System and method for digitally analyzing and discriminating quality of rice husks for brewing wine

Abstract

The invention discloses a wine brewing rice hull quality digital analysis and discrimination system and method, and relates to the technical field of wine brewing auxiliary material quality analysis and evaluation, and the system comprises a flavor analysis module, a texture analysis module, a data processing module and a quality digital discrimination module. According to the method, GC-IMS flavor detection and texture analyzer quantitative analysis are introduced into wine brewing rice hull quality evaluation for the first time, multi-dimensional and multi-view quality scientific characterization is achieved, the blank of a traditional evaluation technology is filled up, a rice hull quality digital judgment model is constructed based on machine learning, and the evaluation accuracy is improved. Sensory evaluation depending on experience is converted into a quantitative index system based on flavor composition and texture characteristics, objective analysis and data-driven evaluation of rice hull quality are achieved, the bottleneck of traditional evaluation timeliness and precision is broken through, the efficient and accurate screening requirement is met, a scientific foundation is laid for accurate regulation and control and raw material management and control of the rice hull steaming process, and the method is suitable for large-scale popularization and application. A new technical support is provided for guaranteeing the fermentation stability and the flavor consistency of the wine body.

Description

A digital analysis and discrimination system and method for the quality of rice husks used in brewing Technical Field

[0001] This invention relates to the field of brewing auxiliary material quality analysis and evaluation technology, specifically to a digital analysis and discrimination system and method for brewing rice husk quality. Background Technology

[0002] Rice husks are an important filler and loosening agent in the brewing process of baijiu (Chinese liquor), and their quality directly affects the permeability of the mash, fermentation efficiency, and even the final flavor and yield of the baijiu. Currently, brewing companies mainly rely on the sensory experience of veteran brewers (such as sight, touch, and smell) to evaluate and select rice husks. This method is highly subjective, has poor repeatability, is difficult to quantify, and is highly dependent on personal experience, making it impossible to achieve stable, large-scale standardized production. This has become a bottleneck restricting the modernization and upgrading of the baijiu industry.

[0003] Although existing single detection technologies can be used for rice husk analysis, they have significant limitations: for example, focusing only on a single dimension such as flavor or physicochemical properties makes it difficult to comprehensively characterize the overall quality; the raw data is complex and lacks effective feature screening methods, failing to accurately identify key indicators directly related to brewing suitability; and some chemical analysis methods also suffer from cumbersome and time-consuming sample pretreatment, failing to meet the online or rapid detection needs of enterprises.

[0004] As the brewing industry develops towards large-scale, standardized, and intelligent production, the inefficiencies and difficulty in quantification of traditional evaluation methods are becoming increasingly apparent. This leads to risks such as large fluctuations in raw material quality and poor batch-to-batch consistency, hindering improvements in production efficiency and quality stability control. The industry urgently needs a digital technology that can integrate multi-dimensional quality indicators, achieve objective quantitative analysis, and provide rapid and accurate judgment. This technology would overcome the limitations of traditional methods, establish a quality control system for the production and use of raw and auxiliary materials, and provide technical support for the scientific screening and efficient application of brewing rice husks. Against this backdrop, developing a digital analysis and judgment method for brewing rice husk quality, enabling precise quantification and rapid judgment of quality indicators, has become a key requirement for improving quality and efficiency in the brewing industry. Summary of the Invention

[0005] In view of the above-mentioned shortcomings in the prior art, the digital analysis and discrimination system and method for brewing rice husk quality provided by the present invention solves the problems of low efficiency, high error and difficulty in quantification of traditional brewing rice husk evaluation methods.

[0006] To achieve the aforementioned objectives, the present invention provides a digital analysis and discrimination system for the quality of brewing rice husks, comprising: a flavor analysis module employing gas chromatography-ion mobility spectrometry (GC-IMS) to acquire volatile organic compound (VOC) data from brewing rice husk samples; a texture analysis module employing texture analysis techniques to acquire multi-dimensional texture parameter data of brewing rice husk samples, including hardness, cohesion, elasticity index, adhesiveness, chewiness, and chewiness index; a data processing module for fusing and preprocessing the VOC data and multi-dimensional texture parameter data, and for selecting a set of key quality characteristic indicators for judging the quality of brewing rice husks based on multivariate statistical analysis and machine learning feature selection algorithms; and a digital quality discrimination module for automatically judging and outputting the quality grade of brewing rice husk samples based on the set of key quality characteristic indicators and using a machine learning classification model.

[0007] Furthermore, the machine learning classification model is a pre-trained support vector machine. During training, all brewing rice husk samples have their original labels removed and are labeled with random codes. The order of the brewing rice husk samples follows a randomization principle. Sensory evaluation quantitative data of the brewing rice husk samples are obtained by a predetermined number of sensory evaluators with predetermined qualifications. The machine learning classification model is trained with a set of key quality characteristic indicators as independent variables and sensory evaluation quantitative data as dependent variables. The sensory evaluation quantitative data includes two types of indicators: quantitative scoring and qualitative judgment. The quantitative evaluation uses a 0-9 point scale to independently score the core sensory dimensions of rice husks, where 0 points correspond to "very poor quality" and 9 points correspond to "very good quality". The qualitative judgment uses a binary "yes" or "no" standard, where "yes" indicates that the sample meets the requirements for use in brewing production and "no" indicates that it does not meet the requirements for use in brewing production. The average of the quantitative scores of all sensory evaluators for the same brewing rice husk sample is taken as the quantitative score result of that brewing rice husk sample.

[0008] A method for a digital analysis and discrimination system of brewing rice husk quality is provided, comprising the following steps: S1, using gas chromatography-ion mobility spectrometry to acquire volatile organic compound (VOC) data of brewing rice husk samples; S2, using texture analysis to acquire multi-dimensional texture parameter data of brewing rice husk samples, including hardness, cohesion, elasticity index, adhesiveness, chewiness, and chewiness index; S3, fusing and preprocessing the VOC data and multi-dimensional texture parameter data, and selecting a set of key quality characteristic indicators for judging the quality of brewing rice husks based on multivariate statistical analysis and machine learning feature selection algorithms; S4, using a machine learning classification model based on the set of key quality characteristic indicators to automatically judge and output the quality grade of brewing rice husk samples.

[0009] Furthermore, when using gas chromatography-ion mobility spectrometry (GC-IMS), the gas chromatographic parameters for GC-IMS are as follows: column: WAX capillary column, 30m*0.53mm*1μm; analysis time: 30min; column temperature: 60℃; carrier gas: N2; carrier gas flow rate: initial flow rate is 2mL / min, flow rate is 10mL / min at 10 minutes, and when it reaches 100mL / min, the flow rate is maintained for 10 minutes.

[0010] Furthermore, when using gas chromatography-ion mobility spectrometry (GC-IMS), the ion migration conditions for GC-IMS are as follows: N2 is used as the drift gas for IMS, the drift gas flow rate is 150 mL / min, and the drift tube temperature is 45 °C; automatic headspace sampling is used for injection, the injection volume is 500 μL, the incubation time is 15 min, the incubation temperature is 60 °C, the injection needle temperature is 85 °C, and the incubation rotation speed is 500 rpm.

[0011] Furthermore, the volatile organic compound (VOC) data of the brewing rice husk samples were obtained by qualitative and relative quantitative detection of the volatile compounds in the samples using VOCal analysis software and the NIST and IMS databases built into the VOCal software, based on gas phase retention time and ion migration time.

[0012] Furthermore, the specific method for obtaining multi-dimensional textural parameter data of brewing rice husk samples using texture analysis technology includes: using a texture analyzer equipped with a cylindrical test probe to determine the hardness, cohesiveness, elasticity index, adhesiveness, chewiness, and chewing index characteristics of 10.0g of rice husk. The measurement method is compression, the test standard is TPA, the test endpoint is displacement, the target value is 7mm, the trigger point is 10g, and the test speed is 1mm / s.

[0013] Furthermore, after obtaining the multidimensional texture parameter data of the brewing rice husk samples, the data and charts were viewed using the DATA / GRAPH VIEWS function in the Texture analysis software, and the data and results were analyzed using the ANALYSIS function.

[0014] Furthermore, the key quality characteristic indicators of brewing rice husks include key flavor factors and key textural characteristics factors that affect the quality of brewing rice husks. Among them, the key flavor factors affecting the quality of brewing rice husks are screened using partial least squares discriminant analysis and random forest analysis, with VIP≥1 and VIM≥Median as the criteria; the key textural characteristics factors affecting the quality of brewing rice husks are screened using correlation analysis, with p≤0.05 as the basis.

[0015] Furthermore, the machine learning classification model is a support vector machine, whose input is a set of key quality feature indicators and whose output is the quality grade of the brewing rice husk sample.

[0016] The beneficial effects of the present invention are: (1) For the first time, GC-IMS flavor detection and texture analyzer quantitative analysis are introduced into the quality evaluation of rice husks for brewing, realizing multi-dimensional and multi-perspective quality scientific characterization, filling the gap in traditional evaluation technology.

[0017] (2) A digital discrimination model for rice husk quality was constructed based on machine learning. The model's qualitative discrimination accuracy rate was 88.9%, and the entire process detection time could be shortened to less than 20 minutes, breaking through the bottleneck of traditional evaluation timeliness and accuracy, and meeting the needs of efficient and accurate screening.

[0018] (3) This invention breaks through the limitations of traditional sensory evaluation, transforms the experience-based sensory evaluation into a quantitative index system based on flavor composition and texture characteristics, realizes objective analysis and data-driven evaluation of rice husk quality, lays a scientific foundation for precise control of rice husk steaming process and raw material management, and provides new technology support for ensuring fermentation stability and consistency of wine flavor. Attached Figure Description

[0019] Figure 1 shows the architecture of this system; Figure 2 shows the flowchart of this method; Figure 3 shows the sensory evaluation results of rice husks; Figure 4 shows the difference in gas phase ion migration spectra of rice husk samples; Figure 5 shows the flavor packing diagram of rice husk samples; Figure 6 shows the permutation test diagram of partial least squares discriminant analysis; Figure 7 shows the cluster analysis results of rice husk samples based on flavor; Figure 8 shows the VIP score diagram of partial least squares discriminant analysis factors; Figure 9 shows the ROC curve of the prediction results of the random forest model; Figure 10 shows the VIM score diagram of random forest factors; Figure 11 shows the box plot of key flavor factors of rice husk samples; Figure 12 shows the box plot of key texture factors of rice husk samples; Figure 13 shows the prediction effect of the qualitative discriminant model for rice husk quality. Detailed Implementation

[0020] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0021] As shown in Figure 1, the digital analysis and discrimination system for brewing rice husk quality includes: a flavor analysis module, which uses gas chromatography-ion mobility spectrometry (GC-IMS) to acquire volatile organic compound (VOC) data of brewing rice husk samples; a texture analysis module, which uses texture analysis techniques to acquire multi-dimensional texture parameter data of brewing rice husk samples, including hardness, cohesion, elasticity index, adhesiveness, chewiness, and chewiness index; a data processing module, which fuses and preprocesses the VOC data and multi-dimensional texture parameter data, and selects a set of key quality characteristic indicators for judging the quality of brewing rice husks based on multivariate statistical analysis and machine learning feature selection algorithms; and a digital quality discrimination module, which uses a machine learning classification model to automatically judge and output the quality grade of brewing rice husk samples based on the set of key quality characteristic indicators.

[0022] In this embodiment, the machine learning classification model is a pre-trained support vector machine. During training, all brewing rice husk samples have their original labels removed and are labeled with random codes. The order of the brewing rice husk samples follows a randomization principle. Sensory evaluation quantitative data of the brewing rice husk samples are obtained by a preset number of sensory evaluators with preset qualifications. The machine learning classification model is trained with a set of key quality characteristic indicators as independent variables and sensory evaluation quantitative data as dependent variables. The sensory evaluation quantitative data includes two types of indicators: quantitative scoring and qualitative judgment. The quantitative evaluation uses a 0-9 point scale to independently score the core sensory dimensions of the rice husk, where 0 points correspond to "very poor quality" and 9 points correspond to "very good quality". The qualitative judgment uses a binary "yes" or "no" standard, where "yes" indicates that the sample meets the requirements for use in brewing production and "no" indicates that it does not meet the requirements for use in brewing production. The average of the quantitative scores of all sensory evaluators for the same brewing rice husk sample is taken as the quantitative score result of that brewing rice husk sample.

[0023] As shown in Figure 2, the method for the digital analysis and discrimination system of brewing rice husk quality includes the following steps: S1, using gas chromatography-ion mobility spectrometry to obtain volatile organic compound (VOC) data of brewing rice husk samples; S2, using texture analysis to obtain multi-dimensional texture parameter data of brewing rice husk samples, including hardness, cohesion, elasticity index, adhesiveness, chewiness, and chewiness index; S3, fusing and preprocessing the VOC data and multi-dimensional texture parameter data, and selecting a set of key quality characteristic indicators for judging the quality of brewing rice husks based on multivariate statistical analysis and machine learning feature selection algorithms; S4, using a machine learning classification model based on the set of key quality characteristic indicators to automatically judge and output the quality grade of brewing rice husk samples.

[0024] In one embodiment of the present invention, the sample to be tested is rice husks provided by a company in Jiangsu Province with different steaming times, including rice husks steamed for 0 min, 15 min, 30 min, 45 min, 60 min, 75 min, 90 min, 105 min, and 120 min. The relevant equipment and instrument information is as follows: FlavorSpec food flavor analyzer: equipped with an automatic headspace sampler, VOCal analysis software, Reporter plugin, Gallery Plot plugin, and Dynamic PCA plugin.

[0025] CTX texture analyzer: equipped with a variety of probes and TexturePro software.

[0026] Before conducting digital analysis and judgment of the quality of rice husks for brewing, rice husks from different batches with different steaming times were selected. The number of batches was N, where N is greater than or equal to 2. Each batch contained D rice husk samples that passed the sensory evaluation and D+5 samples that failed the sensory evaluation, where D is greater than or equal to 2. M samples were taken from each sample as parallel samples, where M is greater than or equal to 3. The total number of samples was N×M×(2D+5).

[0027] When conducting digital analysis and judgment of rice husk quality for brewing, the sensory evaluation team consisted of several professional tasters with many years of experience in the brewing industry. The evaluation experiments were conducted in a sensory evaluation room with ample lighting, no interfering odors, and a quiet and stable environment. A single-blind design was used: all samples had their original labels removed and were marked with random codes. The sample order followed a randomization principle to ensure that tasters could not access relevant background information about the samples, avoiding subjective bias. The evaluation system included two types of indicators: quantitative scoring and qualitative judgment. Quantitative evaluation used a 0-9 point scale, independently scoring the core sensory dimensions of rice husks (color, odor), where 0 points corresponded to "very poor quality" and 9 points to "very good quality." Qualitative judgment used a binary "yes" or "no" standard, where "yes" indicated that the sample met the requirements for use in brewing production, and "no" indicated that it did not meet the requirements. During the evaluation process, each taster independently completed the scoring and judgment. After all evaluation data was collected, the average score of the quantitative indicators for each sample was calculated.

[0028] The sensory evaluation results are shown in Figure 3. Unsteamed rice husks are pale yellow and have a bran-like, dusty, straw-like, and green taste. As the steaming time increases, the color gradually turns golden yellow and dark yellow, accompanied by a bran-like, tobacco-like, roasted, and smoky taste, with a slight cooked grain aroma. Among them, rice husks steamed for 90 minutes and 105 minutes have the highest sensory scores and meet the requirements for brewing production. Rice husks steamed for other times do not meet the requirements for brewing production.

[0029] When using gas chromatography-ion mobility spectrometry (GC-IMS), rice husks from different batches and steaming times were selected, and 1.0 g of each was placed in a headspace vial for analysis. Parameter settings: column parameters were set as follows: WAX capillary column (30m*0.53mm*1μm), analysis time 30 min, column temperature 60℃, carrier gas N2, and carrier gas flow rate: initial carrier gas flow rate was 2 mL / min, flow rate was 10 mL / min at 10 minutes, and when it was increased to 100 mL / min, the flow rate was maintained for 10 minutes.

[0030] The ion mobility spectrometry parameters were set as follows: N2 was used as the drift gas for IMS, the drift gas flow rate was 150 mL / min, and the drift tube temperature was 45 °C. Automated headspace sampling was used, with an injection volume of 500 μL, an incubation time of 15 min, an incubation temperature of 60 °C, a needle temperature of 85 °C, and an incubation rotation speed of 500 rpm.

[0031] Sample testing: After sample loading into GC-IMS, the sample is automatically headspace-injected after a period of incubation. The sample enters the instrument with the carrier gas, first undergoes primary separation by gas chromatography, and then enters the ion mobility spectrometer. After the analyte molecules are ionized in the ionization region, they migrate under the influence of an electric field and reverse drift, achieving secondary separation and obtaining information on the volatile substances in the sample. GC-IMS combines the advantages of high resolution of gas chromatography and high sensitivity of ion mobility spectrometry, requiring no special sample pretreatment, and has the advantage of rapid detection of volatile organic compounds in samples.

[0032] Data processing: (1) Data was collected and identified using the VOCal analysis software provided with the instrument. The NIST and IMS databases built into the VOCal software were used to perform qualitative analysis of the substances by gas phase retention time and ion migration time. Table 1 shows that there are a total of 31 volatile flavor substances that are stable in different batches of rice husk samples. Among them, the M and D suffixes of substances with the same name represent the monomers and dimers of the same compound, respectively. These include 10 alcohols, 8 aldehydes, 5 esters, 7 ketones, and 1 acid.

[0033] Table 1: Qualitative results of volatile flavor components in rice husks

[0034]

[0035]

[0036] (2) The difference comparison spectrum of the test samples was obtained by the Reporter plugin of the GC-IMS instrument. As shown in Figure 4, the difference between the flavor profiles of rice husk samples with different steaming times can be observed intuitively. The results show that the volatile organic compounds in the samples are concentrated in the 200-1000s region. There is a significant difference in the volatile components between the rice husk samples that pass the sensory evaluation and those that fail the sensory evaluation. The content of most volatile substances in the rice husks that pass the sensory evaluation is significantly lower than that in the rice husks that fail the sensory evaluation.

[0037] (3) The Quantification and Origin functions of the VOCal analysis software were used to analyze the changing trends of rice husk flavor components. As shown in Figure 5, it can be clearly observed that the flavor composition of rice husks differs significantly with different steaming times. The content of substances such as alcohols, aldehydes, ketones, and esters in rice husks changes non-linearly with the extension of time, indicating that longer steaming time is not necessarily better, and there is an optimal steaming time. Among them, the overall flavor substance content of rice husks steamed for 90 min and 105 min is the lowest, indicating that the appropriate steaming time is more conducive to ensuring that the off-flavors of rice husks will not be carried into the raw liquor in the subsequent distillation of the mash. The flavor analysis results are consistent with the sensory evaluation results. The above are the results of GC-IMS volatile flavor component and changing trend analysis. Partial least squares discriminant analysis (PLS-DA) will be used to further screen key differential substances.

[0038] (4) PLS-DA is a supervised statistical analysis method that mainly achieves the analysis, discrimination, and prediction of complex data through data dimensionality reduction. Using volatile flavor compounds as independent variables and rice husk samples as dependent variables, PLS-DA analysis was performed. The performance of the model was evaluated through cross-validation and permutation tests, as shown in Figure 6. Model R... 2 X, R 2 Y, Q 2 All values ​​are greater than 0.5, indicating that the model is reliable and has good stability; the supervised model passed 200 permutation tests, Q... 2 A value less than 0 indicates that the model has not overfitted and the model validation is effective.

[0039] A cluster analysis discrimination model was established using the PLS-DA method to extract and model volatile component data. The results are shown in Figure 7. The volatile component characteristics of qualified and unqualified rice husks are significantly distinguished, indicating that the model can efficiently and accurately determine whether the quality of rice husks is qualified.

[0040] The variable importance projection value (VIP) represents the contribution of volatile compounds to sample classification. A VIP > 1 indicates that the variable has an important role, and the larger the VIP value, the more significant the difference between samples for that volatile compound. Through PLS-DA analysis, a total of 7 volatile compounds with a VIP score > 1.0 were screened (including monomers and dimers), as shown in Figure 8. These compounds are ethyl acetate (D), ethyl butyrate (M), butyl acetate (M), hexanal (M, D), heptanal (M, D), n-octanal, and 2-pentanone (M).

[0041] (5) RF is an ensemble machine learning algorithm that constructs multiple decision trees and integrates the results to achieve high-precision classification, regression, and variable importance assessment. Volatile flavor compounds are used as independent variables and rice husk samples are used as dependent variables. RF modeling and important variable screening are performed. Multiple decision trees are generated by Bootstrap resampling. The out-of-bag (OOB) sample prediction results of all trees are summarized, and the overall out-of-bag AUC value of the model is calculated to evaluate the classification performance of the model. Based on this model, the Mean Decrease Accuracy method is used to quantify the contribution of each independent variable to the classification accuracy of the model. The larger the value, the more important the variable. As shown in Figures 9 and 10, the AUC value of the model was 1 after RF analysis, indicating that the model has good classification performance. Based on the importance scores of the variables, six key differential volatile flavor compounds (including monomers and dimers) with VIM≥Median were selected, namely ethyl hexanoate (M), propanol (D), 1-pentanol (D), sec-butanol (M), heptanal (M, D), and 2-pentanone (M).

[0042] (6) In summary, the key flavor factors affecting the quality of rice husks for brewing were screened for VIP>1 and VIM≥Median. As shown in Figure 11, two flavor components (including monomers and dimers) were obtained, namely heptanal (M, D) and 2-pentanone (M). Their contents were relatively lower and more stable in rice husks that were steamed for 105 min.

[0043] When using texture analysis, the parameters are set as follows: sample mass 10.0g, measurement method is compression, test standard is TPA, test endpoint is displacement, target value is 7mm, trigger point is 10g, and test speed is 1mm / s.

[0044] Sample Testing: After placing the rice husk sample on the testing platform of the texture analyzer, the measurement program is started. The probe compresses the sample according to the set parameters. The sample deforms under the mechanical action of the probe, and the force sensor inside the instrument records the force changes in real time during the compression process and plots a force-time curve. By analyzing the characteristic curve, the mechanical properties parameters of rice husk, such as hardness, cohesiveness, elasticity index, adhesiveness, chewiness, and chewiness index, can be obtained. Texture analysis technology can transform the mechanical properties of food into objective, quantitative numerical indicators by simulating the subjective feeling of touching, squeezing, and spreading. It combines the advantages of high repeatability and high precision of instrument measurement with the high correlation of sensory evaluation, and requires no complex sample pretreatment, enabling rapid and accurate assessment of rice husk texture quality.

[0045] Data processing: (1) Data was collected and analyzed using the TexturePro analysis software provided with the instrument. Figure 12 shows that the textural properties of rice husks steamed for 90 min and 105 min fluctuated less, indicating that the quality of rice husks was relatively stable. After steaming for an appropriate time, the rice husks had relatively lower hardness, adhesiveness, and chewiness, while maintaining moderate cohesion and elasticity. This indicates that the skeleton structure of rice husks was enhanced to a certain extent, breaking through the problems of high hardness and insufficient elasticity of skeleton structure when rice husks are used as auxiliary materials in the fermentation of baijiu. This is more conducive to promoting the fermentation efficiency of baijiu and ensuring the quality of baijiu products.

[0046] (2) Using correlation analysis, key textural characteristics factors affecting the quality of rice husks for brewing were screened based on p≤0.05. The results showed that hardness, cohesion, elasticity index, adhesiveness, chewiness, and chewing index were textural characteristics factors affecting the quality of rice husks.

[0047] Based on the key quality characteristic indicators of rice husks selected for brewing and the corresponding sensory evaluation results, a qualitative analysis and discrimination model for rice husk quality was constructed using a support vector machine (SVM) with N×M×(2D+5) samples as the sample set. The results showed that the model had an AUC of 0.935, a recall of 0.852, and a precision of 0.845, indicating good model performance. By collecting key quality factor data from new batches of rice husks and inputting it into the constructed model, and using sensory evaluation results as validation, as shown in Figure 13, the accuracy of the qualitative discrimination model for rice husk quality was 88.9%, demonstrating good discrimination performance.

[0048] The above examples demonstrate that using heptanal (M, D), 2-pentanone (M), hardness, cohesiveness, elasticity index, adhesiveness, chewiness, and chewing index as key quality factors in rice husk quality assessment is effective. Raw rice husks often have a distinct musty, grassy, ​​and muddy odor, and easily absorb unpleasant odors from the environment. These factors directly affect the flavor and quality of baijiu (Chinese liquor). Therefore, in the baijiu brewing process, rice husks need to undergo pretreatment such as steaming to effectively remove the rotten and strawy smells and enhance their "boning strength," thus providing ideal material conditions for subsequent fermentation and distillation. However, current methods relying on manual sensory evaluation are insufficient for accurate quality assessment of rice husks. Therefore, this invention quantifies the flavor and textural characteristics of rice husks and constructs a digital evaluation system for rice husk quality, aiming to provide a standardized system and method for rapid and accurate quality analysis and identification.

Claims

1. A digital analysis and discrimination system for the quality of rice husks used in brewing, characterized in that, include: The flavor analysis module uses gas chromatography-ion mobility spectrometry to acquire volatile organic compound data from rice husk samples used for brewing. The texture analysis module employs texture analysis technology to acquire multi-dimensional texture parameter data of brewing rice husk samples, including hardness, cohesion, elasticity index, adhesiveness, chewiness, and chewiness index. The data processing module is used to fuse and preprocess volatile organic compound data and multi-dimensional texture parameter data, and based on multivariate statistical analysis and machine learning feature selection algorithms, to screen out a set of key quality characteristic indicators for judging the quality of brewing rice husks. The quality digitization judgment module is used to automatically judge and output the quality grade of brewing rice husk samples based on the set of key quality characteristic indicators and using a machine learning classification model.

2. The digital analysis and discrimination system for brewing rice husk quality according to claim 1, characterized in that, The machine learning classification model is a pre-trained support vector machine. During training, all brewing rice husk samples have their original labels removed and are labeled with random codes. The order of the brewing rice husk samples follows a randomization principle. Sensory evaluation quantitative data of the brewing rice husk samples are obtained by a predetermined number of qualified sensory evaluators. The machine learning classification model is trained with a set of key quality characteristic indicators as independent variables and sensory evaluation quantitative data as dependent variables. The sensory evaluation quantitative data includes two types of indicators: quantitative scoring and qualitative judgment. The quantitative evaluation uses a 0-9 point scale to independently score the core sensory dimensions of rice husks, where 0 points correspond to "very poor quality" and 9 points correspond to "very good quality". The qualitative judgment uses a binary "yes" or "no" standard, where "yes" indicates that the sample meets the requirements for use in brewing production and "no" indicates that it does not meet the requirements for use in brewing production. The average of the quantitative scores of all sensory evaluators for the same brewing rice husk sample is taken as the quantitative score result for that brewing rice husk sample.

3. A method for the digital analysis and discrimination system for brewing rice husk quality as described in claim 1, characterized in that, Includes the following steps: S1. Gas chromatography-ion mobility spectrometry (GC-IMS) was used to obtain volatile organic compound (VOC) data from brewing rice husk samples. S2. Texture analysis was used to obtain multi-dimensional texture parameter data from brewing rice husk samples, including hardness, cohesiveness, elasticity index, adhesiveness, chewiness, and chewiness index. S3. The VOC data and multi-dimensional texture parameter data were fused and preprocessed, and a set of key quality characteristic indicators for judging the quality of brewing rice husks was selected based on multivariate statistical analysis and machine learning feature selection algorithms. S4. Based on the set of key quality characteristic indicators, use a machine learning classification model to automatically identify and output the quality grade of brewing rice husk samples.

4. The method according to claim 3, characterized in that, When using gas chromatography-ion mobility spectrometry (GC-IMS), the gas chromatographic parameters for GC-IMS are as follows: column: WAX capillary column, 30m*0.53mm*1μm; analysis time: 30min; column temperature: 60℃; carrier gas: N2; carrier gas flow rate: initial flow rate 2mL / min, flow rate 10mL / min at 10 minutes, and when it reaches 100mL / min, the flow rate is maintained for 10 minutes.

5. The method according to claim 4, characterized in that, When using gas chromatography-ion mobility spectrometry (GC-IMS), the ion migration conditions for GC-IMS are as follows: N2 is used as the drift gas for IMS, the drift gas flow rate is 150 mL / min, and the drift tube temperature is 45 °C; automatic headspace sampling is used for injection, the injection volume is 500 μL, the incubation time is 15 min, the incubation temperature is 60 °C, the injection needle temperature is 85 °C, and the incubation rotation speed is 500 rpm.

6. The method according to claim 5, characterized in that, Volatile organic compound (VOC) data from rice husk samples for brewing were obtained through qualitative and relative quantitative analysis using VOCal analysis software and the NIST and IMS databases built into the software, based on gas phase retention time and ion migration time.

7. The method according to claim 3, characterized in that, The specific method for obtaining multi-dimensional textural parameter data of rice husk samples for brewing using texture analysis technology includes: using a texture analyzer equipped with a cylindrical test probe to determine the hardness, cohesiveness, elasticity index, adhesiveness, chewiness, and chewing index characteristics of 10.0g rice husk. The measurement method is compression, the test standard is TPA, the test endpoint is displacement, the target value is 7mm, the trigger point is 10g, and the test speed is 1mm / s.

8. The method according to claim 3, characterized in that, After obtaining the multidimensional texture parameter data of the brewing rice husk samples, the data and charts were viewed using the DATA / GRAPH VIEWS function in the Texture analysis software, and the data and results were analyzed using the ANALYSIS function.

9. The method according to claim 3, characterized in that, The key quality characteristics indicators of brewing rice husks include key flavor factors and key textural characteristics factors that affect the quality of brewing rice husks. Among them, the key flavor factors affecting the quality of brewing rice husks were screened using partial least squares discriminant analysis and random forest analysis, with VIP≥1 and VIM≥Median as the criteria. The key textural characteristics factors affecting the quality of brewing rice husks were screened using correlation analysis, with p≤0.05 as the basis.

10. The method according to claim 3, characterized in that, The machine learning classification model is a support vector machine, whose input is a set of key quality feature indicators and whose output is the quality grade of the rice husk sample for brewing.

Images