GC-IMS-based food freshness detection method, device and equipment

Abstract

The invention relates to the technical field of food detection, in particular to a GC-IMS-based food freshness detection method, device and equipment, and the method comprises the following steps: obtaining headspace sampling gas of a to-be-detected food sample; according to the type of the to-be-detected food sample, automatically matching preset chromatographic conditions and ion mobility spectrometry conditions from a graded food type database; the headspace sampling gas is injected into GC-IMS, analysis is carried out based on matched chromatographic conditions and ion mobility spectrometry conditions, and a volatile organic compound fingerprint spectrum of the sample is obtained; and extracting preset key feature information from the fingerprint spectrum, and inputting the key feature information into a food freshness evaluation model to obtain a freshness evaluation result of the to-be-detected food sample. Therefore, the problems that in the prior art, detection parameter curing lacks adaptability, key feature analysis is not accurate, and model prediction accuracy is insufficient are solved.

Description

Technical Field

[0001] The present application relates to the field of food detection technology, and in particular to a food freshness detection method, device and equipment based on GC-IMS. Background Art

[0002] As a core element of food safety and quality, food freshness is directly related to consumer health and the efficient operation of the food supply chain. In the modern food industry, accurate and efficient testing of food freshness, from production and circulation to sales, can effectively prevent spoiled food from entering the market, ensure consumer safety, optimize inventory management, reduce resource waste, and is of great significance to the sustainable development of the industry. However, traditional testing methods have significant limitations: sensory evaluation is highly dependent on human experience, has large subjective deviations, and is difficult to standardize; physical and chemical index testing is complex and time-consuming, requires professional equipment and reagents, and cannot meet the actual needs of rapid food testing. More advanced technologies are urgently needed to break through bottlenecks.

[0003] GC-IMS (Gas Chromatography-Ion Mobility Spectrometry) offers a new approach for food freshness assessment, thanks to its efficient separation and precise detection of VOCs (Volatile Organic Compounds). During food spoilage, microbial metabolism and enzymatic reactions cause regular changes in the types and contents of VOCs. GC-IMS can capture these subtle differences, theoretically making it a powerful tool for freshness detection. However, the existing technology is still imperfect: customized detection solutions for different food types are lacking, and test results under general conditions are susceptible to interference. Sample feature extraction and model construction fail to fully tap the data value, resulting in inaccurate screening of key spoilage markers. The adaptability and accuracy of freshness assessment models are insufficient to meet complex real-world scenarios, hindering the in-depth application of GC-IMS technology in food freshness testing. Summary of the Invention

[0004] The present application provides a food freshness detection method, device and equipment based on GC-IMS to solve the problems in the existing technology such as lack of adaptability of fixed detection parameters, inaccurate analysis of key features, and insufficient model prediction accuracy.

[0005] The first aspect of the present application provides a GC-IMS-based food freshness detection method, comprising the following steps: obtaining headspace sampling gas of a food sample to be tested; automatically matching preset chromatographic conditions and ion mobility spectrum conditions from a graded food type database according to the type of the food sample to be tested; injecting the headspace sampling gas into the GC-IMS, analyzing the gas based on the matched chromatographic conditions and ion mobility spectrum conditions, and obtaining a volatile organic compound fingerprint of the sample; extracting preset key feature information from the fingerprint, inputting the key feature information into a food freshness assessment model, and obtaining a freshness assessment result of the food sample to be tested, wherein the freshness assessment result includes at least one of a freshness grade, a freshness score, a predicted specific corruption index value, or a predicted remaining shelf life.

[0006] Optionally, the hierarchical food type database is constructed by hierarchical classification according to the type, processing method and storage conditions of food, including: establishing a food type classification framework, including primary classification and secondary classification, collecting GC-IMS full spectrum data, volatile organic compound concentration and sensory score of the primary and secondary classified foods in the spoilage process; classifying and labeling foods according to processing methods, analyzing their effects on volatile organic compounds and freshness, and forming a characteristic parameter set; establishing a multidimensional model based on storage conditions, collecting data to construct a storage and freshness mapping relationship, and then integrating food type and processing method data to construct a spoilage dynamics prediction matrix; integrating the data related to the food type, processing method and storage condition, assigning weights through the entropy weight method, and generating hierarchical nodes; based on the correlation between the hierarchical nodes, constructing a bidirectional index between food type, processing method, storage conditions and GC-IMS detection parameters and freshness evaluation model.

[0007] Optionally, preset key feature information is extracted from the fingerprint, including: preprocessing the GC-IMS fingerprints of similar food samples with known different freshness levels, including baseline correction, peak alignment and normalization; using random forest feature importance scoring to screen out characteristic peaks with significant differences between different freshness levels, and performing cluster analysis to identify characteristic peak clusters with synergistic change trends, which are defined as characteristic regions; calculating the intensity ratio, peak area ratio or correlation coefficient of each characteristic peak in the characteristic region, and constructing a multidimensional feature vector as the key feature information.

[0008] Optionally, the calculation process of the food freshness assessment model includes: The corruption reaction base rate constant is calculated based on the key characteristic information, where the formula is:

[0009] The formula for the storage temperature compensation term is:

[0010] The output evaluation results are:

[0011]

[0012] in, is the base rate constant of the corruption reaction, is the weight coefficient of key feature information, is the current strength of the i-th key feature information, is the baseline strength of the i-th key feature information, is the number of key feature information, is the activation energy, is the corruption reaction rate constant after storage temperature compensation, is the gas constant, is the actual storage temperature, is the reference temperature, For the remaining shelf life, is the initial freshness characteristic, To store the detection value of the feature quantity after t time, is the corruption risk index, is the rate constant corresponding to the national standard safety limit.

[0013] Optionally, the sampling temperature of the head space sampling gas is 30-60° C., the sampling time is 10-30 min, and the ratio of the volume of the head space sampling bottle to the sample mass is 4-10 mL / g.

[0014] Optionally, the preset chromatographic conditions include chromatographic column type, column oven temperature program, and carrier gas flow rate; the preset ion mobility spectrometry conditions include migration tube temperature, migration voltage, and drift gas flow rate.

[0015] Optionally, after obtaining the freshness evaluation result of the food sample to be tested, the method further includes: generating corresponding warning information according to a preset freshness threshold, wherein the warning information includes prompting the food to be processed as soon as possible or sold safely.

[0016] Optionally, before injecting the head space sampling gas into the GC-IMS, the method further includes filtering and drying the head space sampling gas.

[0017] The second aspect of the present application provides a GC-IMS-based food freshness detection device, including: an acquisition module for acquiring headspace sampling gas of a food sample to be tested; a matching module for automatically matching preset chromatographic conditions and ion mobility spectrum conditions from a graded food type database according to the type of the food sample to be tested; an analysis module for injecting the headspace sampling gas into the GC-IMS, analyzing it based on the matched chromatographic conditions and ion mobility spectrum conditions, and obtaining a volatile organic compound fingerprint of the sample; a calculation module for extracting preset key feature information from the fingerprint, inputting the key feature information into a food freshness assessment model, and obtaining a freshness assessment result of the food sample to be tested, wherein the freshness assessment result includes at least one of a freshness grade, a freshness score, a predicted specific corruption index value, or a predicted remaining shelf life.

[0018] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to perform the GC-IMS-based food freshness detection method as described in the above embodiment.

[0019] Therefore, this application has at least the following beneficial effects: The embodiment of the present application, by constructing a hierarchical food type database, automatically matches GC-IMS detection parameters according to food types, processing methods, and storage conditions, completely breaks the limitations of traditional solidification parameters and achieves accurate adaptation of different food types. At the same time, the standardized headspace sampling process and automated sampling design are adopted, combined with the characteristics of rapid separation and detection of GC-IMS coupling technology, which significantly shortens the single detection cycle and can efficiently meet the time requirements of real-time monitoring of production lines and rapid sampling inspections of warehouses. In addition, the headspace gas analysis mode can obtain volatile organic compound information without destroying the sample, which can not only fully preserve the original quality of the food, but also be suitable for online continuous monitoring of fresh food; the combination of standardized sampling parameters and automated headspace sampling device effectively avoids the problem of poor consistency of sample processing in on-site detection, ensuring that the detection data of different batches and different scenarios are reliable and comparable; the embodiment of the present application combines random forest feature importance scoring and cluster analysis to accurately screen key feature areas from the volatile organic compound fingerprint map, and constructs multidimensional feature vectors through characteristic peak intensity ratios, peak area ratios, etc., which can more comprehensively and accurately capture hidden information related to corruption. This solves problems such as lack of adaptability of fixed detection parameters, inaccurate analysis of key features, and insufficient accuracy of model predictions.

[0020] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which: Figure 1 Flowchart of a food freshness detection method based on GC-IMS according to an embodiment of the present application; Figure 2 A comparison diagram of standardized headspace sampling provided according to one embodiment of the present application; Figure 3 This is a block diagram of a food freshness detection device based on GC-IMS according to an embodiment of the present application; Figure 4 A schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0023] The following describes a method, device and equipment for detecting the freshness of food based on GC-IMS according to an embodiment of the present application with reference to the accompanying drawings. In response to the problem of solidified detection parameters mentioned in the above background technology, the present application provides a method for detecting the freshness of food based on GC-IMS. In this method, by constructing a hierarchical food type database, GC-IMS detection parameters are automatically matched according to food type, processing method and storage conditions, completely breaking the limitations of traditional solidified parameters and achieving accurate adaptation of different food types. At the same time, the standardized headspace sampling process and automated sampling design are adopted, combined with the characteristics of rapid separation and detection of GC-IMS coupling technology, which significantly shortens the single detection cycle and can efficiently meet the time requirements of real-time monitoring of production lines and rapid random inspections in warehouses. In addition, the headspace gas analysis mode can obtain volatile organic compound information without destroying the sample, which can not only fully preserve the original quality of the food, but also be suitable for online continuous monitoring of fresh food; the combination of standardized sampling parameters and automated headspace sampling devices effectively circumvents the problem of poor consistency in sample processing during on-site testing, ensuring that the test data of different batches and different scenarios are reliably comparable; the embodiment of the present application combines random forest feature importance scoring and cluster analysis to accurately screen key feature areas from the volatile organic compound fingerprint map, and constructs a multidimensional feature vector through the characteristic peak intensity ratio, peak area ratio, etc., which can more comprehensively and accurately capture hidden information related to corruption. In this way, problems such as the lack of adaptability of solidified detection parameters, inaccurate analysis of key features, and insufficient model prediction accuracy are solved.

[0024] The following describes a method, device, and apparatus for detecting food freshness based on GC-IMS according to an embodiment of the present application with reference to the accompanying drawings.

[0025] Specifically, Figure 1 A flow chart of a food freshness detection method based on GC-IMS provided in an embodiment of the present application.

[0026] like Figure 1 As shown, the food freshness detection method based on GC-IMS includes the following steps: In step S101, the headspace sampling gas of the food sample to be tested is obtained.

[0027] The sampling temperature of the headspace sampling gas is 30-60°C, the sampling time is 10-30 min, and the ratio of the volume of the headspace sampling bottle to the sample mass is 4-10 mL / g.

[0028] It can be understood that in the embodiments of the present application, standardized headspace sampling is performed on the food samples to be tested. By precisely controlling the sampling temperature, time, and the ratio of the headspace sampling bottle volume to the sample mass, efficient and non-destructive collection of headspace gas is achieved. The standardized process ensures that the release amount of VOCs in the sample is in a stable correspondence with the freshness state, avoiding deviations in the test results caused by fluctuations in sampling conditions, and providing a data basis for subsequent GC-IMS analysis. In addition, headspace sampling does not require the destruction of the sample itself, and is suitable for online continuous monitoring and on-site sampling of fresh food. Combined with an automated sampling device, it can significantly improve the detection efficiency while ensuring the circulation of food after testing.

[0029] Specifically, 5 g of fresh beef chunks were placed in a 20 mL headspace injection vial (volume to mass ratio 4 mL / g) and allowed to stand in a constant temperature environment at 50°C for 20 min.

[0030] Therefore, at 50°C, the concentration of aldehydes (such as hexanal) and sulfur-containing compounds (such as methyl mercaptan) produced by beef spoilage in the headspace increases by 30% compared with room temperature sampling, ensuring that the correlation coefficient between the characteristic peak intensity and TVB-N content in the GC-IMS spectrum reaches 0.92 (traditional room temperature sampling is only 0.75); after sampling, the beef can be directly returned to the production line, and with the help of an automated sampling device, continuous testing of 30 samples per hour can be achieved, which is 10 times more efficient than the traditional destructive microbial culture method.

[0031] In the embodiment of the present application, before the head space sampling gas is injected into the GC-IMS, the process further includes: filtering and drying the head space sampling gas.

[0032] Specifically, a 0.22 μm polytetrafluoroethylene microporous membrane filter (installed between the sampler outlet and the drying tube) was connected in series to remove particles ≥ 0.22 μm, and then the gas was passed through a drying tube (inner diameter 8 mm × length 150 mm) filled with 4A molecular sieve at a flow rate of 50 mL / min for adsorption drying. During this period, a dew point meter was connected through a bypass branch to monitor the moisture removal effect in real time (target dew point ≤ -40°C). The treated clean dry gas was directly introduced into the GC-IMS inlet through a PEEK pipeline to complete the pretreatment process.

[0033] In step S102, according to the type of the food sample to be tested, the preset chromatographic conditions and ion mobility spectrometry conditions are automatically matched from the graded food type database.

[0034] The preset chromatographic conditions may include the type of chromatographic column, column oven temperature program, and carrier gas flow rate; the preset ion mobility spectrometry conditions may include migration tube temperature, migration voltage, and drift gas flow rate.

[0035] Specifically, the preset chromatographic conditions include a three-step temperature increase program for the gas chromatography column, with an initial temperature of 35-45°C, maintained for 2-5 minutes, heated to 150-180°C at a rate of 5-10°C / min, and then heated to 200-250°C at a rate of 15-20°C / min. The preset ion mobility spectrometry conditions include an ion migration tube temperature of 40-60°C, a drift gas of high-purity nitrogen, a drift gas flow rate of 100-200 mL / min, an ionization source of corona discharge ionization source, and an ionization voltage of 2000-3000 V.

[0036] It is understood that the three-step temperature ramp in the chromatographic conditions of the present embodiment enables a step-by-step, refined separation of volatile organic compounds with different boiling points, avoiding detection interference from high- and low-boiling-point substances and improving separation efficiency. Precise settings of the column oven temperature and carrier gas flow further optimize the retention and elution of substances in the chromatographic column. In the ion mobility spectrometry conditions, regulation of the migration tube temperature and drift gas flow ensures stable ion migration within the migration tube. The corona discharge ionization source and specific ionization voltage efficiently ionize the target substance to form ions, improving detection sensitivity.

[0037] Specifically, the three-stage heating program (initial temperature of 35-45°C maintained for 2-5 minutes, and then gradually increased to 200-250°C) can achieve efficient separation based on the boiling point differences of volatile organic compounds in food. Combined with the ion mobility spectrometry conditions (drift gas flow rate of 100-200mL / min, ionization voltage of 2000-3000V), the resolution of characteristic peaks is improved by more than 20%, and the detection time is shortened by 30% compared with traditional methods. At the same time, the database-driven parameter matching mechanism solves the pain point of manual debugging of different food testing conditions, and supports "plug and play" in multiple scenarios such as meat cold chain and fruit and vegetable sampling, providing an intelligent solution for the large-scale application of GC-IMS technology in the entire food industry chain.

[0038] During rapid testing of the chilled salmon production line, the test object is chilled salmon fillets, which are identified as belonging to the "aquatic products - medium to high fat content" category. The optimal parameter combination for fish spoilage characteristics (such as trimethylamine and volatile sulfur compounds) is automatically selected: Chromatographic conditions: A three-step temperature program was set as follows: initially at 40°C for 3 min, then at 8°C / min to 160°C, and then at 18°C / min to 220°C (optimized for high-boiling-point sulfur compounds in fish). Ion mobility spectrometry conditions: ion migration tube temperature 50 °C, drift gas flow rate 150 mL / min, ionization voltage 2500 V (to enhance the ionization efficiency of polar small molecules).

[0039] Compared with traditional fixed parameters (such as constant temperature of 180°C), the separation between the characteristic peak of trimethylamine and adjacent peaks is improved from 1.2 to 1.8 (achieving baseline separation), and the detection time is shortened from 12 minutes to 8 minutes; the database matching parameters reduce the root mean square error of TVB-N prediction from 2.1 mg / 100g to 1.3 mg / 100g, meeting the high-precision requirements of freshness grading of aquatic products.

[0040] In an embodiment of the present application, a hierarchical food type database is constructed by hierarchical classification based on the type, processing method, and storage conditions of food, including: establishing a food type classification framework, including primary classification and secondary classification, collecting GC-IMS full spectrum data, volatile organic compound concentration and sensory score of primary and secondary classified foods in the spoilage process; classifying and labeling foods according to processing methods, analyzing their effects on volatile organic compounds and freshness, and forming a characteristic parameter set; establishing a multidimensional model based on storage conditions, collecting data to construct a storage and freshness mapping relationship, and then integrating food type and processing method data to construct a spoilage dynamics prediction matrix; integrating data related to food type, processing method, and storage conditions, assigning weights through the entropy weight method, and generating hierarchical nodes; constructing a bidirectional index between food type, processing method, storage conditions and GC-IMS detection parameters and freshness evaluation model based on the association relationship between hierarchical nodes.

[0041] It can be understood that the embodiments of the present application are based on the hierarchical classification of food types, processing methods, and storage conditions, systematically integrating GC-IMS full spectrum data, volatile organic compound concentrations and sensory scores in the spoilage process, and deeply analyzing the impact mechanism of different factors on freshness; by establishing a multidimensional model and a spoilage dynamics prediction matrix, quantifying the interaction of storage conditions, processing methods and food types on the spoilage process; using the entropy weight method to scientifically allocate weights, highlighting the importance of each factor in freshness detection, and generating hierarchical nodes; the construction of a bidirectional index realizes the rapid matching of GC-IMS detection parameters and freshness evaluation models from food characteristics, and the reverse correlation of food characteristics from the detection results. Bidirectional precise mapping.

[0042] Specifically, taking fresh chicken and cured meat as an example, there are significant differences between the two in terms of food types, processing methods, and storage conditions. In the process of constructing this hierarchical food type database: Data Collection and Analysis: GC-IMS full-spectrum data was collected for fresh chicken during its decay process. It was found that as decay progressed, the concentrations of volatile organic compounds (VOCs) such as trimethylamine and hydrogen sulfide increased significantly, and sensory scores were used to indicate the freshness of the meat at different stages. In contrast, due to salting and smoking during processing, the VOC changes in cured meat during the decay process were primarily characterized by changes in organic acids and aldehydes. This analysis revealed that the processing methods had distinct effects on the VOCs and freshness of the two, resulting in the development of unique characteristic parameter sets for each.

[0043] Model Construction and Weight Assignment: Regarding storage conditions, fresh chicken is typically refrigerated, so a multidimensional model was developed linking refrigeration temperature (e.g., 0-4°C), storage time, and freshness. Cured meats are typically stored in a dry, cool place, so a mapping relationship between temperature, humidity, and freshness was constructed. By integrating data on food type and processing methods for both, different spoilage dynamics prediction matrices were constructed. Using the entropy weighting method, storage temperature was given a higher weight for fresh chicken due to its significant impact on microbial growth. Processing methods (such as salt content and degree of smoking) for cured meats were given a higher weight due to their direct impact on spoilage.

[0044] Bidirectional indexing application: When testing fresh chicken, the database automatically matches the chromatographic column type, low-temperature column oven temperature program, and appropriate ion mobility spectrometry conditions suitable for the separation of low-boiling-point volatile amines based on information such as food type and refrigerated storage conditions, and simultaneously calls the corresponding freshness assessment model; for cured bacon, it quickly matches the detection parameters and exclusive assessment models suitable for the separation of aldehydes and organic acids, achieving accurate detection of the freshness of different types of food.

[0045] In step S103, the headspace sampling gas is injected into the GC-IMS, and is analyzed based on the matched chromatographic conditions and ion mobility spectrometry conditions to obtain a volatile organic compound fingerprint of the sample.

[0046] It is understood that the present embodiment, which injects headspace sample gas into a GC-IMS and analyzes it based on matching conditions, enables efficient separation and precise detection of volatile organic compounds in food. Parameters such as the column oven program and carrier gas flow rate in the chromatographic conditions enable gradient elution and separation of volatile organic compounds based on the characteristics of different foods, avoiding interference from overlapping peaks. Parameters such as the drift tube temperature and drift gas flow rate in the ion mobility spectrometry conditions ensure stable ion migration in the electric field, enhancing detection sensitivity.

[0047] In step S104, preset key feature information is extracted from the fingerprint, and the key feature information is input into a food freshness assessment model to obtain a freshness assessment result of the food sample to be tested.

[0048] Among them, the key feature information includes the intensity, area, or signal pattern of a specific feature area of at least one characteristic peak, and the freshness assessment result includes at least one of the freshness grade, freshness score, predicted specific corruption indicator value or predicted remaining shelf life.

[0049] It is understood that the embodiments of the present application screen key feature regions through random forest feature importance scoring and cluster analysis, eliminating redundant information, focusing on volatile organic compound features closely related to the spoilage process, and avoiding interference from secondary signals. The multidimensional feature vectors constructed can quantify the synergistic change trends between features and comprehensively reflect the dynamic changes in food freshness. By inputting this key feature information into the food freshness assessment model and combining it with the spoilage reaction benchmark rate constant, temperature compensation mechanism, and spoilage dynamics prediction matrix, the food spoilage rate can be scientifically quantified and the freshness grade, score, remaining shelf life, and other results can be accurately calculated.

[0050] In an embodiment of the present application, preset key feature information is extracted from the fingerprint, including: preprocessing the GC-IMS fingerprints of similar food samples with known different freshness levels; using random forest feature importance scoring to screen out characteristic peaks with significant differences between different freshness levels, and performing cluster analysis to identify characteristic peak clusters with a synergistic change trend, which are defined as characteristic regions; calculating the intensity ratio, peak area ratio or correlation coefficient of each characteristic peak in the characteristic region, and constructing a multidimensional feature vector as key feature information.

[0051] Among them, preprocessing can include baseline correction, peak alignment and normalization processing. The characteristic peak cluster is a group of characteristic peaks with a synergistic change trend obtained through cluster analysis. The characteristic region is the distribution area of the characteristic peak cluster in the GC-IMS two-dimensional spectrum defined as the characteristic region. The intensity ratio is the ratio of the intensity values of two characteristic peaks in the characteristic region, such as the ratio of the hexanal peak intensity to the heptanal peak intensity in the aldehyde characteristic region. The peak area ratio is the ratio of the total area of a peak cluster in the characteristic region to the total area of another peak cluster, such as the ratio of the lipid oxidation peak cluster area to the protein decomposition peak cluster area. The correlation coefficient is the Pearson correlation coefficient of the peak intensities in the characteristic region, which measures the degree of synergistic change between peaks.

[0052] It can be understood that the embodiments of the present application eliminate instrument noise and batch differences through baseline correction, peak alignment and normalization processing to ensure the consistency of the spectral data; use the random forest algorithm to screen out characteristic peaks with significant differences between different freshness levels, and combine cluster analysis to identify characteristic peak clusters with synergistic change patterns (such as aldehyde peak clusters in meat corruption and ethylene-alcohol peak groups in fruit and vegetable ripening), breaking through the limitations of traditional single peak analysis; finally, by calculating parameters such as the intensity ratio and peak area correlation within the characteristic region, a multidimensional feature vector containing spatial distribution and dynamic correlation is constructed, which effectively solves the problems of one-sided feature analysis and insufficient model generalization ability in the prior art.

[0053] Specifically, in the salmon sample spectrum, the low-frequency baseline drift caused by the instability of the ionization source can be eliminated by fitting a fifth-order polynomial, reducing the intensity error of the trimethylamine characteristic peak from ±15% to ±3%.

[0054] For 100 groups of salmon samples of different freshness (TVB-N5-30mg / 100g), the random forest algorithm calculates the Gini impurity reduction of 200+ characteristic peaks: Hexanal peak (retention time 6.8 min): ΔGini = 0.28 (highest), indicating that it contributes most to TVB-N classification; Trimethylamine peak (7.0 min): ΔGini = 0.25, the second highest; Background peaks (such as the unknown peak at 10 min): ΔGini < 0.05, are eliminated.

[0055] Screening results: The first 20 important peaks were retained, including characteristic peaks such as aldehydes (hexanal, heptanal), sulfur-containing compounds (trimethylamine, methyl mercaptan), etc.

[0056] The Pearson correlation coefficients of the first 20 peaks were calculated, and the correlation coefficients of hexanal with heptanal were r = 0.93, and with octanal r = 0.89; Ward linkage clustering was used, and when the intra-group sum of square increments exceeded the threshold, it was clustered into an “aldehyde characteristic cluster” (retention time 6.5-8.5 min, migration time 8-12 ms); The cluster peak intensity increases exponentially with the increase of TVB-N (R 2 =0.91), reflecting the fat oxidation process of salmon.

[0057] The formula for calculating the intensity ratio is:

[0058] in, is the peak intensity of heptanal, is the peak intensity of hexanal.

[0059] The ratio increased from 1.2 (fresh) to 2.3 (corrupted) with the increase of TVB-N, which was higher than the single hexanal peak intensity (linear correlation R 2 =0.78) and TVB-N had a stronger nonlinear correlation (R 2 =0.89).

[0060] Lipid oxidation peak clusters: hexanal, heptanal, octanal, area and A lipid =35000a.u.; Protein decomposition peak clusters: trimethylamine, methyl mercaptan, area and A protein =12000a.u.; AreaRatio=2.92 AreaRatio is the instantaneous value of the sample. This ratio decreases from 3.5 (fresh) to 1.2 (spoilage) during the salmon spoilage process, reflecting that the dominant spoilage pathway shifts from fat oxidation to protein decomposition.

[0061] The correlation coefficient between hexanal and heptanal was r = 0.94, indicating that both were synchronously regulated by lipoxygenase and could be input into the model as a composite feature (e.g., [1.47, 2.92, 0.94]).

[0062] In summary, the embodiment of the present application performs 5th-order polynomial baseline correction (error reduced from ±15% to ±3%) and dynamic time warping peak alignment on the salmon GC-IMS spectrum, and combines random forest to screen out the top 20 characteristic peaks such as hexanal (ΔGini=0.28) and trimethylamine, and constructs an aldehyde characteristic cluster (R²=0.91) through Ward clustering. Then, the intensity ratio (R²=0.89), area ratio (reflecting the conversion of corruption pathways) and correlation coefficient (r=0.94) are calculated, and a three-dimensional feature vector is constructed to reduce the TVB-N prediction error to 1.3 mg / 100 g (an improvement of 38% compared with a single peak), achieving a leap from single-point detection to corruption mechanism analysis, and providing a quantitative solution for on-site rapid detection of aquatic product freshness.

[0063] In the embodiment of the present application, the calculation process of the food freshness evaluation model includes: The corruption reaction base rate constant is calculated based on the key characteristic information, where the formula is:

[0064] The formula for the storage temperature compensation term is:

[0065] The output evaluation results are:

[0066]

[0067] in, is the base rate constant of the corruption reaction, is the weight coefficient of key feature information, is the current strength of the i-th key feature information, is the baseline strength of the i-th key feature information, is the number of key feature information, is the activation energy, is the corruption reaction rate constant after storage temperature compensation, is the gas constant, is the actual storage temperature, is the reference temperature, For the remaining shelf life, is the initial freshness characteristic, To store the detection value of the feature quantity after t time, is the corruption risk index, is the rate constant corresponding to the national standard safety limit.

[0068] In an embodiment of the present application, after obtaining the freshness evaluation result of the food sample to be tested, the method further includes: generating corresponding warning information according to a preset freshness threshold.

[0069] Among them, early warning information may include prompting food to be processed as soon as possible or sold safely.

[0070] It can be understood that after obtaining the freshness assessment results of the food samples to be tested, the embodiments of the present application generate corresponding warning information based on the preset freshness threshold. Through accurate and differentiated information output, it provides clear action guidance for all participants in the food chain, which can effectively assist in timely decision-making, help optimize inventory turnover and reduce food losses.

[0071] Specifically, a cold chain storage company stores 50 tons of salmon and uses GC-IMS testing equipment to assess the freshness of inventory samples every 2 hours. When the model predicts that the TVB-N of a batch of salmon reaches 14mg / 100g (close to the lower limit of the freshness threshold of 15mg / 100g), the system triggers a "near expiration date warning" and marks the batch as "recommended to be shipped within 48 hours"; if the TVB-N reaches 20mg / 100g (semi-fresh range), a "need to be processed as soon as possible warning" is generated and pushed to the sorting department, with priority arranged for processing into raw materials such as surimi; if the TVB-N is less than 15mg / 100g and the total bacterial count is less than 10 5 CFU / g, output "safe sales warning", synchronize to the sales system, mark as "high-quality fresh product" and put it on the shelf quickly.

[0072] In summary, before the application of early warning, the monthly loss rate reached 8% due to delayed inventory turnover. After activation, the monthly loss rate dropped to 3% through precise scheduling based on early warning, the inventory turnover cycle was shortened by 2.5 days, and the customer complaint rate dropped by 12% due to the priority sales of fresh products.

[0073] According to the examples of this application, a GC-IMS-based food freshness detection method proposed by this application constructs a hierarchical food type database and automatically matches GC-IMS detection parameters based on food type, processing method, and storage conditions. This method completely overcomes the limitations of traditional fixed parameters and achieves precise adaptation for different food types. Furthermore, the use of a standardized headspace sampling process and automated sampling design, combined with the rapid separation and detection characteristics of GC-IMS coupling technology, significantly shortens the single detection cycle, effectively meeting the timeliness requirements of real-time production line monitoring and rapid warehouse sampling. In addition, the headspace gas analysis mode can obtain volatile organic compound information without destroying the sample, which can not only fully preserve the original quality of the food, but also be suitable for online continuous monitoring of fresh food; the combination of standardized sampling parameters and automated headspace sampling devices effectively circumvents the problem of poor consistency in sample processing during on-site testing, ensuring that the test data of different batches and different scenarios are reliably comparable; the embodiment of the present application combines random forest feature importance scoring and cluster analysis to accurately screen key feature areas from the volatile organic compound fingerprint map, and constructs a multidimensional feature vector through the characteristic peak intensity ratio, peak area ratio, etc., which can more comprehensively and accurately capture hidden information related to corruption. In this way, problems such as the lack of adaptability of solidified detection parameters, inaccurate analysis of key features, and insufficient model prediction accuracy are solved.

[0074] The following is an example of a GC-IMS-based food freshness detection method. Taking the freshness detection of chilled beef as an example, the specific content is as follows: 1. Standardized headspace sampling: 40°C constant temperature activation of meat characteristic volatile compounds Sample preparation: 5 g of fresh beef tenderloin (TVB-N measured value 8 mg / 100 g, fresh state) was placed in a 20 mL headspace injection vial (volume to mass ratio 4 mL / g).

[0075] Modeling set: 80 groups of beef samples were prepared, covering the TVB-N value range of 5-30 mg / 100 g.

[0076] Independent validation set: Contains fresh beef samples (TVB-N = 8 mg / 100 g).

[0077] Place an appropriate amount of beef sample (ensuring that the volume to mass ratio in the headspace vial is 4 mL / g) in a 20 mL headspace injection vial.

[0078] Thermostatic activation: Constant temperature: 40℃, this temperature is conducive to the efficient release of key volatile organic compounds in beef, such as aldehydes and sulfur-containing compounds.

[0079] Standing time: 20 min. This time helps reduce the interference of myoglobin oxidation on headspace components.

[0080] Constant temperature: 50°C (to optimize the volatilization efficiency of aldehydes and sulfur compounds in beef); Standing time: 20 min (to reduce the interference of myoglobin oxidation on headspace components).

[0081] like Figure 2 As shown, the concentration of hexanal in the headspace reached 980 a.u. (750 a.u. for sampling at room temperature), the concentration of trimethylamine was 620 a.u., and the correlation coefficient between the characteristic peak intensity and TVB-N was increased to 0.91 (only 0.78 for traditional sampling).

[0082] GC-IMS Analysis: Database-Driven Parameter Matching Enhances Separation Efficiency Food type identification: The database automatically retrieves unique detection parameters through a three-level classification of "livestock, red meat, and beef tenderloin"; Chromatographic conditions (three-step heating): initially at 40°C for 3 min; heating at 8°C / min to 160°C; heating at 18°C / min to 220°C; Ion mobility spectrometry conditions: drift tube temperature 50 °C, high-purity nitrogen flow rate 150 mL / min, ionization voltage 2500 V (to enhance the ionization efficiency of nitrogen-containing compounds); Therefore, the separation between the characteristic peak of hexanal (retention time 6.5 min) and the nonanal peak was improved from 1.8 to 2.3 (baseline separation), and the detection cycle was shortened from 7 min to 6 min.

[0083] 3. Graph Data Processing: From Meat Spoilage Markers to Multidimensional Feature Construction Pretreatment: 7th-order polynomial baseline correction: eliminates hemoglobin interference in meat samples, reducing the trimethylamine peak intensity error from ±12% to ±2%; dynamic time warping peak alignment: aldehyde peak position deviation is reduced from ±0.5min to ±0.08min; Random forest screening: The importance of characteristic peaks was calculated for 80 groups of beef samples (TVB-N5-30mg / 100g), and the top 15 key peaks were retained, including: Hexanal (ΔGini=0.32, a marker of fat oxidation), Trimethylamine (ΔGini=0.29, protein degradation marker), Methyl mercaptan (ΔGini = 0.21, degradation product of sulfur-containing amino acids); Hierarchical clustering: Hexanal and nonanal had a correlation coefficient of r=0.90, clustering into a "fat oxidation characteristic cluster" (retention time 6.0-8.0 min); Multidimensional feature construction: Intensity ratio: hexanal / nonanal = 1.54; Area ratio: fat oxidation peak cluster / sulfur compound peak cluster = 2.8; Correlation coefficient: hexanal-trimethylamine = 0.91.

[0084] IV. Model Prediction and Grade Determination: SVR Fitting and Livestock and Poultry Meat Threshold System Model training: Freshness prediction: Input feature vector [1.54, 2.8, 0.91], model output TVB-N = 21.2 mg / 100 g (sub-fresh level, because > 15 mg / 100 g); total colony count = 7 × 10 5 CFU / g; Grade determination: Based on the threshold for livestock and poultry meat, it is determined to be "sub-fresh" grade.

[0085] 5. Warning Generation and Cold Chain Application: Dynamic Scheduling of Beef Processing Lines Warning trigger: When TVB-N of a batch of beef is 14.5mg / 100g (close to the freshness threshold), a "processing warning within 48 hours" is generated; In the embodiment of the present application, TVB-N=21.2mg / 100g triggered a "downgrade processing warning" and was automatically pushed to the minced meat production line. The batch of beef was dispatched to be minced into minced meat within 1.5 hours, which was 8 times faster than manual inspection. The monthly loss rate of cold chain storage was reduced from 7% to 2.5%. The premium rate of high-end steak raw materials increased by 10% due to accurate grading.

[0086] Next, a food freshness detection device based on GC-IMS proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0087] Figure 3Schematic diagram of a food freshness detection device based on GC-IMS according to an embodiment of the present application.

[0088] like Figure 3 As shown, the food freshness detection device 10 based on GC-IMS includes: an acquisition module 100 , a matching module 200 , an analysis module 300 and a calculation module 400 .

[0089] Among them, the acquisition module 100 is used to obtain the headspace sampling gas of the food sample to be tested; the matching module 200 is used to automatically match the preset chromatographic conditions and ion mobility spectrum conditions from the graded food type database according to the type of the food sample to be tested; the analysis module 300 is used to inject the headspace sampling gas into the GC-IMS, analyze it based on the matched chromatographic conditions and ion mobility spectrum conditions, and obtain the volatile organic compound fingerprint of the sample; the calculation module 400 is used to extract preset key feature information from the fingerprint, input the key feature information into the food freshness assessment model, and obtain the freshness assessment result of the food sample to be tested, wherein the freshness assessment result includes at least one of the freshness grade, freshness score, predicted specific corruption index value or predicted remaining shelf life.

[0090] It should be noted that the aforementioned explanation of the embodiment of the food freshness detection method based on GC-IMS is also applicable to the food freshness detection device based on GC-IMS in this embodiment, and will not be repeated here.

[0091] According to the embodiments of this application, a GC-IMS-based food freshness detection device, by constructing a hierarchical food type database, automatically matches GC-IMS detection parameters according to food type, processing method, and storage conditions, completely breaking the limitations of traditional fixed parameters and achieving precise adaptation for different food types. At the same time, the use of a standardized headspace sampling process and automated sampling design, combined with the rapid separation and detection characteristics of GC-IMS coupling technology, significantly shortens the single detection cycle, effectively meeting the timeliness requirements of real-time production line monitoring and rapid warehouse sampling. In addition, the headspace gas analysis mode can obtain volatile organic compound information without destroying the sample, which can not only fully preserve the original quality of the food, but also be suitable for online continuous monitoring of fresh food; the combination of standardized sampling parameters and automated headspace sampling devices effectively circumvents the problem of poor consistency in sample processing during on-site testing, ensuring that the test data of different batches and different scenarios are reliably comparable; the embodiment of the present application combines random forest feature importance scoring and cluster analysis to accurately screen key feature areas from the volatile organic compound fingerprint map, and constructs a multidimensional feature vector through the characteristic peak intensity ratio, peak area ratio, etc., which can more comprehensively and accurately capture hidden information related to corruption. In this way, problems such as the lack of adaptability of solidified detection parameters, inaccurate analysis of key features, and insufficient model prediction accuracy are solved.

[0092] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include: Memory 401 , processor 402 , and computer programs stored in the memory 401 and executable on the processor 402 .

[0093] When the processor 402 executes the program, the food freshness detection method based on GC-IMS provided in the above embodiment is implemented.

[0094] Furthermore, the electronic device further includes: The communication interface 403 is used for communication between the memory 401 and the processor 402 .

[0095] The memory 401 is used to store computer programs that can be run on the processor 402 .

[0096] The memory 401 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0097] If the memory 401, processor 402, and communication interface 403 are implemented independently, the communication interface 403, memory 401, and processor 402 can be interconnected via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 4 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0098] Optionally, in a specific implementation, if the memory 401, the processor 402 and the communication interface 403 are integrated on a chip, the memory 401, the processor 402 and the communication interface 403 can communicate with each other through an internal interface.

[0099] The processor 402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0100] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0101] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0102] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing a custom logical function or process step, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed in a different order than shown or discussed, including performing functions in a substantially simultaneous manner or in a reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0103] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using any one or a combination of the following technologies known in the art: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0104] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

Claims

1. A food freshness detection method based on GC-IMS, characterized in that: The following steps are involved: Obtaining headspace sampling gas of the food sample to be tested; Automatically matching preset chromatographic conditions and ion mobility spectrometry conditions from a graded food type database according to the type of the food sample to be tested; Injecting the headspace sampling gas into the GC-IMS, performing analysis based on matching chromatographic conditions and ion mobility spectrometry conditions, and obtaining a volatile organic compound fingerprint of the sample; Preset key feature information is extracted from the fingerprint, and the key feature information is input into a food freshness assessment model to obtain a freshness assessment result of the food sample to be tested, wherein the freshness assessment result includes at least one of a freshness grade, a freshness score, a predicted specific corruption index value, or a predicted remaining shelf life.

2. The food freshness detection method based on GC-IMS according to claim 1, characterized in that, The hierarchical food type database is constructed by hierarchical classification based on the type, processing method, and storage conditions of food, including: Establish a food classification framework, including primary and secondary classifications, and collect GC-IMS full spectrum data, volatile organic compound concentrations, and sensory scores of foods in the primary and secondary classifications during the spoilage process; Classify and label foods by processing method, analyze their impact on volatile organic compounds and freshness, and form a characteristic parameter set; A multidimensional model was established based on storage conditions, data was collected to construct a mapping relationship between storage and freshness, and then data on food types and processing methods were integrated to construct a spoilage dynamics prediction matrix; Integrate the data related to the food types, processing methods, and storage conditions, assign weights using the entropy weight method, and generate hierarchical nodes; Based on the association relationship between the hierarchical nodes, a bidirectional index is constructed between food type, processing method, storage conditions and GC-IMS detection parameters and freshness evaluation model.

3. The food freshness detection method based on GC-IMS according to claim 1, characterized in that, Extracting preset key feature information from the fingerprint map includes: Preprocessing of GC-IMS fingerprints of similar food samples with different known freshness levels; Random forest feature importance scoring was used to screen out characteristic peaks with significant differences between different freshness levels. Cluster analysis was then performed to identify characteristic peak clusters with a synergistic change trend, which were defined as characteristic regions. The intensity ratio, peak area ratio or correlation coefficient of each characteristic peak in the characteristic region is calculated to construct a multidimensional characteristic vector as the key characteristic information.

4. The food freshness detection method based on GC-IMS according to claim 1, characterized in that, The calculation process of the food freshness evaluation model includes: The corruption reaction base rate constant is calculated based on the key characteristic information, where the formula is: , the formula for storing the temperature compensation term is: , the output evaluation result is: , ,in, is the base rate constant of the corruption reaction, is the weight coefficient of key feature information, is the current strength of the i-th key feature information, is the baseline strength of the i-th key feature information, is the number of key feature information, is the activation energy, is the corruption reaction rate constant after storage temperature compensation, is the gas constant, is the actual storage temperature, is the reference temperature, For the remaining shelf life, is the initial freshness characteristic, To store the detection value of the feature quantity after t time, is the corruption risk index, is the rate constant corresponding to the national standard safety limit.

5. The food freshness detection method based on GC-IMS according to claim 1, characterized in that, The sampling temperature of the headspace sampling gas is 30-60° C., the sampling time is 10-30 min, and the ratio of the volume of the headspace sampling bottle to the sample mass is 4-10 mL / g.

6. The food freshness detection method based on GC-IMS according to claim 1, characterized in that: The preset chromatographic conditions include chromatographic column type, column oven temperature program, and carrier gas flow rate; the ion mobility spectrometry conditions include migration tube temperature, migration voltage, and drift gas flow rate.

7. The food freshness detection method based on GC-IMS according to claim 1, characterized in that: After obtaining the freshness evaluation result of the food sample to be tested, the method further includes: generating corresponding warning information according to a preset freshness threshold, wherein the warning information includes prompting the food to be processed as soon as possible or sold safely.

8. The food freshness detection method based on GC-IMS according to claim 1, characterized in that: Before injecting the head space sampling gas into the GC-IMS, the method further includes: filtering and drying the head space sampling gas.

9. A food freshness detection device based on GC-IMS, characterized in that: include: An acquisition module, used for acquiring headspace sampling gas of the food sample to be tested; A matching module, configured to automatically match preset chromatographic conditions and ion mobility spectrometry conditions from a hierarchical food type database according to the type of the food sample to be tested; An analysis module is used to inject the headspace sampling gas into the GC-IMS, analyze it based on the matching chromatographic conditions and ion mobility spectrometry conditions, and obtain a volatile organic compound fingerprint of the sample; a computing module, configured to extract preset key feature information from the fingerprint, input the key feature information into a food freshness assessment model, and obtain a freshness assessment result of the food sample to be tested, wherein the freshness assessment result includes at least one of a freshness grade, a freshness score, a predicted specific corruption index value, or a predicted remaining shelf life.

10. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the food freshness detection method based on GC-IMS according to any one of claims 1 to 8.

Images