Determination Method and Application of Salt and Sugar Contents in Beverages

Abstract

The present invention relates to a method for determining the salt and sugar content in beverages and its application. The determination method includes: (1) obtaining the beverage to be determined and collecting the ingredient and classification information of the beverage; (2) pre-treating the beverage to prepare a beverage sample; (3) measuring the beverage sample to obtain the approximate salinity and sugar content values of the beverage; (4) establishing a correction model for the salt and sugar content in the beverage; (5) correcting the approximate salinity and sugar content values of the beverage based on the correction model for the salt and sugar content in the beverage to obtain the accurate salinity and sugar content values of the beverage. The method for determining the salt and sugar content in beverages proposed by the present invention is convenient, has relatively high accuracy and low cost, and can achieve the quantitative determination of the salt and sugar content in beverages by ordinary people in daily life. Compared with the standard detection method in the laboratory, this method has the advantages of low cost and easy operation; compared with the existing portable detection methods, the measurement results obtained are more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of food ingredient detection, and more particularly, to a method and application for determining the salt and sugar content in beverages. Background Art

[0002] The intake of salt and sugar by the human body in daily life is closely related to health. Excessive intake will increase the risks of hypertension and cardiovascular diseases and may trigger various chronic diseases. Therefore, the "Dietary Guidelines for Chinese Residents (2016)" and the "Dietary Reference Intakes for Chinese Residents" recommend that, generally, the daily salt intake for adults should not exceed 6 grams and the sugar intake should not exceed 50 grams for good health. To promote these health standards in daily life, it is necessary to accurately quantify the intake of salt and sugar in daily diets.

[0003] Beverages have become an important part of daily diets. At the dinner table, we often come into contact with beverages without packaging or ingredient information (such as homemade beverages). To quantify the salt and sugar content in these beverages while enjoying them, a convenient, fast, and portable determination method is needed to accurately measure the salt and sugar content in beverages.

[0004] Currently, the accurate determination of the salt and sugar content in foods mainly follows the methods for determining sodium ions in the national standard GB5009.91-2017 and reducing sugars in GB5009.7-2016; among them, atomic absorption spectrometry is mainly used for sodium ions; chemical titration is mainly used for determining reducing sugars. The implementation of these determination methods requires a large number of professional reagents and laboratory equipment, is operated by professionals, and takes a long time (at least 30 minutes), which is not suitable for ordinary people to operate in daily life. And accurately measuring the salt and sugar content in beverages by ordinary people in daily life has always been a challenge.

[0005] Currently, the determination of the salt and sugar content in beverages in daily life is mainly achieved by using portable salinometers and saccharimeters; portable salinometers and saccharimeters usually use the conductivity method and the refractive index method respectively to measure the conductivity and refractive index of beverage samples, and then calculate the measured values of beverage salinity and sugar content. Among them, the refractive index method mainly utilizes the refraction phenomenon that occurs when light enters from one medium to another due to different medium densities, and the ratio of the sines of the incident angles is always a constant value, and this ratio is called the refractive index. However, there may be a large number of non-salt and non-sugar components in beverages that interfere with conductivity and refractive index and have not been excluded. This results in certain deviations in the salt and sugar content measured using these portable devices. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention proposes a method for determining the salt and sugar content in beverages that is convenient, highly accurate, and low-cost, aiming to achieve the quantitative determination of the salt and sugar content in beverages by ordinary people in their daily lives. Compared with the standard detection methods in laboratories, this method has the advantages of low cost and easy operation; compared with the existing portable detection methods, the measurement results obtained are more accurate.

[0007] To achieve the above object, the technical solution of the present invention is as follows:

[0008] The present invention first provides a method for determining the salt and sugar content in beverages, including the following steps: (1) Obtain a prepackaged beverage to be measured, and collect the ingredient information and classification information of the beverage; (2) Pretreat the prepackaged beverage to prepare a beverage sample: (3) Measure the beverage sample to obtain a rough salinity and sugar content value of the beverage; (4) Establish a correction model for the salt and sugar content in the beverage; (5) Based on the correction model for the salt and sugar content in the beverage, correct the rough salinity and sugar content values of the beverage to obtain the accurate salinity and sugar content values of the beverage. The present invention roughly measures the beverage sample by using a portable salinity and sugar content meter, combines the ingredient and classification information of the beverage, corrects the rough salinity and sugar content values of the beverage, and calculates to obtain more accurate salinity and sugar content values of the beverage. It has the advantages of low cost and convenient operation, and can meet the rapid, accurate, and convenient determination of the salt and sugar content in beverages in daily life.

[0009] In some embodiments, the collection of the ingredient and classification information of the beverage includes: (101) Extract the ingredient information of the beverage, and use the quantity of the beverage ingredients to represent the ingredient information; (102) Classify the beverage according to the latest version of the "Food Production License Classification Catalog", and use one-hot encoding to represent the classification information. The present invention combines the publicity on the beverage (prepackaged bottle / bag) and the authoritative food classification catalog, and excludes interference by collecting the ingredient and classification information of the beverage, thereby improving the accuracy of model correction.

[0010] In some embodiments, the pretreatment includes: (201) Shake the non-carbonated beverage well, and shake or let the carbonated beverage stand until no bubbles are generated; (202) Extract the beverage sample, place it in a container and let it stand until it reaches a steady state to obtain a test solution. The present invention combines the characteristics of different types of beverages, conducts targeted pretreatment on different types of beverages, improves the accuracy of salinity and sugar content measurement, and excludes the interference of the physical state of the beverage on the measurement.

[0011] In some embodiments, the measurement of the beverage sample includes: (301) Use a salinity meter and a sugar content meter to measure the pretreated beverage sample respectively, and record the measured rough salinity and sugar content values of the beverage. The salinity meter and the sugar content meter have low usage cost and convenient operation, and can be operated by ordinary people in their daily lives.

[0012] In some embodiments, the establishment of the correction model for the salt and sugar content of beverages includes: (401) preparing a training and test data set for the correction model; (402) establishing a correction model for the salt and sugar content of beverages based on the data set. By preparing a sample data set of example beverages and establishing a correction model for the salt and sugar content of beverages accordingly, the present invention ensures the credibility of the correction model.

[0013] In some embodiments, the preparation of the training and test data set for the correction model includes: (4011) selecting multiple types and varieties of prepackaged beverages; (4012) collecting the ingredient and classification information of the beverages according to step (1), preparing beverage samples according to step (2), and measuring the approximate salinity and sugar content values of the beverages according to step (3); (4013) collecting the sodium ion and carbohydrate contents per 100 mL of beverage in the nutrition label of each prepackaged beverage as the accurate values of the salinity and sugar content of each beverage; (4014) recording the beverage information in steps (4012) and (4013) and preparing a beverage sample data set; (4015) dividing the beverage sample data set into two parts: one part is used as the training sample set for the correction model of the salt and sugar content, and the other part is used as the verification data set for the model performance. By selecting multiple types and varieties of prepackaged beverages and collecting the accurate values of the salinity and sugar content of each beverage for training the correction model, the present invention ensures the accuracy of the model.

[0014] In some embodiments, the establishment of the correction model for the salt content of beverages includes: (4021) establishing a correction model for the salt content using a neural network; (4022) using the ingredient and classification information of the beverages collected in step (4012) and the measured approximate salinity and sugar content data of the beverages as the independent variables of the model, and using the sodium ion content per 100 mL of beverage collected in step (4013) as the accurate value of the salt content of the beverage; (4023) training and optimizing the parameters of the neural network model using the backpropagation method to obtain a correction model for the salt content of beverages. The results of subsequent cross-validation experiments show that the neural network model can optimally correct the measurement error of the salt content of beverages. Therefore, the present invention uses a neural network to establish a correction model for the salt content, combines the accurate values of the salinity and sugar content of each beverage collected, and pre-trains a correction model through machine learning to consider and eliminate the interfering components in the beverages in a data-driven manner, so as to obtain the most accurate correction effect of the beverage salinity value subsequently.

[0015] In some embodiments, the establishment of the correction model for the sugar content of beverages includes: (4021) establishing a correction model for the sugar content using a multiple linear model; the model function is: f(x 1 ,...,x 10 )=w 0 +w 1 x 1 +...+w 10 x 10 =w 0+wTx(1); where x 1 and x 2 are the approximate salinity and sugar content of the beverage respectively, and x 3 is the ingredient information, and x 4 to x 10 is the one-hot encoding for beverage classification, and w 0 to w 10 are the model parameters determined through training in formula (1); (4022) Use the beverage ingredients, classification information, and the measured approximate salinity and sugar content data of the beverage collected in step (4012) as the independent variables of the model, and use the carbohydrate content in every 100 mL of beverage collected in step (4013) as the accurate value of the sugar content of the beverage; (4023) Use the pseudo-inverse matrix to solve the model parameters w 0 to w 10 , and obtain the sugar content correction model for beverages. The subsequent cross-validation experiment results show that the multiple linear model can optimally correct the measurement error of the sugar content of beverages. Therefore, the present invention uses the multiple linear model to establish the sugar content correction model, combines the accurate values of the salinity and sugar content of each beverage collected, and obtains the correction model through machine learning pre-training, and considers and excludes the interfering components in the beverage in a data-driven manner, so as to obtain the most accurate correction effect of the beverage sugar content value subsequently.

[0016] In some embodiments, the correction of the approximate salinity and sugar content values of the beverage includes: (501) Use the beverage ingredients and classification information collected in step (1), and the approximate salinity and sugar content values of the beverage measured in step (3) as the independent variables of the correction model; (502) Use the independent variables determined in step (501) as the input and the accurate salt and sugar content of the beverage as the output, and through the beverage salt and sugar content correction model established in step (4), correct the approximate salinity and sugar content values measured in step (3) to accurate salinity and sugar content values. The present invention uses the collected beverage ingredients and classification information as the input of the independent variables, and corrects the salt and sugar content measurement values through the correction model; compared with directly measuring with a salinity meter and a sugar meter alone and other models, the measurement accuracy is greatly improved, and the operation time can be basically maintained unchanged, ensuring the convenience and operability of use.

[0017] The present invention also provides an application of the described determination method in determining the accurate salinity and sugar content values of prepackaged beverages. It can be widely applied to daily household measurements, laboratory tests, and tests in industrial production processes.

[0018] The beneficial effects of the present invention compared with the prior art are:

[0019] A method for measuring the salt and sugar content in beverages proposed by the present invention collects the ingredient and classification information of the beverages. First, a portable salinometer and saccharimeter are used to roughly measure the beverage samples; thereafter, in combination with the ingredient and classification information of the beverages, the rough salinity and sugar content values of the beverages are corrected, and more accurate salinity and sugar content values of the beverages are calculated.

[0020] This method has the advantages of low cost and convenience, and can be operated by ordinary people in daily life; while the accuracy of measuring the salinity and sugar content of beverage samples is greatly improved, the operation time can be basically maintained unchanged; when detecting each beverage sample, the operation time of the salinometer is about 10 seconds, the operation time of the saccharimeter is about 30 seconds, and the calculation time of the numerical correction model is less than 1 second. The total time required from measurement to obtaining accurate salinity and sugar content readings is less than 1 minute; it can meet the requirements of quickly, accurately and conveniently measuring the salt and sugar content in beverages in daily life.

[0021] It should be understood that the implementation of any embodiment of the present invention does not mean that multiple or all of the above beneficial effects need to be simultaneously achieved or reached. Description of the Drawings

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary, and for those of ordinary skill in the art, without creative efforts, other implementation drawings can be obtained according to the provided drawings.

[0023] The structures, ratios, sizes, etc. illustrated in this specification are only used to cooperate with the content disclosed in the specification for those who are familiar with this technology to understand and read, and are not used to limit the limited conditions for the implementation of the present invention. Therefore, they do not have technical substance. Any modification of the structure, change of the proportional relationship or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope covered by the technical content disclosed in the present invention.

[0024] Figure 1 It is the standardized flowchart of the measurement method of an embodiment of the present invention;

[0025] Figure 2 It is the modular schematic diagram of the measurement method of an embodiment of the present invention;

[0026] Figure 3 It is the structure diagram of the neural network model for correcting the salt content of beverages in an embodiment of the present invention; where h represents the hidden layer of the neural network, and w and b are the parameters of the neural network model. Detailed Embodiments

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer and more understandable, the embodiments of the present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings. Those skilled in the art will understand that the following described embodiments are some, but not all, of the embodiments of the present invention, and are only used to illustrate the present invention and should not be construed as limiting the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention. For those conditions not specified in the embodiments, they are carried out according to the conventional conditions. The instruments and equipment used are all conventional products that can be obtained through commercial purchase. The implementation of the present invention will be described in detail in combination with the following embodiments.

[0028] It should be understood that the terms "include / comprise", "consist of", or any other variant thereof are intended to cover non-exclusive inclusion, such that a product, device, process, or method comprising a series of elements not only includes those elements but also, when necessary, includes other elements not explicitly listed, or further includes elements inherent to such product, device, process, or method. Without further limitation, the elements defined by the statement "include / comprise..." or "consist of" do not exclude the presence of additional identical elements in the product, device, process, or method comprising the said elements.

[0029] In view of the deficiencies of the prior art, the present invention proposes a novel, low-cost, convenient, and operable method for ordinary people to measure the salt and sugar content in beverages in daily life. This method calculates more accurate beverage salinity and sugar content values by using a portable salinity and sugar meter in combination with beverage classification and ingredient information.

[0030] The implementation of the present invention will be elaborated in detail below in combination with the preferred embodiments and specific illustrations.

[0031] See Figure 1 、 Figure 2 , a method for measuring the salt and sugar content in beverages, comprising: (1) obtaining a prepackaged beverage to be measured and collecting the ingredient and classification information of the beverage; (2) preprocessing the prepackaged beverage to prepare a beverage sample; (3) measuring the beverage sample to obtain a rough salinity and sugar content value of the beverage; (4) establishing a correction model for the salt and sugar content of the beverage; (5) correcting the rough salinity and sugar content value of the beverage based on the correction model for the salt and sugar content of the beverage to obtain the accurate salinity and sugar content value of the beverage. Compared with the prior art, the method for measuring the salt and sugar content in beverages provided by the present invention corrects the rough salinity and sugar content value of the beverage (such as the readings measured by a portable salinity meter and sugar meter) by using beverage ingredient and classification information, so as to calculate a more accurate salt and sugar content value of the beverage.

[0032] In some embodiments, (1) collecting the ingredient and classification information of the beverage includes:

[0033] (101) Extracting the ingredient information of the beverage on the pre-packaging, and using the quantity of the ingredients on the pre-packaging to represent the ingredient information;

[0034] (102) Classifying the beverage according to the latest version of the "Food Production License Classification Catalog" issued by the State Administration for Market Regulation and Supervision, and using one-hot encoding to represent the classification information.

[0035] The present invention combines the publicity on the pre-packaging bottle / bag of the beverage and the authoritative food classification catalog, and provides information on the possible interfering components in the model beverage by collecting the ingredient information and classification information of the beverage, thereby helping the model to eliminate interference and improve the correction accuracy.

[0036] In some embodiments, (2) preprocessing the beverage includes:

[0037] (201) Shaking the non-carbonated beverage well, and shaking or leaving the carbonated beverage still until no bubbles are generated;

[0038] (202) Extracting a beverage sample, placing it in a container and leaving it still until it reaches a steady state to obtain a test solution.

[0039] The present invention combines the beverage classification and type characteristics, and performs targeted preprocessing on different types of beverages. For common non-carbonated beverages, they are shaken well, and carbonated beverages are shaken or left still until no bubbles are generated, improving the accuracy of salinity and sugar content measurement and eliminating the interference of the physical state of the beverage on the measurement.

[0040] In some embodiments, (3) measuring the beverage sample includes:

[0041] (301) Using a portable salinometer and a refractometer, and according to the product instruction manuals of the salinometer and the refractometer, measuring the beverage sample prepared in step (202) respectively; reading the readings; and recording the measured approximate salinity and sugar content values of the beverage sample.

[0042] The salinometer and the refractometer have low usage costs and are convenient to operate, and can be operated by ordinary people in daily life.

[0043] In some embodiments, (4) establishing a correction model for the salt and sugar content of the beverage includes:

[0044] (401) Making a correction model training and test data set;

[0045] (402) Establishing a correction model for the salt and sugar content of the beverage based on the data set.

[0046] The present invention measures the approximate salt and sugar contents in beverages using a portable salinometer and saccharimeter, creates a beverage sample dataset, and then uses a correction model obtained through machine learning pre-training to consider and exclude interfering components in the beverages in a data-driven manner, ensuring the credibility of the correction model and enabling the calculation of more accurate salt and sugar contents in the beverages.

[0047] In some embodiments, (401) creating the correction model training and test datasets includes:

[0048] (4011) Selecting multiple categories and various types of prepackaged beverages;

[0049] Select 108 types of prepackaged beverages in 5 major categories, including 22 types of carbonated beverages, 20 types of tea beverages, 26 types of fruit and vegetable beverages, 22 types of protein and dairy products, and 18 types of coffee, sports, energy, etc. beverages;

[0050] (4012) Collecting beverage ingredient and classification information according to step (1), preparing beverage samples according to step (2), and measuring the approximate salinity and sugar content values of the beverages according to step (3);

[0051] (4013) Collecting the sodium ion and carbohydrate contents per 100 mL of beverage in the nutrition facts table of each prepackaged beverage as the accurate values of the salinity and sugar content of each beverage; using the accurate values of the salinity and sugar content of each beverage collected for training the correction model to ensure the accuracy of the model;

[0052] (4014) Recording the beverage information in steps (4012) and (4013) and creating a beverage sample dataset;

[0053] (4015) Dividing the beverage sample dataset into two parts; one part is used as the training sample set for the salinity and sugar content correction model (N = 107, N is the number of beverage types); the other part is used as the verification dataset for the model performance (N = 1).

[0054] The present invention selects multiple categories and various types of prepackaged beverages, specifically 108 types of prepackaged beverages in 5 major categories, including carbonated beverages, tea beverages, fruit and vegetable beverages, protein and dairy products, and coffee, sports, energy, etc. beverages, covering comprehensively. By collecting the accurate values of the salinity and sugar content of each beverage for training, screening, and verifying the correction model, the accuracy of the model is ensured.

[0055] In some embodiments, (402) establishing the beverage salinity correction model includes:

[0056] (4021) Using a neural network to establish a salinity correction model to obtain the most accurate correction effect of the beverage salinity value. The model structure is shown in Figure 3 ;

[0057] (4022) Use the beverage ingredients, classification information, and the measured approximate salinity and sugar content data of the beverage collected in step (4012) as the independent variables of the model, and use the sodium ion content in every 100 mL of the beverage collected in step (4013) as the accurate value of the salt content of the beverage;

[0058] (4023) Use the backpropagation method to train and optimize the model parameters to obtain a beverage salt content correction model.

[0059] The present invention uses a neural network to establish a salt content correction model. Combining the accurate values of the salinity and sugar content of each beverage collected, a correction model is obtained through machine learning pre-training. The interference components in the beverage are considered and excluded in a data-driven manner, so that the most accurate correction effect of the beverage salinity value can be obtained subsequently.

[0060] In some embodiments, (402) establishing a beverage sugar content correction model includes:

[0061] (4021) Use a multiple linear model to establish a sugar content correction model to obtain the most accurate correction effect of the beverage sugar content value. The model function is:

[0062] f(x 1 ,...,x 10 )=w 0 +w 1 x 1 +...+w 10 x 10 =w 0 +wTx (1)

[0063] where x 1 , x 2 are the approximate salinity and sugar content of the beverage respectively, x 3 is the ingredient information, and x 4 to x 10 are the one-hot encodings of the beverage classification, and w 0 to w 10 are the model parameters determined through training in formula (1);

[0064] (4022) Use the beverage ingredients, classification information, and the measured approximate salinity and sugar content data of the beverage collected in step (4012) as the independent variables of the model, and use the carbohydrate content in every 100 mL of the beverage collected in step (4013) as the accurate value of the sugar content of the beverage;

[0065] (4023) Use the pseudo-inverse matrix to solve the model parameters w 0 to w 10 in formula (1) to obtain a beverage sugar content correction model.

[0066] The present invention uses a multiple linear model to establish a sugar content correction model. By combining the accurate values of the salinity and sugar content of each beverage collected, a correction model is obtained through pre-training of machine learning. The interfering components in the beverage are considered and excluded in a data-driven manner, so as to obtain the most accurate correction effect of the beverage sugar content value subsequently.

[0067] In addition, in the salinity correction model and the sugar content correction model, the rough salinity and sugar content data of the beverage measured by the salinometer and the saccharimeter, as well as the beverage ingredient and classification information, are integrated, and the model parameters are optimized in a data-driven manner to obtain the optimal salinity correction model to achieve the best salinity correction performance.

[0068] In some embodiments, (5) correcting the rough salinity and sugar content values of the beverage includes:

[0069] (501) Using the beverage ingredient and classification information collected in step (1), and the rough salinity and sugar content values of the beverage measured in step (3) as the independent variables of the correction model;

[0070] (502) Using the independent variables determined in step (501) as the input and the accurate salt and sugar content of the beverage as the output, through the beverage salt and sugar content correction model established in step (4), the rough salinity and sugar content values measured in step (3) are corrected to accurate salinity and sugar content values.

[0071] The performance of the baseline, several commonly used, and the salt and sugar content correction models involved in the present invention is compared, and the comparison results and the average calculation time for each reading correction are shown in Table 1 and Table 2. The mean absolute error is the average of the absolute values of the differences between the measured values and the true values of various beverages in the dataset.

[0072] Table 1 Cross-validation results of the baseline, several commonly used models, and the salinity correction model of the present invention

[0073]

[0074] * represents the model proposed by the present invention

[0075] Table 2 Cross-validation results of the baseline, several commonly used models, and the sugar content correction model of the present invention

[0076]

[0077] * represents the model proposed by the present invention

[0078] In the table, except for the model proposed in the present invention and the use of a salinometer and a saccharimeter alone, the other several models are relatively commonly used and recognized as having good performance. As can be seen from Table 1 and Table 2, on a dataset consisting of 108 different beverages in 5 categories, compared with the baseline methods of separately using a salinometer and a saccharimeter to measure the salinity (mean absolute error of 13.55 mg) and sugar content (mean absolute error of 1.54 g) of beverage samples, the salt (mean absolute error of 7.94 mg) and sugar (mean absolute error of 1.24 g) measurement accuracies of the model proposed in the present invention obtained by the relatively scientific leave-one-out cross-validation in machine learning are respectively improved by 70.7% and 24.2%; at the same time, because the reading correction time is less than 1 second, the operation time (measurement time plus measurement reading correction time) can be maintained basically unchanged, ensuring the convenience of using this method; therefore, the model proposed in the present invention exhibits good measurement accuracy and operability.

[0079] Table 3 Cross-validation results of the baseline, several common models, and the salt and sugar content correction models of the present invention

[0080]

[0081] Note: A proportion greater than 0% indicates performance improvement, and vice versa indicates performance degradation.

[0082] As can be seen from Table 3, the method proposed in the present invention has a greater improvement in accuracy compared with the baseline method when measuring the salt and sugar contents of carbonated, fruit and vegetable juice, and protein-containing milk beverages.

[0083] Therefore, a method for measuring the salt and sugar contents in beverages provided by the present invention can meet the requirements of quickly, accurately, and conveniently measuring the salt and sugar contents in some beverages in daily life.

[0084] Example:

[0085] The present invention also demonstrates an application of the described measurement method in determining the precise salt and sugar content values of pre-packaged fruit juice beverages.

[0086] A total of 26 pre-packaged fruit juice beverages such as different orange juices, apple juices, and grape juices were collected; after shaking the (non-carbonated) fruit juice beverages and allowing them to stand for 10 seconds, 50 ml was taken and placed in a container as the sample to be measured; according to the product user manuals of the portable salinometer and saccharimeter, the equipment was operated to respectively read the approximate salt and sugar content values of the beverage samples, and the readings were recorded; at the same time, the number of ingredients and classification categories of the beverages were recorded; the pre-processed data was respectively imported into the pre-trained salt and sugar content correction models, and the more accurate measured values of the salt and sugar contents of the beverages were calculated through computer programs. The process is as Figure 1 shown.

[0087] The performance of the model was compared and verified using the above operation process and the following several different models and parameter combinations. Tables 4 and 5 show the verification results and the average calculation time for each reading correction.

[0088] Table 4 Cross-validation results of the baseline, several commonly used models, and the salt content correction model of the present invention

[0089]

[0090] * represents the model proposed by the present invention

[0091] Table 5 Cross-validation results of the baseline, several commonly used models, and the sugar content correction model of the present invention

[0092]

[0093] * represents the model proposed by the present invention

[0094] As can be seen from Tables 4 and 5, on a dataset consisting of 26 different fruit juice beverages, compared with the baseline method of separately measuring the salt (average absolute error of 23.82 mg) and sugar content (average absolute error of 1.42 g) of beverage samples using a salinometer and a saccharimeter respectively, the measurement accuracies of the salt (average absolute error of 10.77 mg) and sugar (average absolute error of 0.84 g) in beverages of the model proposed by the present invention are improved by 121.2% and 69.0% respectively, and the operation time remains basically unchanged.

[0095] Using the determination method provided by the present invention in the determination of the precise salt and sugar content values of prepackaged beverages can be widely applied to household daily measurements, laboratory tests, and detections in industrial production processes.

[0096] It is easy for those skilled in the art to understand that, on the premise of no conflict, the above preferred solutions can be freely combined and superimposed.

[0097] At this point, those skilled in the art should recognize that although the exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications that conform to the principles of the present invention can still be directly determined or derived from the content disclosed in the present invention without departing from the spirit and scope of the present invention. Therefore, the scope of the present invention should be understood and determined to cover all these other variations or modifications.

Claims

1. A method for determining the salt and sugar content in a beverage, characterized in that, it comprises the following steps: (1) Obtain the beverage to be measured, and collect the ingredient and classification information of the beverage; (2) Pretreat the beverage to prepare a beverage sample; (3) Measure the beverage sample to obtain the approximate salinity and sugar content values of the beverage; (4) Establish a correction model for the salt and sugar content in the beverage; (5) Based on the correction model for the salt and sugar content in the beverage, correct the approximate salinity and sugar content values of the beverage to obtain the accurate salinity and sugar content values of the beverage; where establishing the correction model for the salt and sugar content in the beverage includes: (401) Prepare a training and testing data set for the correction model; (402) Based on the data set, establish a correction model for the salt and sugar content in the beverage; Preparing a training and testing data set for the correction model includes: (4011) Select multiple types and varieties of prepackaged beverages; (4012) According to step (1), collect the ingredient and classification information of the beverage, according to step (2), prepare a beverage sample, and according to step (3), obtain the approximate salinity and sugar content values of the beverage; (4013) Collect the sodium ion and carbohydrate content per 100 mL of beverage in the nutrition label of each prepackaged beverage as the accurate values of the salinity and sugar content of each beverage; (4014) Record the beverage information in steps (4012) and (4013), and prepare a beverage sample data set; (4015) Divide the beverage sample data set into two parts: one part is used as the training data set for the salt and sugar content correction model, and the other part is used as the test data set for the model performance; Establishing a correction model for the salt content in the beverage includes: (4021) Use a neural network to establish a correction model for the salt content; (4022) Use the ingredient and classification information of the beverage collected in step (4012) and the approximate salinity and sugar content data of the beverage as the independent variables of the model, and use the sodium ion content per 100 mL of beverage collected in step (4013) as the accurate value of the salt content of the beverage; (4023) Use the backpropagation method to train and optimize the neural network model parameters to obtain a correction model for the salt content in the beverage; Establishing a correction model for the sugar content in the beverage includes: (4031) Use a multiple linear model to establish a correction model for the sugar content; the model function is: （1） Among them x 1 and x 2 are the approximate salinity and sugar content of the beverage, respectively, x 3 is the ingredient information, x 4 to x 10 is the one-hot encoding for beverage classification, w 0 to w 10 are the model parameters determined by training in formula (1); (4032) Use the ingredient and classification information of the beverage collected in step (4012) and the approximate salinity and sugar content data of the beverage as the independent variables of the model, and use the carbohydrate content per 100 mL of beverage collected in step (4013) as the accurate value of the sugar content of the beverage; (4033) Solve the model parameters in formula (1) using the pseudo-inverse matrix w 0 to w 10 , and obtain the corrected model for the sugar content of the beverage 2. The determination method according to claim 1, characterized in that, the collection of the ingredient and classification information of the beverage includes: (101) Extract the ingredient information of the beverage, and use the quantity of the beverage ingredients to represent the ingredient information; (102) According to the latest version of the "Food Production License Classification Catalog", classify the beverage, and use one-hot encoding to represent the classification information.

3. The determination method according to claim 1, characterized in that, the pretreatment includes: (201) Shake the non-carbonated beverage well, and shake or let the carbonated beverage stand until no bubbles are produced; (202) Extract the beverage sample, place it in a container and let it stand until it reaches a steady state to obtain a test solution.

4. The measurement method according to claim 1, characterized in that, the measurement of the beverage sample includes: (301) Measuring the pretreated beverage sample using a salinometer and a saccharimeter respectively, and recording the measured rough salinity and sugar content values of the beverage.

5. The measurement method according to claim 1, characterized in that, the correction of the rough salinity and sugar content values of the beverage includes: (501) Using the beverage ingredients and classification information collected in step (1), and the rough salinity and sugar content values of the beverage measured in step (3) as independent variables of the correction model; (502) Using the independent variables determined in step (501) as input and the accurate salt and sugar content of the beverage as output, through the beverage salt and sugar content correction model established in step (4), correcting the rough salinity and sugar content values measured in step (3) to accurate salinity and sugar content values.

6. An application of the measurement method according to any one of claims 1 to 5 in measuring the accurate salinity and sugar content values of prepackaged beverages.

Images