A method for rapid and accurate detection of flavors and fragrances by portable density meter
Through the optimization of cleaning agents and the establishment of linear regression model of portable densitometer, the problem of deviation in the quality inspection of flavor and fragrance is solved, and the rapid and accurate detection of flavor and fragrance density is achieved, and the fragrance base, material base, leaf liquid, Stability and consistency of tobacco surface fragrance.
Patent Information
- Application Number
- CN202210939244.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-08-05
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2042-08-05
AI Technical Summary
In the prior art, the quality stability inspection of flavors and fragrances depends on sensory verification and physical indicators, such as relative density and refractive index, resulting in deviations in the detection results and quality risks, making it difficult to achieve rapid and scientific verification.
Through the method of quickly and accurately detecting flavors and fragrances with a portable densitometer, we study the impact of cleaning agent concentration and cleaning agent sequence on the detection value, establish a linear regression prediction model, eliminate system errors, and realize the accurate matching of the detection data of the portable densitometer and the density refractive integrated machine. .
It realizes rapid and accurate detection of the density and relative density of flavors, reduces detection time and cost, improves quality control capabilities, and ensures the stability and consistency of fragrance base, material base, leaf liquid, and tobacco surface fragrance.
Smart Images

Figure CN115468873B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of flavors and fragrances, and in particular to a method for quickly and accurately detecting flavors and fragrances by using a portable density meter. Background Art
[0002] Tobacco flavor refers to an additive for tobacco products, which is composed of two or more tobacco flavors in a certain ratio and added with applicable auxiliary materials and appropriate solvents. It is specially used for flavoring and correcting the flavor of various tobacco products, so that they can produce beautiful aroma and comfortable taste when they are re-burned and smoked.
[0003] Tobacco flavoring gives tobacco products a beautiful and comfortable smell, gives tobacco products a characteristic aroma, eliminates the differences between tobacco leaves of different grades, coordinates the aroma of different types of tobacco leaves, reduces the irritation of smoke, masks impurities, makes the smoke soft and mellow, improves the aftertaste, increases or enhances certain aromas, and makes the smoke fuller. The accuracy of the blending of flavors and fragrances directly affects the flavoring effect.
[0004] At present, the quality stability test of our company's flavors and fragrances is mainly based on the physical indicators of relative density and refractive index. The silk-making workshop uses portable machines to complete the quality inspection of process ingredients, and the process quality department uses integrated machines for supervision inspection and key brand inspection. During the inspection process, there was a difference in the detection accuracy of the DMA501 portable density meter and the DMA4500-Abbmat 300 density and refraction integrated machine. The testers used two different types of instruments to test the results, and there was a deviation in the values of the two. The quality verification of each brand of leaf material and surface fragrance is completed by sensory comparison of standard samples. The quality of ingredients between batches relies only on sensory verification and process supervision, which poses certain quality risks.
[0005] This application aims to verify the quality stability of flavors and fragrances more quickly and scientifically. Summary of the invention
[0006] To solve the above problems, the present application provides a method for quickly and accurately detecting flavors and fragrances through a portable density meter, which can realize the transformation from sensory verification to physical indicator detection, and the transformation from extensive data accumulation to precise data control, thereby making the intrinsic quality of the aroma base, material base, leaf (stem) material liquid, and tobacco (stem) surface aroma entering the production link more stable and accurate, thereby improving the stability and consistency of the sensory indicators of cigarettes on this basis.
[0007] The technical solution adopted by the present invention to solve the technical problem is:
[0008] Methods for quickly and accurately testing flavors and fragrances using a portable density meter include:
[0009] S1, Study on the influence of cleaning agent concentration on the detection value of portable density meter
[0010] By increasing the concentration of cleaning agent I: alcohol from 75% to 100%, the effect of the concentration of cleaning agent I on the detection value of the portable density meter was studied:
[0011] When the alcohol concentration increases from 75% to 95%, the value detected by the portable density meter changes significantly;
[0012] When the alcohol concentration increases from 95% to 100%, the value detected by the portable density meter remains basically unchanged;
[0013] Determine the concentration of cleaning agent I: alcohol is 95%;
[0014] Cleaning agent II: pure water;
[0015] S2, Study on the influence of cleaning agent sequence on the detection value of portable density meter
[0016] Select standard samples, use cleaning agents I and II in different use orders to clean the portable density meter, test the density of the standard samples after cleaning, and obtain the test values under different cleaning agent use orders;
[0017] Compare and determine which cleaning agent usage sequence results in a smaller deviation between the test value of the standard sample and the theoretical value;
[0018] The final determined order of use of cleaning agent I and cleaning agent II is obtained;
[0019] S3, sample density detection
[0020] N low / medium / high density flavor and fragrance samples are respectively sampled, and their densities are tested by a portable densitometer and a density refractometer to obtain N groups of sample density test data;
[0021] S4, Data analysis and processing
[0022] S4.1. Data Analysis
[0023] The density comparison data of N groups of samples were analyzed, and the deviation between the density detection data of each group of portable density meter and density refractometer was studied. The deviation was determined to have the following rules and trends: for samples with a standard density greater than that of pure water, the detection value of the portable density meter was smaller than that of the density refractometer; for samples with a standard density less than that of pure water, the detection value of the portable density meter was larger than that of the density refractometer;
[0024] Determine the correlation between the test values of portable density meter and density refractometer;
[0025] S4.2 Data processing
[0026] Eliminate abnormal data in N groups of sample density detection data;
[0027] S4.3, Data Grouping
[0028] Based on the results of S4.1 data analysis, the sample density detection data is grouped again:
[0029] The test data of samples with density greater than pure water are grouped together;
[0030] The test data of samples with density less than pure water are grouped together;
[0031] S5, Construction of prediction model
[0032] S5.1, Construction of the split model
[0033] The test data processed by S4.2 and S4.3 were grouped and entered into the mini-tab software to verify whether the data conformed to the normal distribution;
[0034] Select statistical regression and fit linear graph in the data statistics command to establish a linear regression prediction model;
[0035] For samples with a density greater than that of pure water, the linear regression prediction split model established is model 1-1;
[0036] For samples with a density lower than that of pure water, the linear regression prediction split model established is Model 1-2;
[0037] S5.2. Construction of the overall model
[0038] Enter the test data processed by S4.2 into the mini-tab software to verify whether the data conforms to the normal distribution;
[0039] Select statistical regression in the data statistics command and fit the linear graph to establish a linear regression prediction overall model;
[0040] The overall linear regression prediction model established is model 2;
[0041] S6. Validation of the prediction model
[0042] Select several standard samples to verify the prediction accuracy of Model 1-1, Model 1-2, and Model 2;
[0043] A portable densitometer was used to detect the density of the standard sample, and the detected values were entered into each prediction model respectively to obtain the predicted value of the standard sample when detected by a density refractometer under each prediction model. The predicted value was compared with the measured value to obtain the deviation value, and the prediction accuracy and reliability of each prediction model were verified by deviation analysis.
[0044] The beneficial effects brought by the present invention are:
[0045] This method changes the traditional sensory verification method into physical indicator detection and verification. By studying the analysis of the numerical deviation of the portable density meter and the density refractometer, the optimal cleaning agent concentration and cleaning sequence are selected for the portable density meter. The test data of the two instruments are used to establish a data prediction model to achieve the purpose of eliminating systematic errors. The portable density meter can be used to quickly and accurately detect the relative density of flavors and fragrances.
[0046] This method can quickly detect the density and relative density of flavor and fragrance samples. The actual operation process is relatively simple and convenient. The predicted value obtained by the prediction model is compared with the measured value of the density refractometer. The numerical deviation between the two is extremely small, indicating that the prediction result of this method is accurate and reliable; that is, this method can be used for the factory inspection of flavors and fragrances and the quality inspection of the production process, improve the inspection efficiency, reduce the inspection cost, use a simple and efficient detection method, facilitate the scientific judgment of quality management personnel, and comprehensively improve the quality control ability and level.
[0047] The outstanding contributions of the present invention are: first, the density of flavors and fragrances can be quickly detected; second, the accuracy of the predicted values can be guaranteed to reach the level of a density refraction integrated machine. The present invention realizes the transformation from sensory verification to physical index detection, and the transformation from extensive data accumulation to precise data control. Under the premise of ensuring accuracy, the detection time of a sample is reduced from 18 minutes to 5 minutes, thereby making the internal quality of the aroma base, material base, leaf (stem) material liquid, and tobacco (stem) surface aroma entering the production link more rapid, stable, and accurate, and on this basis, the stability and consistency of the sensory indicators of cigarettes can be improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.
[0049] Figure 1 This is the linear regression prediction model of this method: the fitted linear graph of model 1-1;
[0050] Figure 2 This is the linear regression prediction model of this method: the fitted linear graph of model 1-2;
[0051] Figure 3 Linear regression prediction model of this method: fitted linear plot of model 2. DETAILED DESCRIPTION
[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0053] Example 1
[0054] Methods for quickly and accurately testing flavors and fragrances using a portable density meter include:
[0055] S1, Study on the influence of cleaning agent concentration on the detection value of portable density meter
[0056] By contrast experiment, the concentration of cleaning agent I: alcohol was increased from 75% to 100% to study the effect of the concentration of cleaning agent I on the detection value of the portable density meter:
[0057] When the alcohol concentration increases from 75% to 95%, the value detected by the portable density meter changes significantly;
[0058] When the alcohol concentration increases from 95% to 100%, the value detected by the portable density meter remains basically unchanged;
[0059] Because: Compared with 100% alcohol, 95% alcohol has a special alcohol storage room, fixed purchase approval procedures, a more advantageous price, a higher safety factor, and a fixed source and supplier. 100% pure alcohol needs to be purchased temporarily, has a long purchase cycle, is risky to store, and has a relatively high price.
[0060] Determine the concentration of cleaning agent I: alcohol is 95%;
[0061] Cleaning agent II: pure water;
[0062] S2, Study on the influence of cleaning agent sequence on the detection value of portable density meter
[0063] By comparing the effects of the order of cleaning agents on the test values, it was found that changing the order of using cleaning agents I and II would have a significant impact on the test values, and it was determined that the order of using cleaning agents would have an impact on the test values;
[0064] Select standard samples, use cleaning agents I and II in different use orders to clean the portable density meter, test the density of the standard samples after cleaning, and obtain the test values under different cleaning agent use orders;
[0065] Compare and determine which cleaning agent usage sequence results in a smaller deviation between the test value of the standard sample and the theoretical value;
[0066] The final determined order of use of cleaning agent I and cleaning agent II is obtained;
[0067] The final order of cleaning agent use is:
[0068] First use 95% alcohol, then pure water;
[0069] S3, sample density detection
[0070] N low / medium / high density flavor and fragrance samples are respectively sampled, and the sample densities are respectively tested by a portable densitometer and a density refractometer to obtain N groups of sample density test data;
[0071] The medium-density flavor and fragrance samples here mainly refer to samples with a density close to that of pure water, which mainly emphasizes the continuity of the data and has no actual distinction.
[0072] S4, Data analysis and processing
[0073] S4.1. Data Analysis
[0074] The density comparison data of N groups of samples were analyzed, and the deviation between the density detection data of each group of portable densitometers and density refractometers was studied. The analysis determined that the deviation had the following rules and trends:
[0075] For samples with a standard density greater than that of pure water, the detection value of the portable density meter is smaller than that of the density refractometer;
[0076] For samples with a standard density lower than that of pure water, the detection value of the portable density meter is greater than that of the density refractometer;
[0077] Therefore, it is determined that there is a correlation between the test values of the portable density meter and the density refractometer;
[0078] S4.2 Data processing
[0079] Eliminate abnormal data in N groups of sample density detection data;
[0080] Abnormal data includes:
[0081] Data whose trend is opposite to other data trends and is identified by data classification;
[0082] The data trend is consistent with other data trends, but the deviation of the test data is much greater than that of other data;
[0083] S4.3, Data Grouping
[0084] Based on the results of S4.1 data analysis, the sample density detection data is grouped again:
[0085] The test data of samples with density greater than pure water are grouped together;
[0086] The test data of samples with density less than pure water are grouped together;
[0087] S5, Construction of prediction model
[0088] Taking the density of pure water as the boundary, modeling is performed according to different data combinations:
[0089] S5.1, Construction of the split model
[0090] The test data processed by S4.2 and S4.3 were grouped and entered into the mini-tab software to verify whether the data conformed to the normal distribution;
[0091] In accordance with the normal distribution, statistical regression and linear graph fitting were selected in the data statistics command to establish a linear regression prediction model;
[0092] For samples with a density greater than that of pure water, the linear regression prediction split model established is model 1-1;
[0093] For samples with a density lower than that of pure water, the linear regression prediction split model established is Model 1-2;
[0094] S5.2. Construction of the overall model
[0095] Enter the test data processed by S4.2 into the mini-tab software to verify whether the data conforms to the normal distribution;
[0096] In line with the normal distribution, select statistical regression in the data statistics command and fit the linear graph to establish a linear regression prediction overall model;
[0097] The overall linear regression prediction model established is model 2;
[0098] S6. Validation of the prediction model
[0099] Select multiple standard samples to verify the prediction accuracy of Model 1-1, Model 1-2, and Model 2;
[0100] A portable densitometer was used to detect the density of the standard sample, and the detected values were entered into each prediction model respectively to obtain the predicted value of the standard sample when detected by a density refractometer under each prediction model. The predicted value was compared with the measured value to obtain the deviation value, and the prediction accuracy and reliability of each prediction model were verified by deviation analysis.
[0101] Example 2
[0102] A method for predicting the detection value of flavors and fragrances by a target high-precision detector quickly and accurately using a portable density meter, including:
[0103] S1, improve the cleaning operation method of portable density meter, and use experimental comparison method to confirm that the alcohol concentration of 95% is the best;
[0104] S2, using an experimental comparison method to confirm the test values of the two cleaning agents in different cleaning sequences, obtain the deviation between the test value of the standard sample and the standard value, and determine the cleaning sequence with the smaller deviation as the optimal cleaning sequence of the cleaning agent;
[0105] S3, after determining the alcohol concentration of 95% and the cleaning sequence of the cleaning agent, a plurality of samples are tested respectively by a portable density meter and a target high-precision detector to obtain a plurality of groups of test comparison data;
[0106] S4, modeling according to different data combinations;
[0107] Data combinations include:
[0108] First combination:
[0109] A set of comparative data formed by testing high-density samples is formed by using a portable density meter and a target high-precision detector;
[0110] A set of comparative data formed by testing low-density samples using a portable density meter and a target high-precision detector;
[0111] The standard for distinguishing high-density samples from low-density samples is the density of pure water;
[0112] The purpose is to build the data model in sections:
[0113] Second combination:
[0114] Multiple sets of comparative data are formed by testing low-density, medium-density and high-density samples using a portable density meter and a target high-precision detector;
[0115] The intermediate density here mainly refers to the value close to the density of water, which mainly emphasizes the continuity of the data and has no actual distinction.
[0116] S5.1, modeling by the first combination method:
[0117] For a set of comparative data formed by high-density samples, a linear model 1-1 was established;
[0118] For a set of comparative data formed by low-density samples, linear model 1-2 was established;
[0119] S5.2, modeling by the second combination method:
[0120] Two instruments were used to detect multiple groups of test values of low, medium and high density flavors and fragrances, and a linear model 2 was established;
[0121] S6, verify the reliability and confidence of each model.
[0122] When this prediction method is applied, the corresponding model is first selected according to the prediction range and prediction accuracy of each prediction model, and then the detection value of a certain flavor and fragrance sample when detected by another target high-precision detector can be quickly and accurately obtained through the detection value of the portable density meter.
[0123] Example 3
[0124] Verify the effects of 75% alcohol, 95% alcohol, and 100% alcohol on the test values. The results are shown in Table 1:
[0125] Table 1
[0126] Alcohol concentration 75% 95% 100% Alcohol concentration 75% 95% 100% Sample 1 1.1545 1.1535 1.1533 Sample 6 1.1545 1.1535 1.1533 Sample 2 1.1239 1.1230 1.1230 Sample 7 1.1239 1.1230 1.1230 Sample 3 0.9922 0.9911 0.9911 Sample 8 0.9922 0.9911 0.9911 Sample 4 0.9502 0.9491 0.9491 Sample 9 0.9502 0.9491 0.9491 Sample 5 1.1545 1.1535 1.1535 Sample 10 1.1215 1.1225 1.1225
[0127] It can be seen that alcohol concentration has a significant impact on the test value, with the value changing significantly from 75% to 95%, and basically no change from 95% to 100%.
[0128] Example 4
[0129] We collected 70 samples of low, medium and high density flavors and fragrances respectively, and compared the test data of portable density meter (B) and density refractometer (Y). The test data are shown in Table 2:
[0130] Table 2
[0131] B Y B Y B Y B Y B Y 0.9016 0.9004 0.9454 0.9456 0.9016 0.9004 1.0756 1.0770 1.0021 0.9989 0.9194 0.9174 0.9454 0.9456 0.9824 0.9822 1.1020 1.1071 1.2791 1.2847 0.9029 0.9015 0.9000 0.8988 0.9249 0.9221 1.1234 1.1251 1.0644 1.0671 0.9249 0.9221 0.9177 0.9158 0.9029 0.9015 1.0437 1.0443 1.0434 1.0461 0.9841 0.9839 0.9013 0.8999 0.9045 0.9029 1.0532 1.0539 1.0756 1.0770 0.9016 0.9004 0.9233 0.9205 0.8093 0.8073 1.0394 1.0396 1.1020 1.1071 0.9194 0.9174 0.9054 0.9045 0.8123 0.8107 1.0276 1.0280 1.1234 1.1251 0.9029 0.9015 0.9029 0.9045 0.8540 0.8529 1.0634 1.0640 1.0437 1.0443 0.9249 0.9221 0.9824 0.9800 0.8615 0.8601 1.0241 1.0220 1.0532 1.0539 0.9841 0.9839 0.9077 0.9061 0.8992 0.8978 1.0853 1.0883 1.0533 1.0571 1.0394 1.0396 1.2024 1.2009 1.0241 1.0220 1.0906 1.0908 1.2814 1.2847 1.0276 1.0280 1.1764 1.1784 1.0872 1.0883 1.0948 1.0953 1.0664 1.0671 1.0634 1.0640 1.0069 1.0072 1.0039 0.9989 1.1732 1.1744 1.0453 1.0461 1.0017 1.0021 1.1622 1.1635 1.1556 1.1563 1.0573 1.0578 0.9454 0.9456
[0132] Example 5
[0133] Taking the density of pure water as the dividing line, a linear regression prediction model is established:
[0134] (1) A data model is established for the detection data with a density higher than that of pure water in Table 2, which is called Model 1-1;
[0135] 43 groups of flavor and fragrance sample test values with density higher than pure water were screened out, and 3 groups of abnormal data were eliminated to establish Table 3:
[0136] Table 3
[0137] B Y B Y B Y B Y 1.0756 1.0770 1.0021 0.9989 1.0276 1.0280 1.0069 1.0072 1.1020 1.1071 1.0644 1.0671 1.0634 1.0640 1.0964 1.1140 1.1234 1.1251 1.0434 1.0461 1.0241 1.0220 1.0906 1.0908 1.0437 1.0443 1.0756 1.0770 1.0872 1.0883 1.0948 1.0953 1.0532 1.0539 1.1020 1.1071 1.0039 0.9989 1.1732 1.1744 1.0394 1.0396 1.1234 1.1251 1.2814 1.2847 1.0573 1.0578 1.0276 1.0280 1.0437 1.0443 1.0664 1.0671 1.1556 1.1563 1.0634 1.0640 1.0532 1.0539 1.0453 1.0461 1.1622 1.1635 1.0241 1.0220 1.0533 1.0571 1.0017 1.0021 1.2752 1.2782 1.0853 1.0883 1.0394 1.0396 1.1764 1.1784 1.0205 1.0211
[0138] Using the data in Table 3, enter it into the mini-tab software to verify that the data conform to a normal distribution;
[0139] Reference Figure 1, then select statistical regression in the data statistics command and fit the linear graph to establish the linear regression prediction split model 1-1: Y=-0.01685+1.017B.
[0140] (2) A data model is established for the test data with a density lower than that of pure water in Table 2, which is called Model 1-2;
[0141] 30 groups of flavor and fragrance sample test values with density lower than pure water were screened out to establish Table 4:
[0142] Table 4
[0143] B Y B Y B Y 0.9016 0.9004 0.9454 0.9456 0.9016 0.9004 0.9194 0.9174 0.9454 0.9456 0.9824 0.9822 0.9029 0.9015 0.9000 0.8988 0.9249 0.9221 0.9249 0.9221 0.9177 0.9158 0.9029 0.9015 0.9841 0.9839 0.9013 0.8999 0.9045 0.9029 0.9016 0.9004 0.9233 0.9205 0.8093 0.8073 0.9194 0.9174 0.9054 0.9045 0.8123 0.8107 0.9029 0.9015 0.9029 0.9045 0.8540 0.8529 0.9249 0.9221 0.9824 0.9800 0.8615 0.8601 0.9841 0.9839 0.9077 0.9061 0.8992 0.8978
[0144] Using the data in Table 4, enter it into the mini-tab software to verify that the data conform to a normal distribution;
[0145] Reference Figure 2 , then select statistical regression in the data statistics command and fit the linear graph to establish the linear regression prediction split model 1-2: Y = -0.005638 + 1.005B.
[0146] Example 6
[0147] Based on the low / medium / high density flavor and fragrance detection data in Table 2, a linear regression prediction overall model is established:
[0148] After eliminating abnormal data, enter them into mini-tab software to verify that the data conform to normal distribution;
[0149] Reference Figure 3 , then select statistical regression in the data statistics command and fit the linear graph to establish the linear regression prediction overall model 2: Y = -0.01335 + 1.013B.
[0150] Example 7
[0151] Five flavor and fragrance samples with a standard density smaller than that of pure water and five flavor and fragrance samples with a standard density larger than that of pure water were selected to verify the reliability of the prediction models of Examples 4 and 5.
[0152] Table 5
[0153] Sample No. B Y Model prediction value deviation 1 0.9924 0.9922 0.9925 0.0003 2 0.9029 0.9015 0.9025 0.0011 3 0.8615 0.8600 0.8601 0.0001 4 0.8992 0.8978 0.8980 0.0002 5 0.9116 0.9105 0.9101 0.0004 6 1.1017 1.1021 1.1023 0.0002 7 1.1562 1.1554 1.1557 0.0003 8 1.0205 1.0215 1.0219 0.0004 9 1.0766 1.0770 1.0773 0.0003 10 1.1234 1.1254 1.1249 0.0004
[0154] Comparing the detection value of the density refractometer with the predicted value obtained by the prediction split model, the deviation between the two data is shown in Table 5. It is found that the deviation of each group of data is within 0.005, indicating that the prediction split model is effective and has high prediction accuracy.
[0155] The same method was used to verify the reliability of the prediction model of Example 6.
[0156] The results show that the prediction accuracy of prediction model 1-1 and prediction model 1-2 in Examples 4 and 5 is better than that of prediction model 2. In the actual application process, the corresponding linear regression prediction split model is selected according to the density of the flavor and fragrance.
[0157] Example 8
[0158] A system for rapid and accurate testing of flavors and fragrances using a portable density meter, including:
[0159] Sample density detection unit: It is used to extract N low / medium / high density flavor and fragrance samples respectively, and respectively detect the sample density by a portable densitometer and a density refractometer to obtain N groups of sample density detection data;
[0160] Data analysis and processing unit: It is used to analyze the density test data of N groups of samples, study the deviations and their laws and trends between the density test data of each group of portable densitometers and density refractometers; it is also used to eliminate abnormal data in the density test data of N groups of samples; it is also used to group the sample density test data again:
[0161] The test data of samples with density greater than pure water are grouped together;
[0162] The test data of samples with density less than pure water are grouped together;
[0163] Prediction model building unit: It is used to build models according to different data combinations:
[0164] 1) Combination method 1
[0165] The test data processed by the data analysis and processing unit are grouped and entered into the mini-tab software to verify whether the data conforms to the normal distribution. Then, statistical regression and linear graph fitting are selected in the data statistics command to establish a linear regression prediction model.
[0166] For samples with a density greater than that of pure water, the linear regression prediction split model established is model 1-1;
[0167] For samples with a density lower than that of pure water, the linear regression prediction split model established is Model 1-2;
[0168] 2) Combination method 2
[0169] Enter the test data processed by the data analysis and processing unit into the mini-tab software to verify whether the data conforms to the normal distribution; then select statistical regression in the data statistics command and fit the linear graph to establish a linear regression prediction overall model; the established linear regression prediction overall model is model 2;
[0170] Prediction model verification unit: It is used to verify the prediction accuracy of model 1-1, model 1-2, and model 2.
[0171] It should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for quickly and accurately detecting flavors and fragrances by a portable density meter, characterized in that: include S1, Study on the influence of cleaning agent concentration on the detection value of portable density meter By increasing the concentration of cleaning agent I: alcohol from 75% to 100%, the effect of the concentration of cleaning agent I on the detection value of the portable density meter was studied: Determine the optimal concentration of cleaning agent I: alcohol; Cleaning agent II: pure water; S2, Study on the influence of cleaning agent sequence on the detection value of portable density meter Select standard samples, use cleaning agents I and II in different use orders to clean the portable density meter, test the density of the standard samples after cleaning, and obtain the test values under different cleaning agent use orders; Compare and determine which cleaning agent usage sequence results in a smaller deviation between the test value of the standard sample and the theoretical value; The final determined order of use of cleaning agent I and cleaning agent II is obtained; S3, sample density detection N low / medium / high density flavor and fragrance samples are respectively sampled, and tested by a portable density meter and a density refractometer to obtain N groups of sample density test data; S4, Data analysis and processing S4.
1. Data Analysis The N groups of sample comparison data were analyzed to study the deviation between the test data of each group of portable density meters and density refractometers. The deviation was determined to have the following rules and trends: For samples with a standard density greater than that of pure water, the detection value of the portable density meter is smaller than that of the density refractometer; For samples with a standard density lower than that of pure water, the detection value of the portable density meter is greater than that of the density refractometer; Determine the correlation between the test values of portable density meter and density refractometer; S4.2 Data processing Eliminate abnormal data in N groups of sample density detection data; S4.3, Data Grouping Based on the results of S4.1 data analysis, the sample density detection data is grouped again: The test data of samples with density greater than that of pure water are grouped together; The test data of samples with density less than pure water are grouped together; S5, Construction of prediction model S5.1, Construction of the split model The test data processed by S4.2 and S4.3 are grouped to establish a linear regression prediction model; For samples with a density greater than that of pure water, the linear regression prediction split model established is model 1-1; For samples with a density lower than that of pure water, the linear regression prediction split model established is Model 1-2; S5.
2. Construction of the overall model The detection data processed by S4.2 are used to establish a linear regression prediction overall model; The overall linear regression prediction model established is model 2; S6. Validation of the prediction model Select several standard samples to verify the prediction accuracy of Model 1-1, Model 1-2, and Model 2; S7, testing the relative density of flavors and fragrances by a portable density meter; Select the corresponding linear regression prediction split model according to the density of spices and flavors; The input of the linear regression prediction split model is the detection value of the portable density meter, and the output is the predicted value of the detection value of the density refractometer.
2. The method for quickly and accurately detecting flavors and fragrances by a portable density meter according to claim 1, characterized in that: Step S1: When the alcohol concentration increases from 75% to 95%, the value detected by the portable density meter changes significantly; When the alcohol concentration increases from 95% to 100%, the value detected by the portable density meter remains basically unchanged; Determine the concentration of Cleaner I: Alcohol to be 95%.
3. The method for quickly and accurately detecting flavors and fragrances by a portable density meter according to claim 1, characterized in that: Step S2: The final order of using cleaning agent I and cleaning agent II is: alcohol first, then pure water.
4. The method for quickly and accurately detecting flavors and fragrances by a portable density meter according to claim 1, characterized in that: Step S5.1, construction of split model The test data processed by S4.2 and S4.3 were grouped and entered into the mini-tab software to verify whether the data conformed to the normal distribution; Select statistical regression and fit linear graph in the data statistics command to establish a linear regression prediction model; For samples with a density greater than that of pure water, the linear regression prediction split model established is model 1-1; For samples with density less than pure water, the established linear regression prediction split model is Model 1-2.
5. The method for quickly and accurately detecting flavors and fragrances by a portable density meter according to claim 1, characterized in that: Step S5.2, construction of the overall model Enter the test data processed by S4.2 into the mini-tab software to verify whether the data conforms to the normal distribution; Select statistical regression in the data statistics command and fit the linear graph to establish a linear regression prediction overall model; The established linear regression prediction overall model is Model 2.
6. The method for quickly and accurately detecting flavors and fragrances by a portable density meter according to claim 1, characterized in that: Step S6 includes: A portable densitometer was used to detect the density of the standard sample, and the detected values were entered into each prediction model respectively to obtain the predicted value of the standard sample when detected by a density refractometer under each prediction model. The predicted value was compared with the measured value to obtain the deviation value, and the prediction accuracy and reliability of each prediction model were verified by deviation analysis.
