A dynamic regulation method for tea blending scheme based on fuzzy rule improvement
Patent Information
- Application Number
- CN202610533085.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-21
- Publication Date
- 2026-08-18
AI Technical Summary
为实现这种物理兼容性约束，而现有方法通常直接剔除违反约束的方案，或施加一个极大的固定惩罚值；缺少对轻微偏离、严重违反等不同状态的梯度响应能力
[0013] According to the present invention, when generating tea blending schemes using the particle swarm optimization (PSO) algorithm, intelligent dynamic trade-offs among multiple objectives can be achieved, improving the accuracy of blending scheme optimization; and the feasibility subdivision of blending schemes overcomes the drawbacks of coarse constraint processing, achieving refined feasibility of schemes and improving practicality.
Smart Images
Figure CN122596455A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of tea processing technology, and more specifically, to a dynamic control method for tea blending schemes based on fuzzy rules. Background Technology
[0002] Tea blending is a crucial step in tea processing, aiming to optimize the sensory quality, physicochemical stability, and other indicators of the finished tea by scientifically proportioning different raw tea leaves while controlling costs. Traditional tea blending methods rely heavily on the personal experience of blenders, resulting in strong subjectivity, poor consistency, and low efficiency. With technological advancements, intelligent optimization algorithms have been gradually introduced into the field of tea blending. Among them, the Particle Swarm Optimization (PSO) algorithm, due to its strong global search capability and few parameters, has been used to generate tea blending schemes. For example, the invention patent "A Method for Determining Tea Blending Strategy Based on Particle Swarm Optimization Algorithm" (application number 202511931668.2) calculates the quality score of the raw tea leaves. Cost rating Stability score The matching schemes are scored and selected by combining a comprehensive evaluation function with fixed weights.
[0003] However, the fixed weight formula used in the Particle Swarm Optimization (PSO) algorithm cannot adapt to the dynamic conflicts and uncertainties among multiple objectives (such as cost, quality, and stability) during the blending process. For example, when the cost of the blending scheme is too high, the fixed weight may still overemphasize the quality advantage, causing the algorithm to get stuck in a high-cost local optimum and failing to guide the search direction to adjust to a more economical region. Moreover, the singularity of the fixed weight function causes the PSO algorithm to quickly converge to a seemingly excellent but obviously weak local optimum in the early stages of iteration, making it impossible to achieve intelligent adjustment of the search. When the scheme results in a contradictory state of high cost and high quality, the fixed weight will still overemphasize the quality advantage, causing the algorithm to continue searching in the high-cost region. In addition, tea blending has strict physical compatibility constraints, that is, it is necessary to determine the two raw teas with the largest and smallest physical data as raw tea 1 and raw tea 2 from the tea raw materials. During the initial screening, the following must be satisfied simultaneously: |Moisture content of raw tea 1 - Moisture content of raw tea 2| ≤ 2% and |pH value of raw tea 1 - pH value of raw tea 2| ≤ 0.5. To achieve this physical compatibility constraint, existing methods typically either directly eliminate solutions that violate the constraint or impose a very large fixed penalty value; they also lack the ability to provide gradient responses to different states such as slight deviations and severe violations. These problems all reflect the inadequacy of dynamic control capabilities in tea blending design.
[0004] Therefore, a technical solution is needed to address the problems of rigid search direction, coarse constraint handling, and premature convergence in existing particle swarm optimization algorithms, so as to achieve dynamic control of tea blending design and obtain efficient and reliable blending schemes. Summary of the Invention
[0005] To achieve the above objectives, this application provides a dynamic control method for tea blending schemes based on fuzzy rules, comprising the following steps: Load the tea blending formula library and obtain the formula, tea quality, and physical compatibility of the target tea blending scheme from the tea blending formula library; among them, the target tea blending scheme is a number of blending schemes that can be achieved based on the actual available raw materials, and it is necessary to evaluate and select the best among the multiple blending schemes; Obtain the tea cost of the target tea blending scheme; tea cost, tea quality, and physical compatibility constitute the attribute set of the target tea blending scheme; Load the blending dynamic control model, input the attribute set into the blending dynamic control model, and obtain the weighted reorganization of the target tea blending scheme; the weighted reorganization of the target tea blending scheme consists of weighted variables of quality, cost, and stability. , , constitute; The score of the target tea blending scheme is calculated based on the weighted reorganization, thereby enabling dynamic adjustment of the tea blending scheme.
[0006] Before loading the tea blending formula library, the tea blending formula library is constructed. The tea blending formula library is used to systematically store and manage historical blending strategies and corresponding tea quality and physical compatibility. Tea quality includes the overall fitness value of each historical blending strategy; Physical compatibility includes the maximum / minimum moisture content and pH value of the raw tea in each solution.
[0007] Furthermore, before loading the dynamic control model for assembly, a dynamic control model for assembly is constructed; The construction of the dynamic control model for the assembly includes the following steps: Define fuzzy rules for regulatory indicators; the regulatory indicators include: tea cost, tea quality, and physical compatibility; Define weight adjustment factors, including: cost weight adjustment factor. Quality weight adjustment factor Stability weight adjustment factor ; A fuzzy control rule is determined to define the weight adjustment factor; the fuzzy control rule consists of m rules. Based on fuzzy control rules, the weight adjustment factors are defuzzified to output the weight variables of quality, cost, and stability. , , .
[0008] Among them, the fuzzy rules for defining the cost of tea include: The cost evaluation level of tea is defined as having multiple levels, including the lowest level, the highest level, and one or more intermediate levels; Define standard values at multiple levels, namely, the minimum standard value, the maximum standard value, and one or more intermediate standard values; Calculate the proportion of tea cost = Actual cost / Budgeted cost; Define the membership function for tea cost. Triangle-trapezoidal mixed membership function; membership degree function Based on the proportion of tea cost It is expressed as follows: When the proportion of tea cost When the value is less than the minimum standard value or greater than the maximum standard value, the membership function Represented as a trapezoidal function, it is expressed as: =1; otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are the coefficients and the constant term, respectively. Based on the membership function, the output cost level fuzzy set contains the membership function value corresponding to each fuzzy category. Cost activation intensity.
[0009] Fuzzy rules for defining tea quality include: The quality of tea is defined as having multiple grades, including the worst grade, the best grade, and one or more intermediate grades. Define standard values at multiple levels, namely, the minimum standard value, the maximum standard value, and one or more intermediate standard values; Calculating the difference in tea quality =Target quality score / Actual quality score; Define the membership function of tea quality Triangle-trapezoidal mixed membership function; membership degree function Based on the differences in tea quality It is expressed as follows: When the quality difference of tea When the membership function is below the minimum standard value or above the maximum standard value, Represented as a trapezoidal function, it is expressed as: =1; otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are the coefficients and the constant term, respectively. Based on the membership function, the output fuzzy set of quality evaluation levels, and the membership function value corresponding to each fuzzy classification. The quality activation intensity.
[0010] The fuzzy rules defining the constraint deviation of tea physical compatibility include: Set physical compatibility constraints for tea: moisture content difference = Moisture content - Moisture content, moisture content difference ≤ Moisture content constraint value, pH difference = pH value - pH value, pH difference ≤pH value constraint; The constraint calculation results are normalized, and the constraint violation rate is designed. for: , ; Define the overall violation degree of the fitting constraint. , is represented as: = ; Calculate constraint deviation =Overall violation of matching constraints ; Define membership function Using a triangular-trapezoidal hybrid membership function to assemble the overall violation degree of constraints. Given the conditions, complete the fuzzy set classification of tea constraint deviation. when When the membership function meets or significantly deviates from the compliance criterion threshold, it exhibits a trapezoidal function, expressed as follows: Otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are the coefficients and the constant term, respectively. Calculate the membership function value The activation strength is the stability factor.
[0011] Furthermore, the fuzzy control rules for the weight adjustment factors include a quality-first mode and a cost-first mode; The fuzzy control rules under the quality-first mode refer to: increasing the cost weight and decreasing the quality weight to control the budget when the cost is high; increasing the stability weight to correct feasibility when there is a deviation from the constraint; completely rejecting the region when there is a serious deviation; and increasing the quality weight in the opposite direction when the cost is low and the quality is low, so as to guide the particle swarm to adaptively converge to the region that is economically feasible and of balanced quality. The fuzzy control rules under the cost-first model refer to: when the cost is low and the quality is low, the weight of quality is greatly increased, and cost is sacrificed to adjust constraints when necessary; when the cost is low and the quality is moderate, the cost advantage is strengthened, and quality is sacrificed to ensure feasibility when deviation occurs; when the cost is high and the quality is high, the cost is compressed to the extreme, and excess quality can be sacrificed to cooperate with constraint adjustment.
[0012] Furthermore, weight variables , , The calculation methods include: , , , In the formula, For different rules The corresponding activation level is represented as: ,in, , Calculating the score for the target tea blending scheme involves the following steps: Define dynamic cost weights Dynamic weighting of quality Stability dynamic weights They are respectively , , , In the formula, Weighted based on cost; As a basic weight for quality; As the basic weight for stability; The score of the target tea blending scheme The calculation method is as follows: In the formula, , , The scores are: raw tea quality score, cost score, and stability score.
[0013] According to the present invention, when generating tea blending schemes using the particle swarm optimization (PSO) algorithm, intelligent dynamic trade-offs among multiple objectives can be achieved, improving the accuracy of blending scheme optimization; and the feasibility subdivision of blending schemes overcomes the drawbacks of coarse constraint processing, achieving refined feasibility of schemes and improving practicality. Attached Figure Description
[0014] Figure 1 This is a flowchart illustrating the steps of a dynamic control method for tea blending schemes provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the cost membership function provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the quality membership function provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the constraint deviation membership function provided in an embodiment of the present invention. Detailed Implementation
[0015] This invention provides a dynamic control method for tea blending schemes based on fuzzy rules. By establishing a dynamic control model for blending, it solves the problems of rigid search direction, coarse constraint processing, and premature convergence caused by fixed weight strategies in the prior art. It also guides the particle swarm algorithm to bypass coupling problems such as high cost and constraint violation, thereby realizing dynamic control of the blending scheme.
[0016] The specific implementation of the present invention will now be described in detail with reference to the accompanying drawings.
[0017] The dynamic control method for tea blending schemes based on fuzzy rules provided by this invention includes the following steps: Figure 1 As shown, it includes: Step S100: Construct a tea blending formula library; the tea blending formula library is used to systematically store and manage historical blending strategies and corresponding tea quality and physical compatibility.
[0018] 1) The historical blending strategy is the raw material combination scheme optimized and converged by the PSO algorithm in the existing technology. Each strategy record contains a unique identifier, blending mode, list of raw material types and ratio vector of each element, and completely retains the optimal position information of the algorithm particles. It can be used for direct reuse or as the initial population for iterative optimization. Among them, the blending modes include cost priority, rich and mellow, fresh and crisp and high aroma. 2) Tea quality includes the overall adaptability value of each historical blending strategy, including: rich and mellow quality score, fresh and crisp quality score, and highly aromatic quality score; 3) Physical compatibility includes the maximum / minimum moisture content and pH value of the raw tea in each scheme, and indicates whether the range meets the threshold (e.g., moisture content difference ≤ 2% and pH difference ≤ 0.5).
[0019] The tea blending formula library is used to store blending strategies generated through experience and algorithm optimization, enabling the reusability and iterative nature of blending knowledge.
[0020] Step S110: Load the tea blending formula library and obtain the formula, tea quality and physical compatibility of the target tea blending scheme from the tea blending formula library; In this step, the target tea blending scheme is one of several blending schemes that can be achieved based on the actual available raw materials, and it is necessary to evaluate and select the best among the multiple blending schemes.
[0021] Step S120: Obtain the tea cost of the target tea blending scheme; in this step, the current cost can be calculated based on the current price of raw materials as the tea cost; Since raw material values vary at different times, this step calculates the current cost by combining the current raw material price or the purchase price of materials in storage. The current raw material price or the purchase price of materials in storage can be obtained through the company's internal management system or procurement system. Similarly, if the quality and physical compatibility of the raw materials for the target tea blending scheme change, these can be retrieved again in this step.
[0022] At this point, the tea cost, tea quality, and physical compatibility of the target tea blending scheme have been obtained, forming the attribute set of the target tea blending scheme. The attribute values are the specific numerical values of tea cost, tea quality, and physical compatibility.
[0023] Step S130: Load the blending dynamic control model and obtain the weighted reorganization of the target tea blending scheme; The dynamic control model includes input variable fuzzification, fuzzy rule definition, and fuzzy rule defuzzification; inputting an attribute set into the dynamic control model yields a weighted reorganization composed of the weights corresponding to each attribute output by the dynamic control model.
[0024] Step S131: First, construct the dynamic control model for the assembly, including the following steps: Step S132: Define the fuzzy rules for the control indicators; In this step, the control indicators are the attributes of the target tea blending scheme, including tea cost, tea quality, and physical compatibility; fuzzy rules convert the specific values of the attributes into fuzzy results with expert experience.
[0025] 1) The fuzzy rules for defining the cost of tea specifically include: The cost evaluation level of tea is defined as having multiple levels, including the lowest level, the highest level, and one or more intermediate levels; Define standard values at multiple levels, namely, the minimum standard value, the maximum standard value, and one or more intermediate standard values; Calculate the proportion of tea cost = Actual cost / Budgeted cost; Define the membership function for tea cost. Triangle-trapezoidal mixed membership function; membership degree function Based on the proportion of tea cost It is expressed as follows: When the proportion of tea cost When the value is less than the minimum standard value or greater than the maximum standard value, the membership function Represented as a trapezoidal function, it is expressed as: =1; otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are coefficients and constants, which can be set according to the actual application. By using the membership function and taking the proportion of tea cost as known conditions, a fuzzy set classification of tea cost evaluation level is achieved.
[0026] The present invention provides specific examples as follows: The levels of tea cost evaluation are defined as [very low, low, medium, high, very high]; Define membership function , ; specifically as Figure 2 As shown: Based on the evaluation level of tea cost, the standard values for the standard levels of tea cost ratio are defined as 0.7, 0.85, 1.0, 1.15 and 1.3 respectively; The membership function is as follows: A is used to evaluate the cost level of tea at a very low level (VL), and the membership function is a trapezoidal function; when Membership function ; when Membership function .
[0027] B is used to evaluate the level of tea cost as low (L), and the membership function is a trigonometric function; when Membership function ; when Membership function .
[0028] C is used to evaluate the level of tea cost as medium (M), and the membership function is a trigonometric function; when Membership function ; when Membership function .
[0029] D is used to evaluate the level of tea cost as a high (H) grade, and the membership function is a trigonometric function; when Membership function ; when Membership function .
[0030] E is used to evaluate the cost level of tea as a very high level (VH), and the membership function is a trapezoidal function; when Membership function ; when Membership function .
[0031] In practical applications, the cost ratio of each matching scheme is obtained based on the membership function of ABCDE. Calculate the membership function values of the corresponding ABCDE, and output the cost level fuzzy set as a vector {extremely low:} ;Low: ;medium: ;high: Extremely high: }, the membership function value corresponding to each fuzzy classification The activation intensity is determined by cost and is used to guide subsequent fuzzy instructions for fuzzy control.
[0032] 2) Defining the fuzzy rules for tea quality scoring includes the following steps: First, the quality of tea is defined into multiple levels, including the worst level, the best level, and one or more intermediate levels. Define standard values at multiple levels, namely, the minimum standard value, the maximum standard value, and one or more intermediate standard values; Calculating the difference in tea quality =Target quality score / Actual quality score; Define the membership function of tea quality Triangle-trapezoidal mixed membership function; membership degree function Based on the differences in tea quality It is expressed as follows: When the quality difference of tea When the value is below the minimum standard value or above the maximum standard value, the membership function Represented as a trapezoidal function, it is expressed as: =1; otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are coefficients and constants, which can be set according to the actual application.
[0033] This invention provides specific implementation examples. Figure 3 As shown: The quality evaluation levels for tea are defined as [Very Poor, Poor, Average, Good, Excellent]. : A is the grade used to evaluate the quality level of tea, represented by the range (VP), which is expressed as a trapezoidal function. Membership function ; when Membership function .
[0034] B is used to evaluate the grade of tea quality, which is represented by the difference (P), and can be expressed as a triangular function. Membership function ; when Membership function .
[0035] C is used to evaluate the quality level of tea as medium (M), which is represented by a trigonometric function. Membership function ; when Membership function .
[0036] D is used to evaluate the quality level of tea as "good" (G), which is represented by a triangular function. Membership function ; when Membership function .
[0037] E, used to evaluate the quality level of tea, is rated as Excellent (VG), and is represented by a trapezoidal function. Membership function ; when Membership function .
[0038] Based on the membership functions of ABCDE, the quality gap of each matching scheme is obtained, the membership function values of ABCDE are calculated, and the output quality evaluation level fuzzy set is a vector {range: ;Difference: ;medium: ;good: Excellent: }, the membership function value corresponding to each fuzzy classification The quality activation strength is used to guide subsequent fuzzy instructions.
[0039] 3) Define fuzzy rules for the constraint deviation of the physical compatibility of tea leaves; In this step, three compliance thresholds of 0.05, 0.1, and 0.2 are defined for the tea constraint deviation, corresponding to the tea constraint deviation levels [satisfied, slightly deviated, severely deviated]. These levels are used to measure whether the blending scheme generated by the particle swarm optimization algorithm violates the physical compatibility of the tea. The physical compatibility constraints for the tea are set as follows: moisture content difference = Moisture content - Moisture content, moisture content difference ≤ Moisture content constraint value (moisture content difference constraint value is set to 2%); x pH difference = pH value - pH value, pH difference ≤pH value constraint (pH difference constraint is set to 0.5).
[0040] The constraint calculation results are normalized, and the constraint violation rate is designed. for: , ,For example: , ; Define the overall violation degree of the fitting constraint. ,in = ; Calculate constraint deviation =Overall violation of matching constraints ; Define membership function Using a triangular-trapezoidal hybrid membership function to assemble the overall violation degree of constraints. Given the conditions, complete the fuzzy set classification of tea constraint deviation, specifically as follows: Figure 4 As shown: when When the membership function meets or significantly deviates from the compliance criterion threshold, it exhibits a trapezoidal function, expressed as follows: ; Otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are coefficients and constants, which can be set according to the actual application.
[0041] In the case provided by this invention, the definition is... ; A is used for the tea constraint deviation to satisfy (S), which is reflected as a trapezoidal function, when Membership function ; when Membership function .
[0042] B is used when the deviation of the tea constraint is slight (SD), and is represented by a trigonometric function; when Membership function ; when Membership function .
[0043] C is used for the tea constraint deviation degree as severe deviation (VD), which is reflected as a trapezoidal function; when Membership function ; when Membership function .
[0044] Membership function value for each fuzzy classification The activation strength is used to stabilize the fuzzy instructions.
[0045] Step S133: Define the weight adjustment factor; In this step, cost weight adjustment factors are defined respectively. Quality weight adjustment factor Stability weight adjustment factor The value ranges are as follows: , , ; Define fuzzy linguistic variables as negative large, negative small, zero, positive small, and positive large, respectively. These fuzzy linguistic variables are then compared with cost weight adjustment factors. Quality weight adjustment factor The relationships are shown in Table 1: Table 1 Fuzzy linguistic variables and , Numerical mapping
[0046] Fuzzy linguistic variables and stability weight adjustment factors The relationships are shown in Table 2: Table 2 Fuzzy linguistic variables and Numerical mapping
[0047] Step S134: Determine the fuzzy control rules for the weight adjustment factor; In this step, set + This indicates more stringent cost control, leading to cost reduction. This indicates a decrease in sensitivity to cost fluctuations, allowing costs to rise within a certain range. This indicates a greater encouragement for quality requirements. This indicates a reduction in quality requirements.
[0048] Specifically, fuzzy control rules include quality-first mode and cost-first mode.
[0049] 1) The fuzzy control rules under the quality-first mode refer to: increasing the cost weight and decreasing the quality weight to control the budget when the cost is high; increasing the stability weight to correct feasibility when there is a constraint deviation; completely rejecting the region when there is a serious deviation; and increasing the quality weight in the opposite direction when the cost is low and the quality is low, so as to guide the particle swarm to adaptively converge to the region that is economically feasible and of balanced quality.
[0050] The specific rules are shown in Table 3: Table 3 Fuzzy Control Rule Settings
[0051] Specifically, Rule 1: This solution has high cost and good quality, but the cost is too high. To reduce costs, a slight reduction in quality requirements is acceptable. Significantly increase the weight of cost in the navigation score, and slightly reduce the weight of quality to guide the particle swarm algorithm to relax its obsession with extreme quality, thereby prioritizing leaving the high-cost peak area. Rule 2: This solution offers high cost and high quality, but there's a slight deviation in constraint deviation. In this case, besides addressing the excessive cost, the feasibility of the solution must be prioritized. Constraints should be corrected immediately, even at the cost of significantly sacrificing quality requirements. Similarly, significantly increase the weight of cost in the navigation score, slightly increase the weight of stability, and significantly decrease the weight of quality. Rule 3: This scheme has high cost and high quality, but the constraint deviation is severe, meaning that this matching scheme is completely incompatible and practically infeasible. Therefore, the particle swarm algorithm's search in this direction must be completely rejected, the quality weight must be reduced across the board, and the navigation score must be made extremely low, thereby forcing the entire particle swarm to deviate from this area. Rule 4: This scheme results in moderate cost and poor quality. Therefore, by keeping the cost weight unchanged and increasing the quality weight, the optimization potential of particle swarm search is directed towards significantly improving tea quality. Rule 5: This solution has moderate cost, poor quality, and slight deviation from the constraint. Therefore, simultaneous optimization is required, accepting a slight increase in cost, slightly incentivizing quality improvement, and slightly increasing the stability weight. Rule 6: This approach has high cost and moderate quality. Therefore, while maintaining quality, we should guide the particle swarm optimization algorithm to search for lower-cost solutions. Rule 7: This solution has high cost, moderate quality, and slight deviations. Therefore, the constraints should be modified first, slightly increasing the stability weight while sacrificing some quality weight to strengthen cost control.
[0052] 2) Fuzzy control rules under the cost-first model refer to: when the cost is low and the quality is low, the weight of quality is greatly increased, and cost is sacrificed to correct constraints when necessary; when the cost is low and the quality is moderate, the cost advantage is strengthened, and quality is sacrificed to ensure feasibility when deviation occurs; when the cost is high and the quality is high, the cost is compressed to the extreme, and excess quality can be sacrificed to cooperate with constraint correction.
[0053] The specific fuzzy control rules under the cost-first mode are shown in Table 4. Table 4 Fuzzy Control Rule Settings
[0054] Rule 8 states that this solution results in extremely low costs but extremely poor quality. Therefore, the weighting for quality is significantly increased.
[0055] Rule 9: This solution has extremely low cost but extremely poor quality, and the constraint deviation is slightly off. Therefore, a comprehensive revision is needed, sacrificing some cost requirements to correct the constraints while slightly increasing the quality weight.
[0056] Rule 10: This solution offers low cost and moderate quality. Continue to strengthen cost control while maintaining quality requirements.
[0057] Rule 11: This solution offers low cost and moderate quality, but there is a slight deviation from the constraints. Therefore, it is necessary to prioritize risk mitigation, accepting a decrease in quality while maintaining cost advantages, in order to improve the feasibility of the solution.
[0058] Rule 12: This solution offers high cost but high quality. Therefore, significantly increase the cost weighting, maximizing the disadvantages of high cost, and sacrificing some quality to achieve cost reduction.
[0059] Rule 13: This solution is costly, of high quality, and has slight deviations. Therefore, excess quality is sacrificed to align with corrective constraints and significantly reduce costs.
[0060] Step S135: Based on the fuzzy control rules, defuzzify the weight adjustment factor and output weight recombination; In this step, the center-of-gravity method is used, and the output weight variables for quality, cost, and stability are set as follows: , , , is represented as: , , , In the formula, For different rules The corresponding activation level is represented as: .
[0061] For example, based on the evaluation criteria in the first step, the cost activation output of a certain combination scheme is: {Very low: 0; Low: 0; Medium: 0.2; High: 0.8; Very high: 0}; the quality activation output is: {Very poor: 0; Poor: 0.3; Medium: 0.7; Good: 0; Excellent: 0}; the stability activation output is: {Satisfactory: 0; Slight deviation: 1; Severe deviation: 0}.
[0062] This scheme triggers two rules: Rule 5: Medium cost (0.2), poor quality (0.3), slight deviation from constraint (1).
[0063] Under this rule, the activation level of rule 5 is obtained. =0.2.
[0064] Rule 6: High cost (0.8), medium quality (0.7), slight deviation from constraint (1).
[0065] Under this rule, the activation level of rule 6 is obtained. =0.7.
[0066] Therefore, calculate the cost output weight variable. for Similarly, we can obtain the quality output weight variable. With stability output weight variables This achieves defuzzification of weight variables and generates... , , This constitutes a rights reorganization.
[0067] Step S140: Calculate the score of the target tea blending scheme based on the weighted reorganization to achieve dynamic control of the tea blending scheme.
[0068] Calculating the score for the target tea blending scheme involves the following steps: 1) Define dynamic cost weights Dynamic weighting of quality Stability dynamic weights They are respectively , , , In the formula, Weighted based on cost; As a basic weight for quality; The basic weights are for stability.
[0069] 2) Define the score for the target tea blending scheme. The calculation method is expressed as follows: In the formula, , , The scores are: raw tea quality score, cost score, and stability score.
[0070] The score is calculated in this step. It is used to guide the particle swarm algorithm to iterate and optimize the blending scheme, ultimately achieving dynamic control of the tea blending scheme and guiding the optimization direction of the scheme generation.
[0071] This invention constructs a decision model based on fuzzy rule-based particle swarm optimization. Through real-time fuzzy diagnosis and dynamic weight adjustment of constraint deviations in cost, quality, and physical compatibility, the search direction is dynamically reshaped. This enables the system to intelligently identify and prioritize multi-objective conflicts in the matching scheme, achieving adaptive trade-offs among multiple objectives and effectively preventing premature convergence due to local optima. This method realizes intelligent dynamic trade-offs among multiple objectives, improving the accuracy of matching scheme optimization. Simultaneously, it significantly shortens the search time and ensures superior overall performance of the final solution. Furthermore, this invention designs a constraint hierarchical response mechanism based on fuzzy reasoning and its corresponding fuzzy rules. By quantifying constraint deviations into three levels—satisfied, slightly deviated, and severely deviated—and matching differentiated strategies, the feasibility of the matching scheme is subdivided, overcoming the drawbacks of coarse-grained constraint processing and achieving refined feasibility assessment, thus improving practicality.
[0072] The above-disclosed embodiments are merely a few specific examples of the present invention. However, the present invention is not limited thereto, and any variations that can be conceived by those skilled in the art should fall within the protection scope of the present invention.
Claims
1. A dynamic control method for tea blending schemes based on fuzzy rules, characterized in that, Includes the following steps: Load the tea blending formula library, and obtain the formula, tea quality and physical compatibility of the target tea blending scheme from the tea blending formula library; wherein, the target tea blending scheme is a number of blending schemes that can be achieved based on the actual available raw materials, and it is necessary to evaluate and select the best among the multiple blending schemes; Obtain the tea cost of the target tea blending scheme; tea cost, tea quality, and physical compatibility constitute the attribute set of the target tea blending scheme; Load the blending dynamic control model, input the attribute set into the blending dynamic control model, and obtain the weighted reorganization of the target tea blending scheme; the weighted reorganization of the target tea blending scheme consists of weighted variables of quality, cost, and stability. , , constitute; The score of the target tea blending scheme is calculated based on the weighted reorganization, thereby enabling dynamic adjustment of the tea blending scheme.
2. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 1, characterized in that, Before loading the tea blending formula library, a tea blending formula library is constructed. The tea blending formula library is used to systematically store and manage historical blending strategies and corresponding tea quality and physical compatibility. The tea quality includes the overall fitness value for each historical blending strategy; The physical compatibility includes the maximum / minimum moisture content and pH value of the raw tea in each scheme.
3. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 1, characterized in that, Before loading the dynamic control model for assembly, construct the dynamic control model for assembly. The construction of the dynamic control model for the assembly includes the following steps: Define fuzzy rules for regulatory indicators; the regulatory indicators include: tea cost, tea quality, and physical compatibility; Define a weighting adjustment factor, which includes: a cost weighting adjustment factor. Quality weight adjustment factor Stability weight adjustment factor ; A fuzzy control rule is determined to define the weight adjustment factor; the fuzzy control rule consists of m rules. Based on fuzzy control rules, the weight adjustment factors are defuzzified to output the weight variables of quality, cost, and stability. , , .
4. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 3, characterized in that, The fuzzy rules defining the cost of the tea include: The cost evaluation level of tea is defined as having multiple levels, including the lowest level, the highest level, and one or more intermediate levels; Define standard values at multiple levels, namely, the minimum standard value, the maximum standard value, and one or more intermediate standard values; Calculate the proportion of tea cost = Actual cost / Budgeted cost; Define the membership function for tea cost. Triangle-trapezoidal mixed membership function; membership degree function Based on the proportion of tea cost It is expressed as follows: When the proportion of tea cost When the membership function is less than the minimum standard value or greater than the maximum standard value, Represented as a trapezoidal function, it is expressed as: =1; otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are the coefficients and the constant term, respectively. Based on the membership function, the output cost level fuzzy set contains the membership function value corresponding to each fuzzy category. Cost activation intensity.
5. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 3, characterized in that, The fuzzy rules defining the quality of the tea include: The quality of tea is defined as having multiple grades, including the worst grade, the best grade, and one or more intermediate grades. Define standard values at multiple levels, namely, the minimum standard value, the maximum standard value, and one or more intermediate standard values; Calculating the difference in tea quality =Target quality score / Actual quality score; Define the membership function of tea quality Triangle-trapezoidal mixed membership function; membership degree function Based on the differences in tea quality It is expressed as follows: When the quality difference of tea When the membership function is below the minimum standard value or above the maximum standard value, Represented as a trapezoidal function, it is expressed as: =1; otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are the coefficients and the constant term, respectively. Based on the membership function, the output fuzzy set of quality evaluation levels, and the membership function value corresponding to each fuzzy classification. The quality activation intensity.
6. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 3, characterized in that, The fuzzy rules defining the constraint deviation of the physical compatibility of the tea include: Set physical compatibility constraints for tea: moisture content difference = Moisture content - Moisture content, moisture content difference ≤ Moisture content constraint value, pH difference = pH value - pH value, pH difference ≤pH value constraint; The constraint calculation results are normalized, and the constraint violation rate is designed. for: ， ； Define the overall violation degree of the fitting constraint. , is represented as: = ; Calculate constraint deviation =Overall violation of matching constraints ; Define membership function Using a triangular-trapezoidal hybrid membership function to assemble the overall violation degree of constraints. Given the conditions, complete the fuzzy set classification of tea constraint deviation. when When the membership function meets or significantly deviates from the compliance criterion threshold, it exhibits a trapezoidal function, expressed as follows: Otherwise, membership function Trigonometric functions are represented as follows: = ,in, , These are the coefficients and the constant term, respectively. Calculate the membership function value The activation strength is the stability factor.
7. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 3, characterized in that, The fuzzy control rules for the weight adjustment factors include a quality-first mode and a cost-first mode. The fuzzy control rules under the quality-first mode refer to: increasing the cost weight and decreasing the quality weight to control the budget when costs are high; and increasing the stability weight to correct the feasibility when constraint deviations occur. If there is a serious deviation, the entire region will be rejected; when the cost is low and the quality is low, the quality weight will be increased in reverse to guide the particle swarm to adaptively converge to a region that is economically feasible and of balanced quality. The fuzzy control rules under the cost-first model refer to: when the cost is low and the quality is low, the weight of quality is greatly increased, and cost is sacrificed to adjust constraints when necessary; when the cost is low and the quality is moderate, the cost advantage is strengthened, and quality is sacrificed to ensure feasibility when deviation occurs; when the cost is high and the quality is high, the cost is compressed to the extreme, and excess quality can be sacrificed to cooperate with constraint adjustment.
8. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 3, characterized in that, The weight variables , , The calculation methods include: ， ， ， In the formula, For different rules The corresponding activation level is represented as: ,in, .
9. The dynamic control method for tea blending schemes based on fuzzy rules as described in claim 1, characterized in that, The calculation of the score for the target tea blending scheme includes the following steps: Define dynamic cost weights Dynamic weighting of quality Stability dynamic weights They are respectively ， ， ， In the formula, Weighted based on cost; As a basic weight for quality; As the basic weight for stability; The score of the target tea blending scheme The calculation method is as follows: In the formula, , , The scores are: raw tea quality score, cost score, and stability score.