Evaluation method suitable for pepper processing suitability analysis and application

Abstract

The invention belongs to the technical field of agricultural technology popularization service, and discloses an evaluation method suitable for pepper processing suitability analysis. An analysis basis is formed by constructing a feature knowledge base containing raw material texture features and producing area environment information and defining a strategy rule base of a mapping relation between a raw material feature mode and processing technology parameters, an evaluation algorithm specially designed for pepper features is adopted, and specific texture parameters such as the proportion of closed seeds and the like are introduced. Environment similar samples are matched according to the producing area information of the target raw materials, texture contours of the environment similar samples are estimated, adaptability scores in different machining directions are calculated, and customized detailed machining strategies are matched and generated from the rule base. The limitation that traditional processing depends on fixed regulations or experience is effectively overcome, accurate matching of the processing strategy and the personalized characteristics of the raw materials is achieved, and the flavor retention rate, the product quality stability and the processing efficiency of pepper processing are remarkably improved.

Description

Technical Field

[0001] This application belongs to the field of agricultural technology extension service technology, and in particular relates to an evaluation method and application suitable for the suitability analysis of Sichuan pepper processing. Background Technology

[0002] In existing technologies, the processing of seasoning ingredients such as Sichuan peppercorns largely relies on fixed process procedures or the personal experience of processing personnel. A common approach is to use a set of roughly fixed processing parameters for a specific origin or variety, lacking a systematic analysis of the raw material's inherent characteristics, particularly its physical texture and chemical quality. Furthermore, existing evaluation systems are often too generic, failing to fully consider the significant differences among different varieties of Sichuan peppercorns in key flavor compounds such as numbing substances, volatile oil content, and physical structures like peel thickness and closed-eye seed rate, as well as the potential intrinsic correlation between these differences and environmental factors such as climate and soil of the raw material's origin. This one-size-fits-all or experience-dependent approach easily leads to a mismatch between processing technology and raw material characteristics, resulting in flavor loss, unstable quality, or increased processing energy consumption.

[0003] Furthermore, while some advanced processing technologies attempt to incorporate raw material testing data, their analytical models are mostly based on general algorithms and fail to be deeply adapted to the characteristics of Sichuan pepper. For example, when assessing whether raw materials are suitable for whole-grain drying or pulverization and extraction, general algorithms may treat all texture parameters equally, failing to significantly increase the evaluation weight of key indicators that determine the core value of Sichuan pepper, such as the potential for retaining numbing substances and the ability to enrich aroma components. This may cause the final processing recommendations to deviate from the optimal solution, failing to maximize the flavor value and economic benefits of the raw materials.

[0004] Therefore, there is an urgent need in this field for an intelligent processing suitability analysis and strategy generation method that closely aligns with the characteristics of Sichuan pepper raw materials. This method should be able to systematically integrate information on the origin environment and detailed textural data of the raw materials, and through built-in analysis logic optimized for the characteristics of Sichuan pepper, scientifically assess its adaptability in various processing directions, and ultimately output customized and precise processing strategies to overcome the shortcomings of existing technologies that rely on experience, ignore the individualization of raw materials, and have inaccurate processing quality control. Summary of the Invention

[0005] To achieve the above objectives, this application provides the following technical solution: An evaluation method for the suitability analysis of Sichuan pepper processing, applicable to round-shaped seasoning raw materials, including green Sichuan pepper, red Sichuan pepper, and black pepper, comprising the following steps: S1. Establish and maintain a raw material characteristic knowledge base, storing sample records of multiple known varieties of granular seasoning raw materials. The sample records are associated with the sample's identification information, texture characteristic data set, and growth environment characteristic data set of the sample's place of origin. S2, Construct a processing strategy rule base, which stores processing strategy rules and defines the correspondence between the raw material characteristic pattern composed of the range of texture characteristic data and the range of growth environment characteristic data and the recommended set of processing parameters; S3, receive input information for the target raw material, the input information including at least the variety and origin information of the target raw material; S4. Based on the variety type and origin information of the target raw material, retrieve sample records from the raw material characteristic knowledge base that are the same or similar to the variety type and match the characteristics of the growing environment of the origin, and form a reference sample set. S5. Based on the textural feature data of the samples in the reference sample set, and combined with the preset processing suitability evaluation algorithm, for the pepper variety, the content of numbing substances and volatile oil in the textural feature data set are assigned a weight coefficient higher than that of the basic textural feature items to calculate the adaptability score of the target raw material in multiple predefined processing directions. S6. Based on the growth environment characteristic data corresponding to the origin information of the target raw material and the calculated adaptability score, the applicable processing strategy rules are matched from the processing strategy rule base. S7. Based on the matched processing strategy rules, generate a processing adaptation strategy for the target raw material. The processing adaptation strategy includes at least the recommended processing technology type, the range of key process parameters, and the expected product quality target. S8, output the processing adaptation strategy to guide the processing and production of the target raw material.

[0006] Furthermore, when the target raw material is a new variety that has not yet been entered into the raw material characteristic knowledge base, steps S4 and S5 are replaced by the following steps: S4a, based on the variety type and origin information of the target raw material, retrieve known variety sample records from the raw material feature knowledge base that belong to the same category as the target raw material and whose origin growth environment characteristics have a similarity exceeding a preset similarity threshold, and form a prediction reference set; S5a, integrate and analyze the textural feature data of all samples in the prediction reference set, and use the pepper feature weighted clustering algorithm to generate the predicted textural feature profile of the target raw material. When calculating the distance between samples, the pepper feature weighted clustering algorithm applies higher weights to the dimensions of numbing substance content and volatile oil content. S5b, based on the generated predicted texture feature profile and combined with the preset processing suitability evaluation algorithm, calculate the suitability score of the target raw material in multiple predefined processing directions.

[0007] Furthermore, the construction and execution of the preset processing suitability evaluation algorithm in step S5 specifically includes: Define the basic processing directions, including granulation and shape-preserving drying, crushing and extraction, and fine grinding. An ideal raw material texture feature vector is set for each of the basic processing directions. Each feature dimension in the ideal raw material texture feature vector has an expected value or expected range. For the processing direction involving Sichuan pepper, the ideal raw material texture feature vector includes the expected range of the content of numbing substances and the content of volatile oils. Obtain the statistical distribution of the textural feature data of the reference sample set, and generate the estimated textural feature profile of the target raw material based on the statistical distribution. The estimated textural feature profile includes the median estimate and confidence interval of each feature. For each basic processing direction, an adaptability score calculation sub-step is performed, which includes: The feature matching degree between the estimated textural feature profile of the target raw material and the ideal raw material textural feature vector for the corresponding processing direction is calculated. The feature matching degree is calculated using the weighted Manhattan distance algorithm. For Sichuan pepper varieties, the weights of the numbing substance content and volatile oil content feature dimensions are set to configurable boost values. These boost values ​​are dynamically adjusted according to the sensitivity of the processing direction to aroma and numbing intensity. The confidence interval width of each feature in the estimated texture feature profile is evaluated and compared with the tolerance range of the process parameters in the corresponding processing direction to generate a feature stability score. The reciprocal of the feature matching degree is linearly weighted and fused with the feature stability score to generate the final fitness score, wherein the weight of the feature stability score will increase accordingly when the number of samples in the reference sample set is less than a threshold.

[0008] Furthermore, the method also includes a regional preference adaptation step performed before step S7, specifically including: Maintain a regional dietary preference knowledge base, which stores descriptions of core flavor and taste preferences for processed condiment products corresponding to different regional identifiers; The target geographical identifier is determined based on the target sales market information of the target raw material; Based on the target region identifier, obtain the corresponding flavor and taste preference description from the regional dietary preference knowledge base; The obtained flavor and taste preference descriptions are used as adjustment factors and input into the strategy generation process; When generating a processing adaptation strategy based on the matched processing strategy rules, the set of processing process parameters recommended by the rules is fine-tuned according to the adjustment factor.

[0009] Furthermore, the method also includes a collaborative optimization mechanism between the knowledge base and the rule base. This mechanism is triggered after each execution of processes S1 to S8 and obtaining actual production feedback, and includes the following steps: Collect feedback data, which includes records of key process parameters executed during actual production based on the processing adaptation strategy, as well as quality inspection data of the final product. The execution records of the key process parameters are compared with the range of process parameters recommended in the processing adaptation strategy to calculate the process execution compliance. The quality inspection data of the final product is compared with the expected product quality target in the processing adaptation strategy to calculate the degree of achievement of the quality target; When the process execution compliance is higher than a preset compliance threshold while the quality target achievement is lower than a preset achievement threshold, it is determined that the current recommended strategy has room for optimization, and the analysis process is initiated. The analysis process includes: By comparing the quality data of the actual product with the typical quality data of the samples in the reference sample set, and combining the growth environment characteristic data of the target raw material, the key textural or environmental characteristics that cause the deviation are identified. Based on the analysis results, optimization instructions for the processing strategy rule base are generated; The approved optimization instructions are applied to the processing strategy rule base, and the raw material characteristics and origin environment combinations associated with this optimization are recorded simultaneously to enrich the empirical data of the raw material characteristic knowledge base.

[0010] Furthermore, the construction process of the processing strategy rule base specifically includes: Collect historical successful processing cases, each case including complete textural characteristics data of raw materials, growth environment characteristics data of the place of origin, specific processing technology and detailed process parameters used, and quality evaluation report of the final product; The collected cases are processed using a hierarchical clustering algorithm based on feature importance. The first-level classification is based on the type of raw material. Within the same type, the second-level clustering is based on the core textural features and core environmental features that have the greatest impact on processing quality. Cases with similar processing paths are grouped into the same case cluster.

[0011] Furthermore, when the texture feature data set is applied to the processing suitability evaluation algorithm, the algorithm parameters are configured differently according to the type of raw material, specifically as follows: For Sichuan pepper varieties, the texture feature data set is divided into a key quality feature group and a basic physical feature group; In the processing suitability evaluation algorithm, different feature weight vectors are configured for different processing directions. When evaluating processing directions involving the preservation of Sichuan pepper aroma and flavor, the total weight assigned to the key quality feature group is higher than that of the basic physical feature group. When evaluating processing directions mainly involving changes in physical morphology, the total weight assigned to the basic physical feature group is higher than that of the key quality feature group. The weight vector is predefined based on the correlation analysis results between different features and processing results in historical cases, and is iteratively fine-tuned based on feedback data in the collaborative optimization mechanism.

[0012] Further, in step S7, a processing adaptation strategy for the target raw material is generated based on the matched processing strategy rules, including: S71, parse one or more processing strategy rules matched from the processing strategy rule base, extract the raw material feature pattern conditions defined by each rule, the corresponding standard processing strategy template, and the historical call success rate of the rule, and put the parsed rule into the candidate rule set; S72, Multidimensional matching degree calculation and ranking: For each rule in the candidate rule set, calculate the matching degree between the current information of the target raw material and the raw material feature pattern conditions defined by the rule; S73, Strategy Template Selection and Fusion: Select the top K rules with the highest overall matching degree as core reference rules; if K=1, directly use the standard processing strategy template corresponding to that rule as the base template; if K>1, initiate the template fusion logic, which includes: By comparing the standard processing strategy templates corresponding to each core reference rule, the common and different parts of their process step sequences can be identified. For the common parts, the intersection or weighted average range of the process parameter ranges of each template is taken as the parameters of the merged template; For the difference part, based on the comprehensive matching degree weight of each rule, the step in the template with the highest weight is selected as the backbone of the fusion template, and the advantage parameters of the difference steps in other templates are added to the fusion template as optional or annotation. S74, based on the selected or merged basic template, makes the following adjustments and enrichments: If a regional preference adjustment factor exists, the preset parameter adjustment mapping table is called, and the corresponding process parameters in the template are offset and adjusted according to the specific preference description item to generate a fine-tuned strategy version. The supplementary strategy metadata includes: the source of the rules on which the strategy is generated, the hypothesis on the key characteristics of the target raw materials, the parameter adjustment suggestions under different production scales or equipment conditions, and the monitoring suggestions for core quality control points. S75 organizes the finalized processing technology type, process step sequence, key process parameter range, expected quality target, and supplementary metadata into a complete processing adaptation strategy document according to a predefined structured format.

[0013] Furthermore, the method sets an ideal raw material texture feature vector for the granulation and drying direction, the crushing and extraction processing direction, and the fine grinding processing direction, and the specific feature terms are dynamically adjusted according to whether the target raw material is Sichuan pepper: When the target raw material is Sichuan pepper, in the ideal raw material texture feature vector of the whole grain shape-preserving drying direction, the expected range of the numbing flavor substance content feature item is set to ensure that it is above the threshold required for commercial grade, and the expected range of the volatile oil content feature item is set to be sufficient to support the level of characteristic aroma. In the ideal raw material texture feature vector of the crushing and extraction processing direction, the expected range of the volatile oil content feature term is given the highest priority. In the ideal raw material texture feature vector of the fine grinding processing direction, the hardness and brittleness features in the basic physical feature group are given higher weights, while the proportion of closed-eye seeds is required to be below a certain threshold. The processing suitability evaluation algorithm automatically loads the corresponding ideal raw material texture feature vector set according to the variety type of the target raw material for calculation, thereby ensuring that the scoring model closely matches the variety characteristics of the raw material.

[0014] According to a second aspect of the present invention, the present invention requests the application of the evaluation method for suitability analysis of Sichuan pepper processing in the evaluation of suitability analysis of Sichuan pepper processing.

[0015] This invention relates to the field of agricultural technology extension services and discloses an evaluation method for suitability analysis of Sichuan pepper processing. By constructing a feature knowledge base containing raw material texture characteristics and origin environment information, and a strategy rule base defining the mapping relationship between raw material characteristic patterns and processing parameters, an analytical foundation is formed. An evaluation algorithm specifically designed for Sichuan pepper characteristics is employed, incorporating unique texture parameters such as the proportion of closed-eye seeds. Based on the origin information of the target raw material, environmentally similar samples are matched to predict its texture profile and calculate its suitability score in different processing directions. Customized detailed processing strategies are then generated by matching from the rule base. This invention effectively overcomes the limitations of traditional processing methods that rely on fixed procedures or experience, achieving precise matching between processing strategies and the individual characteristics of raw materials, significantly improving the flavor retention rate, product quality stability, and processing efficiency of Sichuan pepper processing. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the workflow of an evaluation method for suitability analysis of Sichuan pepper processing, as claimed in an embodiment of the present invention. Figure 2 The second workflow diagram is shown for an evaluation method for suitability analysis of Sichuan pepper processing, as claimed in an embodiment of the present invention. Figure 3 The third workflow diagram is shown for an evaluation method for suitability analysis of Sichuan pepper processing, as claimed in an embodiment of the present invention. Figure 4 The fourth flowchart of an evaluation method for suitability analysis of Sichuan pepper processing, as claimed in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the effect of puncture distance on the hardness determination of fresh green Sichuan peppercorns, according to an evaluation method for analyzing the suitability of Sichuan peppercorn processing as claimed in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of those features. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications in the embodiments of this application, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationships and movements between components in a specific orientation as shown in the accompanying drawings. If the specific orientation changes, the directional indications will change accordingly. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0019] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] According to the first embodiment of the present invention, referring to Figure 1 This invention claims protection for an evaluation method suitable for the suitability analysis of Sichuan pepper processing. The method is applicable to round-shaped seasoning raw materials, including green Sichuan pepper, red Sichuan pepper, and black pepper, and includes the following steps: S1. Establish and maintain a raw material characteristic knowledge base, storing sample records of multiple known varieties of granular seasoning raw materials, and associating the sample records with the sample identification information, texture characteristic data set, and growth environment characteristic data set of the sample origin. S2, Construct a processing strategy rule base, which stores processing strategy rules and defines the correspondence between the raw material characteristic pattern composed of the range of texture characteristic data and the range of growth environment characteristic data and the recommended set of processing parameters; S3, receive input information for the target raw material, the input information including at least the variety and origin information of the target raw material; S4. Based on the variety and origin information of the target raw material, retrieve sample records from the raw material characteristic knowledge base that are the same or similar in variety and match the characteristics of the growing environment of the origin, and form a reference sample set. S5. Based on the textural feature data of the samples in the reference sample set, combined with the preset processing suitability evaluation algorithm, for the pepper variety, the content of numbing substances and volatile oil in the textural feature data set are assigned a weight coefficient higher than that of the basic textural feature items to calculate the adaptability score of the target raw material in multiple predefined processing directions. S6. Based on the growth environment characteristic data corresponding to the origin information of the target raw material and the calculated adaptability score, the applicable processing strategy rules are matched from the processing strategy rule base. S7. Based on the matched processing strategy rules, generate a processing adaptation strategy for the target raw material. The processing adaptation strategy shall include at least the recommended processing technology type, the range of key process parameters, and the expected product quality target. S8 outputs a processing adaptation strategy to guide the processing and production of target raw materials.

[0021] In this embodiment, the relevant material involves fresh green Sichuan peppercorns, and the specific details of each sample are shown in Table 1. Sample packaging and appearance. Sample storage conditions are as per product instructions; samples should be thawed in a microwave oven before use.

[0022] Table 1 Information on Fresh Green Sichuan Pepper Samples ; A structured raw material characteristic knowledge base is established and maintained to persistently store and manage sample data of known varieties obtained from multiple channels. Specifically, the physical texture and chemical quality data of each sample are systematically measured and recorded using laboratory testing equipment, such as physical property analyzers and chromatographs, forming a set of textural characteristic data for that sample. For Sichuan pepper samples, this set explicitly includes the total content of amide-based numbing substances determined by high-performance liquid chromatography, the volatile oil content determined by steam distillation, and the pericarp cracking force and shell hardness determined by a texture analyzer puncture test. Simultaneously, by linking with a geographic information system and a professional agricultural environmental database, a set of growth environment characteristic data corresponding to the precise origin coordinates of each sample is obtained. This set includes multi-year average temperature, annual effective accumulated temperature, and annual precipitation interpolated from standardized meteorological station data, as well as soil pH and soil organic matter content data extracted from a soil survey database. The identification information of each sample, the aforementioned complete set of textural characteristic data, and the set of growth environment characteristic data are combined into an indivisible record, indexed, and stored in the raw material characteristic knowledge base.

[0023] A well-defined processing strategy rule base is constructed, based on reverse engineering and pattern extraction from numerous historical successful production cases. Each processing strategy rule consists of two parts: conditions and suggestions. The conditions part is a specific raw material characteristic pattern, which characterizes a class of raw materials with common properties by defining a combination of textural characteristic data, such as the content of numbing substances between A and B mg / g, and growth environment characteristic data, such as altitude between C and D meters. The suggestions part is a set of recommended processing parameters corresponding to this pattern, specifically covering the dominant processing technology type, such as low-temperature slow drying, the sequence of key processes, the target range of core control parameters for each step, such as the temperature of the first drying stage, E to F degrees Celsius, and the duration, G to H hours, as well as the desired product quality target, such as a numbing substance retention rate of over 1%. The rule base is stored and managed in the form of database tables or configuration files.

[0024] The system receives user input information for a specific batch of raw materials through a pre-designed application programming interface (API) or graphical user interface. The input information must clearly specify the variety of the raw material (e.g., Jiangjin green peppercorns) and its specific origin (e.g., XX township, Jiangjin District, Chongqing City, Sichuan Province). The system will validate the format of the input information and may conduct preliminary verification of the origin information's authenticity by linking to an external database.

[0025] Upon receiving compliant input information, the query logic is initiated. Based on the type of the target raw material, the system first defines a range of similar varieties or substitutes considered to have similar processing characteristics within the raw material characteristic knowledge base. Then, based on the origin information of the target raw material, the system compares it with the growth environment characteristic data of sample records in the knowledge base, filtering out sample records whose origin environmental factors, such as average annual temperature and soil pH, fall within a certain threshold range centered on the target origin data. Finally, sample records that simultaneously meet the conditions of variety relevance and origin environmental similarity are compiled to form a reference sample set for subsequent analysis.

[0026] The algorithm invokes a pre-defined processing suitability evaluation algorithm. Its core task is to infer the processing potential of the target raw material based on a reference sample set. First, the algorithm analyzes the textural characteristics of all samples in the reference sample set, calculating the statistical distribution of various characteristics such as the content of numbing substances, volatile oil content, and pericarp cracking force (mean and quartiles), thereby generating a predicted profile describing the range of possible textural properties of the target raw material. Subsequently, for several pre-defined basic processing directions, such as whole-grain drying, crushing and extraction, and powdering, the algorithm calculates the degree of matching between this predicted profile and the ideal raw material textural template defined for each processing direction. When calculating the matching degree, the algorithm internally configures different feature weight schemes according to the raw material type. When processing Sichuan pepper, the algorithm actively increases the weight coefficients of the two core quality characteristics, numbing substance content and volatile oil content, in the comprehensive matching degree calculation to reflect the principle of prioritizing the flavor value of Sichuan pepper. Finally, the algorithm outputs a quantitative suitability score for each processing direction.

[0027] The adaptability score calculated in step S5 is combined with the target raw material's growing environment characteristics data input in step S3, serving as query conditions for strategy matching within the processing strategy rule base. The matching process involves traversing each rule in the rule base and checking whether the predicted texture profile of the target raw material, reflected by the score, and the actual environmental data, conform to the raw material characteristic pattern defined by a certain rule's condition. A degree of ambiguity is permissible in the matching process; typically, a comprehensive matching score is calculated, and rules with matching scores exceeding a set threshold are selected as applicable processing strategy rules. One or more rules may be matched simultaneously.

[0028] The strategy generation engine synthesizes strategies based on one or more processing strategy rules matched in step S6. If only one rule is matched, its suggested process parameter set is used as the blueprint. If multiple rules are matched, the strategy generation engine analyzes the commonalities and differences in the suggested parts of these rules, and may use logic such as parameter intersection, weighted averaging, or selection based on rule priority to fuse and generate a unified processing adaptation strategy for the current target raw material. This strategy is a detailed process guidance document that clearly indicates the recommended processing path, such as recommending a two-stage gradient drying process for granulation and shape preservation, lists the recommended parameter control ranges for key processes, and sets clear and quantifiable product quality targets, such as the volatile oil retention rate of the finished product should be greater than J.

[0029] The processing adaptation strategy generated in step S7 is delivered through the designated output module. The output can be a structured text report, a data file importable into the production management system, or a visual representation on a graphical user interface. The core purpose of the output is to provide clear, specific, data-driven, and rule-based decision support for the production department's process development and operations, thereby guiding the processing and production activities for this batch of target raw materials.

[0030] Furthermore, referring to Figure 2 When the target raw material is a new variety that has not yet been entered into the raw material characteristic knowledge base, steps S4 and S5 are replaced by the following steps: S4a, based on the variety and origin information of the target raw material, retrieve known variety sample records from the raw material feature knowledge base that belong to the same category as the target raw material and whose similarity to the growth environment characteristics of the origin exceeds the preset similarity threshold, and form a prediction reference set; S5a integrates and analyzes the textural feature data of all samples in the prediction reference set, and uses the pepper feature weighted clustering algorithm to generate the predicted textural feature profile of the target raw material. When calculating the distance between samples, the pepper feature weighted clustering algorithm applies higher weights to the dimensions of numbing substance content and volatile oil content. S5b, based on the generated predicted texture feature profile and combined with the preset processing suitability evaluation algorithm, calculates the suitability score of the target raw material in multiple predefined processing directions.

[0031] In this embodiment, the strict requirements for variety matching are relaxed, and initial screening is conducted based on the broad category of the target raw material, such as Sichuan pepper. Then, the system relies heavily on the environmental characteristics data of the production area to retrieve known variety sample records from the raw material characteristic knowledge base that have growth environment characteristics highly similar to those of the target production area. The determination of high similarity is based on calculating the comprehensive similarity score between environmental feature vectors, which must exceed a preset, high similarity threshold, such as 0.85. In this way, the system aggregates a sample set that may contain multiple different varieties but with similar environmental backgrounds, called the prediction reference set.

[0032] A deep integration analysis is performed on the textural feature data of all samples in the prediction reference set. The analysis process is not a simple averaging, but rather employs a logic called Sichuan pepper feature-weighted clustering analysis. This logic, when analyzing the similarity of textural features among samples for grouping, assigns greater influence to key flavor indicators of Sichuan pepper, such as the content of numbing substances and volatile oils. This means that two samples that are similar in these key flavor indicators are more likely to be grouped together, even if they differ slightly in some physical textural indicators. Through this weighted clustering, the system infers several typical combinations of textural features that the raw material may exhibit under similar environmental conditions. Combined with the sample proportions of each category, it ultimately generates a predicted textural feature profile with probability distribution properties to describe the target raw material, rather than just a single estimate.

[0033] The generated predicted texture profile is input into the pre-defined processing suitability evaluation algorithm, just like the actual predicted profile. Based on this predicted profile, the algorithm executes the same logic as step S5, calculating the suitability score of the target raw material in various predefined processing directions. Thus, even in the absence of direct detection data for the target raw material, the system can still complete a preliminary assessment of processing suitability based on environmental similarity and commonalities among major product categories.

[0034] Furthermore, the construction and execution of the preset processing suitability evaluation algorithm in step S5 specifically includes: Define the basic processing directions, including granulation and shape-preserving drying, crushing and extraction, and fine grinding. For each basic processing direction, an ideal raw material texture feature vector is set. Each feature dimension in the ideal raw material texture feature vector has an expected value or expected range. For the processing direction involving Sichuan pepper, the ideal raw material texture feature vector includes the expected range of the content of numbing substances and volatile oil content. Obtain the statistical distribution of the textural feature data of the reference sample set, and generate the estimated textural feature profile of the target raw material based on the statistical distribution. The estimated textural feature profile includes the median estimate and confidence interval of each feature. For each basic processing direction, an adaptability score calculation sub-step is performed, which includes: The feature matching degree between the estimated textural feature profile of the target raw material and the ideal raw material textural feature vector for the corresponding processing direction is calculated. The feature matching degree is calculated using the weighted Manhattan distance algorithm. For Sichuan pepper varieties, the weights of the numbing substance content and volatile oil content feature dimensions are set to configurable boost values, which are dynamically adjusted according to the sensitivity of the processing direction to aroma and numbing intensity. Assess the confidence interval width of each feature in the predicted texture feature profile and compare it with the tolerance range of the process parameters in the corresponding processing direction to generate a feature stability score; The inverse of the feature matching degree is linearly weighted and fused with the feature stability score to generate the final fitness score. The weight of the feature stability score will increase accordingly when the number of samples in the reference sample set is less than a threshold.

[0035] In this embodiment, the algorithm's design begins with an abstraction of common industrial processing routes, explicitly defining at least three basic processing directions: whole-grain drying, aimed at preserving the fruit's intact shape and dehydrating it; crushing and extraction, aimed at disrupting cell structure to extract flavor oil components; and fine grinding, aimed at pulverizing raw materials to a specific particle size. Each processing direction has its inherent, specific requirements regarding the physical and chemical properties of the raw materials.

[0036] To quantify these requirements, the algorithm sets an ideal raw material texture feature vector for each basic processing direction. This vector is a multi-dimensional data structure, with each dimension corresponding to a texture feature such as hardness or numbing substance content, and setting an ideal value or ideal range for that feature under the corresponding processing direction. For example, for the whole-grain shape-preserving drying direction, the pericarp cracking force dimension in its ideal vector will have a moderately high expected range to ensure that it is not easily broken during the drying process; at the same time, the numbing substance content dimension will have a minimum threshold to ensure commercial value. For the direction related to Sichuan pepper, its ideal vector must explicitly include the expected range of numbing substance content and volatile oil content, which is a key design point for the algorithm to closely integrate with the characteristics of Sichuan pepper.

[0037] The core computational process of the algorithm begins with the analysis of a reference sample set. Instead of simply mixing all sample data, the algorithm first calculates the statistical distribution of each textural characteristic in the set. Based on this distribution, the algorithm generates a predicted textural characteristic profile of the target raw material. This profile includes not only an estimate of the central tendency of each characteristic, such as the median, but also a confidence interval describing the uncertainty of that estimate, such as the range of possible fluctuations in the characteristic values ​​calculated using statistical methods.

[0038] Subsequently, for each basic processing direction to be evaluated, the algorithm initiates an independent adaptive scoring calculation sub-process. This sub-process contains two core calculation units and one scoring fusion unit.

[0039] The first computational unit is responsible for calculating the feature matching degree. Its input is the estimated textural feature profile of the target raw material, mainly the center estimate, and the ideal raw material textural feature vector for the current processing direction. During calculation, the algorithm compares the actual estimated value of each feature dimension with the ideal value or the midpoint of the ideal range. To reflect the different importance of different features, the algorithm multiplies each dimension by a preset weight coefficient when summarizing these differences. This is the core of the algorithm's representation of Sichuan pepper characteristics: when the raw material is identified as Sichuan pepper, the algorithm automatically calls a preset weight configuration table, in which the weight coefficients of the numbing substance content and volatile oil content dimensions are significantly increased. The specific increase can even be related to the sensitivity of the processing direction to aroma or numbing sensation. For example, for the crushing and extraction processing direction, the weight increase value of volatile oil content will be set higher than that for the whole grain drying direction.

[0040] The second calculation unit is responsible for calculating the feature stability score. Its evaluation is based on the estimated confidence interval width in the texture feature profile. This unit compares the confidence interval width for each feature dimension with a preset tolerance range for process parameters specific to the current processing direction and that feature dimension. A wide confidence interval indicates high uncertainty in that feature of the target raw material, potentially exceeding the ability to adjust the process, thus lowering the stability score. Conversely, a narrower confidence interval results in a higher score.

[0041] Finally, the scoring fusion unit integrates the outputs of the first two units. It converts the feature matching degree calculated by the first unit (generally, a smaller value indicates a better match) into a positive score, for example, by taking its reciprocal. This positive score is then linearly weighted and summed with the feature stability score calculated by the second unit according to a configurable weight ratio to generate the final fitness score. This final score comprehensively reflects both the suitability of the target raw material in terms of texture and the degree of certainty of suitability. Specifically, when the number of samples in the reference sample set is small, the algorithm automatically increases the weight of the feature stability score to reflect the risk of insufficient data.

[0042] Furthermore, the method also includes a regional preference adaptation step performed before step S7, specifically including: Maintain a regional dietary preference knowledge base, which stores descriptions of core flavor and taste preferences for processed condiment products corresponding to different regional identifiers; Determine the target geographic identifier based on the target sales market information for the target raw material; Based on the target region identifier, retrieve the corresponding flavor and taste preference descriptions from the regional dietary preference knowledge base; The obtained flavor and taste preference descriptions are used as adjustment factors and input into the strategy generation process; When generating a processing adaptation strategy based on the matched processing strategy rules, the set of processing parameters recommended by the rules is fine-tuned according to the adjustment factor.

[0043] In this embodiment, a separate regional dietary preference knowledge base is first maintained. This knowledge base structurally stores the mapping relationships between different regional identifiers and corresponding countries, provinces, specific culinary regions, and a series of researched and summarized descriptions of product flavor and texture preferences. For Sichuan pepper products, these preference descriptions are specific and actionable, at least broken down into graded or categorized descriptions of numbing intensity (e.g., a preference for strong stimulation or a mild numbing sensation), aroma type (e.g., a preference for fresh fruity or rich woody aroma), and crispness for whole-pepper products.

[0044] When the system starts up or a user specifies a processing strategy recommendation task, in addition to providing target raw material information, it may also receive a target region identifier, which indicates the main market area where the processed products are planned to be sold.

[0045] Based on this target region identifier, as a separate query, the corresponding flavor and texture preference description is precisely retrieved from the regional dietary preference knowledge base. This description is then packaged into a structured adjustment factor data package.

[0046] In the subsequent S7 step, when the strategy generation engine begins constructing specific processing adaptation strategies based on the matched processing strategy rules, this adjustment factor data package is used as one of the key inputs. The strategy generation engine internally includes a mapping logic that maps preference descriptions to process parameter adjustment tendencies. Specifically, this fine-tuning process manifests as follows: If the preference description specifies a need for high retention of numbing sensation, the strategy generation engine, when integrating or selecting recommended process parameters, will prioritize parameter settings that have been proven effective in reducing the loss of heat-sensitive numbing substances such as hydroxy-α-sanshool in historical cases. For the drying process, this might manifest as recommending a lower constant drying temperature, or designing a staged drying temperature profile with a low initial temperature and a gradual increase later, to minimize the degradation of amide substances while dehydrating.

[0047] If the preference description specifies a need for a rich aroma, the strategy generation engine will adjust the priority of parameter selection. For crushing and extraction processing involving aroma extraction, it will tend to recommend process parameter combinations aimed at increasing the total extraction rate of volatile oils, such as adjusting the solvent ratio, increasing the extraction pressure, or appropriately increasing the extraction temperature within a safe range. For whole-grain drying, while ensuring basic flaky texture retention, the engine may allow or recommend using a relatively high temperature for a short period at a specific stage of the drying process to promote Maillard reactions and other reactions of aroma precursors in the raw material, thereby generating richer characteristic aroma compounds.

[0048] If the preference description specifies a need for high crispness primarily for products manufactured using whole-grain drying with shape preservation, the strategy generation engine will optimize parameters for the later dehydration stage of the drying process. For example, it might recommend adjusting the heating rate and dehumidification airflow in the later stages of drying to promote rapid and even release of moisture from the inside of the peppercorn shell, thereby forming more microporous structures and ultimately achieving a crisper texture.

[0049] Through the above rule-based parameter fine-tuning, the final processing adaptation strategy can theoretically make the produced pepper products more in line with the expectations of consumers in the target market, realizing a strategy upgrade from raw material-driven to a dual-driven strategy of raw materials and market.

[0050] Furthermore, referring to Figure 3 The method also includes a collaborative optimization mechanism between the knowledge base and the rule base. This mechanism is triggered after each execution of processes S1 to S8 and after receiving actual production feedback, and includes the following steps: Collect feedback data, which includes records of key process parameters executed during actual production based on the processing adaptation strategy, as well as quality inspection data of the final product. Compare the execution records of key process parameters with the recommended range of process parameters in the processing adaptation strategy, and calculate the process execution compliance. The quality inspection data of the final product are compared with the expected product quality targets in the processing adaptation strategy to calculate the degree of achievement of the quality targets. When the process execution compliance is higher than the preset compliance threshold but the quality target achievement is lower than the preset achievement threshold, it is determined that there is room for optimization in the current recommended strategy, and the analysis process is initiated. The analysis process includes: By comparing the quality data of the actual product with the typical quality data of the samples in the reference sample set, and combining the growth environment characteristics data of the target raw material, the key textural or environmental characteristics that cause the deviation are identified. Based on the analysis results, optimized instructions for the processing strategy rule base are generated; The approved optimization instructions are applied to the processing strategy rule base, and the raw material characteristics and origin environment combinations associated with this optimization are recorded simultaneously to enrich the empirical data of the raw material characteristic knowledge base.

[0051] In this embodiment, two key feedback data are collected through interfaces or manual input: first, process log data, which records the actual process parameters executed by the equipment at each key workstation when the factory strictly follows or partially deviates from the processing adaptation strategy in actual production, such as the actual temperature curve of drying room 1 and the actual pressure value of the extraction tank; second, result evaluation data, which is the quality data obtained after testing the final product according to standard methods, such as the content of numbing substances in the finished product measured by the laboratory and the aroma score given by the sensory evaluation team.

[0052] The collected actual process log data is compared item by item with the process parameter ranges explicitly recommended in the initially generated processing adaptation strategy. For each key parameter, the proportion of its measured value sequence falling within the recommended range or the average deviation is calculated. Finally, a process execution compliance metric score is obtained to measure the degree to which the production end faithfully executes the recommended strategy.

[0053] The evaluation results are compared with the pre-set expected product quality targets in the processing adaptation strategy. For example, if the strategy target is a flaking retention rate of >75%, but the actual product test value is 70%, then this target has not been achieved. By comprehensively comparing the actual achievement of all pre-set quality targets, a quality target achievement metric score is calculated.

[0054] A built-in judgment logic is in place: when the process execution compliance score is higher than a preset high threshold, indicating that the manufacturer is basically strictly following the recommended strategy, but the quality target achievement score is lower than another preset pass threshold, a meaningful deviation situation is identified—that is, the correct method was followed, but the expected good result was not obtained. This situation strongly suggests that the underlying rules or knowledge currently used to generate the strategy may be biased or inappropriate. Once this condition is met, the system automatically initiates the subsequent in-depth analysis process.

[0055] Analysts or intelligent analysis modules will perform the following operations: They will conduct a horizontal comparison of the final quality data of the actual product with the typical quality data of those samples in the reference sample set upon which the strategy was initially generated, to observe whether there are any systematic differences. Simultaneously, they will thoroughly examine the environmental characteristics of the target raw materials, especially those factors that may significantly differ from the reference samples. The focus of the analysis is to attempt to identify which one or more key textural characteristics, such as the actual content of numbing substances being far lower than estimated, or environmental characteristics, such as abnormal predictions or insufficient consideration of certain trace elements in the soil, led to the quality deviation in the final product.

[0056] Based on the deviation analysis results from the previous step, the system generates specific optimization suggestion instructions. These instructions directly target the processing strategy rule base. Possible instruction types include: a) Correcting rules: For cases where mismatches occur due to the overly broad or narrow definition of the raw material characteristic pattern in a rule, specific suggestions are made to modify the boundary of the specific characteristic data range in that rule; b) Adjusting rules: For cases where the set of process parameters recommended by a rule is not entirely optimal for that type of raw material, suggestions are made to adjust one or more process parameters in that rule; c) Adding rules: If it is found that the current raw material characteristic combination is a completely new success or failure pattern not covered by existing rules, a draft suggestion to add a new processing strategy rule is proposed.

[0057] The generated optimization instructions do not take effect automatically; instead, they are submitted to domain experts, such as senior process engineers, for review and confirmation. Once approved, the administrator formally applies these changes to the processing strategy rule base of the production system. Simultaneously, the raw material characteristic data, environmental data, actual process data, and result data associated with this optimization event can be selectively added to the raw material characteristic knowledge base as a new, practically validated case study, thereby continuously enriching the system's knowledge reserves. Through this mechanism, the entire system can continuously improve the accuracy and reliability of its recommended strategies as usage time and practical cases accumulate.

[0058] Furthermore, the construction process of the processing strategy rule base specifically includes: Collect historical successful processing cases, each case including complete textural characteristics data of raw materials, growth environment characteristics data of the place of origin, specific processing technology and detailed process parameters used, and quality evaluation report of the final product; The collected cases are processed using a hierarchical clustering algorithm based on feature importance. The first-level classification is based on the type of raw material. Within the same type, the second-level clustering is based on the core textural features and core environmental features that have the greatest impact on processing quality. Cases with similar processing paths are grouped into the same case cluster.

[0059] In this embodiment, successful processing cases from different production areas, varieties, and processing techniques, and which have obtained market or quality inspection approval, are extensively collected. Each case must be a complete data package, containing at least: 1) the raw material's identity file, i.e., its variety and origin; 2) the raw material's health check report, i.e., a complete set of textural characteristic data that must include key indicators of Sichuan pepper; 3) the raw material's growth background, i.e., a set of standardized growth environment characteristic data of the production area; 4) the production operation manual, i.e., a detailed description of the actual processing technology process and records of key process parameters for each step; 5) the final report card, i.e., a comprehensive quality evaluation report of the finished product, including physicochemical indicators and sensory evaluation.

[0060] The key steps in extracting general patterns from specific cases are as follows: The system employs a hierarchical clustering logic based on feature importance to process the large number of collected cases. First, a coarse first-level classification is performed based on the variety of raw materials, such as red Sichuan pepper and green Sichuan pepper, because different varieties have different baseline processing characteristics. Then, within the same major variety category, a more refined second-level clustering is performed. The second-level clustering is based on the combination of core features that most significantly affect the processing quality of that variety. For Sichuan pepper, based on industry experience and data analysis, its core textural features typically include the content of numbing substances and the pericarp cracking force, representing intrinsic flavor value and physical processing tolerance, respectively. Core environmental features typically include altitude and average annual temperature difference, which have a significant impact on the accumulation of flavor substances. The algorithm automatically merges cases with similar values ​​on these core features and similar final processing paths, such as drying methods and extraction methods, into the same case cluster. Each case cluster represents a typical scenario with common raw material attributes and successful processing routes.

[0061] For each case cluster formed in the previous step, the system performs standardized data refinement operations to generate a prototype rule: First, it analyzes the raw material texture characteristics data of all cases within the cluster. For each characteristic, it statistically analyzes the distribution of that characteristic value across all cases, and then selects a numerical range that covers most cases, such as the 80th percentile interval, as the condition range for that characteristic in the raw material characteristic pattern of future rules. Second, using the same method, it analyzes the origin environment characteristics data of all cases within the cluster, extracting the condition ranges for key environmental factors such as altitude and soil pH. Finally, it analyzes the processing parameters used in all cases within the cluster. It extracts the common and essential process steps shared by all cases, and statistically analyzes the value distribution of key parameters for each step, such as temperature and time, taking the median and common fluctuation ranges to form a standard, reusable processing strategy template.

[0062] The raw material characteristic patterns, texture ranges, and environmental ranges extracted for each case cluster are linked one-to-one with the processing strategy templates, process steps, and parameters. The causal pair where, if the raw material matches pattern A, template B is recommended, constitutes an initial processing strategy rule. After manual verification and necessary merging and deduplication, all rules generated from all case clusters are formally stored in the processing strategy rule base for querying and invocation by the main workflow. This construction process can be executed periodically to incorporate new successful cases, enabling the rule base to expand and update.

[0063] Furthermore, when the texture feature dataset is applied to the processing suitability evaluation algorithm, the algorithm parameters are configured differently according to the type of raw material, specifically as follows: For Sichuan pepper varieties, the texture characteristic dataset is divided into key quality characteristic group and basic physical characteristic group; In the processing suitability evaluation algorithm, different feature weight vectors are configured for different processing directions. When the evaluation involves the preservation of Sichuan pepper aroma and flavor, the total weight of the key quality feature group is higher than that of the basic physical feature group. When the evaluation mainly involves the processing direction of physical morphological changes, the total weight of the basic physical feature group is higher than that of the key quality feature group. The weight vector is predefined based on the correlation analysis results between different features and processing results in historical cases, and is iteratively fine-tuned based on feedback data in the collaborative optimization mechanism.

[0064] In this embodiment, a set of textural feature grouping schemes and corresponding weight configuration templates are predefined for each major raw material category, such as Sichuan pepper and black pepper. Taking Sichuan pepper as an example, its textural feature data set is logically divided into two groups within the algorithm: a key quality feature group and a basic physical feature group. The key quality feature group must at least include the content of numbing substances and volatile oils that directly determine its core flavor value. The basic physical feature group must at least include hardness, elasticity, and breaking force, which describe its physical processing properties.

[0065] When calculating the suitability score, the algorithm dynamically loads a corresponding weight configuration template based on two pieces of information: the current processing direction being evaluated and the target raw material variety. Each textural feature dimension in the template has a assigned weight coefficient. The configuration logic is as follows: when the primary goal of the evaluated processing direction is to preserve or extract the flavor compounds of Sichuan pepper (e.g., in crushing and extraction processing or whole-grain drying focusing on the retention of numbing sensation), the template loaded by the algorithm will assign a significantly higher total weight to features in the key quality feature group, especially the content of numbing substances and volatile oils, while assigning relatively lower weights to features in the basic physical feature group. Conversely, when the evaluated processing direction mainly involves changes in the physical form of the raw material with a low risk of flavor degradation (e.g., in the evaluation of certain fine grinding processing directions), the loaded template will increase the total weight of the basic physical feature group and appropriately decrease the weight of the key quality feature group.

[0066] The initial values ​​of these predefined weight configuration templates are not arbitrarily set, but rather derived from the results of statistical analysis of a large amount of historical case data. By analyzing the correlation between different feature values ​​and the ultimate success or failure of different processing directions, the initial weight base of each feature is determined. More importantly, this weight configuration is not fixed; it can be covered by a collaborative optimization mechanism. When the system finds that the importance of certain features is overestimated or underestimated based on production feedback, the weight configuration templates for specific varieties and specific processing directions can be fine-tuned under the review of domain experts, thereby enabling the evaluation algorithm to continuously evolve and become more accurate.

[0067] Furthermore, referring to Figure 4 In step S7, a processing adaptation strategy for the target raw material is generated based on the matched processing strategy rules, including: S71, parse one or more processing strategy rules matched from the processing strategy rule base, extract the raw material feature pattern conditions defined by each rule, the corresponding standard processing strategy template, and the historical call success rate of the rule, and put the parsed rule into the candidate rule set; S72, Multidimensional matching degree calculation and ranking: For each rule in the candidate rule set, calculate the matching degree between the current information of the target raw material and the raw material feature pattern conditions defined by the rule; S73, Strategy Template Selection and Fusion: Select the top K rules with the highest comprehensive matching degree as core reference rules. If K=1, directly use the standard processing strategy template corresponding to the rule as the base template; if K>1, start the template fusion logic. Template fusion logic includes: By comparing the standard processing strategy templates corresponding to each core reference rule, the common and different parts of their process step sequences can be identified. For the common parts, the intersection or weighted average range of the process parameter ranges of each template is taken as the parameters of the merged template; For the difference part, based on the comprehensive matching degree weight of each rule, the step in the template with the highest weight is selected as the backbone of the fusion template, and the advantage parameters of the difference steps in other templates are added to the fusion template as optional or annotation. S74 uses the selected or merged basic template as a blueprint for adjustment and enrichment: Specific adjustments and additions include: If a regional preference adjustment factor exists, the preset parameter adjustment mapping table is called, and the corresponding process parameters in the template are offset and adjusted according to the specific preference description item to generate a fine-tuned strategy version. The supplementary strategy metadata includes: the source of the rules on which the strategy is generated, the hypothesis on the key characteristics of the target raw materials, the parameter adjustment suggestions under different production scales or equipment conditions, and the monitoring suggestions for core quality control points. S75 organizes the finalized processing technology type, process step sequence, key process parameter range, expected quality target, and supplementary metadata into a complete processing adaptation strategy document according to a predefined structured format.

[0068] In this embodiment, the system first parses one or more processing strategy rules matched in step S6. The parsing includes: extracting the specific textural and environmental feature ranges (i.e., raw material feature patterns) defined in the conditional part of each rule; extracting the complete content of the standard processing strategy template corresponding to its suggestion part; and simultaneously, querying the historical call records of each rule to calculate its historical application success rate or confidence index. After parsing, all this information is encapsulated into a structured object and placed into a candidate rule set for subsequent processing steps.

[0069] For each rule in the candidate rule set, calculate a comprehensive matching score. This score is derived by a weighted sum of sub-scores across three dimensions: The first dimension, matching score, calculates the matching score between the current textural information of the target raw material and the range of textural features in the rules. If there is measured data for the target raw material, the measured values ​​are used for comparison; if a predicted profile is used, the predicted central trend value is used for comparison. The degree of conformity for each feature dimension is considered during the calculation, and a weighted sum is applied.

[0070] The second dimension, matching score, calculates the matching score between the target raw material's origin and growing environment characteristics and the environmental characteristic range in the rules. The calculation logic is similar to the first dimension, but may use different feature sets and weights.

[0071] The third dimension, matching score, is only enabled when the geographic preference adaptation step has been performed. It calculates the degree of fit between the standard processing strategy template corresponding to the current rule, which typically produces product quality styles such as high numbing intensity or high aroma, and the adjustment factors obtained from the preference knowledge base, i.e., the specific preference description of the target market. For example, if the market prefers high numbing intensity, and the historical products produced by a certain rule typically have high numbing intensity, then its third dimension matching score will be high.

[0072] Weighting coefficients were preset for these three dimensions, and the three sub-scores were weighted and summed to obtain the comprehensive matching score for each rule. Subsequently, the candidate rule set was sorted in descending order based on this score.

[0073] Based on the ranking results, the top K rules with the highest overall matching degree are selected as core reference rules. K is typically 1-3. If K=1, strategy generation is simplest, directly using the standard processing strategy template corresponding to this unique core rule as the base template for subsequent processing. If K>1, the system initiates template fusion logic. The fusion process is as follows: First, compare the standard processing strategy templates of these core rules to identify their common process step sequences and the differences. For the common parts, a convergence strategy is adopted, such as taking the intersection of the same parameter ranges in several templates, or calculating a weighted average based on the matching degree weight, as the parameter for that step in the fused template. For the differences, an optimization strategy is adopted: usually, the process steps and parameters in the rule with the highest overall matching degree are selected as the backbone of the fusion template. At the same time, the differentiated and distinctive steps or parameters in other highly matching rules are added to the corresponding positions of the fusion template as alternative solutions or optimization suggestions for production personnel to refer to. Finally, a base template that integrates the advantages of multiple rules is generated.

[0074] Using the basic template obtained in the previous step as a blueprint, the system performs final customized adjustments and information enrichment. First, if a regional preference adjustment factor exists, the system will call a preset parameter adjustment mapping table. This mapping table defines the correspondence between different preference descriptions, such as increasing roughness, and process parameter adjustment actions, such as lowering the lower limit of the first stage drying temperature by 5°C. Based on the specific adjustment factor content, the system automatically offsets and adjusts the relevant process parameters in the basic template, generating a fine-tuned strategy version. Second, the system supplements the strategy with rich meta-information. This information does not directly guide operations but increases the interpretability and practicality of the strategy, including: traceability of the main rule IDs on which the production cost strategy is based; assumptions regarding key characteristics of the target raw materials, especially predictive characteristics; parameter adjustment tolerance suggestions considering differences in equipment conditions in different factories; and suggestions for key quality control points that need to be monitored during processing.

[0075] All the determined information—the final processing technology type, the detailed sequence of process steps, the key parameter control range for each step, the expected product quality target value, and all the supplementary metadata mentioned above—is organized and packaged in a pre-designed, machine-readable and human-readable structured format such as JSON, XML, or a specific document template. This packaged, complete data package constitutes the final processing adaptation strategy and will be passed to step S8 for output.

[0076] Furthermore, this method sets ideal raw material texture feature vectors for the granulation and drying direction, the crushing and extraction processing direction, and the fine grinding processing direction. The specific feature terms are dynamically adjusted according to whether the target raw material is Sichuan pepper. When the target raw material is Sichuan pepper, in the ideal raw material texture feature vector of whole grain drying, the expected range of the numbing substance content feature item is set to ensure that it is above the threshold required for commercial grade, and the expected range of the volatile oil content feature item is set to be sufficient to support the level of characteristic aroma. In the ideal raw material texture feature vector for crushing and extraction processing, the expected range of the volatile oil content feature term is given the highest priority. In the ideal raw material texture feature vector for fine grinding processing, the hardness and brittleness features in the basic physical feature group are given higher weights, while the proportion of closed-eye seeds is required to be below a certain threshold. The processing suitability evaluation algorithm automatically loads the corresponding ideal raw material texture feature vector set according to the variety type of the target raw material for calculation, thereby ensuring that the scoring model closely matches the variety characteristics of the raw material.

[0077] In this embodiment, after each generated processing adaptation strategy is actually applied to the production line, subsequent feedback data is collected actively or passively. This data is mainly divided into two parts: first, actual processing log data, which is recorded and provided by the production line controllers or operators, detailing the parameters actually used in each process step, such as the actual baking temperature curve and the actual pulverizer speed; second, quality evaluation data of the final product, which can come from laboratory instrument testing reports, such as texture retesting and flavor substance content determination of the finished product, or from scores by a standardized sensory evaluation team.

[0078] Through single-factor experiments, referring to parameter settings in relevant literature and the instrument's default parameters, the following operations were performed: the texture analyzer used single compression, with a pre-test speed of 3.0 mm / s, a test speed of 1.0 mm / s, a post-test speed of 3.0 mm / s, a trigger force of 20.0 g, a target mode of deformation, and a deformation rate of 30%. This was repeated 15 times. The coefficient of variation (COP) of a single hardness measurement of the same fresh green peppercorn was compared, i.e., (standard deviation / mean) × 100%. A lower COP indicated better stability.

[0079] Using the optimal probe from the above experiments, a single-factor experiment on compression was designed with deformation values ​​of 30%, 40%, 50%, 60%, 70%, 80%, and 90%. Other parameters were: single compression test mode; pre-test speed 3 mm / s; test speed 1 mm / s; post-test speed 3 mm / s; trigger force 20 g. Hardness was measured and repeated 15 times. The coefficients of variation (CVA) of the three hardness measurements of the same sample under different parameters were compared. A lower CVA indicates better stability. The average CVA of these 15 CVA values ​​was then calculated, reflecting the stability of the measurement method itself or the uniformity of a local area.

[0080] Using the optimal probe and parameters determined in the above experiments, the test speeds were set to 2, 1, and 0.5 mm / s, and the triggering forces were set to 10, 20, and 30 g, with other parameters remaining unchanged. A total of 6 single-factor experiments were conducted to determine the hardness of fresh green peppercorns. Each fresh green peppercorn was measured once, and the results were repeated 15 times. The average hardness and coefficient of variation were calculated in the same way.

[0081] Triggering forces of 5, 10, 20, and 30 g were used to determine the hardness of fresh green Sichuan peppercorns in four single-factor experiments, with all other parameters remaining unchanged. Each fresh green Sichuan peppercorn was measured once, and the results were repeated 15 times. The average hardness and coefficient of variation were calculated in the same way.

[0082] One-way ANOVA was performed using SPSS 26.0, with P < 0.05 indicating a significant difference. Experimental data were statistically analyzed and charts were generated using Excel 2010. The Baosheng Texture Data Analysis System was also used.

[0083] The TA2 probe measures greater hardness than the TA36 probe, but its overlap is poor. It can measure pericarp strength and pericarp toughness. The TA2 probe simulates the biting action of incisors. Therefore, the TA2 probe better reflects the texture of whole green Sichuan peppercorns, while the TA36 probe simulates the chewing action of teeth and is more convenient for analyzing the texture of the seeds.

[0084] Based on the overlap of hardness change curves measured by a series of deformation tests, 30% deformation showed the best repeatability. Furthermore, 30% deformation completely crushed the seed husks, achieving the testing objective. Other deformation rates, such as 40% and above, only made the fresh green peppercorns more fragmented, with 80% becoming flakes. 90% deformation was unsuitable for testing, often resulting in overload and making the test impossible to complete. Therefore, 30% deformation was the optimal result.

[0085] Reference Figure 5 Based on the test results and the damage to the fresh green Sichuan peppercorns, a puncture distance of 1.5 mm is optimal. A puncture distance of 0.5 mm will not penetrate the fruit skin, a puncture distance of 1.0 mm will damage most of the peel and may not reach the inside of the seed, while a puncture distance of 3.0 mm may result in over-testing. Therefore, a puncture distance of 1.5 mm is the most suitable parameter.

[0086] The results showed that changes in triggering force within a certain range had some impact on the hardness and coefficient of variation of fresh green Sichuan pepper. Aside from differences between triggering forces of 20.0 g and 15 g, there were no differences compared to other treatments. Based on existing research, excessively high triggering forces can lead to large data abrupt changes, making it impossible to analyze some textural characteristics. Therefore, the triggering force should not be too high. This study determined a triggering force of 20.0 g as the optimal parameter.

[0087] Table 2. Results of the effect of different triggering forces on the hardness determination of fresh green Sichuan pepper. ; When the system determines that the target raw material is Sichuan pepper, including both green and red varieties, the algorithm loads a set of ideal raw material texture feature vectors specifically customized for Sichuan pepper from the configuration library before starting. The core difference between this customized set and the general set is that it assigns specific and prioritized expected values ​​to the key chemical indicators unique to Sichuan pepper.

[0088] For whole-grain drying, the customized vector for Sichuan peppercorns, while setting the expected range of physical characteristics such as skin cracking force, will particularly emphasize its chemical characteristics: the lower limit of the expected range of numbing substance content must be set no less than the minimum standard value required to ensure the commodity grade of this variety, such as the industry standard for premium red Sichuan peppercorns, while the expected range of volatile oil content is set at a level sufficient to support the perception of its typical aroma characteristics, rather than a general value.

[0089] For the crushing and extraction processing direction, when setting the customized vector for Sichuan pepper, the expected range of the volatile oil content characteristic is given the highest priority, because this is the main target of the extraction process, and its ideal range is set high to pursue economic benefits. At the same time, since the extraction process can usually also extract numbing substances, the expected range of numbing substance content is also set at a high level to ensure the overall flavor value of the extract.

[0090] For the fine grinding and processing direction, while the customized vector for Sichuan peppercorns focuses more on physical properties, it also reflects the unique characteristics of Sichuan peppercorns. In the basic physical characteristic group, it assigns higher weights to hardness and brittleness, as these two characteristics directly affect grinding efficiency and particle size distribution. Furthermore, it introduces an additional characteristic crucial for whole-peppercorn processing—the proportion of closed-eye seeds—and sets a strict upper limit for it, such as <5%, because too many seeds can negatively impact the texture and flavor of the powder.

[0091] Throughout the evaluation algorithm's operation, the system automatically selects and loads a set of ideal raw material texture feature vectors rich in expected varietal characteristics based on the target raw material's variety type identifier. This mechanism ensures that the ideal benchmark used by the algorithm in calculating the matching degree and scoring is highly correlated with the raw material's characteristics, thereby enabling the final fitness score to more realistically and accurately reflect the raw material's actual potential in the corresponding processing direction.

[0092] According to a second embodiment of the present invention, the application of an evaluation method for suitability analysis of Sichuan pepper processing claimed in the present invention is described in the evaluation of suitability analysis of Sichuan pepper processing.

[0093] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. An evaluation method suitable for the suitability analysis of Sichuan pepper processing, characterized in that, The method is applicable to granular seasoning ingredients, including green Sichuan peppercorns, red Sichuan peppercorns, and black pepper, and includes the following steps: S1. Establish and maintain a raw material characteristic knowledge base, storing sample records of multiple known varieties of granular seasoning raw materials. The sample records are associated with the sample's identification information, texture characteristic data set, and growth environment characteristic data set of the sample's place of origin. S2, Construct a processing strategy rule base, which stores processing strategy rules and defines the correspondence between the raw material characteristic pattern composed of the range of texture characteristic data and the range of growth environment characteristic data and the recommended set of processing parameters; S3, receive input information for the target raw material, the input information including at least the variety and origin information of the target raw material; S4. Based on the variety type and origin information of the target raw material, retrieve sample records from the raw material characteristic knowledge base that are the same or similar to the variety type and match the characteristics of the growing environment of the origin, and form a reference sample set. S5. Based on the textural feature data of the samples in the reference sample set, and combined with the preset processing suitability evaluation algorithm, for the pepper variety, the content of numbing substances and volatile oil in the textural feature data set are assigned a weight coefficient higher than that of the basic textural feature items to calculate the adaptability score of the target raw material in multiple predefined processing directions. S6. Based on the growth environment characteristic data corresponding to the origin information of the target raw material and the calculated adaptability score, the applicable processing strategy rules are matched from the processing strategy rule base. S7. Based on the matched processing strategy rules, generate a processing adaptation strategy for the target raw material. The processing adaptation strategy includes at least the recommended processing technology type, the range of key process parameters, and the expected product quality target. S8, output the processing adaptation strategy to guide the processing and production of the target raw material.

2. The method according to claim 1, characterized in that, When the target raw material is a new variety that has not yet been entered into the raw material characteristic knowledge base, steps S4 and S5 are replaced by the following steps: S4a, based on the variety type and origin information of the target raw material, retrieve known variety sample records from the raw material feature knowledge base that belong to the same category as the target raw material and whose origin growth environment characteristics have a similarity exceeding a preset similarity threshold, and form a prediction reference set; S5a, integrate and analyze the textural feature data of all samples in the prediction reference set, and use the pepper feature weighted clustering algorithm to generate the predicted textural feature profile of the target raw material. When calculating the distance between samples, the pepper feature weighted clustering algorithm applies higher weights to the dimensions of numbing substance content and volatile oil content. S5b, based on the generated predicted texture feature profile and combined with the preset processing suitability evaluation algorithm, calculate the suitability score of the target raw material in multiple predefined processing directions.

3. The method according to claim 1, characterized in that, The construction and execution of the preset processing suitability evaluation algorithm in step S5 specifically includes: Define the basic processing directions, including granulation and shape-preserving drying, crushing and extraction, and fine grinding. An ideal raw material texture feature vector is set for each of the basic processing directions. Each feature dimension in the ideal raw material texture feature vector has an expected value or expected range. For the processing direction involving Sichuan pepper, the ideal raw material texture feature vector includes the expected range of the content of numbing substances and the content of volatile oils. Obtain the statistical distribution of the textural feature data of the reference sample set, and generate the estimated textural feature profile of the target raw material based on the statistical distribution. The estimated textural feature profile includes the median estimate and confidence interval of each feature. For each basic processing direction, an adaptability score calculation sub-step is performed, which includes: The feature matching degree between the estimated textural feature profile of the target raw material and the ideal raw material textural feature vector for the corresponding processing direction is calculated. The feature matching degree is calculated using the weighted Manhattan distance algorithm. For Sichuan pepper varieties, the weights of the numbing substance content and volatile oil content feature dimensions are set to configurable boost values. These boost values ​​are dynamically adjusted according to the sensitivity of the processing direction to aroma and numbing intensity. The confidence interval width of each feature in the estimated texture feature profile is evaluated and compared with the tolerance range of the process parameters in the corresponding processing direction to generate a feature stability score. The reciprocal of the feature matching degree is linearly weighted and fused with the feature stability score to generate the final fitness score, wherein the weight of the feature stability score will increase accordingly when the number of samples in the reference sample set is less than a threshold.

4. The method according to claim 1, characterized in that, The method also includes a regional preference adaptation step performed before step S7, specifically including: Maintain a regional dietary preference knowledge base, which stores descriptions of core flavor and taste preferences for processed condiment products corresponding to different regional identifiers; The target geographical identifier is determined based on the target sales market information of the target raw material; Based on the target region identifier, obtain the corresponding flavor and taste preference description from the regional dietary preference knowledge base; The obtained flavor and taste preference descriptions are used as adjustment factors and input into the strategy generation process; When generating a processing adaptation strategy based on the matched processing strategy rules, the set of processing process parameters recommended by the rules is fine-tuned according to the adjustment factor.

5. The method according to claim 1, characterized in that, The method also includes a collaborative optimization mechanism between the knowledge base and the rule base. This mechanism is triggered after each execution of processes S1 to S8 and after receiving actual production feedback, and includes the following steps: Collect feedback data, which includes records of key process parameters executed during actual production based on the processing adaptation strategy, as well as quality inspection data of the final product. The execution records of the key process parameters are compared with the range of process parameters recommended in the processing adaptation strategy to calculate the process execution compliance. The quality inspection data of the final product is compared with the expected product quality target in the processing adaptation strategy to calculate the degree of achievement of the quality target; When the process execution compliance is higher than a preset compliance threshold while the quality target achievement is lower than a preset achievement threshold, it is determined that the current recommended strategy has room for optimization, and the analysis process is initiated. The analysis process includes: By comparing the quality data of the actual product with the typical quality data of the samples in the reference sample set, and combining the growth environment characteristic data of the target raw material, the key textural or environmental characteristics that cause the deviation are identified. Based on the analysis results, optimization instructions for the processing strategy rule base are generated; The approved optimization instructions are applied to the processing strategy rule base, and the raw material characteristics and origin environment combinations associated with this optimization are recorded simultaneously to enrich the empirical data of the raw material characteristic knowledge base.

6. The method according to claim 1, characterized in that, The construction process of the processing strategy rule base specifically includes: Collect historical successful processing cases, each case including complete textural characteristics data of raw materials, growth environment characteristics data of the place of origin, specific processing technology and detailed process parameters used, and quality evaluation report of the final product; The collected cases are processed using a hierarchical clustering algorithm based on feature importance. The first-level classification is based on the type of raw material. Within the same type, the second-level clustering is based on the core textural features and core environmental features that have the greatest impact on processing quality. Cases with similar processing paths are grouped into the same case cluster.

7. The method according to claim 1, characterized in that, When the texture feature data set is applied to the processing suitability evaluation algorithm, the algorithm parameters are configured differently according to the type of raw material, specifically as follows: For Sichuan pepper varieties, the texture feature data set is divided into a key quality feature group and a basic physical feature group; In the processing suitability evaluation algorithm, different feature weight vectors are configured for different processing directions. When evaluating processing directions involving the preservation of Sichuan pepper aroma and flavor, the total weight assigned to the key quality feature group is higher than that of the basic physical feature group. When evaluating processing directions mainly involving changes in physical morphology, the total weight assigned to the basic physical feature group is higher than that of the key quality feature group. The weight vector is predefined based on the correlation analysis results between different features and processing results in historical cases, and is iteratively fine-tuned based on feedback data in the collaborative optimization mechanism.

8. The method according to any one of claims 1, 2 or 4, characterized in that, Step S7 generates a processing adaptation strategy for the target raw material based on the matched processing strategy rules, including: S71, parse one or more processing strategy rules matched from the processing strategy rule base, extract the raw material feature pattern conditions defined by each rule, the corresponding standard processing strategy template, and the historical call success rate of the rule, and put the parsed rule into the candidate rule set; S72, Multidimensional matching degree calculation and ranking: For each rule in the candidate rule set, calculate the matching degree between the current information of the target raw material and the raw material feature pattern conditions defined by the rule; S73, Strategy Template Selection and Fusion: Select the top K rules with the highest overall matching degree as core reference rules; if K=1, directly use the standard processing strategy template corresponding to that rule as the base template; if K>1, initiate the template fusion logic, which includes: By comparing the standard processing strategy templates corresponding to each core reference rule, the common and different parts of their process step sequences can be identified. For the common parts, the intersection or weighted average range of the process parameter ranges of each template is taken as the parameters of the merged template; For the difference part, based on the comprehensive matching degree weight of each rule, the step in the template with the highest weight is selected as the backbone of the fusion template, and the advantage parameters of the difference steps in other templates are added to the fusion template as optional or annotation. S74, based on the selected or merged basic template, makes the following adjustments and enrichments: If a regional preference adjustment factor exists, the preset parameter adjustment mapping table is called, and the corresponding process parameters in the template are offset and adjusted according to the specific preference description item to generate a fine-tuned strategy version. The supplementary strategy metadata includes: the source of the rules on which the strategy is generated, the hypothesis on the key characteristics of the target raw materials, the parameter adjustment suggestions under different production scales or equipment conditions, and the monitoring suggestions for core quality control points. S75 organizes the finalized processing technology type, process step sequence, key process parameter range, expected quality target, and supplementary metadata into a complete processing adaptation strategy document according to a predefined structured format.

9. The method according to claim 1, characterized in that, The ideal raw material texture feature vector is set for the granulation and drying direction, the crushing and extraction processing direction, and the fine grinding processing direction. The specific feature terms are dynamically adjusted according to whether the target raw material is Sichuan pepper. When the target raw material is Sichuan pepper, in the ideal raw material texture feature vector of the whole grain shape-preserving drying direction, the expected range of the numbing flavor substance content feature item is set to ensure that it is above the threshold required for commercial grade, and the expected range of the volatile oil content feature item is set to be sufficient to support the level of characteristic aroma. In the ideal raw material texture feature vector of the crushing and extraction processing direction, the expected range of the volatile oil content feature term is given the highest priority. In the ideal raw material texture feature vector of the fine grinding processing direction, the hardness and brittleness features in the basic physical feature group are given higher weights, while the proportion of closed-eye seeds is required to be below a certain threshold. The processing suitability evaluation algorithm automatically loads the corresponding ideal raw material texture feature vector set according to the variety type of the target raw material for calculation, thereby ensuring that the scoring model closely matches the variety characteristics of the raw material.

10. The application of the evaluation method for suitability analysis of Sichuan pepper processing as described in any one of claims 1-9 in the evaluation of suitability analysis of Sichuan pepper processing.

Images