Oriented fermentation and quality control method for sour bamboo shoots based on organic acid pretreatment and multi-platform detection
By combining organic acid pretreatment and multi-platform detection with machine learning, the problems of uncontrollable microbial communities and unstable flavor in pickled bamboo shoot production have been solved, achieving standardization and rapid quality control in pickled bamboo shoot production.
Patent Information
- Application Number
- CN202511488174.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-17
- Publication Date
- 2026-02-13
AI Technical Summary
In current pickled bamboo shoot production, the succession of microbial communities is uncontrollable, the initial fermentation conditions lack scientific design, the detection methods for flavor substances are limited, and there is a lack of quantitative quality control indicators, resulting in unstable product flavor and difficulty in achieving standardized production.
Organic acid pretreatment combined with liquid-sealed fermentation, multi-platform detection and machine learning were employed. Pretreatment methods such as solid-phase microextraction, thin-film solid-phase microextraction and stir bar extraction were used. GC-MS and GC-IMS platforms were used for detection, and robust flavor biomarkers were screened using algorithms such as random forest and support vector machine. A discriminant model was established for quality control.
This approach enables controllability of the fermentation process of pickled bamboo shoots, improves the coverage of flavor substance detection, obtains robust quality control indicators, and ensures the standardization and batch stability of pickled bamboo shoot production.
Smart Images
Figure CN121522071A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of food fermentation and quality control technology, in particular to an acid bamboo directional fermentation and flavor quality control method based on organic acid pretreatment, multi-pretreatment method, cross-platform detection and multi-algorithm screening. The method belongs to the field of cross-application of food science and fermentation engineering, and is especially suitable for process optimization and standardized quality control of plant lactic acid fermented food. BACKGROUND
[0002] As a characteristic lactic acid fermented food in southwest China, acid bamboo is loved by consumers because of its unique sour and "smelly" flavor. However, the existing acid bamboo production relies on natural fermentation or empirical process control, which has the following shortcomings:
[0003] Firstly, the succession of microbial community is uncontrollable. In natural fermentation, the source of microorganisms is complex, and the dominant microbial community takes over at an uncertain time, which easily leads to contamination by miscellaneous bacteria or batch differences, resulting in unstable product flavor.
[0004] Secondly, the initial fermentation conditions lack scientific design. At present, there is a general lack of regulation of initial acidity or organic acid ratio in production, which makes it difficult for target lactobacillus (such as lactobacillus plantarum) to enrich in time, resulting in a high failure rate of fermentation.
[0005] Thirdly, the flavor substance detection means is limited. The existing detection mainly relies on a single platform (such as GC-MS), which lacks sensitivity to small polar molecules and low threshold sulfur-containing substances, and cannot fully reflect the overall flavor of the flavor substances, resulting in incomplete flavor analysis.
[0006] In addition, there is a lack of quality control indicators. At present, there is a lack of quantifiable core flavor markers and batch discrimination models, and the release mainly relies on artificial sensory experience, which makes it difficult to realize standardization and industrialization.
[0007] Therefore, there is an urgent need for a method that integrates process regulation, cross-platform detection and flavor marker screening, which can not only directionally control the microbial community structure of acid bamboo fermentation, but also comprehensively analyze the flavor characteristics and stably control the quality, in order to meet the needs of the industrialization development of acid bamboo and other plant fermented foods. SUMMARY
[0008] The present application provides an acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection, which aims to solve the problems of uncontrollable microbial community succession, significant flavor difference between batches, insufficient detection means and lack of quantitative quality control indicators in the existing acid bamboo fermentation technology. The present application can not only achieve directional microbial community enrichment through process regulation, but also comprehensively analyze flavor substances through multi-platform detection, and obtain stable markers through machine learning screening, so as to realize the standardization and rapid quality control of acid bamboo production.
[0009] To achieve the above objectives, the present invention adopts the following technical solution:
[0010] A method for directional fermentation and quality control of pickled bamboo shoots based on organic acid pretreatment and multi-platform detection includes the following steps:
[0011] (1) Apply organic acid initial stress to the pickled bamboo shoot raw material before fermentation. The organic acid is one of lactic acid and acetic acid or a mixture of the two. When the two are mixed, the mass concentration ratio of lactic acid to acetic acid is 2:1-5:1, and the total acidity is controlled at 10-20g / L.
[0012] (2) Anaerobic fermentation is carried out under liquid-sealed conditions, and the community take-over time is shortened by the organic acid stress regulation, and the target lactic acid bacteria are enriched;
[0013] (3) The fermentation samples were pretreated by solid phase microextraction (SPME), thin film solid phase microextraction (TF), stir bar extraction (SBSE) and its combination (TF+SBSE) to obtain volatile flavor substances with different polarities and volatility;
[0014] (4) The pretreated sample was introduced into a gas chromatography-mass spectrometry (GC-MS) and gas chromatography-ion mobility spectrometry (GC-IMS) platform for detection to obtain full spectrum data of fermentation flavor;
[0015] (5) Input the detection data into the Random Forest (RF), Support Vector Machine (SVM) and Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) models for cross-selection of multiple algorithms;
[0016] (6) Based on the principles of cross-algorithm consistency and cross-platform repeated detection, a robust set of flavor biomarkers is identified;
[0017] (7) Based on the set of markers, establish a pickled bamboo shoot flavor discrimination model and set a quality control threshold to realize batch stability evaluation and abnormal batch early warning.
[0018] Further, the organic acid mentioned in step (1) is a mixture of lactic acid and acetic acid, with a lactic acid concentration of 8-15 g / L, an acetic acid concentration of 2-6 g / L, and a total acidity preferably of 12-18 g / L.
[0019] Furthermore, in step (2), the liquid sealing method is water sealing, oxygen barrier film covering or oxygen-free valve, preferably water sealing, to ensure a low-oxygen state during the fermentation process.
[0020] Furthermore, in step (2), on the second day of fermentation, the relative abundance of Firmicutes reaches ≥98%, with the target genera being Lactiplantibacillus and Levilactobacillus, and the total abundance not less than 90%.
[0021] Further, the combination of pretreatment methods in step (3) is preferably TF+SBSE to simultaneously enrich polar and non-polar compounds, increasing the number of compounds detected by 20-30% compared to a single method.
[0022] Further, the gas chromatography-mass spectrometry in step (4) is used to detect semi-volatile aromatic compounds, phenols and medium-long chain aldehydes, and gas chromatography-ion mobility spectrometry is used to detect light polar small molecules and sulfur-containing low threshold compounds, both of which have complementary coverage of more than 30%.
[0023] Further, the marker set in step (7) includes one or more of benzyl alcohol, p-cresol, m-cresol, o-cresol, 2,6-dimethylpyrazine, hexanal, nonanal, decanal, acetic acid, ethyl acetate, and dimethyl sulfide.
[0024] Further, the classification accuracy of the discriminant model in step (7) is ≥95%, and maintains Q 2 >0.85 in 200 permutation tests.
[0025] Further, the concentration-time curve of p-cresol in step (7) is used as a release criterion, and when its relative abundance is increased by ≥500 times compared to the initial value, it is determined that the fermentation is complete.
[0026] Further, the quality control threshold in step (4) is set based on the relative abundance ratio of the markers, including the benzyl alcohol / aldehyde ratio and the ethyl acetate / acid ratio, for batch difference detection.
[0027] Further, the method is also applicable to the flavor detection and quality control of pickles, sour green beans, sour radishes and other plant-based fermented foods dominated by lactic acid bacteria.
[0028] Compared with existing processes that rely on natural fermentation and single detection methods, the present application has the following advantages:
[0029] (1) Controllable fermentation process: The present application realizes the stable takeover of Firmicutes dominant flora (≥98%) within 48 hours through organic acid pretreatment combined with liquid sealing, significantly shortening the succession time and reducing the interference of miscellaneous bacteria.
[0030] (2) More comprehensive detection coverage: The combination of multi-method pretreatment, GC-MS and GC-IMS dual platforms in the present application achieves an increase of more than 30% in flavor compound coverage, overcoming the detection limitations of single technology.
[0031] (3) Stable and reliable markers: The present application obtains a batch of stable markers such as benzyl alcohol, cresols, 2,6-dimethylpyrazine, aldehydes, ethyl acetate and dimethyl sulfide through consistent screening across platforms and algorithms, which are highly related to the characteristic flavor of acid bamboo.
[0032] (4) The quality control model is efficient and accurate: the discrimination model established by the application has an accuracy of ≥95%, and the p-cresol concentration threshold is used as a release criterion, providing a quantitative index for batch quality control of sour bamboo shoots.
[0033] (5) Wide application range: In addition to sour bamboo shoots, the method of the application is also applicable to the flavor detection and quality control of various plant lactic acid fermented foods such as pickles, sour green beans, and sour radishes, providing a new operable method for the standardized production and rapid quality control of sour bamboo shoots and other plant fermented foods. BRIEF DESCRIPTION OF DRAWINGS
[0034] Figure 1 is the phylum level bacterial community structure, showing that Firmicutes accounted for ≥98% on the second day of fermentation, achieving rapid takeover.
[0035] Figure 2 is the genus level bacterial community structure, with Lactiplantibacillus as the dominant genus, and Levilactobacillus in the mixed acid group cooperatively dominating the chart.
[0036] Figure 3 is a chart of volatile substance differences caused by inoculation and extraction methods. Among them Figure 3 (A) is a volcano plot under inoculation conditions; Figure 3 (B) is a heat map under inoculation conditions; Figure 3 (C) is a volcano plot under extraction conditions; Figure 3 (D) is a heat map under extraction conditions. The differential compound screening criteria are VIP>1, |log2FC|≥1, and p<0.05.
[0037] Figure 4 is a chart of discrimination model accuracy and bacterial community-flavor matching degree verification. Among them Figure 4 (A): The samples are randomly divided into 70% training set and 30% test set, and the SVM model discrimination rate reaches 100%; Figure 4 (B): The matching ratio of dominant genera and key characteristic flavor substances with extremely significant positive correlation reaches 95.65%. DETAILED DESCRIPTION
[0038] In the embodiments of the application, a sour bamboo shoot directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection includes the following steps:
[0039] Organic acid pretreatment: lactic acid, acetic acid or a mixture of lactic acid and acetic acid with a mass ratio of 2:1-5:1 is added to the acid bamboo material before fermentation, the total acidity is controlled at 10-20 g / L, and anaerobic fermentation is carried out under liquid sealing conditions. This step can selectively enrich acid-tolerant Lactobacillus by acid stress in the early stage of fermentation, shorten the time of taking over, and inhibit the growth of miscellaneous bacteria.
[0040] Multi-method pretreatment: Take the fermentation sample, and respectively use solid-phase microextraction (SPME), thin-film solid-phase microextraction (TF), stirring rod extraction (SBSE) and their combination (TF+SBSE) for pretreatment, so as to enrich polar, non-polar, light and hydrophobic flavors, and improve the coverage of detected substances.
[0041] Cross-platform detection: The pretreated sample is respectively introduced into a gas chromatography-mass spectrometry (GC-MS) and a gas chromatography-ion mobility spectrometry (GC-IMS) platform, and the complementarity of the two in the detection of semi-volatile aromatic substances and light polar small molecules is utilized, so as to obtain a more comprehensive flavor spectrum.
[0042] Machine learning screening: The double-platform detection data is input into multi-algorithm models such as orthogonal partial least squares discriminant analysis (OPLS-DA), random forest (RF) and support vector machine (SVM), the threshold of VIP>1.0 and p<0.05 is set, and the cross-platform and cross-algorithm results are required to be consistent, and finally a stable flavor marker set is screened out.
[0043] Quality control criterion: A discriminant model is constructed based on the marker set, and the p-cresol concentration-time curve is used as a release index to determine whether the acid bamboo fermentation meets the standard, so as to realize batch quality control and abnormal early warning.
[0044] In order to make the present disclosure more complete, the following is described by more specific examples.
[0045] Example 1: Effect of organic acid pretreatment on microbial succession
[0046] Fresh acid bamboo produced in Guangxi is divided into a lactic acid group (11.59 g / L), an acetic acid group (3.57 g / L), a mixed acid group (lactic acid and acetic acid 3:1, total acidity 15 g / L) and a control group, and all are fermented under liquid sealing conditions.
[0047] The results show (as shown in Figure 1 , Figure 2Firmicutes accounted for ≥98% of the total bacteria on the 2nd day of fermentation under lactic acid and acetic acid pretreatment conditions, indicating that it took over at an early stage. At the genus level, Lactiplantibacillus was the main dominant genus, and Levilactobacillus accounted for about 41% of the total bacteria in the mixed acid group. In the control group, Lactococcus and Weissella proliferated abnormally, and the community succession process fluctuated greatly. This shows that organic acid pretreatment combined with liquid-sealed fermentation can quickly establish an acid-tolerant lactic acid bacterial dominant community through acid stress, thereby shortening the takeover time and improving system stability.
[0048] Example 2: Complementarity of multi-pretreatment and cross-platform detection
[0049] The same batch of acid bamboo samples were pretreated by SPME, TF, SBSE and TF+SBSE, and then introduced into GC-MS and GC-IMS for detection.
[0050] The results show that (Table 1) Figure 3 ): SPME detected the most polar aldehydes (hexanal, octanal), a total of 12 different substances; SBSE had the best enrichment effect on hydrophobic compounds (terpenes, esters); TF was biased towards capturing aromatic and heterocyclic substances; TF+SBSE detected about 25% more substances than single methods; GC-MS was superior in detecting semi-volatile aromatic substances and phenols, while GC-IMS was more sensitive to light polar and sulfur-containing low threshold substances; After cross-platform integration, the coverage rate of flavor compounds increased by more than 30%. As shown in Table 1, 95 volatile signal peaks were resolved by GC-IMS, corresponding to 77 identified compounds, covering acids, esters, alcohols, aldehydes and ketones, and sulfur-containing compounds. The CAS number, retention time, and other qualitative parameters were given, which can be used as a database benchmark for subsequent model and release indicators.
[0051] This shows that different pretreatments and detection platforms have complementarity, and combined use can significantly improve the comprehensiveness of flavor detection.
[0052]
[0053]
[0054]
[0055]
[0056]
[0057] Example 3: Robust marker screening driven by machine learning
[0058] The detection data of GC-MS and GC-IMS in Example 2 were input into OPLS-DA, random forest and SVM models for analysis, with the screening criteria being VIP > 1.0 and p < 0.05, and the requirement being consistent results across platforms and algorithms.
[0059] A total of 11 robust markers were obtained (Table 2), including benzyl alcohol, p-cresol, m-cresol, o-cresol, 2,6-dimethylpyrazine, hexanal, nonanal, decanal, acetic acid, ethyl acetate and dimethyl sulfide. These substances were highly correlated with the caramel, sour and fruity characteristics of acid bamboo shoots and could be attributed to amino acid degradation, lipid oxidation and esterification metabolic pathways. From the results of the matching degree analysis of the bacterial community structure and flavor characteristics, it was found that the established discriminant model had a classification accuracy of ≥ 95% for acid bamboo shoot samples inoculated in different ways, and the matching degree reached 95.65%. Figure 4
[0060] It was shown that the method could achieve stable screening of flavor markers and efficient discrimination of batch quality.
[0061]
Claims
1. A method for the directional fermentation and quality control of acid bamboo based on organic acid pretreatment and multi-platform detection, characterized by, The method comprises the following steps: (1) applying an initial stress of organic acid to the acid bamboo material before fermentation, the organic acid being one or a mixture of lactic acid and acetic acid, when mixed, the mass concentration ratio of lactic acid to acetic acid being 2:1-5:1, and the total acidity being controlled at 10-20 g / L; (2) performing anaerobic fermentation under liquid sealing conditions, shortening the community takeover time and enriching the target Lactobacillus genus through the organic acid stress regulation; (3) performing pretreatment on the fermentation sample by solid-phase microextraction, thin-film solid-phase microextraction, stirring bar extraction and their combinations to obtain volatile flavor substances of different polarity and volatility; (4) introducing the pretreated sample into a gas chromatography-mass spectrometry and gas chromatography-ion mobility spectrometry platform for detection to obtain fermentation flavor full-spectrum data; (5) inputting the detection data into random forest, support vector machine and orthogonal partial least squares discriminant analysis models for multi-algorithm cross screening; (6) locking a robust flavor marker set according to the cross-algorithm consistency and cross-platform repeated detection principle; (7) establishing an acid bamboo flavor discriminant model based on the marker set and setting a quality control threshold to realize batch stability evaluation and abnormal batch early warning.
2. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the organic acid in step (1) is a mixture of lactic acid and acetic acid, the lactic acid concentration is 8-15 g / L, and the acetic acid concentration is 2-6 g / L.
3. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the liquid sealing method in step (2) is water sealing, oxygen barrier film covering or oxygen-free valve to ensure a low-oxygen state during fermentation.
4. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein in step (2), the relative abundance of the Firmicutes phylum reaches ≥98% on the second day of fermentation, the target genus is Lactiplantibacillus and Levilactobacillus, and the total abundance is not less than 90%.
5. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the pretreatment method combination in step (3) is TF+SBSE to simultaneously enrich polar and non-polar compounds, so that the number of compounds detected is increased by 20-30% compared with a single method.
6. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the gas chromatography-mass spectrometry in step (4) is used to detect semi-volatile aromatic substances, phenols and medium-long chain aldehydes, and the gas chromatography-ion mobility spectrometry is used to detect light polar small molecules and sulfur-containing low-threshold compounds, and the complementary coverage of the two is increased by more than 30%.
7. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the marker set in step (7) comprises one or more of benzyl alcohol, p-cresol, m-cresol, o-cresol, 2,6-dimethylpyrazine, hexanal, nonanal, decanal, acetic acid, ethyl acetate and dimethyl sulfide.
8. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the classification accuracy of the discriminant model in step (7) is ≥ 95%, and Q 2 > 0.85 is maintained in 200 times of permutation test.
9. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the quality control threshold in step (7) is set based on the relative abundance ratio of the markers, including the ratio of benzyl alcohol to aldehyde and the ratio of ethyl acetate to acid, for batch difference detection.
10. The acid bamboo directional fermentation and quality control method based on organic acid pretreatment and multi-platform detection according to claim 1, wherein the method is also applicable to the flavor detection and quality control of pickles, sour green beans, sour radishes, and other lactic acid bacteria-dominated plant-based fermented foods.