

[0063]

[0064] Where K(·) is the kernel function (such as the Gaussian kernel), h x and h y It is a bandwidth parameter.

[0065] Finally, we obtain a correlation matrix R:

[0066]

[0067] Where, r ij =i(X) i ;Y j The correlation matrix represents the degree of correlation between the i-th process parameter and the j-th quality indicator. This correlation matrix reflects the strength of the correlation between each process parameter and the quality indicator, providing a foundation for the subsequent construction of a multi-objective prediction model.

[0068] The specific implementation method of step S30 is as follows:

[0069] A random forest algorithm is used to establish the mapping relationship between process parameters and quality indicators. For each quality indicator Y... j Construct a Random Forest (RF) model j Each Random Forest model (RF) jIt consists of T decision trees, and the prediction result of each decision tree is: The final prediction result is the average of all decision tree predictions:

[0070]

[0071] The construction process of each decision tree is as follows:

[0072] The Bootstrap sample set is formed by randomly selecting n samples with replacement from the n samples.

[0073] Randomly select m try Features (usually) ).

[0074] Find the optimal split point among the selected features to maximize the purity of the child nodes after splitting.

[0075] Repeat steps 2 and 3 until a stopping condition is met (such as maximum depth or minimum number of leaf node samples).

[0076] The splitting criterion for a decision tree can be either the Gini index or information gain. Taking the Gini index as an example, the Gini index of node t is defined as:

[0077]

[0078] Where, p j It is the proportion of samples belonging to category j in node t.

[0079] Information gain is defined as the impurity of the parent node minus the weighted impurity of the child nodes.

[0080]

[0081] Where N is the number of samples in the parent node, N i It is the number of samples in the i-th child node.

[0082] The Random Forest (RF) model established through the above steps j This constitutes the process parameters to quality index Y. j The initial mapping relationship.

[0083] The specific implementation method of step S40 is as follows:

[0084] To augment the original dataset, the following data augmentation methods can be used:

[0085] Add Gaussian noise:

[0086] For each original sample (x) i ,y i ), generate new sample (x) i′,y i ′):

[0087]

[0088] Where, σ x and σ y This is a noise intensity parameter, which needs to be adjusted according to the actual situation.

[0089] Linear interpolation:

[0090] For any two original samples (x) i ,y i ) and (x j ,y j ), can generate new samples (x) new ,y new ):

[0091] x new =λx i +(1-λ)x j ;

[0092] y new =λy i +(1-λ)y j ;

[0093] Where λ∈[0,1] are the interpolation coefficients.

[0094] SMOTE (Synthetic Minority Over-sampling Technique):

[0095] For minority class samples, the SMOTE algorithm can be used to generate new samples. For sample x i Find its k nearest neighbors, and then randomly select a nearest neighbor x. j Generate a new sample x new :

[0096] x new =x i +α(x j -x i );

[0097] Where α∈[0,1] is a random number.

[0098] Using the methods described above, the original dataset can be expanded to N samples, forming an augmented dataset D. aug ={(x1,y1),(x2,y2),…,(x N ,y NThis newly generated data can further enrich the correlation between process parameters and quality indicators, laying the foundation for subsequent applications of optimal transport theory.

[0099] The specific implementation method of step S50 is as follows:

[0100] First, define the source distribution μ (process parameter space) and the target distribution ν (quality index space) as empirical distributions:

[0101]

[0102] Where, δ x This represents the Dirac delta function.

[0103] Then define the transmission cost function c(x,y):

[0104]

[0105] Where, w i It is the weight of the i-th parameter, which can be determined based on the previously calculated correlation.

[0106] The optimal transmission problem can be expressed as:

[0107] min γ∈Π(μ,v) ∫ X×Y c(x,y)dγ(x,y);

[0108] Where Π(μ,v) is the set of all joint distributions that transmit μ to v.

[0109] To solve this problem, the Sinkhorn algorithm can be used. First, the problem is discretized:

[0110]

[0111] st∑ i P ij =μ i ,∑ i P ij =v j ,P ij ≥0;

[0112] Where, C ij =c(x i ,y j ) is the cost matrix, P ij It's a transmission plan.

[0113] The iterative steps of the Sinkhorn algorithm are as follows:

[0114] Initialize u (0) =v (0) =1.

[0115] For t = 1, 2, ..., repeat the following steps until convergence:

[0116] u (t) =μ. / K(v (t-1) );

[0117] v (t) =v. / (K) T u (t) );

[0118] Where K = e -C / ∈ It is the Gibbs kernel, ∈ is the regularization parameter, and . / represents element-wise division.

[0119] The final transmission plan is P ij =u i K ij v j Based on the transmission plan, the optimal adjustment path from the current parameters to the target quality can be obtained. The path that meets the preset quality target is selected, and a feasibility assessment is conducted based on the preliminary mapping relationship, ultimately yielding a set of candidate process parameters.

[0120] The specific implementation method of step S60 is as follows:

[0121] First, the K-means algorithm is used to cluster the candidate process parameter groups, resulting in multiple cluster centers. The objective function of the K-means algorithm is:

[0122]

[0123] Where K is the number of clusters, It is the i-th sample belonging to the j-th cluster, μ j It is the center of the j-th cluster.

[0124] The steps of the K-means algorithm are as follows:

[0125] Randomly initialize K cluster centers μ1, μ2, ..., μ K .

[0126] Repeat the following steps until convergence:

[0127] For each sample x i Assign it to the nearest cluster center:

[0128] c i =argmin j ||x i -μ j || 2 ;

[0129] Update each cluster center:

[0130]

[0131] Where, C j It is the set of all samples belonging to the j-th cluster.

[0132] To determine the optimal number of clusters K, the silhouette coefficient can be used:

[0133]

[0134] Where a(i) is the average distance between sample i and other samples in the same cluster, and b(i) is the average distance between sample i and the nearest sample in another cluster.

[0135] The overall silhouette coefficient is the average of the silhouette coefficients of all samples.

[0136]

[0137] Choose the K value that maximizes S as the optimal number of clusters.

[0138] For each cluster center μ j Small-batch experiments were conducted to obtain experimental data E. j =(e j1 ,e j2 ,…,e jk ), where e ji This represents the experimental value of the i-th quality index under the process parameters corresponding to the j-th cluster center. Step S70 compares the performance of the quality indexes of different experimental data and selects the process parameters corresponding to the optimal experimental data as the final optimization result output.

[0139] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, perform the aforementioned method for optimizing coffee deep processing parameters.

[0140] A third aspect of the present invention provides a coffee deep processing parameter optimization system, wherein the system includes the aforementioned computer-readable storage medium.

[0141] Specifically, the principle of this invention is to fully utilize historical production data and construct a multi-objective prediction model from process parameters to product quality through advanced data analysis and modeling techniques such as Bayesian networks, random forests, and optimal transport theory, and then find the optimal combination of process parameters based on this model.

[0142] Specifically, in step S10, process parameter data and corresponding quality index data from multiple historical coffee deep processing processes were collected to form a process parameter matrix P and a quality index matrix Q. This historical data provides the foundation for subsequent correlation analysis and modeling.

[0143] In step S20, a Bayesian network structure learning algorithm was used to construct the correlation network topology between process parameters and quality indicators. Then, mutual information was used to quantitatively evaluate the correlation degree between each pair of parameters and quality indicators, resulting in the correlation degree matrix R. The purpose of this step is to uncover the potential dependencies between parameters and quality indicators, laying the foundation for building a predictive model.

[0144] Based on the correlation information obtained in step S20, step S30 uses the random forest algorithm to establish a multi-objective prediction model from process parameters to quality indicators. Random forest is an ensemble learning method composed of multiple decision trees, which can effectively fit complex nonlinear relationships. By training the random forest model, a preliminary parameter-quality mapping relationship can be obtained.

[0145] To further enrich this mapping relationship, step S40 employs data augmentation methods, utilizing techniques such as Gaussian noise, linear interpolation, and SMOTE to expand the original dataset and generate more samples. These additional samples enable the prediction model to better summarize the complex relationship between parameters and quality.

[0146] With the preliminary prediction model and augmented data, step S50 applies optimal transport theory to find the best adjustment path from the current parameters to the target quality. The optimal transport problem seeks the optimal transport plan between the source distribution (process parameter space) and the target distribution (quality index space) to minimize the total transport cost. By solving this optimization problem, the optimal combination of process parameters that satisfies each quality objective can be obtained.

[0147] Finally, step S60 performs cluster analysis on these candidate parameter schemes to identify key cluster centers and conducts small-batch experimental verification. The purpose of this step is to further confirm the final optimal process parameters and provide feasible optimization results for production practice.

[0148] In summary, the core innovations of this invention are: first, the use of advanced data analysis methods such as Bayesian networks and random forests can better uncover complex parameter-quality relationships; second, the introduction of optimal transport theory enables the synergistic optimization of multiple quality factors; and third, through data augmentation and experimental verification, a final optimization result with strong operability is presented.

[0149] To better understand and implement this invention, a specific application scenario is provided below: The R&D team of a coffee company has been committed to improving the quality of its coffee products to meet the needs of different consumer groups. The company's production management department collected and organized historical data on coffee deep processing over the past five years, including process parameters and corresponding quality index data for each process such as roasting, extraction, and post-processing. Through analysis and modeling of this data, the R&D team decided to adopt the coffee deep processing process parameter optimization method proposed in this invention in order to find a better process parameter formula.

[0150] First, in step S10, the R&D team compiled 100 sets of data from the past 5 years. Each set of data contains 5 process parameters and 4 quality indicators, forming the process parameter matrix P and the quality indicator matrix Q as shown in Table 1.

[0151] Table 1 Overview of Historical Production Data

[0152]

[0153] In step S20, the R&D team first constructed a Bayesian network structure containing 5 process parameter nodes and 4 quality indicator nodes. The K2 algorithm was used to learn this network structure and determine the dependencies between the nodes.

[0154] Then, the R&D team calculated the mutual information value between each pair of process parameters and quality indicators, obtaining the correlation matrix R shown in Table 2, and established the corresponding parameter-quality correlation matrix, as shown in Table 2. Figure 2 As shown, baking temperature and time have a strong positive correlation with acidity and bitterness, extraction temperature and pressure have a strong positive correlation with aroma concentration and body shape, while packaging humidity has a relatively weak correlation with each quality indicator.

[0155] Table 2 Parameter-Quality Correlation Matrix

[0156] acidity Suffering Aroma concentration posture Baking temperature 0.78 0.71 0.46 0.39 Baking time 0.72 0.65 0.41 0.35 Extraction temperature 0.52 0.48 0.81 0.76 Extraction pressure 0.47 0.43 0.75 0.71 Packaging humidity 0.29 0.26 0.22 0.19

[0157] Based on the above analysis results, the R&D team constructed four random forest models in step S30 to predict acidity, bitterness, aroma concentration, and body shape, respectively. Taking acidity prediction as an example, the specific construction process of the random forest model is as follows:

[0158] First, randomly select 100 samples with replacement from the initial 100 samples to form the bootstrap sample set. Then, randomly select three features (baking temperature, baking time, and extraction temperature) for splitting. Among the selected features, find the optimal split point that maximizes the purity of the child nodes after splitting. Repeat the above steps until the maximum tree depth of 5 or the minimum number of leaf node samples of 5 is reached.

[0159] This random forest model consists of 100 such decision trees, and each tree makes a prediction. The final prediction is the average of all tree predictions:

[0160]

[0161] Similarly, the R&D team constructed three random forest models to predict bitterness, aroma concentration, and body shape. By training these models, a preliminary mapping relationship from process parameters to quality indicators was obtained.

[0162] To further enrich this mapping relationship, the research team employed data augmentation techniques in step S40. First, Gaussian noise with a mean of 0 and a standard deviation of 0.2 was added to the original 100 samples, generating 100 new samples. Then, 10 pairs were randomly selected from the original samples, and linear interpolation was used to generate 10 new samples. Finally, for minority class samples (such as samples with acidity below 4.0), the SMOTE algorithm was used to generate 20 new samples.

[0163] Combining the above three methods, the research team obtained 230 augmented samples, forming a new dataset D. aug These new samples further enrich the relationship between parameters and quality, laying the foundation for subsequent optimal transmission optimization.

[0164] With the preliminary predictive model and augmented data, the R&D team applied optimal transport theory in step S50 to find the best adjustment path from the current parameters to the target quality. First, the process parameter space and quality index space were defined as empirical distributions μ and v, respectively:

[0165]

[0166] Where, x i The process parameters for the i-th sample are y. i It is the quality index of the i-th sample.

[0167] Then, the transmission cost function is defined as:

[0168] c(x,y)=0.4(x1-y1) 2 +0.3(x²-y²) 2 +0.2(x³-y³) 2 +0.1(x⁴-y⁴) 2 +0.05(x5-y5) 2 ;

[0169] The weights of each parameter are determined based on the correlation matrix R obtained in step S20.

[0170] Finally, the R&D team used the Sinkhorn algorithm to solve the optimal transmission problem, obtaining the best adjustment path from the current parameter combination to the target quality index. Based on this path, the R&D team selected three candidate parameter schemes that meet the preset quality target and are within the process feasibility range, as shown in Table 3.

[0171] Table 3. Three sets of candidate process parameters

[0172]

[0173] In step S60, the R&D team performed K-means clustering analysis on these three sets of candidate parameters, setting the number of clusters to 2. Figure 3 As shown, Figure 3 In this context, Centroids represents the cluster centers.

[0174] After iterative calculations, two cluster centers were finally obtained, and the specific parameters are shown in Table 4.

[0175] Table 4. Process parameters for the two cluster centers

[0176]

[0177] For these two cluster centers, the R&D team conducted small-batch experiments to test the corresponding quality index data, and the results are shown in Table 5.

[0178] Table 5. Experimental data on the quality indicators of the two cluster centers.

[0179] Cluster Center acidity Suffering Aroma concentration posture 1 4.0 4.3 4.2 4.6 2 4.1 4.4 4.3 4.7

[0180] By comparing these two sets of experimental data, the research team determined that cluster center 2 was the optimal parameter scheme, namely:

[0181] Baking temperature 211℃, baking time 20min, extraction temperature 96℃, extraction pressure 1.4MPa, packaging humidity 58%.

[0182] This optimization result not only met the R&D team's overall target requirements for each quality indicator (acidity ≥ 4.0, bitterness ≥ 4.3, aroma concentration ≥ 4.2, body ≥ 4.5), but also verified the feasibility of the process. Compared with previous empirical adjustment methods, this optimization method based on Bayesian networks, random forests, and optimal transport theory has significant advantages in both theoretical analysis and practical application.

[0183] It can more accurately uncover complex parameter-quality relationships, improving the accuracy and robustness of prediction models.

[0184] It achieves synergistic optimization of multiple quality indicators and provides the optimal combination of process parameters to meet various quality objectives.

[0185] The optimization results are highly operable and provide direct guidance for production practice.

[0186] This method has good versatility and scalability, and is not only applicable to the coffee industry, but can also be extended to the optimization of processing technology for other agricultural products.

[0187] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing coffee process parameters for coffee intensive processing, characterized in that, The method comprises the following steps: S10, obtaining historical process parameters of multiple processes in a coffee deep processing process and corresponding quality indexes of coffee finished products; S20, calculating the correlation degree of each process parameter and each quality index by using a Bayesian network and a mutual information algorithm; S30, constructing a coffee quality multi-objective prediction model based on the calculated correlation degree, establishing a mapping relationship between the parameters and the quality indexes by using a random forest algorithm, and recording the mapping relationship as a preliminary mapping; S40, obtaining augmented data by using a data augmentation method based on the preliminary mapping and the process parameters and corresponding quality indexes of coffee finished products; S50, obtaining candidate process parameter groups by applying optimal transport theory to the augmented data; S60, clustering the candidate process parameter groups to obtain multiple cluster centers, and obtaining quality index data of coffee finished products corresponding to the process parameters of each cluster center obtained through small-batch experiments, which is recorded as experimental data; S70, selecting the process parameters corresponding to the optimal experimental data as the optimized target process parameters and outputting the process parameters.

2. A coffee process parameter optimization method according to claim 1, characterized in that, The process parameters include roasting process parameters, extraction process parameters, and post-processing process parameters.

3. A coffee process parameter optimization method according to claim 2, characterized in that, The roasting process parameters include roasting temperature, roasting time, roasting machine rotation speed, feed amount, and hot air speed.

4. A coffee process parameter optimization method according to claim 3, characterized in that, The extraction process parameters include extraction temperature, extraction pressure, extraction time, water-to-coffee ratio, and grinding particle size.

5. A coffee process parameter optimization method according to claim 4, characterized in that, The post-processing process parameters include cooling temperature, drying time, packaging gas composition, storage temperature, and storage humidity.

6. A coffee process parameter optimization method according to claim 5, characterized in that, The quality indexes include acidity, bitterness, aroma concentration, body, and cleanliness.

7. A coffee process parameter optimization method according to claim 6, characterized in that, The optimal transport theory applied to the augmented data is as follows: the parameter space generated by all the process parameters in the augmented data is regarded as the source distribution, and the index space generated by all the quality indexes in the augmented data is regarded as the target distribution; a transmission cost function is constructed by using the Euclidean distance and parameter weight to reflect the relationship between process parameter adjustment and quality index improvement; an optimal transport plan is solved to obtain an adjustment path from the current parameters to the target quality; according to the preset target quality index, the adjustment path and the corresponding process parameter combination that meet the quality requirements are selected, and a feasibility evaluation is performed in combination with the preliminary mapping relationship, and finally the candidate process parameter groups are obtained.

8. A coffee process parameter optimization method according to claim 7, characterized in that, The candidate process parameter groups are specifically combinations of process parameters that meet the quality target and are within the feasible range of the process.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores program instructions, and the program instructions run in the computer to execute the coffee deep processing process parameter optimization method of any one of claims 1-8.

10. A coffee intensive processing process parameter optimization system characterized by, The computer readable storage medium comprises the computer readable storage medium of claim 9.

Images