A method for rapidly determining DPPH scavenging rate of pear juice and pear paste based on electronic tongue technology

Abstract

The present application belongs to the technical field of food physical and chemical detection, and relates to a method for rapidly detecting DPPH clearance rate of pear juice and pear paste based on an electronic tongue. In view of the defects of traditional spectrophotometric method for determining DPPH clearance rate of pear products, such as complicated operation, strong reagent dependence and difficulty in batch detection, the present application prepares pear juice and pear paste samples to be tested through a standardized sample pretreatment process, takes the DPPH clearance rate measured by the spectrophotometric method (traditional standard method) as the reference true value, collects 9 kinds of taste characteristic values by using a SA-402B electronic tongue, adopts a least squares support vector machine (LS-SVM) algorithm to construct a radial basis kernel function regression prediction model and optimize key parameters, and realizes rapid and accurate prediction of the DPPH clearance rate of the sample to be tested. The method for determining the DPPH clearance rate provided by the present application is simple to operate, does not need complicated chemical reagent preparation, is suitable for batch detection requirements, and can be widely applied to scenes such as antioxidant activity evaluation, quality grading and processing technology optimization of pear products.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of intelligent sensory detection and food functional component analysis, specifically involving a method for rapidly detecting the antioxidant capacity of pear juice and pear paste based on biomimetic sensing and pattern recognition technology. Background Technology

[0002] Pear juice and pear syrup, traditional Chinese medicine and food, are rich in polyphenols, flavonoids, amino acids, and other natural antioxidants. These components not only carry the unique flavor and core nutritional value of pear products but also exert anti-inflammatory, anti-cancer, and anti-aging effects by scavenging free radicals and inhibiting oxidative stress, thus playing a vital role in human health. With the rapid development of the functional food market, the antioxidant capacity of pear juice and pear syrup has become a core indicator for evaluating product quality, pricing, and market competitiveness. Among these, DPPH scavenging rate is a key parameter for quantifying antioxidant activity, directly reflecting product quality and functional value, and serving as an important basis for raw material selection, process optimization, and market pricing of pear products.

[0003] Currently, the detection of DPPH scavenging rate in pear juice and pear paste mainly relies on the traditional DPPH free radical scavenging method (spectrophotometry). This method requires multiple cumbersome steps, including precise reagent preparation, gradient reactions, and absorbance measurement. It not only consumes large amounts of chemical reagents and easily causes environmental pollution, but also has a detection cycle of over 2 hours, making it difficult to meet the needs of batch testing. Furthermore, existing auxiliary detection methods have significant shortcomings: while high-performance liquid chromatography (HPLC) can indirectly estimate the content of components related to DPPH scavenging rate, it requires complex sample pretreatment, has high equipment costs, and low detection efficiency, making it unsuitable for real-time monitoring scenarios on production lines; human sensory evaluation relies entirely on the evaluator's subjective experience, failing to establish a quantitative correlation with DPPH scavenging rate and making objective and accurate detection difficult.

[0004] Electronic tongues, as an intelligent detection technology that simulates human taste perception, simultaneously capture the signal responses of characteristic substances in samples through a multi-channel sensor array. Combined with pattern recognition algorithms, they can achieve rapid quantitative analysis of sample attributes, offering significant advantages such as fast detection speed, strong objectivity of results, low operating costs, and the ability to perform batch testing. They have shown promising application prospects in fields such as food flavor evaluation and quality grading. However, there are currently no reports on the application of electronic tongue technology to the detection of DPPH clearance rates in pear juice and pear paste. The core technical bottleneck lies in the complex composition of pear products; the intrinsic correlation between their taste signals and DPPH clearance rates has not been revealed, and there is a lack of targeted model building and parameter optimization methods. This prevents electronic tongue technology from being effectively transformed into a highly efficient detection method for DPPH clearance rates.

[0005] To address the shortcomings of existing DPPH scavenging rate detection methods, such as long detection cycles, cumbersome operations, significant reagent contamination, high professional barriers, and difficulty in batch testing, this invention provides a rapid detection method for DPPH scavenging rate of pear juice and pear paste based on electronic tongue technology. By integrating standardized sample pretreatment, electronic tongue feature data acquisition, and least squares support vector machine (LS-SVM) modeling technology, a precise correlation between taste characteristics and DPPH scavenging rate is established, enabling rapid and accurate quantitative analysis of the antioxidant activity of pear products. This fills a gap in existing technology and meets the batch testing needs of food production sites. Summary of the Invention

[0006] Technical solution: The technical solution of the present invention is achieved through the following steps. The parameters of each step are optimized and screened by the system to ensure the synergistic improvement of detection accuracy and efficiency. Specifically, it includes experimental preparation, sample pretreatment, reference data determination, electronic tongue data acquisition, model construction and target sample detection.

[0007] 1. Preparation of experimental materials and instruments 1.1 Sample screening and pretreatment: Pear fruits of different varieties such as white pear, sand pear, and autumn pear were collected. The screening criteria were no mechanical damage, no pests or diseases, and no rot or deterioration. After harvesting, the fruits were refrigerated at 4 ℃ for no more than one month to ensure the freshness and representativeness of the samples and to avoid affecting the accuracy of the test results due to sample deterioration. 1.2 Main Instruments and Reagents: Instruments include a HU24FR3L multi-functional juicer, an LC-EA3S electric ceramic stove, a UV-2600 UV-Vis spectrophotometer, a PAL-α refractometer, a PAL-1 handheld digital saccharimeter, an SA-402B electronic tongue, and a CR22GⅢ centrifuge; reagents include 1,1-diphenyl-2-trinitrophenylhydrazine (DPPH, purity ≥98%) and anhydrous ethanol. All reagents and instruments are commercially available conventional products, reducing the cost of promoting and applying the method.

[0008] 2. Standardization Pretreatment of Pear Juice and Pear Paste Samples 2.1 Pear juice sample preparation: Selected pear fruits were washed, cored, and cut into pieces. Juice was extracted using a HU24FR3L multi-functional juicer. The crude juice was transferred to 50 mL centrifuge tubes and centrifuged at 4 ℃ and 6000 r / min for 10 min. The supernatant was collected and sealed at 4 ℃. The analysis should be completed within 24 h to avoid sample oxidation and deterioration affecting the accuracy of the analysis. 2.2 Preparation of Pear Syrup Samples: Selected pear fruits were washed, cored, cut into pieces, and juiced. The juice was then passed through a 100-mesh sieve to remove pulp residue and other impurities. The filtered pear juice was placed on an LC-EA3S electric ceramic stove for gradient temperature control cooking. The cooking parameters were precisely set as follows: initial heat 550 W, cooking until the sugar content of the juice reached 40%.0 When Brix, adjust the heat to 350 W and continue simmering; when the sugar content of the juice reaches 55... 0 When braising, adjust the heat to 150 W until the juice reaches a sugar content of 70%. 0 The cooking process ends at Brix; take 10 g of pear paste sample, dilute it with deionized water at a volume ratio of 1:6, transfer it to a 50 mL centrifuge tube, centrifuge at 4 ℃ and 6000 r / min for 10 min, take the supernatant and seal it at 4 ℃, and complete the detection within 24 h.

[0009] 3. Determination of DPPH scavenging rate by conventional spectrophotometry (reference method) This step provides accurate reference data for model building. The specific operational process is standardized as follows: 3.1 Reagent preparation: Accurately weigh 0.025 g DPPH, dissolve it in anhydrous ethanol by sonication and bring the volume to 1000 mL to prepare a DPPH stock solution with a concentration of 60 μmol / L. Store the stock solution at 4 ℃ protected from light. Before use, it needs to be equilibrated to room temperature (25±1 ℃) and should be prepared fresh each time to avoid reagent failure affecting the test results. 3.2 Reaction System Construction: 100 μL of the sample to be tested (pear juice or pear paste supernatant) was placed in a 10 mL test tube, and 3.0 mL of DPPH stock solution was added. The mixture was vortexed for 10 s to ensure homogeneity. The mixture was then reacted at room temperature (25±1 ℃) in the dark for 30 min to ensure sufficient reaction between DPPH and the antioxidant components in the sample. Two control experiments were simultaneously set up to eliminate systematic errors: ① Blank control group: 100 μL anhydrous ethanol + 3.0 mL DPPH stock solution; ② Sample control group: 100 μL of the sample to be tested + 3.0 mL anhydrous ethanol. Both control groups were subjected to the same shaking and light-protected reaction conditions as described above. 3.3 Absorbance Measurement: Using anhydrous ethanol as a reference, the absorbance of the blank control group, sample control group and sample group were measured sequentially at a wavelength of 517 nm using a UV-2600 UV-Vis spectrophotometer. Each sample was measured in parallel 3 times and the average value was taken as the final absorbance data to reduce random errors. 3.4 Calculation of DPPH clearance rate: The DPPH clearance rate of the sample is calculated according to the following formula: DPPH clearance rate (%) = [A0 - (A1 - AS)] / A0 × 100%; where A0 is the absorbance of the blank control group, A1 is the absorbance of the sample group, and AS is the absorbance of the sample control group. This formula can effectively eliminate the interference of the sample's own color on the absorbance measurement, ensuring that the calculation results are accurate and reliable.

[0010] 4. Electronic tongue feature data acquisition 4.1 Electronic tongue sample preparation: Take 35.0 mL of the supernatant of pear juice and pear paste after centrifugation, transfer it into the special detection cup for electronic tongue, and cover it to prevent contamination; place the detection cup in an environment of 25±1 ℃ for equilibration for 10 min to stabilize the sample temperature and avoid temperature fluctuations affecting the stability of the sensor response signal. 4.2 Electronic tongue sensor data acquisition: The SA-402B electronic tongue (Insent Corporation, Japan) was used to detect the taste sensory information of the sample. The device is equipped with 7 sensors: AAE (umami), CTO (salty), CA0 (sour), C00 (bitter), AE1 (astringent), AC0 (umami auxiliary), and ANO (sweet), and can output 9 kinds of taste information (including the astringent aftertaste corresponding to AE1 and the bitter aftertaste information corresponding to C00). Before testing, a strict sensor self-test and pre-processing procedure was performed: the sensor was first cleaned in the cleaning solution for 90 seconds, then cleaned in two different reference solutions for 120 seconds each, and finally the sensor was zeroed at the equilibrium position for 30 seconds to ensure that the sensor was in a stable working state; after reaching the equilibrium condition, the test was started, and the initial taste value was output after 30 seconds of testing; then, the sensor was briefly cleaned in the two sets of reference solutions for 3 seconds each, and the aftertaste was tested for 30 seconds after inserting a new reference solution; the test was repeated 4 times, the unstable data of the first cycle was discarded, and the average value of the results of the last 3 tests was taken. The potential signal was converted into a standardized electronic tongue taste characteristic value by the system's built-in software.

[0011] 5. Construction and Optimization of LS-SVM Prediction Model 5.1 Basic Data for Model Construction: Using nine taste characteristics collected by the electronic tongue as independent variables and the DPPH clearance rate determined by traditional spectrophotometry as the dependent variable, a dataset containing 131 samples was constructed to ensure that the data covers pear products of different varieties and processing methods, thereby improving the model's generalization ability. 5.2 Model Construction and Parameter Optimization: A regression prediction model was built using the Least Squares Support Vector Machine (LS-SVM) algorithm. The dataset was loaded into the LS SVMlabv1 8R2009b_R2011a software and randomly divided into a training set (88 groups) and a prediction set (43 groups) at a ratio of 3:1. The radial basis function (RBF) kernel function was selected, and the key parameters of the model (regularization parameter gam and kernel parameter sig2) were optimized by leave-one-out cross-validation. The optimal parameter combination was finally determined to be gam=42.43 and sig2=2.07. The regularization parameter was used to adjust the model complexity to suppress overfitting, and the kernel parameter was used to optimize the feature space mapping characteristics. 5.3 Model Performance Validation: The model was constructed using the training set, and its generalization ability was validated using the prediction set. The coefficient of determination (R²), root mean square error (RMSE), and relative analytical error (RPD) were used as the core evaluation metrics. Validation results showed that the training set R²c = 0.99 and RMSE = 1.66; the prediction set R²c = 0.99 and RMSE = 1.66. p =0.99, RPD=8.7 (industry standard: RPD>2.0 indicates that the model has effective prediction ability, RPD>8.0 indicates that the model has excellent detection accuracy), which proves that the stability, reliability and generalization ability of the model constructed by this invention meet the actual needs of batch detection.

[0012] 6. Rapid detection of DPPH scavenging rate of target samples 6.1 Sample pretreatment: Take the pear juice or pear paste sample to be tested and pretreat it strictly according to the standardized sample preparation method in step 2 above (centrifugation, filtration or dilution) to obtain the test solution that meets the testing requirements. 6.2 Electronic tongue feature data acquisition: Following the instrument debugging, preprocessing and detection process in step 4.2, acquire nine electronic tongue taste feature data of the target sample to ensure that the data acquisition conditions are consistent with the model construction stage and avoid systematic errors. 6.3 Output of detection results: The collected electronic tongue feature data is input into the established LS-SVM prediction model. The model automatically calculates and outputs the predicted value of DPPH clearance rate of the target sample, which can realize the rapid detection of batch samples.

[0013] This invention, through standardized sample pretreatment procedures, precise optimization of electronic tongue detection parameters and modeling algorithms, establishes for the first time a precise correlation model between electronic tongue taste characteristic data and DPPH scavenging rates of pear juice and pear paste, enabling rapid quantitative detection of the antioxidant activity of pear products. Compared with existing technologies, it possesses the following core advantages: 1. Highly efficient testing process: It can meet the needs of rapid testing of large batches of samples and is suitable for real-time quality monitoring scenarios in food production and processing. 2. Simple operation and avoidance of environmental pollution: No need for precise preparation of complex chemical reagents and long-term light-protected reaction, standardized sample pretreatment process, high degree of automation of electronic tongue detection, which greatly reduces the professional threshold of operators, and avoids the pollution of the environment by chemical reagents, making it suitable for rapid on-site detection. 3. Excellent and stable detection accuracy: The prediction model satisfies R². p With an RPD of ≥0.98 and RPD ≥8.0, the test results show good consistency and small error with traditional spectrophotometry. It can accurately quantify the antioxidant activity of pear products, providing an efficient and reliable technical means for quality evaluation, grading and screening, and optimization of processing technology of pear products, and has a wide range of applications. Attached image description: Figure 1 : Schematic diagram of pear fruit raw materials and processed products (pear juice and pear paste); This diagram clearly shows the specific morphology of the test objects of this invention, including pear fruits of different varieties, standardized pear juice and pear paste samples, and clarifies the basic morphological standards for sample screening and preparation, providing an intuitive reference for sample preparation steps. Figure 2 : Flowchart for establishing a quantitative detection model for a rapid determination of DPPH clearance rate in pear juice and pear paste; This diagram systematically presents the core technical process of this invention, including the complete logical chain of sample preparation, electronic tongue data acquisition, traditional method reference determination, model construction and model verification, and intuitively demonstrates the connection relationship and core technology of each step. Figure 3 : Fitting graph of the true and predicted values ​​of DPPH scavenging rates of pear juice and pear paste; This graph is a key attached figure for model performance verification. The fitting curve intuitively shows the high consistency between the predicted values ​​of the LS-SVM prediction model and the true values ​​measured by traditional spectrophotometry, directly confirming the detection accuracy and reliability of the model.

[0014] Detailed Implementation: The present invention will be further described in detail below with reference to specific embodiments, accompanying drawings, and experimental data. Experimental conditions not explicitly defined in this embodiment are performed in accordance with standard operating procedures in the food testing field; reagents and instruments not labeled with their source are all commercially available conventional products. This embodiment involves the appendix... Figure 1 (Sample illustration) Figure 2 (Overall flowchart of the method) Figure 3 (Model fitting result diagram), the specific steps are as follows:

[0015] 1. Preparation of experimental materials and instruments 1.1 Sample screening and pretreatment: Fruits from 22 different pear varieties, including white pear (Ya pear, Crown pear), sand pear (Nanshui pear, Golden pear, Huashan pear), and autumn pear (Fragrant pear, Nanguo pear, An pear), were collected. The screening criteria were no mechanical damage, no pests or diseases, and no rot or spoilage. After harvesting, the samples were refrigerated at 4 ℃ for no more than one month to ensure the freshness and representativeness of the samples. 1.2 Main Instruments and Reagents: Instruments included a HU24FR3L multi-functional juicer (Huiren Electronics Co., Ltd., Korea), an LC-EA3S electric ceramic stove (Shunde Zhongchen Electric Appliance Co., Ltd., Guangdong), a UV-2600 UV-Vis spectrophotometer (Shimadzu Corporation, Japan), a PAL-α refractometer, a PAL-1 handheld digital saccharimeter (both from ATAGO Corporation, Japan), an SA-402B electronic tongue (Insent Corporation, Japan), and a CR22GⅢ centrifuge (Hitachi Koki Corporation, Japan); reagents included 1,1-diphenyl-2-trinitrophenylhydrazine (DPPH, purity ≥98%, Shanghai Yuanye Biotechnology Co., Ltd.) and anhydrous ethanol (Sinopharm Chemical Reagent Co., Ltd.).

[0016] 2. Preparation of pear juice and pear paste samples 2.1 Preparation of pear juice samples: 22 varieties of pear fruits were taken, washed, cored, and cut into pieces, and juiced using a HU24FR3L multi-functional juicer. The crude juice was transferred to 50 mL centrifuge tubes and centrifuged at 4 ℃ and 6000 r / min for 10 min. The supernatant was collected and sealed at 4 ℃ and stored. The analysis should be completed within 24 h. 2.2 Preparation of Pear Syrup Samples: Pears of 22 varieties were collected, washed, cored, and chopped. Juice was extracted using a HU24FR3L multi-functional juicer, and the juice was then passed through a 100-mesh sieve to remove pulp residue. The filtered pear juice was then cooked on an LC-EA3S electric ceramic stove with the following parameters: initial heat 550 W, cooking until the sugar content reached 40%. 0 When Brix, adjust the heat to 350W and continue simmering; when the sugar content of the juice reaches 55... 0 When braising, adjust the heat to 150 W until the juice reaches a sugar content of 70%. 0 The cooking process ends at Brix. Take 10 g of the above pear paste sample and dilute it with deionized water at a volume ratio of 1:6. Transfer the diluted pear paste sample to a 50 mL centrifuge tube and centrifuge at 4 ℃ and 6000 r / min for 10 min. Take the supernatant and seal it at 4 ℃. The test must be completed within 24 hours.

[0017] 3. Determination of DPPH scavenging rate by conventional spectrophotometry (reference method) 3.1 Reagent preparation: Accurately weigh 0.025 g DPPH, dissolve it in anhydrous ethanol by sonication and bring the volume to 1000 mL to prepare a DPPH stock solution with a concentration of 60 μmol / L. Store the stock solution at 4 ℃ protected from light. Equilibrate to room temperature (25±1℃) before use and prepare fresh before use. 3.2 Construction of reaction system: 100 μL of the sample to be tested was placed in a 10 mL test tube, 3.0 mL of DPPH stock solution was added, and the mixture was vortexed for 10 s. The mixture was then reacted for 30 min at room temperature (25±1 ℃) in the dark. A blank control group (100 μL of anhydrous ethanol + 3.0 mL of DPPH stock solution) and a sample control group (100 μL of the sample to be tested + 3.0 mL of anhydrous ethanol) were set up simultaneously. Both control groups were subjected to the same shaking and light-protected reaction conditions as described above. 3.3 Absorbance measurement: Using anhydrous ethanol as a reference, the absorbance of each group was measured at a wavelength of 517 nm using a UV-2600 UV-Vis spectrophotometer. Each sample was measured in triplicate, and the average value was taken. 3.4 Scavenging rate calculation: DPPH scavenging rate (%) is calculated according to the formula: DPPH scavenging rate (%) = [A0 - (A1 - AS)] / A0 × 100%, where A0 is the absorbance of the blank control group, A1 is the absorbance of the sample group, and AS is the absorbance of the sample control group.

[0018] 4. Electronic tongue data acquisition 4.1 Electronic tongue sample preparation: Take 35.0 mL of the supernatant of pear juice and pear paste after centrifugation, transfer it into the special test cup for electronic tongue, cover it to prevent contamination, and place it in an environment of 25±1 ℃ for 10 min to equilibrate before testing. 4.2 Electronic tongue sensor data acquisition: The SA-402B electronic tongue (Insent Corporation, Japan) was used to detect the taste sensory information of the sample. The device is equipped with 7 sensors: AAE (umami), CT0 (salty), CA0 (sour), C00 (bitter), AE1 (astringent), AC0 (umami auxiliary), and AN0 (sweet), which can output 9 kinds of taste information (including the astringent aftertaste corresponding to AE1 and the bitter aftertaste corresponding to C00). Before testing, a sensor preprocessing procedure is performed: the sensor is first cleaned in a cleaning solution for 90 seconds, then cleaned in two different reference solutions for 120 seconds each, and finally zeroed at the equilibrium position for 30 seconds; after reaching the equilibrium condition, the test begins, and the initial taste value is output after 30 seconds of testing; then, the sensor is briefly cleaned in two sets of reference solutions for 3 seconds each, and the aftertaste is tested for 30 seconds after inserting a new reference solution; the test is repeated 4 times, the unstable data from the first cycle is discarded, and the average value of the results of the last 3 tests is taken. The potential signal is converted into electronic tongue taste characteristic value by the system's built-in software.

[0019] 5. Data Matching and Model Building: Basic Data Preparation The DPPH clearance rate of 22 varieties of pear juice and pear paste samples was determined using the traditional method in step 3. Simultaneously, data on nine taste characteristics of the electronic tongue were collected for each sample using the method in step 4. A one-to-one correspondence dataset of "taste characteristic value - DPPH clearance rate" was established. Partial detection data are shown in the table below:

[0020] 6. Model Building and Performance Validation 6.1 Model Construction and Parameter Optimization: A regression prediction model was established using the least squares support vector machine (LS-SVM) algorithm. The above-mentioned "taste feature value-DPPH clearance rate" dataset was loaded into the LS SVMlabv1 8R2009b_R2011a software. During program execution, the dataset was randomly divided into a training set (88 groups) and a prediction set (43 groups) at a ratio of 3:1. The radial basis function (RBF) kernel function was selected. The key parameters of the model were optimized by leave-one-out cross-validation. The optimal parameter combination was finally determined to be regularization parameter gam=42.43 (to control model complexity and suppress overfitting) and kernel parameter sig2=2.07 (to optimize feature space mapping characteristics). 6.2 Model Performance Validation: The model was constructed using the training set, and its generalization ability was validated using the prediction set. The coefficient of determination (R²), root mean square error (RMSE), and relative analytical error (RPD) were used as the core evaluation metrics. Validation results showed: training set R²c = 0.99, RMSE = 1.66; prediction set R²c = 0.99, RMSE = 1.66. p =0.99, RPD=8.7 (RPD>8.0 indicates that the model has excellent detection accuracy); the fitting effect between the true and predicted values ​​on the training and prediction sets is as follows: Figure 3 As shown, the model's stability, reliability, and generalization ability all meet the requirements for batch detection.

[0021] 7. Rapid detection of DPPH scavenging rate of target samples 7.1 Sample pretreatment: Take the pear juice or pear paste sample to be tested and pretreat it according to the standardized sample preparation method in step 2 (centrifugation, filtration or dilution) to obtain the test solution that meets the testing requirements. 7.2 Electronic tongue feature data acquisition: Following the instrument debugging, preprocessing and detection process in step 4.2, collect nine electronic tongue taste feature data of the target sample and input them into the LS-SVM prediction model constructed in step 6. 7.3 Detection result output: The model automatically calculates and outputs the predicted DPPH clearance rate of the target sample; it can realize the rapid detection of batch samples.

[0022] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for rapidly determining the DPPH scavenging rate of pear juice and pear paste based on electronic tongue technology, characterized in that, The method includes the following steps: S1. Standardized Sample Preparation: Select fresh pears free from mechanical damage, pests, diseases, and spoilage. After washing, core removal, and cutting into pieces, juice is extracted. The crude juice is centrifuged to obtain the pear juice sample for testing. Alternatively, the extracted pear juice is filtered to remove impurities and then boiled using a gradient temperature control method until the sugar content reaches 70%. 0 Brix prepared pear paste, and then diluted and centrifuged the pear paste to obtain a pear paste sample for testing. S2. Electronic tongue feature data acquisition: Take the sample to be tested prepared in S1, place it in the special cup for electronic tongue detection, and use SA-402B type electronic tongue to detect and extract 9 taste feature values, including sour taste, bitter taste, astringent taste, bitter aftertaste, astringent aftertaste, umami taste, richness, salty taste, and sweet taste. S3. Construction of LS-SVM Prediction Model: A dataset was constructed using the nine taste features collected in S2 as independent variables and the DPPH clearance rate measured by spectrophotometry as the dependent variable. The least squares support vector machine (LS-SVM) algorithm was used to establish a regression prediction model. The dataset was randomly divided into training and prediction sets in a 3:1 ratio. A radial basis function kernel was selected, and the model parameters were optimized by leave-one-out cross-validation. The optimal parameter combination was determined to be regularization parameter gam=42.43 and kernel parameter sig2=2.

07. S4. Rapid detection of target sample: Take the pear juice or pear paste sample to be tested, prepare the test solution according to the standardized method in S1, and then collect the electronic tongue feature data according to the process in S2. Input the feature data into the LS-SVM prediction model constructed in S3. The model automatically outputs the predicted value of DPPH clearance rate.

2. The method according to claim 1, characterized in that, The specific preparation parameters for the pear juice sample in step S1 are as follows: place the extracted pear juice in an environment of 4 ℃ and centrifuge it at a speed of 6000 r / min for 10 min. After centrifugation, take the supernatant as the pear juice sample. The supernatant needs to be sealed and stored at 4 ℃ and the test should be completed within 24 h.

3. The method according to claim 1, characterized in that, The specific parameters for cooking the pear syrup in step S1 are as follows: First, filter the pear juice through a 100-mesh sieve to remove impurities, then cook it using a gradient temperature control method; the initial heat setting is 550 W, and cooking continues until the pear juice reaches a sugar content of 40%. 0 When Brix, adjust the heat to 350 W, and when the sugar content reaches 55%. 0 When braising, adjust the heat to 150 W and continue simmering until the pear juice reaches a sugar content of 70%. 0 Stop cooking when Brix is ​​reached, and you will get pear syrup.

4. The method according to claim 1, characterized in that, The specific preparation parameters for the pear paste sample in step S1 are as follows: take a sample with a sugar content of 70... 0 Brix pear syrup was diluted with deionized water at a volume ratio of 1:6 and centrifuged at 6000 r / min for 10 min at 4 ℃. After centrifugation, the supernatant was taken as the pear syrup sample to be tested. The supernatant was sealed and stored at 4 ℃ and the test was completed within 24 h.

5. The method according to claim 1, characterized in that, The sensor preprocessing and detection process of the electronic tongue in step S2 is as follows: Before detection, the sensor is first cleaned in the cleaning solution for 90 seconds, then cleaned in two reference solutions for 120 seconds each, and finally zeroed at the equilibrium position for 30 seconds; after reaching a stable working state, the test begins, and the initial taste value is output after 30 seconds of testing; then, it is cleaned in two sets of reference solutions for 3 seconds each, and the aftertaste is tested for 30 seconds after inserting a new reference solution; the test is repeated 4 times, the unstable data of the first test is discarded, and the average value of the last 3 test results is taken. The potential value is converted into the 9 taste characteristic values ​​by the system's built-in software; before detection, the sample to be tested needs to be placed in an environment of 20±1 ℃ for 10 minutes to equilibrate.

6. The method according to claim 1, characterized in that, The performance validation metrics and pass / fail criteria for the prediction model in step S3 are: training set coefficient of determination R²c = 0.99, root mean square error RMSEC = 1.66; prediction set coefficient of determination R²c = 0.

99. p The model is considered to have effective predictive ability if it meets the above indicators: ≥0.98 and relative analysis error RPD≥8.0.

Images