A composite juice freshness real-time monitoring method

Abstract

The application provides a composite juice freshness real-time monitoring method, comprising: for a single kind of juice, performing physical and chemical characteristic analysis to obtain accurate juice characteristic parameters, including acidity, sugar content, nutritional ingredients and microbial activity; associating the taste data of the juice with the shelf life of the juice to determine the correlation between different taste attributes and the shelf life; obtaining the taste prediction result of the composite juice through a composite juice taste prediction model, and obtaining the shelf life prediction result of the composite juice according to a juice shelf life prediction model; comparing the prediction result with the set shelf life standard and taste standard, and using chi-square test and T test statistical analysis methods to determine whether the composite juice meets the standard; if the shelf life of the composite juice does not meet the standard, or the taste does not meet the expectation, the formula of the composite juice is changed, the characteristic analysis and taste evaluation are performed again, and the prediction step is entered again until the prediction result meeting the standard is obtained.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for real-time monitoring of the freshness of compound fruit juice. Background Technology

[0002] With the improvement of people's living standards and the enhancement of health awareness, compound fruit juices, as beverages rich in various nutrients and with a diverse taste, have become increasingly popular among consumers. However, due to the presence of multiple different fruit juice components, assessing and controlling their shelf life has become a challenge. Traditional shelf-life assessment methods mainly rely on experience and sensory judgment, which introduces certain uncertainties and subjectivity into the prediction of shelf life and taste evaluation of compound fruit juices. Moreover, the different components in compound fruit juices may interact during storage, leading to changes in shelf life. Therefore, a scientific and accurate method is needed to assess the shelf life and taste of compound fruit juices to ensure their quality and safety. Summary of the Invention

[0003] This invention provides a method for real-time monitoring of the freshness of compound fruit juice, mainly including:

[0004] For a single type of fruit juice, physical and chemical characteristic analysis is performed to obtain precise juice characteristic parameters, including acidity, sugar content, nutritional components, and microbial activity. Enzyme-based sensors are used to precisely calibrate the acidity and sugar content chemical characteristics of the juice. Simultaneously, electronic tongue sensors and high-frequency oscillation technology are applied to measure the sweetness, acidity, and aroma of the juice, forming a detailed taste dataset. The taste data of the juice is correlated with its shelf life to determine the relationship between different taste attributes and shelf life. Based on the type of juice, the taste data, and the mixing ratio of different types of juice, a multilayer perceptron is used to train a model to obtain complex fruit juice flavors. The system employs a predictive model, utilizing both linear regression and fruit juice flavor data to construct a fruit juice shelf-life prediction model. A compound fruit juice flavor prediction model is used to obtain the flavor prediction results, and a fruit juice shelf-life prediction model is used to obtain the shelf-life prediction results. The predicted results are compared with the established shelf-life and flavor standards, and statistical analysis methods such as chi-square and t-tests are used to determine whether the compound fruit juice meets the standards. If the shelf-life of the compound fruit juice does not meet the standards, or the flavor does not meet expectations, the formula of the compound fruit juice is changed, and the characteristic analysis and flavor evaluation are repeated, and the prediction process is restarted until a prediction result that meets the standards is obtained.

[0005] In one embodiment, the physical and chemical characterization analysis of a single type of fruit juice to obtain precise fruit juice characteristic parameters, including acidity, sugar content, nutritional components, and microbial activity, includes:

[0006] Collect and label samples of each type of juice; use acid-base titration to mix the juice sample with an acid-base indicator, and calculate the acidity value of the juice sample by measuring the volume of the acidity index point; after filtering and diluting the juice sample, inject it into a chromatograph, use a chromatographic column to separate the sugars in the sample, and then use a detector to detect the area of ​​the sugar peaks to calculate the sugar content of the juice sample; use a nutrient analysis instrument to hydrolyze the sample and determine the vitamin, mineral, and amino acid components in the sample; dilute the sample, inoculate it on a nutrient-containing culture medium, and incubate it in petri dishes or culture flasks for a preset time, calculate the number of microbial colonies, and determine the microbial activity of each sample.

[0007] In one embodiment, an enzyme-based sensor is used to precisely calibrate the chemical characteristics of the fruit juice, including its acidity and sugar content. Simultaneously, an electronic tongue sensor and high-frequency oscillation technology are applied to measure the sweetness, acidity, and aroma of the fruit juice, forming a detailed taste dataset, including:

[0008] An enzyme-based sensor was designed based on the acidity and sugar content of the fruit juice. Standard acidity and sugar content solutions of different concentrations were reacted with the enzyme-based sensor to obtain the electrical signals of the sensor response. By comparing with a standard curve, the sensor signals were converted into corresponding acidity and sugar content concentrations for precise calibration. An electronic tongue sensor was used to measure the sweetness and acidity of the fruit juice sample and obtain the corresponding sensor response data. The fruit juice sample was subjected to high-frequency oscillation to release aroma components into the gas phase. Gas chromatography was used to separate and quantitatively analyze the aroma components, obtaining detailed aroma data. The data obtained from the enzyme-based sensor, electronic tongue, and high-frequency oscillation technology were integrated and correlated to establish a taste dataset.

[0009] In one implementation, the step of associating the flavor data of the juice with its shelf life, and determining the correlation between different flavor attributes and shelf life, includes:

[0010] This process involves acquiring taste data from different types of fruit juice samples, including sweetness, acidity, and aroma, while recording the shelf life information for each sample. The acquired taste data is normalized and cleaned to remove missing and outlier values. Based on the shelf life information, the fruit juice samples are divided into different shelf-life groups: long-term, medium-term, and short-term. Analysis of variance (ANOVA) is used to compare the taste attributes of different shelf-life groups to determine if there are significant differences between the attributes. Based on the statistical analysis results, the differences in taste attributes among different shelf-life groups are assessed to identify the taste attributes that affect shelf life. Correlation coefficients are calculated to determine the relationship between different taste attributes and shelf life. The process also includes: using ANOVA to compare the taste attributes of different shelf-life groups to identify the taste attributes that affect shelf life; and calculating correlation coefficients to determine the relationship between different taste attributes and shelf life.

[0011] The method of using analysis of variance to compare the taste attributes of different shelf-life groups and determine the taste attributes that affect the shelf-life specifically includes:

[0012] Taste data, including sweetness, acidity, and aroma, were obtained from fruit juice samples of different shelf-life groups. Analysis of variance and t-tests were used to compare the differences in taste attributes between different shelf-life groups, determining the significance level of these differences. Based on the statistical analysis results, it was determined whether significant differences existed between the different shelf-life groups. If the p-value was less than the preset significance level, a significant difference was considered to exist between the different shelf-life groups; if the p-value was greater than the preset significance level, it was impossible to determine whether a significant difference existed between the different groups. If a significant difference existed, the mean differences of each taste attribute were compared between the different groups to identify the attributes with significant differences. Based on the mean differences and dispersion between the different shelf-life groups, the differences in taste attributes between the different shelf-life groups were determined, and taste attributes that might affect shelf life were identified.

[0013] The method of determining the correlation between different flavor attributes and shelf life by calculating correlation coefficients specifically includes:

[0014] Taste data, including sweetness, acidity, and aroma, along with corresponding shelf-life data, were obtained from different fruit juice samples. The collected taste and shelf-life data were cleaned, denoised, and normalized to remove outliers and noise interference. Based on the taste and shelf-life data, the taste characteristics that significantly affect shelf-life were identified. Feature engineering methods were used to extract features, combining sweetness, acidity, and aroma into a comprehensive taste index. Correlation coefficients were calculated to determine the relationship between different taste characteristics and shelf-life. Based on the correlation between different taste characteristics and shelf-life, the degree of influence of the taste characteristics on shelf-life was determined.

[0015] In one implementation, a multilayer perceptron is used to train a model based on the types of fruit juice, flavor data, and mixing ratios of different types of fruit juice to obtain a composite fruit juice flavor prediction model. Simultaneously, a linear regression model is used to construct a fruit juice shelf-life prediction model based on the fruit juice flavor data, including:

[0016] This process involves acquiring flavor data for different types of fruit juice, their mixing ratios, and the flavor data of compound fruit juices, and preprocessing these data, including outlier removal and normalization. Based on a multilayer perceptron, a compound fruit juice flavor prediction model is constructed using the flavor data for different types of fruit juice and the flavor data for compound fruit juices. The prediction model is then used to obtain the predicted flavors of the compound fruit juices. Fruit juice flavor data and shelf-life data are also acquired. Based on a linear regression model, a fruit juice shelf-life prediction model is constructed using the flavor data and shelf-life data. Finally, a compound fruit juice flavor prediction model is constructed by training a multilayer perceptron model based on the types of fruit juice, their flavor data, and the mixing ratios of different types of fruit juice.

[0017] The process involves constructing a composite fruit juice flavor prediction model by training a multilayer perceptron model based on the types of fruit juice, flavor data, and mixing ratios of different types of fruit juice. Specifically, this includes:

[0018] Taste data from different types of fruit juices, including sweetness, acidity, and aroma, are collected to determine the composition of compound fruit juices, including their types and mixing ratios. This data is then compiled into training samples, where each sample contains the type of fruit juice, its taste attributes, and the mixing ratio. A compound fruit juice taste prediction model is constructed by training a multilayer perceptron (MLP) model. The number of neurons in the input, hidden, and output layers of the MLP is determined based on the number of taste attributes in the training samples. The model is trained using the training samples, and its weights and biases are continuously updated through backpropagation to gradually improve its accuracy in predicting the taste of compound fruit juices. Using the trained compound fruit juice taste prediction model, new mixing ratios and types of fruit juices are input to obtain the predicted taste of the compound fruit juice.

[0019] In one embodiment, obtaining the flavor prediction result of the compound fruit juice using a compound fruit juice flavor prediction model, and obtaining the shelf life prediction result of the compound fruit juice using a fruit juice shelf life prediction model, includes:

[0020] Using an electronic tongue sensor and high-frequency oscillation technology, the sweetness, acidity, and aroma of the mixed fruit juice are measured to obtain the taste data of the fruit juice to be mixed. Based on the taste data of the fruit juice to be mixed, a mixed fruit juice taste prediction model is used to obtain the taste prediction result of the mixed fruit juice. Based on the predicted taste of the mixed fruit juice, a fruit juice shelf life prediction model is used to obtain the shelf life prediction result of the mixed fruit juice.

[0021] In one implementation, the step of comparing the predicted results with set shelf-life and taste standards, and using statistical analysis methods such as chi-square test and t-test to determine whether the compound fruit juice meets the standards, includes:

[0022] Obtain shelf-life and taste data for the compound fruit juice; calculate a comprehensive taste score by weighted summation based on the sweetness, acidity, and aroma data of each batch; divide the shelf-life into two groups: compliant and non-compliant, and the taste score into two groups: acceptable and unacceptable; use a chi-square test to determine whether there is a significant relationship between the predicted results and the standards for the shelf-life and taste of the compound fruit juice; if the p-value of the chi-square test is less than the set significance level, the predicted results are considered to have a significant relationship with the shelf-life and taste, otherwise, no significant relationship is considered; use a t-test to determine whether the difference between the predicted results and each taste attribute meets the standards for different taste attributes of the compound fruit juice; group each taste attribute of the compound fruit juice, including aroma, sweetness, and acidity, compare the predicted results with each attribute, and use a t-test to determine whether there is a significant difference; if the p-value of the t-test is less than the set significance level, the compound fruit juice meets the standards for that taste attribute, otherwise, it does not meet the standards.

[0023] In one implementation, if the shelf life of the compound fruit juice fails to meet the standard, or the taste does not meet expectations, the formula of the compound fruit juice is changed, the characteristic analysis and taste evaluation are repeated, and the prediction step is entered again until a prediction result that meets the standard is obtained, including:

[0024] The process involves: acquiring shelf-life and taste data for the compound fruit juices to identify which samples failed to meet shelf-life or taste targets; obtaining taste data for the substandard samples, including acidity, sweetness, and aroma; determining whether to adjust acidity or sweetness to achieve the desired shelf-life or taste based on the obtained taste data; gradually increasing or decreasing acidity or sweetness in the substandard compound fruit juices according to the adjustment direction; preparing adjusted compound fruit juice samples based on the adjusted acidity or sweetness; retesting the acidity, sweetness, and shelf-life indicators on the prepared adjusted samples to obtain the adjusted fruit juice taste data; and comparing the component data before and after adjustment with the shelf-life and taste targets to determine if the adjustment was successful. And whether the expected goals have been achieved; if the shelf life and taste cannot be met simultaneously, a juice formula adjustment model is established based on a linear programming algorithm to obtain the final formula combination, so that the compound juice can obtain the best taste while meeting the shelf life requirements; the chi-square test and T test are used to determine whether the adjustment significantly affects the shelf life and taste targets; based on the results of statistical analysis, the optimal adjustment plan is determined to maintain the taste target while meeting the shelf life requirements, or to obtain the best taste of the compound juice while meeting the shelf life requirements; it also includes: establishing a juice formula adjustment model based on a linear programming algorithm to obtain the final formula combination, so that the compound juice can obtain the best taste while meeting the shelf life requirements.

[0025] The method involves establishing a juice formula adjustment model based on linear programming to obtain the final formula combination, ensuring that the compound juice achieves optimal taste while meeting shelf-life requirements. Specifically, this includes:

[0026] Based on the taste prediction model and shelf-life prediction model, a juice formula adjustment model is established. The objective function of the model is the taste prediction result, and the constraint is the shelf-life requirement. The objective function and constraints are transformed into mathematical expressions, and a linear programming algorithm is used to solve for the optimal formula combination. Based on the established juice formula adjustment model, possible juice formula combinations are formulated, including the proportion and adjustment range of different juice types. For each formula combination, the predicted taste and shelf-life values ​​are calculated according to the taste prediction model and the shelf-life prediction model. The merits of each formula combination are evaluated according to the objective function and constraints, and the combination with the best taste quality is selected. A genetic algorithm is used, with the predicted taste value as the objective and the predicted shelf-life value as the constraint, for iterative search. With each iteration, a new juice formula combination is generated according to the algorithm strategy, and its predicted taste and shelf-life values ​​are calculated. Based on the obtained formula combinations, new compound juice samples are prepared, and the taste and shelf-life of the prepared new juice samples are evaluated. The actual taste data and shelf-life data are compared with the previous prediction results. The deviation of the taste data and the degree of conformity with the shelf-life data are compared to determine whether the adjusted formula has achieved the expected effect. If the adjusted formula does not achieve the expected results, update the parameters in the formula adjustment model based on the actual test results, and iterate and optimize until a more accurate formula combination is obtained, so that the compound juice can achieve the best taste while meeting the shelf life requirements, and obtain the final formula combination.

[0027] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0028] This invention discloses a method for real-time monitoring of the freshness of compound fruit juice. First, for a single type of fruit juice, physical and chemical characteristics are analyzed to obtain precise characteristic parameters such as acidity, sugar content, nutritional components, and microbial activity. Next, an enzyme-based sensor is used to precisely calibrate the acidity and sugar content of the fruit juice. Simultaneously, an electronic tongue sensor and high-frequency oscillation technology are applied to measure the sweetness, acidity, and aroma of the fruit juice, forming a detailed taste dataset. Using this taste data, the impact of the sweetness, acidity, and aroma of the fruit juice on its shelf life and taste can be determined. Based on the types of fruit juice, taste data, and the mixing ratio of different types of fruit juice, a multilayer perceptron model is used to train a model to obtain a compound fruit juice taste prediction model. Simultaneously, a linear regression model is used to construct a fruit juice shelf life prediction model based on the fruit juice taste data. The taste prediction results of the compound fruit juice can be obtained through the compound fruit juice taste prediction model. Based on the fruit juice shelf life prediction model, the predicted shelf life of the compound fruit juice can be calculated and obtained. By comparing the predicted results with the set shelf life and taste standards, statistical analysis methods such as chi-square test and t-test are used to determine whether the compound fruit juice meets the standards. If the shelf life of the compound fruit juice does not meet the standards or the taste does not meet expectations, the formula of the compound fruit juice can be changed, the characteristic analysis and taste evaluation can be repeated, and the prediction step can be repeated until a prediction result that meets the standards is obtained. In summary, this invention integrates liquid chromatography technology, enzyme-based sensors, electronic tongue sensors, and high-frequency oscillation technology, and achieves accurate assessment and prediction of the freshness of compound fruit juice through characteristic analysis and taste evaluation methods. This method can ensure that the quality of the compound fruit juice meets the standards, improving the efficiency and quality of juice production. Attached Figure Description

[0029] Figure 1 This is a flowchart of a method for real-time monitoring of the freshness of compound fruit juice according to the present invention.

[0030] Figure 2 This is a schematic diagram of a method for real-time monitoring of the freshness of compound fruit juice according to the present invention.

[0031] Figure 3 This is another schematic diagram of a method for real-time monitoring of the freshness of compound fruit juice according to the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0033] This embodiment of a method for real-time monitoring of the freshness of compound fruit juice may specifically include:

[0034] S101. For a single type of fruit juice, perform physical and chemical characteristic analysis to obtain precise fruit juice characteristic parameters, including acidity, sugar content, nutritional components, and microbial activity.

[0035] For example, collect and label 10 apple juice samples. Using acid-base titration, mix each 10 mL apple juice sample with an acid-base indicator. Calculate the acidity value of the apple juice sample as 15 mol / L NaOH solution by measuring the volume of the acidity index point. After pretreatment by filtration and dilution, inject the apple juice sample into a chromatograph. Separate the sugars from the sample using a chromatographic column, and then detect the peak area of ​​the sugars using a detector. Assume the glucose peak area in the apple juice sample is 2000. Hydrolyze the apple juice sample using a nutrient analysis instrument to determine the vitamin C content. Assume the vitamin C content is 50 mg / 100 mL. Dilute the apple juice sample, inoculate it onto a nutrient-containing culture medium, and incubate it in petri dishes for a preset time of 48 hours. Calculate the number of microbial colonies in the apple juice sample as 500 CFU / mL. For example, collect and label 10 apple juice samples. Using acid-base titration, each 10 mL apple juice sample was mixed with an acid-base indicator. The acidity was calculated to be 15 mol / L by measuring the volume of the acidity index point and assuming an average consumption of 15 mL of 1 mol / L NaOH solution. After pretreatment by filtration and dilution, the apple juice sample was injected into a chromatograph. Sugars were separated using a chromatographic column, and the peak area of ​​the sugars was detected. The glucose peak area was assumed to be 2000. The apple juice sample was hydrolyzed using a nutrient analysis instrument to determine the vitamin C content, which was assumed to be 50 mg / 100 mL. The apple juice sample was diluted and inoculated onto a nutrient-rich culture medium. It was incubated in petri dishes for 48 hours, and the number of microbial colonies in the apple juice sample was calculated to be 500 CFU / mL.

[0036] S102. An enzyme-based sensor is used to precisely calibrate the chemical characteristics of fruit juice, including acidity and sugar content. Simultaneously, an electronic tongue sensor and high-frequency oscillation technology are applied to measure the sweetness, acidity, and aroma of the fruit juice, generating a detailed taste dataset.

[0037] An enzyme-based sensor was designed based on the acidity and sugar content of fruit juice. Standard acidity and sugar content solutions of different concentrations were reacted with the enzyme-based sensor to obtain the electrical signals of the sensor response. By comparing with a standard curve, the sensor signal was converted into corresponding acidity and sugar content concentrations for accurate calibration. An electronic tongue sensor was used to measure the sweetness and acidity of the fruit juice sample, obtaining the corresponding sensor response data. The fruit juice sample was subjected to high-frequency oscillation to release aroma components into the gas phase. Gas chromatography was used to separate and quantitatively analyze the aroma components, obtaining detailed aroma data. The data obtained from the enzyme-based sensor, electronic tongue, and high-frequency oscillation technology were integrated and correlated to establish a taste dataset. For example, assuming the designed enzyme-based sensor can accurately measure the acidity and sugar content of fruit juice, standard acidity and sugar content solutions were reacted to obtain the sensor's electrical signals. Assuming the concentration of the standard acidity solution is 1 mol / L, the sensor's response electrical signal is 10 mV, while the concentration of the standard sugar content solution is 10 g / L, and the sensor's response electrical signal is 20 mV. By comparing the sensor signal with a standard curve, the corresponding acidity and sugar concentration can be converted. Assuming a sensor signal of 8 mV, the corresponding acidity concentration can be determined to be 0.8 mol / L based on the standard curve. Similarly, if the sensor signal is 15 mV, the corresponding sugar concentration can be determined to be 15 g / L. Then, an electronic tongue sensor is used to measure the sweetness and acidity of the juice sample, obtaining the corresponding sensor response data. Assuming the sweetness sensor response is 50 mV and the acidity sensor response is 12 mV, the juice sample is then subjected to high-frequency oscillation to release aroma components into the gas phase. Gas chromatography can then be used to separate and quantify the aroma components, obtaining detailed aroma data. Assuming the concentration of a certain aroma component is 5 ppm, the data obtained from the enzyme-based sensor, electronic tongue, and high-frequency oscillation technique are integrated and correlated. A taste dataset can be created, which includes detailed information on the acidity concentration of a juice sample as 0.8 mol / L, the sugar content concentration as 15 g / L, the sweetness sensor response as 50 mV, the acidity sensor response as 12 mV, and the concentration of a certain aroma component as 5 ppm.

[0038] S103. Correlate the flavor data of the juice with the shelf life of the juice to determine the correlation between different flavor attributes and shelf life.

[0039] Taste data, including sweetness, acidity, and aroma, was acquired from different types of fruit juice samples, along with the shelf life information for each sample. The acquired taste data was normalized, and data cleaning was performed to handle missing and outlier values. Based on the shelf life information, the fruit juice samples were divided into different shelf-life groups: long-term, medium-term, and short-term. Analysis of variance was used to compare the taste attributes of different shelf-life groups to determine if there were significant differences between the attributes. Based on the statistical analysis results, the differences in taste attributes between different shelf-life groups were assessed, identifying the taste attributes that affect shelf life. Correlation coefficients were calculated to determine the association between different taste attributes and shelf life. For example, data from three different fruit juice samples—apple juice, orange juice, and grape juice—were collected. Apple juice had an average sweetness of 8.5, average acidity of 3.2, average aroma of 7.0, and a shelf life of 20 days. Orange juice had an average sweetness of 7.8, average acidity of 2.5, average aroma of 6.5, and a shelf life of 12 days. Grape juice had an average sweetness of 9.2, average acidity of 2.8, average aroma of 8.2, and a shelf life of 28 days. The samples were categorized into long-term, medium-term, and short-term based on their shelf life: longer than 20 days for long-term, 10-20 days for medium-term, and less than 10 days for short-term. Analysis of variance (ANOVA) was used to obtain the F-values ​​and significance levels for sweetness, acidity, and aroma among the different shelf-life groups. The F-value for sweetness was 12.45, with a significance level of 0.002 (less than 0.05), indicating a significant difference in sweetness among the different shelf-life groups. The F-value for acidity was 1.67, with a significance level of 0.213 (greater than 0.05), indicating no significant difference in acidity among different shelf-life groups. The F-value for aroma was 0.95, with a significance level of 0.398 (greater than 0.05), indicating no significant difference in aroma among different shelf-life groups. Analysis of variance confirmed a significant difference in sweetness among different shelf-life groups, while acidity and aroma showed no significant differences. Pearson correlation coefficients were calculated between sweetness, acidity, aroma, and shelf life. The correlation coefficient between sweetness and shelf life was 0.75, indicating a positive correlation; higher sweetness is associated with a longer shelf life. The correlation coefficient between acidity and shelf life was -0.15, indicating a weak negative correlation. The correlation coefficient between aroma and shelf life was 0.05, indicating an extremely weak correlation. Therefore, there is a significant positive correlation between sweetness and shelf life. Fruit juices with high sweetness tend to have a longer shelf life, while acidity and aroma have a smaller impact on shelf life.

[0040] Analysis of variance was used to compare the flavor attributes of different shelf-life groups to determine the flavor attributes that affect shelf-life.

[0041] Taste data, including sweetness, acidity, and aroma, were obtained from fruit juice samples with different shelf-life groups. Analysis of variance and t-tests were used to compare the differences in taste attributes between different shelf-life groups to determine the significance level of these differences. Based on the statistical analysis results, it was determined whether significant differences existed between the different shelf-life groups. If the p-value was less than the preset significance level, a significant difference was considered to exist between the different shelf-life groups; if the p-value was greater than the preset significance level, it was impossible to determine whether a significant difference existed between the different groups. If a significant difference existed, the mean differences of each taste attribute were compared between the different groups to identify the attributes with significant differences. Based on the mean differences and dispersion between the different shelf-life groups, the differences in taste attributes between the different shelf-life groups were determined, and taste attributes that might affect shelf life were identified. For example, apple juice samples with long, medium, and short shelf lives were purchased from the market. Sweetness, acidity, and aroma data were measured for each sample. Analysis of variance (ANOVA) and t-tests were performed on sweetness, acidity, and aroma to compare the differences in taste attributes among different shelf-life groups. Therefore, the ANOVA F-value for sweetness was 12.98, with a significance level p-value of 0.001, which was less than the pre-specified significance level of 0.05, indicating a significant difference in sweetness among different shelf-life groups. The t-values ​​for long-term versus medium-term were 3.23, for long-term versus short-term were 4.89, and for medium-term versus short-term were 2.11. Based on the pre-specified significance level, the t-values ​​were compared with the critical values ​​to determine the significance between different groups. For sweetness, since the p-values ​​of both ANOVA and t-tests were less than 0.05, a significant difference in sweetness among different shelf-life groups could be determined. If significant differences exist, further comparisons of the mean differences of each taste attribute among different shelf-life groups would be conducted. The mean sweetness of the long-term group was 0.45 with a standard deviation of 0.03, the mean sweetness of the mid-term group was 0.39 with a standard deviation of 0.02, and the mean sweetness of the short-term group was 0.35 with a standard deviation of 0.04. This analysis shows a significant difference in sweetness among the different shelf-life groups; apple juice with a long shelf life has a higher sweetness, while apple juice with a short shelf life has a lower sweetness. The same analytical steps were used to analyze the acidity and aroma of the juice, and statistical analysis was used to identify flavor attributes that might affect shelf life in different shelf-life groups.

[0042] By calculating the correlation coefficient, the relationship between different flavor attributes and shelf life was determined.

[0043] Based on different fruit juice samples, taste data including sweetness, acidity, and aroma, along with corresponding shelf-life data, were obtained. The collected taste and shelf-life data were cleaned, denoised, and normalized to eliminate outliers and noise interference. Based on the taste and shelf-life data, the taste characteristics that significantly affect shelf-life were identified. Feature engineering methods were used to extract features, combining sweetness, acidity, and aroma into a comprehensive taste index. The correlation coefficient was calculated to determine the relationship between different taste characteristics and shelf-life. Based on the correlation between different taste characteristics and shelf-life, the degree of influence of taste characteristics on shelf-life was determined. For example, apple juice, orange juice, and grape juice samples were purchased, and sweetness, acidity, and aroma data, along with corresponding shelf-life data, were recorded for each sample. The collected taste and shelf-life data were cleaned to remove potential outliers and noise, and the taste data was normalized to ensure that different attributes had the same scale. Analyzing the cleaned data identified flavor characteristics that significantly impacted shelf life. For example, the relationship between data distribution and shelf life revealed a significant correlation between acidity and shelf life. Feature engineering was employed to combine multiple flavor characteristics, such as sweetness, acidity, and aroma, into a comprehensive flavor index to better represent the flavor attributes of the juice. Correlation coefficients between different flavor characteristics and shelf life were calculated to quantify their association. For instance, the correlation coefficient between acidity and shelf life was -0.78, indicating that juices with higher acidity may have a shorter shelf life. Based on the correlation coefficients between different flavor characteristics and shelf life, the degree of influence of each flavor characteristic on shelf life was determined. For example, the closer the correlation coefficient is to -1 or 1, the greater the influence of the flavor characteristic on shelf life. The analysis concluded that acidity has a significant impact on shelf life, with a clear correlation coefficient, while sweetness and aroma may have a smaller impact on shelf life, with lower correlation coefficients.

[0044] S104. Based on the types of fruit juice, the taste data of the fruit juice, and the mixing ratio of different types of fruit juice, a multilayer perceptron is used to train a model and obtain a compound fruit juice taste prediction model. At the same time, a linear regression model is used to construct a fruit juice shelf life prediction model based on the fruit juice taste data.

[0045] This process involves acquiring taste data for different types of fruit juice, their mixing ratios, and the taste data for compound fruit juices. Preprocessing this data includes outlier removal and normalization. Based on a multilayer perceptron, a compound fruit juice taste prediction model is constructed using the taste data for different types of fruit juice and the taste data for compound fruit juices. The prediction model is then used to obtain the predicted taste results for compound fruit juices. Next, fruit juice taste data and shelf-life data are acquired. Based on a linear regression model, a fruit juice shelf-life prediction model is constructed using the taste data and shelf-life data. For example, taste data for orange juice, strawberry juice, and pineapple juice, including sweetness, acidity, and aroma, are collected, and the shelf-life information for each sample is recorded. The taste data is normalized to ensure each attribute value is between 0 and 1, and the data is cleaned to remove possible outliers and noise. Finally, using the taste data for orange juice, strawberry juice, and pineapple juice as input, a compound fruit juice taste prediction model is constructed based on a multilayer perceptron. Taste data from different types of fruit juice were input into a mixed fruit juice taste prediction model to obtain the predicted taste results, such as a mixed sweetness of 0.75, a mixed acidity of 0.4, and a mixed aroma of 0.6. Taste data and corresponding shelf-life data for orange juice, strawberry juice, and pineapple juice samples were obtained from experiments. Based on a linear regression model, using the taste data from orange juice, strawberry juice, and pineapple juice as input, a fruit juice shelf-life prediction model was constructed.

[0046] Based on the types of fruit juice, the taste data of the fruit juice, and the mixing ratio of different types of fruit juice, a compound fruit juice taste prediction model is constructed by training a multilayer perceptron model.

[0047] Taste data from different types of fruit juices, including sweetness, acidity, and aroma, were collected to determine the composition of compound fruit juices, including the types and mixing ratios. This data was then compiled into training samples, with each sample containing the type of juice, taste attributes, and mixing ratio. A compound fruit juice taste prediction model was constructed by training a multilayer perceptron (MLP) model. The number of neurons in the input, hidden, and output layers of the MLP was determined based on the number of taste attributes in the training samples. The model was trained using the training samples, and the weights and biases were continuously updated using backpropagation to gradually improve the model's accuracy in predicting the taste of compound fruit juices. Using the trained compound fruit juice taste prediction model, new mixing ratios and types of juice were input to obtain the predicted taste of the compound fruit juice. For example, taste data from three different types of fruit juices, including sweetness, acidity, and aroma, were collected. The composition of the compound fruit juice was determined to be 50% apple juice, 30% orange juice, and 20% grape juice. The data is organized into training samples, each containing an encoding for the type of juice (e.g., 1 for apple juice, 2 for orange juice, 3 for grape juice), taste attributes such as sweetness, acidity, and aroma, and the mixing ratio. For example, a training sample might have juice type 1, sweetness 0.8, acidity 0.3, aroma 0.5, and a mixing ratio of 50% apple juice, 30% orange juice, and 20% grape juice. Based on the number of taste attributes (3) in the training samples, the input layer of the multilayer perceptron is set to 3 neurons, the hidden layer to 2 neurons, and the output layer to 1 neuron. The model is trained using the training samples, and the weights and biases are continuously updated through backpropagation to gradually and accurately predict the taste of the compound juice. Using the trained compound juice taste prediction model, a new mixing ratio and juice type are input to obtain the predicted taste of the compound juice.

[0048] S105. Using the compound fruit juice flavor prediction model, obtain the flavor prediction result of the compound fruit juice, and based on the fruit juice shelf life prediction model, obtain the shelf life prediction result of the compound fruit juice.

[0049] Using an electronic tongue sensor and high-frequency oscillation technology, the sweetness, acidity, and aroma of the mixed fruit juices are measured to obtain taste data. Based on this taste data, a mixed fruit juice taste prediction model is used to obtain a predicted taste result. Then, based on the predicted taste, a fruit juice shelf-life prediction model is used to obtain a predicted shelf-life result. For example, using an electronic tongue sensor and high-frequency oscillation technology to test two fruit juices, juice A has a sweetness of 5, acidity of 2, and aroma of 6; juice B has a sweetness of 8, acidity of 9, and aroma of 2. Using the mixed fruit juice taste prediction model, based on the taste data of juices A and B, the predicted taste result for the mixed fruit juice is sweetness of 2, acidity of 1, and aroma of 8. Then, using the fruit juice shelf-life prediction model, based on the taste prediction result, the predicted shelf-life of the mixed fruit juice is obtained, resulting in a shelf life of 2 weeks.

[0050] S106. By comparing the predicted results with the set shelf life standards and taste standards, the statistical analysis methods of chi-square test and T test are used to determine whether the compound fruit juice meets the standards.

[0051] Obtain shelf-life and taste data for the compound fruit juice. Calculate a comprehensive taste score by weighted summation based on sweetness, acidity, and aroma data for each batch. Divide shelf-life into two groups: meeting and not meeting standards. Divide taste scores into two groups: acceptable and unacceptable. Use a chi-square test to determine if there is a significant relationship between the predicted results and the standards for both shelf-life and taste. If the p-value of the chi-square test is less than the set significance level, the predicted results are considered to have a significant relationship with shelf-life and taste; otherwise, no significant relationship is considered. For different taste attributes of the compound fruit juice, use a t-test to determine if the difference between the predicted results and each taste attribute meets the standards. Group each taste attribute of the compound fruit juice, including aroma, sweetness, and acidity. Compare the predicted results with each attribute and use a t-test to determine if there is a significant difference. If the p-value of the t-test is less than the set significance level, the compound fruit juice meets the standards for that taste attribute; otherwise, it does not meet the standards. For example, obtain sample data for different batches of compound fruit juice, including shelf-life and taste data for each batch. Batch 1: Shelf life = 180 days, sweetness = 0.75, acidity = 0.45, aroma = 0.60; Batch 2: Shelf life = 150 days, sweetness = 0.80, acidity = 0.42, aroma = 0.65; Batch 3: Shelf life = 170 days, sweetness = 0.70, acidity = 0.48, aroma

[0052] =0.58; Batch 4: Shelf life = 160 days, sweetness = 0.73, acidity = 0.50, aroma = 0.61; Based on the sweetness, acidity, and aroma attributes of each batch, a weighted summation of the overall taste score is calculated. According to the shelf life standard, a shelf life of 180 days or more is classified as the compliant group, and a shelf life less than 180 days is classified as the non-compliant group. A taste score of 0.7 or more is classified as the qualified group, and a taste score less than 0.7 is classified as the unqualified group. A chi-square test is used to determine whether there is a significant relationship between the predicted results and the shelf life and taste. Assuming the p-value of the chi-square test is 0.02, which is less than the set significance level of 0.05, since the p-value is less than the significance level, it is considered that there is a significant relationship between the predicted results and the shelf life and taste. For different taste attributes—aroma, sweetness, and acidity—a t-test was conducted between the qualified and unqualified groups. It was assumed that the p-value for sweetness was 0.015 (less than the significance level), the p-value for acidity was 0.085 (greater than the significance level), and the p-value for aroma was 0.03 (less than the significance level). The p-values ​​for sweetness and aroma being less than the significance level indicate a significant difference between the qualified and unqualified groups. However, the p-value for acidity being greater than the significance level makes it impossible to determine whether a significant difference exists. Therefore, based on the results of the chi-square test and t-test, it can be concluded that there is a significant relationship between the predicted results of the compound fruit juice and its shelf life and taste, indicating a significant association between taste scores and shelf life. Regarding taste attributes, there are significant differences in sweetness and aroma between the qualified and unqualified groups, while the difference in acidity is not significant.

[0053] S107. If the shelf life of the compound fruit juice does not meet the standard, or the taste does not meet the expectations, change the formula of the compound fruit juice, re-conduct the characteristic analysis and taste evaluation, and re-enter the prediction step until a prediction result that meets the standard is obtained.

[0054] The process involves obtaining shelf-life and taste data for the compound fruit juices to identify which samples failed to meet shelf-life or taste targets. Taste data, including acidity, sweetness, and aroma, is then collected from these samples. Based on this data, it's determined whether acidity or sweetness needs adjustment to achieve the desired shelf-life or taste. Acidity or sweetness is adjusted by gradually increasing or decreasing it in the failed samples, depending on the desired direction. Adjusted samples are then prepared based on the adjusted acidity or sweetness. The prepared adjusted samples are then retested for acidity, sweetness, and shelf-life to obtain taste data. The component data before and after adjustment are compared with the shelf-life and taste targets to determine if the adjustment was successful and if the expected goals were achieved. If both shelf-life and taste cannot be simultaneously met, a fruit juice formulation adjustment model is established using linear programming to obtain the final formulation combination, ensuring the compound fruit juice achieves optimal taste while meeting shelf-life requirements. Chi-square and T-tests are then used to determine if the adjustment significantly affected the shelf-life and taste targets. Based on the statistical analysis results, the optimal adjustment plan was determined to maintain the target taste while meeting the shelf life requirement, or to achieve the best taste for the compound fruit juice while meeting the shelf life requirement. For example, there are two compound fruit juices, A and B, where A has an acidity of 0.8 and a sweetness of 0.6; B has an acidity of 1.2 and a sweetness of 0.4. The preset target values ​​are an acidity between 0.9 and 1.1 and a sweetness between 0.5 and 0.7. It was identified that the acidity of juice B did not meet the target and needed adjustment. First, the acidity was gradually increased, starting from 1.2, and a series of samples were prepared, such as 1.3, 1.4, etc. These samples were retested, and it was found that when the acidity reached 1.3, the shelf life target was met, but the acidity exceeded the expected range. Then, chi-square tests and t-tests were performed to determine whether the effect of adjusting the acidity on the shelf life and sweetness was significant. If both shelf life and taste cannot be met simultaneously, a juice formula adjustment model is established based on a linear programming algorithm to obtain the final formula combination, enabling the compound juice to achieve the best taste while meeting the shelf life requirements. Based on statistical analysis, the optimal adjustment scheme can be determined to ensure that the acidity of juice B reaches the expected target while also meeting the shelf life standard. The final formula combination obtained is the final result.

[0055] Based on the linear programming algorithm, a juice formula adjustment model is established to obtain the final formula combination, so that the compound juice can achieve the best taste while meeting the shelf life requirements.

[0056] Based on the taste prediction model and shelf-life prediction model, a juice formula adjustment model is established. The objective function of the model is the taste prediction result, and the constraint is the shelf-life requirement. The objective function and constraints are transformed into mathematical expressions, and a linear programming algorithm is used to solve for the optimal formula combination. Based on the established juice formula adjustment model, possible juice formula combinations are formulated, including the proportion and adjustment range of different juice types. For each formula combination, the predicted taste and shelf-life values ​​are calculated according to the taste prediction model and the shelf-life prediction model. The merits of each formula combination are evaluated according to the objective function and constraints, and the combination with the best taste quality is selected. A genetic algorithm is used, with the predicted taste value as the objective and the predicted shelf-life value as the constraint, for iterative search. With each iteration, a new juice formula combination is generated according to the algorithm strategy, and its predicted taste and shelf-life values ​​are calculated. Based on the obtained formula combinations, new compound juice samples are prepared, and the taste and shelf-life of the prepared new juice samples are evaluated. The actual taste data and shelf-life data are compared with the previous prediction results. The deviation of the taste data and the degree of conformity with the shelf-life data are compared to determine whether the adjusted formula has achieved the expected effect. If the adjusted formula does not achieve the expected results, the parameters in the formula adjustment model are updated based on the actual test results, and iterative optimization is performed until a more accurate formula combination is obtained, so that the compound juice achieves the best taste while meeting the shelf-life requirements, thus obtaining the final formula combination. For example, we want to adjust a juice formula using apple and orange juice to achieve the best taste quality. A taste prediction model and a shelf-life prediction model have already been established. Now, we use a linear programming algorithm to solve for the optimal formula combination. The objective function is the taste prediction result, aiming to maximize the predicted taste value. The constraint is the shelf-life requirement, aiming for the predicted shelf-life value to be greater than or equal to a given value. The assumed taste prediction model is: taste prediction value = 8 * apple juice + 6 * orange juice; the assumed shelf-life prediction model is: shelf-life prediction value = 5 * apple juice + 3 * orange juice. Now, we use a linear programming algorithm to solve for the optimal formula combination. Let x be the proportion of apple juice and y be the proportion of orange juice. The optimal formula is determined by maximizing the predicted taste value (8x + 6y) under the constraint that the predicted shelf life must be greater than or equal to the required shelf life value (5x + 3y >= 7). There is also a proportional constraint: x + y = 1 (the sum of the apple and orange juice ratios equals 1). By solving this linear programming problem, the optimal formula combination can be obtained, maximizing the predicted taste value while satisfying the shelf life constraints. Let's assume the optimal formula combination is x = 6, y = 4, meaning the ratio of apple to orange juice is 3:2. Based on this optimal formula combination, a new compound juice sample can be prepared. Then, the sample is subjected to taste evaluation and shelf life testing. Assuming that the actual test results show a predicted taste value of 9 and a predicted shelf life value of 8, the results are compared with the predicted values.Based on the deviations in taste data and the degree of conformity with shelf-life data, it can be determined whether the adjusted formula has achieved the expected results. If the expected results are achieved, the optimal formula combination is x=6, y=4. If the expected results are not achieved, the parameters in the juice formula adjustment model can be updated based on actual test results, such as readjusting the weights of the taste prediction model and the shelf-life prediction model. Then, iterative optimization is performed until a more accurate formula combination is obtained, enabling the compound juice to achieve the best taste quality while meeting shelf-life requirements. The final formula combination obtained is the final result.

[0057] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for real-time monitoring of the freshness of compound fruit juice, characterized in that, The method includes: For a single type of fruit juice, physical and chemical characteristic analysis is performed to obtain precise fruit juice characteristic parameters, including acidity and sugar content. Enzyme-based sensors are used to precisely calibrate the chemical characteristics of acidity and sugar content in the fruit juice. Simultaneously, an electronic tongue sensor measures the sweetness and acidity of the fruit juice sample. Detailed aroma data of the fruit juice is acquired using high-frequency oscillation and gas chromatography techniques to form a taste dataset. The taste data of the fruit juice is correlated with its shelf life to determine the correlation between different taste attributes and shelf life. Based on the type of fruit juice, its taste data, and the mixing ratio of different types of fruit juice, a multilayer perceptron is used to train a model to obtain a mixed fruit juice taste prediction model. Simultaneously, a linear regression model is used to construct a fruit juice shelf life prediction model based on the fruit juice taste data. The mixed fruit juice taste prediction results are obtained through the mixed fruit juice taste prediction model, and the shelf life prediction results are obtained based on the fruit juice shelf life prediction model. The prediction results are then used to predict the shelf life of the mixed fruit juice. The results are compared with the set shelf life and taste standards. Statistical analysis methods, including chi-square test and t-test, are used to determine whether the compound fruit juice meets the standards. This includes: obtaining shelf life and taste data of the compound fruit juice and obtaining a comprehensive taste score through weighted summation; grouping the results by shelf life and taste score to determine whether they meet the standards; using chi-square test to determine if there is a significant relationship between the predicted results and the shelf life and taste; if the p-value of the chi-square test is less than the set significance level, a significant relationship exists; for different taste attributes of the compound fruit juice, using t-test to determine if the difference between the predicted results and each taste attribute meets the standards; grouping each taste attribute and comparing the predicted results using t-test to determine if there are significant differences; if the shelf life of the compound fruit juice does not meet the standards or the taste does not meet expectations, the formula of the compound fruit juice is changed, the characteristic analysis and taste evaluation are repeated, and the prediction step is entered again until a predicted result that meets the standards is obtained.

2. The method according to claim 1, wherein, The method involves performing physical and chemical characteristic analysis on a single type of fruit juice to obtain precise fruit juice characteristic parameters, including acidity and sugar content, among others. Physical and chemical characterization also includes nutritional components and microbial activity; For the collected and labeled fruit juice samples, acidity was determined by acid-base titration to obtain acidity data; sugars were separated and detected by chromatography to obtain sugar content; vitamins, minerals and amino acids in the samples were determined by a nutrient analysis instrument to obtain nutrient data; and microbial activity was determined to obtain microbial activity data.

3. The method according to claim 1, wherein, The method employs an enzyme-based sensor to precisely calibrate the chemical characteristics of fruit juice, including acidity and sugar content. Simultaneously, it utilizes an electronic tongue sensor and high-frequency oscillation technology to measure the sweetness, acidity, and aroma of the fruit juice, forming a taste dataset, including: Design an enzyme-based sensor corresponding to the acidity and sugar content of the fruit juice; acquire the electrical signal of the enzyme-based sensor response to achieve accurate calibration of the acidity and sugar content of the fruit juice; integrate and correlate the various data to establish a corresponding fruit juice flavor dataset.

4. The method according to claim 1, wherein, The process of correlating the flavor data of fruit juice with its shelf life, and determining the correlation between different flavor attributes and shelf life, includes: The process involves acquiring taste data and shelf-life information for different types of fruit juice samples; cleaning and normalizing the data; dividing the fruit juice samples into different groups based on the shelf-life information; comparing the taste attributes of different shelf-life groups using analysis of variance; identifying the taste attributes that affect the shelf-life; and determining the correlation between taste attributes and shelf-life based on the calculated correlation coefficient.

5. The method according to claim 1, wherein, Based on the types of fruit juice, their flavor data, and the mixing ratio of different types of fruit juice, a multilayer perceptron is used to train a model and obtain a composite fruit juice flavor prediction model. Simultaneously, a linear regression model is used to construct a fruit juice shelf-life prediction model based on the fruit juice flavor data, including: By collecting taste data and mixing ratios of different types of fruit juice, a dataset of the tastes and mixing ratios of various fruit juices is obtained. From this dataset, taste data of compound fruit juices is obtained. Further data preprocessing is performed on the datasets, including outlier removal and normalization, to determine a dataset of fruit juice tastes and mixing ratios that meet the data processing conditions. Using a multilayer perceptron, a compound fruit juice taste prediction model is constructed based on the determined dataset of fruit juice tastes and mixing ratios that meet the data processing conditions. The prediction results of the compound fruit juice taste are obtained using this model. Finally, by collecting fruit juice taste data and shelf-life data, a fruit juice shelf-life prediction model is constructed using a linear regression model.

6. The method according to claim 1, wherein, The process of obtaining a flavor prediction result for the compound fruit juice using the compound fruit juice flavor prediction model and obtaining a shelf life prediction result for the compound fruit juice using the fruit juice shelf life prediction model includes: The sweetness, acidity, and aroma of the mixed fruit juice were measured using an electronic tongue sensor and high-frequency oscillation technology to obtain the taste data of the fruit juice to be mixed. Based on the obtained taste data, a compound fruit juice taste prediction model was used to obtain the prediction results. Based on the predicted taste of the compound fruit juice, a fruit juice shelf life prediction model was used to obtain the shelf life prediction results.

7. The method according to claim 1, wherein, If the shelf life of the compound fruit juice does not meet the standard, or the taste does not meet expectations, the formula of the compound fruit juice is changed, the characteristic analysis and taste evaluation are repeated, and the prediction step is entered again until a prediction result that meets the standard is obtained, including: Obtain shelf-life and taste data for the compound fruit juice, and identify samples that do not meet the standards. Obtain taste data for the substandard compound fruit juice samples and determine whether acidity or sweetness needs to be adjusted to meet the standards. Adjust the acidity or sweetness of the substandard compound fruit juice samples and prepare adjusted compound fruit juice samples. Test the adjusted compound fruit juice samples, obtain the adjusted taste data, and compare it with the shelf-life and taste targets to determine whether the adjustment was successful. If the shelf-life and taste cannot be met simultaneously, establish a fruit juice formula adjustment model to obtain the final formula combination so that the compound fruit juice achieves the best taste while meeting the shelf-life requirements. Use chi-square test and T-test to determine whether the adjustment significantly affects the shelf-life and taste targets, and determine the optimal adjustment scheme.

Images