CN-122022204-A - Intelligent quantitative detection method for compound synergistic effect of fermented coarse cereals

CN122022204ACN 122022204 ACN122022204 ACN 122022204ACN-122022204-A

Abstract

The invention discloses an intelligent quantitative detection method for a compound synergistic effect of fermented coarse cereals, and belongs to the technical field of food processing and intelligent detection. According to the invention, multi-dimensional characteristic data of all-period time sequence nodes of the mixed fermentation of different coarse cereals are collected, a standardized interval number characteristic matrix containing characteristic fluctuation intervals and time sequence characteristics is constructed through pretreatment, a synergistic effect interval fuzzy comprehensive evaluation system is constructed, an interval fuzzy relation matrix representing characteristics and a synergistic grade membership degree is constructed based on the characteristic matrix, weight calculation and interval fuzzy comprehensive operation are completed, synergistic effect two-dimensional grading calibration is realized by combining dynamic stability indexes, and finally a detection result of a system to be detected is output through a solidified intelligent quantitative detection model. The invention realizes standardized intelligent quantitative detection of the synergistic effect, improves objectivity and comprehensiveness of detection results, adapts to research and development requirements of different products, and provides standardized data support for research and development of a fermented coarse cereal compound formula.

Inventors

QI HUI
ZOU LIANG
BAI JUHONG

Assignees

成都农业科技职业学院

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. The intelligent quantitative detection method for the compound synergistic effect of the fermented coarse cereals is characterized by comprising the following steps of: s1, collecting multi-dimensional characteristic data of fermentation full-period time sequence nodes of different coarse cereal compound groups, preprocessing the collected multi-dimensional characteristic data, and constructing a standardized interval number characteristic matrix, wherein the interval number characteristic matrix comprises fluctuation intervals and time sequence characteristics of each characteristic data in the fermentation full period; S2, constructing a synergistic effect interval fuzzy comprehensive evaluation system, setting a synergistic effect comment set and a corresponding grade boundary, and constructing an interval fuzzy relation matrix based on an interval number feature matrix, wherein the interval fuzzy relation matrix represents membership of each feature data to different synergistic effect grades; S3, calculating a weight distribution result of the characteristic data, executing interval fuzzy comprehensive operation by combining an interval fuzzy relation matrix to obtain a synergistic effect quantized score interval and a basic grade, and calculating a synergistic effect dynamic stability index by combining fermentation full-period time sequence data to complete synergistic effect two-dimensional hierarchical calibration; S4, collecting time sequence characteristic data of the coarse cereal compound system to be tested, performing preprocessing consistent with the S1, inputting the processed data into a solidified intelligent quantitative detection model of the synergistic effect, and outputting a detection result of the synergistic effect of the coarse cereal compound system to be tested.
2. The method according to claim 1, characterized in that step S1 comprises the sub-steps of: s1.1, setting a single-coarse-cereal blank control group and a multi-coarse-cereal compound experimental group, and performing fixed setting on fermentation process parameters corresponding to all groups, wherein the fermentation process parameters comprise fermentation temperature, fermentation time, strain inoculation amount, system solid-liquid ratio and sterilization parameters, and the fermentation process parameters corresponding to all groups adopt consistent set values; S1.2, setting time sequence acquisition nodes based on a coarse cereal fermentation process, and acquiring multi-dimensional characteristic data of all groups of samples at each time sequence acquisition node, wherein the multi-dimensional characteristic data comprise physicochemical characteristic data, microorganism characteristic data, flavor characteristic data, nutrition function characteristic data and finished product quality calibration data; S1.3, performing missing value filling, outlier rejection and standardized normalization preprocessing on the collected multidimensional characteristic data, and constructing a single characteristic interval number based on the preprocessed time sequence data, wherein the single characteristic interval number comprises an interval lower limit, an interval middle point, an interval upper limit and a time sequence fluctuation coefficient; S1.4, constructing a standardized interval number feature matrix based on all single feature interval numbers, wherein rows of the interval number feature matrix correspond to coarse cereal compound samples, and columns of the interval number feature matrix correspond to feature data items.
3. The method according to claim 1, characterized in that step S2 comprises the sub-steps of: s2.1, constructing a secondary evaluation factor set, wherein the secondary evaluation factor set comprises a primary index and a secondary index, the primary index comprises a physicochemical synergistic factor, a microbial synergistic factor, a flavor synergistic factor, a nutrition function synergistic factor and a finished product quality synergistic factor, and the secondary index is a specific characteristic data item corresponding to the primary index; S2.2, setting a synergistic effect comment set and a corresponding grade boundary, wherein the synergistic effect comment set comprises no synergy, weak synergy, medium synergy, strong synergy and super strong synergy, and setting an interval boundary value corresponding to each synergistic grade based on reference data of a single coarse cereal blank control group; S2.3, constructing a membership function by adopting the trapezoidal interval fuzzy number, and calculating interval membership degrees of each secondary index to different cooperative grades based on the membership function; S2.4, constructing an interval fuzzy relation matrix based on interval membership calculation results of all the secondary indexes.
4. The method according to claim 1, characterized in that step S3 comprises the sub-steps of: S3.1, calculating objective basic weights of all secondary indexes based on an interval number feature matrix through an entropy weight method, and completing adjustment of weight distribution results by combining set scene adjustment rules; s3.2, based on the weight distribution result and the interval fuzzy relation matrix, performing double-layer weighting operation of the secondary index and the primary index by adopting an interval number weighting synthesis algorithm to obtain the comprehensive interval membership of the sample to each cooperative level; s3.3, determining a synergistic effect basic grade based on a maximum membership principle, and calculating to obtain a synergistic effect quantized score interval through an interval gravity center method; S3.4, constructing a synergistic effect dynamic evolution curve based on the synergistic effect instantaneous values of all time sequence nodes in the whole fermentation period, calculating to obtain a synergistic effect dynamic stability index, and completing the two-dimensional grading calibration by combining the synergistic effect basic grade.
5. The method according to claim 1, characterized in that step S4 comprises the sub-steps of: S4.1, acquiring characteristic data of a fermentation full-period time sequence node of a coarse cereal compound system to be tested according to an acquisition standard consistent with a training sample, wherein the dimension of the characteristic data is completely matched with the characteristic dimension of an interval number characteristic matrix; s4.2, preprocessing the obtained characteristic data of the coarse cereal compound system to be tested by adopting a preprocessing rule consistent with the training sample, and constructing an interval number characteristic matrix of the sample to be tested; s4.3, inputting the interval number feature matrix of the sample to be detected into a solidified synergistic effect intelligent quantitative detection model, and executing automatic operation through the synergistic effect intelligent quantitative detection model to obtain an operation result; s4.4, outputting a synergistic effect quantization score interval, a synergistic effect grade and a dynamic stability grade of the coarse cereal compound system to be detected, generating a standardized detection result, and completing a detection flow.
6. The method according to claim 2, wherein in step S1.2, the time sequence collection nodes cover a fermentation delay period, a log-phase growth period early period, a log-phase growth period late period, a stationary phase early period, a stationary phase late period and a fermentation end point, the time intervals of the time sequence collection nodes are adaptively adjusted based on the acidity change rate and the reducing sugar content change rate of the coarse cereal fermentation system, each time sequence collection node adopts the same collection flow and detection standard to execute multi-dimensional characteristic data collection, each group sample is provided with a corresponding number of parallel samples, the collected multi-dimensional characteristic data are bound with the corresponding group and the information of the corresponding time sequence collection node one by one, and the bound multi-dimensional characteristic data are matched with the reference data of a single coarse cereal blank control group in dimension, so that a unified-dimension basic data source is provided for the construction of the subsequent single characteristic interval number.
7. The method according to claim 3, wherein in step S2.3, the interval boundary value of the membership function is set based on the theoretical sum value and the set synergy amplitude of each index of the single-grain blank control group, each secondary index corresponds to an independent membership function and the interval boundary value, the membership function of the trapezoid interval fuzzy number adopts a four-segment linear calculation rule, corresponds to an interval with 0 membership, an interval with 1 membership and a linear decline of membership respectively, and the membership calculation of each collaborative level is completed based on the membership function to obtain an interval membership result of the corresponding secondary index, thereby providing a standardized calculation basis for the construction of an interval fuzzy relation matrix.
8. The method according to claim 4, wherein in step S3.1, the adjustment rule includes weight adjustment coefficients corresponding to instant porridge scenes, meal replacement scenes and nutrient enrichment scenes, when objective basic weights are calculated by entropy weight method, information entropy values of the secondary indexes are calculated first, the information entropy values and index distinction degree are in inverse correspondence, basic weight calculation of the secondary indexes is completed based on the information entropy values, then directional adjustment is performed on basic weights of the primary indexes by combining the weight adjustment coefficients of the corresponding scenes, weight normalization processing is performed after adjustment is completed, it is ensured that weight sum of all the primary indexes is kept to be set value, and standardized weight basis is provided for subsequent interval fuzzy comprehensive operation.
9. The method according to claim 4, wherein in step S3.4, the synergistic effect dynamic stability index includes a fluctuation coefficient, a stability coefficient and an effective synergistic duration, the synergistic effect dynamic evolution curve is constructed by taking a fermentation time length as an abscissa and a synergistic effect instantaneous value of each time sequence node as an ordinate, three dynamic stability index calculation is completed based on the synergistic effect dynamic evolution curve, contribution degree values of each characteristic data of different fermentation stages to the synergistic effect are calculated through time sequence sensitivity analysis, driving factors influencing the synergistic effect are screened based on a set contribution degree threshold, one-to-one binding is performed between the screened driving factors and corresponding characteristic data items, and standardized characteristic basis is provided for optimization of a follow-up coarse cereal compound system.
10. The method according to claim 1, wherein in step S4, the synergistic effect intelligent quantization detection model includes a data preprocessing module, an interval number construction module, a fuzzy comprehensive operation module, a dynamic stability calculation module, and a result output module; The intelligent quantitative detection model of the synergistic effect adopts a serial unidirectional hierarchical connection structure, the output end of the data preprocessing module is connected with the input end of the interval number construction module, the output end of the interval number construction module is connected with the input end of the fuzzy comprehensive operation module, the output end of the fuzzy comprehensive operation module is connected with the input end of the dynamic stability calculation module, and the output end of the dynamic stability calculation module is connected with the input end of the result output module; The training step of the synergistic effect intelligent quantitative detection model comprises the steps of taking a marked interval number feature matrix and a synergistic effect calibration result as training samples, setting training iteration times, learning rate and batch size as corresponding set values, taking the synergistic effect calibration result as a supervision label to execute supervision training, executing fitting effect verification and parameter adjustment through a verification set, and completing training and solidification of the synergistic effect intelligent quantitative detection model when the performance index of the synergistic effect intelligent quantitative detection model reaches a set threshold value, wherein the input item of the synergistic effect intelligent quantitative detection model is time sequence feature data of a coarse cereal compound system to be tested, and the output item is a synergistic effect quantitative score interval, a synergistic effect grade and a dynamic stability grade.

Description

Intelligent quantitative detection method for compound synergistic effect of fermented coarse cereals Technical Field The invention relates to the technical field of food processing and intelligent detection, in particular to an intelligent quantitative detection method for a compound synergistic effect of fermented coarse cereals. Background Along with the continuous popularization of resident healthy diet concepts, the coarse cereal food with balanced nutrition and edible convenience becomes one of the core directions of research and development and market expansion of the food industry, wherein the fermented coarse cereal product realizes wide application and rapid development in a plurality of products such as instant food, meal replacement food, nutrition-enhanced food and the like by virtue of the natural nutrition advantages of coarse cereal raw materials and the quality improvement effect brought by a fermentation process. The raw materials of the coarse cereals are rich in dietary fibers, minerals, active polyphenol and various functional components, the complementation of nutritional components and the cooperation of functional characteristics can be realized by the compound combination of different raw materials of the coarse cereals, the taste characteristics of the coarse cereals can be further improved by the microbial fermentation process, the content of the active nutritional components is improved, the flavor expression and the storage stability of the products are optimized, and a complete industry research and development system covering the screening of raw materials of the coarse cereals, the design of compound formula, the optimization of fermentation process parameters and the evaluation of a finished product quality system is formed in the industry at present. The method is characterized in that the method is used for detecting the physical and chemical properties, microorganism metabolism, flavor substance composition, nutrition functional component change and other dimensions of the coarse cereal fermentation process, so that a mature standardized detection method and operation specifications are formed, comprehensive and accurate data representation of the full period state of coarse cereal fermentation can be realized, meanwhile, the digital technologies such as machine learning, intelligent data analysis, fuzzy mathematical comprehensive evaluation and the like are used for realizing mature floor application in the fields of food formula research and development and product quality intelligent evaluation, and a perfect technical basis and application environment are provided for digital and intelligent research and development of fermented coarse cereal products. In the research and development process of a fermented coarse cereal compound system, the prior art is difficult to realize standardized and intelligent quantitative detection of a coarse cereal compound fermentation synergistic effect, the conventional evaluation mode aiming at the coarse cereal compound synergistic effect is mostly dependent on manual detection and subjective sensory evaluation of a single index of a fermentation end point, the inherent correlation between full-period multi-dimensional time sequence characteristic data of fermentation and the compound synergistic effect cannot be fully excavated, and the comprehensive and objective quantitative result is difficult to be formed on the synergistic effect of the compound system. The prior art can only complete the static evaluation of the fermentation end point, can not characterize the dynamic change characteristic and fluctuation rule of the coarse cereal compound synergistic effect in the whole fermentation period, can not realize the two-dimensional comprehensive calibration of the static quantification result and the dynamic stability characteristic of the synergistic effect, and is difficult to completely reflect the real expression of the synergistic effect of the compound system. Meanwhile, the prior art cannot effectively solve the quantization problems of fuzzy index boundaries and data uncertainty in the collaborative effect evaluation process, deviation of an evaluation result is easy to occur, the whole process is difficult to realize standardized and automatic execution of the detection process, the operation process difference of different detection batches and different operators is easy to cause insufficient consistency and comparability of the detection result, and a unified and stable evaluation basis cannot be provided for research and development of a coarse cereal compound formula. In addition, the prior art cannot realize the rapid intelligent detection of the synergistic effect of the coarse cereal compound system to be detected through standardized model construction, training and curing processes, and aiming at a new coarse cereal compound system, the manual detection and subjective evaluation of the complete process are require