A method for analyzing the sensory crispness of puffed food

Abstract

The application discloses a method for analyzing the sensory crispness of puffed food, and belongs to the field of food analysis. The method takes puffed food as the object, and can accurately analyze the crispness of the sample through data collection and conversion, energy frequency interval segmentation and calculation of the low-frequency interval proportion. Compared with other acoustic characteristics, the method has wider applicability, is more accurate and reliable, and can be applied to predicting the sensory crispness difference of products under different puffing time or subtle formula difference when developing new puffed food, so as to further provide the basis for predicting the consumer response.

Description

Technical Field

[0001] This invention relates to a method for analyzing the sensory crispness of puffed foods, belonging to the field of food analysis. Background Technology

[0002] Sound, as a prominent feature of the consumption process of puffed foods, can be used to characterize the crispness of puffed foods. When food is cut, squeezed, deformed, and broken, it releases energy, which is transmitted to the human ear through a medium to produce sound. Therefore, the process of food making sound is essentially a process of energy transfer.

[0003] In recent years, many studies have focused on exploring the correlation between acoustic signals and food texture. For example, the paper "Study on the Detection of Crispness of Fragrant Pear Flesh Using a Synchronous Mechanism; DOI:10.27332 / d.cnki.gshzu.2021.000232." discloses an evaluation of the crispness of fragrant pears based on the characteristics of mechanoacoustic signals. This study used a texture analyzer and a microphone to simultaneously collect the mechanoacoustic response signals when the flesh of a fragrant pear cracks. Then, the peak value method, Fourier power spectrum method, and apparent fractal dimension method were used to measure the serration characteristics of the mechanoacoustic response curve. It was found that among the 43 serration characteristic parameters, 35 characteristic parameters were significantly correlated with the sensory crispness of fragrant pear flesh (P<0.05), which can reflect the changes in the crispness of fragrant pears. Principal component analysis was used to extract the first eight principal components, which explained 85.58% of the information of the original variables. A multiple linear regression model for predicting sensory crispness was constructed using these eight principal components. The model calibration set determination coefficient was 0.862 and the root mean square error was 0.588, indicating that the model has high prediction accuracy and stability, and can provide a research basis for the accurate instrument detection of crispness of pears.

[0004] Patent CN111272875A discloses the feasibility of non-destructive evaluation of apple crispness based on portable acoustic signals. This study uses sensory evaluation methods to assess apple crispness and compares it with non-destructive testing methods using portable acoustic signals to explore the feasibility of non-destructive evaluation of apple crispness based on portable acoustic signals. The acoustic feature values ​​of the acoustic signals are processed, and then the correlation between apple crispness evaluated by sensory evaluation and the acoustic feature values ​​is analyzed. Multiple linear regression (MLR) and artificial neural networks (ANN) are applied to predict apple crispness. The literature "Classification of puffed snacks freshness based on crispness-related mechanical and acoustical properties" discloses the detection of crispness of puffed foods through sound. Compared with traditional texture analysis techniques, the classification accuracy is improved, and an accuracy of up to 92% can be achieved using quadratic support vector machines or artificial neural network algorithms.

[0005] While the aforementioned technologies use acoustic signals to detect the crispness of food, in practical applications, different foods have different reference standards for their acoustic signals due to variations in cavity size, degree of crumbliness, and other factors. Furthermore, these methods are relatively cumbersome and time-consuming, making it difficult to quickly distinguish the crispness of different foods and obtain accurate results when analyzing products with different formulations. Therefore, to more widely utilize acoustic signals to assess the crispness of different types of food, more stable and accurate parameters or methods are needed to achieve precise and rapid identification of food crispness. Summary of the Invention

[0006] To address the aforementioned issues, this invention focuses on puffed foods. After denoising using wavelet transform, acoustic signal features are selected, and the energy range is segmented using Hilbert-Huang Transform (HHT). Based on this segmentation, a "low-frequency range" is defined. Compared to the intuitive basic parameters such as the number of sound peaks, sound pressure level, and waveform used in existing technologies, this invention first determines the energy concentration range of puffed foods. The energy concentration within this range characterizes their sensory crispness, offering broader applicability than other acoustic features and providing new ideas and technical support for developing more accurate and reliable acoustic evaluation methods.

[0007] The first objective of this invention is to provide a method for analyzing the crispness of puffed foods, the method comprising the steps of:

[0008] (1) Data acquisition and conversion: The sample was compressed using a texture analyzer, and the instantaneous frequency energy distribution was obtained from the sound signal of the sample breaking. The instantaneous frequency energy data was then converted into a time-frequency energy distribution map.

[0009] (2) Energy frequency interval segmentation: The energy interval of the time-frequency energy distribution map is segmented, the energy of each frequency band after segmentation is calculated, and the cumulative energy of each frequency band accounts for 80% to 85% of the total time-frequency energy as the threshold to divide the interval. The frequency bands above the threshold are high frequency intervals, and the rest are low frequency intervals.

[0010] (3) Calculate the proportion of the total energy of each frequency band in the low frequency range in the total time-frequency energy. The closer the proportion is to the threshold, the higher the brittleness of the sample.

[0011] In one implementation, the brittleness of the sample can also be characterized by a sensory score obtained from a sensory evaluation; the higher the score, the more brittle the sample.

[0012] In one embodiment, the sensing element of the texture analyzer is in the range of 800–1200 N when compressing the sample.

[0013] In one embodiment, the load force at the trigger point when the texture analyzer compresses the sample is 0.1 to 0.2 N.

[0014] In one embodiment, the velocity sensor moves at a speed of 400–600 mm / min when the texture analyzer compresses the sample.

[0015] In one implementation, the energy range of the time-frequency energy distribution map is divided into segments of 0–1500 Hz.

[0016] In one embodiment, puffed foods include: puffed rice, puffed millet, puffed corn, puffed potatoes, puffed fruit, puffed barley, puffed wheat, and puffed oats.

[0017] In one embodiment, the puffed food includes, but is not limited to, common puffed foods such as potato chips, seaweed rice cakes, Crispy Original Flavor French Fries, sweet potato chips, banana crisps, crispy shrimp chips, molten egg rolls, Wolong rice crust, Garden biscuits, and barley rice sticks.

[0018] In one implementation, MATLAB is used to convert instantaneous frequency energy data into a time-frequency energy distribution map.

[0019] In one embodiment, the method for analyzing the crispness of puffed food includes the steps of:

[0020] (1) Data acquisition and conversion: The sample was compressed to complete breakage using a texture analyzer. The parameters of the texture analyzer were set as follows: the range of the sensing element was 800-1200N, the load force of the trigger point was 0.1-0.2N, and the speed of the velocity sensor was set to 400-600mm / min. The Hilbert Huang Transform (HHT) analysis method was used to obtain the instantaneous frequency energy distribution of each intrinsic modal component (IMF) from the sound signal of the sample breakage through empirical mode decomposition (EMD). The instantaneous frequency energy distribution data was converted into a complete time-frequency energy distribution map using MATLAB.

[0021] (2) Energy frequency interval segmentation: The time-frequency energy distribution map (Hilbert spectrum) obtained in step (1) is divided into energy intervals of 0 to 20000Hz according to each segment of 0 to 1500Hz. The time-frequency energy of each frequency segment and the proportion of the cumulative time-frequency energy of the frequency segment in the total time-frequency energy are calculated. When the cumulative proportion exceeds 80% to 85% of the total time-frequency energy, the frequency segment above this threshold is defined as the high-frequency interval, and the rest are low-frequency intervals.

[0022] (3) Calculate the proportion of the total energy of each frequency band in the low-frequency range of the sample in the total energy. The closer the proportion of the sample is to the threshold, the higher the brittleness.

[0023] A second objective of this invention is to provide the application of the above-described method in detecting the crispness of puffed foods.

[0024] Beneficial effects

[0025] This invention focuses on puffed foods. After denoising using wavelet transform, acoustic signal features are selected, and the energy range is segmented using Hilbert-Huang Transform (HHT). Based on this, a "low-frequency range" is defined. Compared to the intuitive basic parameters such as the number of sound peaks, sound pressure level, and waveform in existing technologies, this invention first determines the energy concentration range of puffed foods. The energy concentration within this range characterizes their sensory crispness, offering broader applicability and greater accuracy than the acoustic features reported in existing technologies. This invention can be applied to the development of new puffed food products to predict subtle differences in formulation or differences in sensory crispness at different puffing times, thereby providing a basis for predicting consumer reactions. Detailed Implementation

[0026] The preferred embodiments of the present invention are described below. It should be understood that the embodiments are for better explanation of the present invention and are not intended to limit the present invention.

[0027] 1. Raw materials:

[0028] The following products were purchased from a supermarket: Want Want Rice Crackers (abbreviated as Rice Crackers), Seaweed Rice Cakes, Crispy Original Flavor Potato Fries, Sweet Potato Chips, Banana Crispy Rolls, Crispy Shrimp Chips, Molten Egg Rolls, Wolong Rice Crackers, Garden Biscuits, and Barley Rice Sticks.

[0029] 2. Experimental Methods

[0030] (1) Signal acquisition and decomposition methods:

[0031] The texture analyzer (TA.XT.Plus texture analyzer, Stable MicroSystems, Godalming, Surrey, UK) compresses the sample at a constant force rate until the sample is completely broken. A single-blade composite shear head is used to break the sample. The sensing element has a range of 1000N, the trigger point load force is 0.15N, and the speed sensor's movement speed is set to 480mm / min.

[0032] Sound signals were collected at 4cm using a 1 / 2-inch free-field microphone (iSV1610, Hangzhou Aiwa Instruments Co., Ltd., 10Hz-20kHz), and then converted into electrical signals for storage in a computer. An acoustic calibration instrument (Hangzhou Aiwa Instruments Co., Ltd., 94.0dB and 114.0dB SPL, 1000Hz) was used for calibration. Sound was recorded at 44100Hz (32-bit). The sound signals were stored and edited using Adobe Audition 2022, and signal processing was performed using MATLAB R2022b.

[0033] The Hilbert Huang Transform (HHT) consists of two parts: Empirical Mode Decomposition (EMD) and Hilbert Transform. EMD decomposes the signal based on its local features, breaking down the acquired acoustic signal into 10th-order Intrinsic Mode Functions (IMFs). By decomposing the signal into a superposition of several IMFs, it can better describe the dynamic changes of nonlinear and non-stationary signals.

[0034] (2) Sensory evaluation methods

[0035] According to ISO 8586-2023 (Sensory analysis - Selection and training of sensory assessors), 42 healthy candidates (male to female ratio, 1:1) (numbered 1-42) were randomly invited to rank the crispness of four different samples (A, B, C, and D represent brown rice rolls, rock slab grills, rice noodles, and crispy shark, respectively) through sensory evaluation.

[0036] Since crispness is not explicitly defined as a parameter, but rather a perceptible characteristic derived from hearing, a standard for evaluating sensory crispness was developed based on the consistency profile analysis of ISO 11035-1994 (Sensory Analysis – Identification and selection of sensory profile descriptors using a multidimensional approach). After extensive discussion among the judges, the evaluation criteria consisted of the pleasantness of the sound produced when the product breaks in the mouth, the desire to continue eating stimulated by the sound, and the freshness conveyed by the sound. Each item was scored out of 20 points, for a total of 60 points.

[0037] Referring to ISO 8587-2006 (Sensory analysis – Methodology – Ranking), the Friedman test was used to determine whether there were significant differences in the evaluation rankings of the candidates. The F-score for each candidate's evaluation result was calculated.

[0038]

[0039] In the formula, P is the sample size, J is the number of evaluators, and R... i The rank sum for each sample.

[0040] A new sample pool was formed by selecting candidates who demonstrated good discriminative ability for different fragility levels, and the evaluation results of the new sample pool were ranked by variance. Finally, a sensory evaluation panel was formed by selecting 10 subjects (male to female ratio of 1:1) whose evaluation results had the smallest deviation from the overall evaluation results.

[0041] Table 1. F-values ​​of sensory outcomes for 42 participants

[0042]

[0043]

[0044] Table 2. Variance of sensory crispness values ​​among 30 subjects

[0045] Subject number variance Subject number variance Subject number variance 2 106.92 19 298.00 31 57.67 3 44.67 20 15.33 32 18.67 7 148.92 22 92.67 34 135.33 8 230.92 23 82.92 35 25.67 9 130.25 24 4.92 36 156.33 10 256.92 25 160.33 37 120.92 11 69.67 26 236.92 38 30.25 16 138.25 27 276.92 39 202.92 17 54.25 28 278.25 40 210.92 18 84.92 30 2.92 41 267.58

[0046] Table 1 shows the F-values ​​of the sensory results from 42 subjects. According to the Friedman rank-sum test approximate critical value table (refer to ISO 8587-2006), the critical values ​​for P, J, and α (corresponding to values ​​of 4, 3, and 0.01 respectively) are 8.20. Therefore, 30 of the 42 subjects considered the four types of samples to have discernible differences in sensory fragility, indicating that the sample selection was successful. Table 2 shows the variance of sensory evaluations from the 30 subjects with good fragility discrimination ability. The selection was based on samples with minimal bias. Using 161.42 as a reference, 10 subjects (male to female ratio of 1:1) with smaller absolute values ​​of difference were selected to form a new sensory evaluation group.

[0047] (2) Preparation of samples with different brittleness

[0048] The scallop samples were placed in a constant temperature incubator (RH-LHP-300L, Runhua Electric Co., Ltd., Changzhou, China) to obtain samples with different humidity absorption rates. The incubator conditions were set as follows: temperature 20℃, humidity 65%, and storage times of 0, 12, 24, 36, 48, 60, and 72 hours (samples were named S0, S12, S24, S46, S48, S60, and S72, respectively).

[0049] Example 1: Rapid evaluation of sensory crispness of puffed foods based on energy analysis

[0050] 1. Taking scallops with different humidity levels as an example, the sensory crispness is quickly evaluated based on energy analysis. The steps are as follows:

[0051] (1) Data acquisition and conversion: The sample was compressed to complete breakage using a texture analyzer. The parameters of the texture analyzer were set as follows: the range of the sensing element was 1000N, the load force of the trigger point was 0.15N, and the speed of the velocity sensor was set to 480mm / min. The Hilbert Huang Transform (HHT) analysis method was used to obtain the instantaneous frequency energy distribution of each intrinsic modal component (IMF) from the sound signal of the sample breakage through empirical mode decomposition (EMD). The instantaneous frequency energy distribution data was converted into a complete time-frequency energy distribution map using MATLAB.

[0052] (2) Energy frequency interval segmentation: The time-frequency energy distribution map (Hilbert spectrum) obtained in step (1) is divided into 0-20000Hz energy intervals according to each segment of 1000Hz. The time-frequency energy of each frequency segment and the proportion of the cumulative time-frequency energy of the frequency segment in the total time-frequency energy are calculated. When the cumulative proportion exceeds 80% of the total time-frequency energy, the frequency segment above this threshold is defined as the high-frequency interval, and the rest are low-frequency intervals.

[0053] (3) Calculate the proportion of the total energy of each frequency band in the low frequency range of the sample in the total energy. The closer the proportion of the sample is to 80%, the higher the brittleness.

[0054] The results are shown in Table 3, which presents the relative energy proportions of scallops with different water absorption times in different frequency ranges under a humidity of 65%. Using scallop samples with 0 hours of moisture absorption as a baseline, and dividing the range by an energy threshold of 80%, the low-frequency range of scallops was determined to be 0-9000Hz. The total energy proportions of samples with different moisture absorption times in the low-frequency range (0-9000Hz) were calculated as S0 (0.805), S12 (0.826), S24 (0.841), S36 (0.873), S48 (0.883), S60 (0.893), and S72 (0.900). It can be seen that the energy proportion of the samples in the low-frequency range increases with increasing moisture absorption time, indicating that the energy released during food breakage in the high-frequency range decreases as brittleness decreases, and the released energy migrates to the low-frequency range. In other words, the higher the energy proportion of the sound signal in the low-frequency range, the lower the sensory crispness value of the food.

[0055] Table 3 Energy of scallop samples at different frequency bands under different moisture absorption times

[0056]

[0057] 2. Brittleness verification

[0058] The rice crackers with different levels of crispness were given to a sensory evaluation group for blind testing, and sensory scores were collected.

[0059] The results are shown in Table 4. The sensory scores of scallops S0 to S72 also decreased with the increase of moisture absorption time. The sensory score results are consistent with the results in 1, which showed a high proportion of low-frequency energy and low brittleness. This means that the rapid evaluation of sensory brittleness based on energy analysis has good accuracy.

[0060] Table 4 Sensory Evaluation Scores

[0061] sample Sensory score S0 55.90±3.77 S12 45.50±3.93 S24 38.00±3.73 S36 29.80±3.85 S48 20.80±2.63 S60 13.10±2.70 S72 10.70±2.73

[0062] Example 2: Rapid evaluation of sensory crispness value of puffed food based on energy analysis

[0063] Based on Example 1, the energy range was divided into segments of 500Hz, the proportion of low-frequency energy was calculated, and the brittleness of the sample was evaluated.

[0064] Table 5. Energy levels of scallop samples at different frequency bands under different moisture absorption times for each energy segment of the 500Hz frequency range.

[0065]

[0066]

[0067] The results are shown in Table 5. Dividing the energy range into 500Hz segments, the energy proportion of the sample in the low-frequency range increases with increasing moisture absorption time. This is consistent with Example 1. The energy released during food breakage in the high-frequency range decreases as brittleness decreases, and the released energy migrates to the low-frequency range. The higher the energy proportion of the acoustic signal in the low-frequency range, the lower the sensory brittleness of the food. In principle, smaller energy ranges yield more accurate results, but the computational load increases accordingly. The theoretical range of the energy range is 0–1000Hz; considering the computational burden, 500Hz is used as an example here.

[0068] Example 3: Low-frequency range detection of other puffed foods

[0069] Based on Example 1, the samples were changed to seaweed rice cakes, crispy original flavor fries, sweet potato chips, banana crisps, crispy shrimp chips, molten egg rolls, Wolong rice crust, Garden biscuits, and barley rice sticks, and the detection method was the same as in Example 1.

[0070] Table 6. Percentage of low-frequency range for different samples

[0071]

[0072] Table 6 shows the percentage of low-frequency energy range in common commercially available puffed foods. Sensory evaluation verified that using 80% as the energy threshold for analysis can effectively distinguish different levels of crispness among the same samples. In other words, using an 80% energy threshold to analyze the crispness of puffed foods has good universality.

[0073] In summary, this invention can be applied to the research and development of new puffed food products to predict subtle differences in formulation or differences in sensory crispness of products under different puffing times, thereby providing a basis for predicting consumer reactions.

[0074] Comparative Example 1: Using 70% as the threshold to divide the low-frequency range

[0075] Based on Example 1, the low-frequency range threshold was changed to 70%, the energy percentage was calculated, and the fragility was evaluated. The results are shown in Table 7.

[0076] Table 7 uses 70% as a threshold to divide the low-frequency range for evaluating sample fragility.

[0077]

[0078] Based on the sensory evaluation of samples S0 to S72 in Example 1, Table 4 shows that when using 70% as the energy threshold for brittleness evaluation, the low-frequency energy range and brittleness detected for samples with different humidity levels are not consistent, making it difficult to distinguish the brittleness values ​​between samples with different humidity levels.

[0079] Comparative Example 2: Energy range divided into segments of 2000Hz.

[0080] Based on Example 1, the energy range was divided into segments of 2000Hz, the proportion of low-frequency energy was calculated, and the brittleness value was evaluated. The results are shown in Table 8.

[0081] Table 8 divides the energy range by 2000Hz.

[0082]

[0083]

[0084] Based on the sensory evaluation of samples S0 to S72 in Example 1, Table 4 shows that when the energy range is divided into segments of 2000Hz, the proportion of the low-frequency range and the brittleness are not consistent, the evaluation of brittleness is inaccurate, and it is difficult to distinguish the brittleness values ​​between samples with different brittleness.

[0085] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make various modifications and alterations without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the claims.

Claims

1. A method for analyzing the crispness of puffed foods, characterized in that, The method includes the following steps: (1) Data acquisition and conversion: The texture analyzer was used to compress the puffed food sample, and the instantaneous frequency energy distribution was obtained from the sound signal of the sample breaking. The instantaneous frequency energy data was then converted into a time-frequency energy distribution map. (2) Energy frequency interval segmentation: The energy interval of the time-frequency energy distribution map is segmented, the energy of each frequency band after segmentation is calculated, and the cumulative energy of each frequency band accounts for 80% to 85% of the total time-frequency energy as the threshold to divide the interval. The frequency bands above the threshold are high frequency intervals, and the rest are low frequency intervals. (3) Calculate the proportion of the total energy of each frequency band in the low frequency range in the total time-frequency energy. The closer the proportion is to the threshold, the higher the brittleness of the sample.

2. The method according to claim 1, characterized in that, The parameters of the sensing element when the texture analyzer compresses the sample are 800-1200N.

3. The method according to claim 1, characterized in that, The load force at the trigger point when the texture analyzer compresses the sample is 0.1 to 0.2 N.

4. The method according to claim 1, characterized in that, The velocity sensor moves at a speed of 400–600 mm / min when the texture analyzer compresses the sample.

5. The method according to claim 1, characterized in that, The energy range of the time-frequency energy distribution map is divided into segments of 0–1500 Hz.

6. The method according to claim 1, characterized in that, Puffed foods include: puffed rice, puffed millet, puffed corn, puffed potatoes, puffed fruit, puffed barley, puffed wheat, and puffed oats.

7. The method according to claim 1, characterized in that, Use MATLAB to convert instantaneous frequency energy data into a time-frequency energy distribution map.

8. The application of the method according to any one of claims 1 to 7 in detecting the crispness of puffed foods.

Images