Biological perception-based evaluation method for digitalized taste information of cured meat

Abstract

This invention discloses a method for evaluating the digital taste information of cured meat based on biosensoring, belonging to the field of food testing technology. The method includes: acquiring the current response matrix and impedance spectrum matrix of the cured meat sample under a preset potential excitation sequence; identifying the characteristic response peaks corresponding to taste-active substances and mapping them to a taste attribute coordinate map to generate an initial taste fingerprint vector; performing segmented frequency domain integration on the impedance spectrum matrix to obtain the impedance accumulation within each taste attribute characteristic frequency band, and correcting the initial taste fingerprint vector accordingly to generate a corrected taste fingerprint vector; and performing similarity matching between the corrected taste fingerprint vector and the standard fingerprint vector in a standard taste fingerprint database to determine the taste level identifier. This method achieves objective quantification and level determination of the taste attributes of cured meat through multi-dimensional electrochemical signal acquisition and fusion.

Description

Technical Field

[0001] This invention relates to the field of food testing technology, specifically to a method for evaluating the digital taste information of cured meats based on biosensing. Background Technology

[0002] The evaluation of the taste quality of cured meats has traditionally relied on human sensory tasting. However, the evaluation results are significantly influenced by individual physiological states, experience differences, and psychological factors, making repeatability difficult to guarantee. With the development of instrumental analysis technology, existing electronic tongue systems use sensor arrays to collect electrochemical response signals, attempting to replace human senses for taste recognition. These existing technologies typically use fixed-potential scanning or pulse sequences to acquire response currents, classifying samples through feature extraction and pattern recognition algorithms. However, in practical applications, the composition of cured meat systems is complex. The matrix effect formed by the interaction of fats, protein degradation products, and various seasonings means that the current signal under a single excitation mode often only reflects the redox information of some active substances on the electrode surface, failing to fully characterize the differentiated effects of various taste-active substances such as acidity, sweetness, bitterness, saltiness, and umami on the electrode interface. Simultaneously, existing technologies do not fully utilize the ion conduction and interface structure information contained in the impedance spectrum, making it difficult to effectively separate matrix background interference when the intensity of taste attributes changes. This results in a deviation between the taste fingerprint vector and the actual taste perception, ultimately limiting the accuracy of taste rating determination. These shortcomings lead to two critical problems that urgently need to be addressed. First, how to simultaneously acquire electrochemical signals reflecting the multidimensional characteristics of taste-active substances in a single measurement, so that the extracted information can correspond to the material basis of different taste attributes, thereby constructing a digital fingerprint with sufficient discriminative power for differences in taste among samples. Second, how to introduce a signal dimension independent of the current response to compensate for fingerprint distortion caused by matrix effects, electrode state fluctuations, and other factors, so that the taste feature vector used for grading matching remains stable in repeated tests and can accurately distinguish subtle quality changes between adjacent taste grades. Summary of the Invention

[0003] This paper provides a digital taste information evaluation method for marinated meat based on biosensing, in order to solve the problems in the existing technology where a single signal dimension cannot fully represent taste attributes and is easily affected by matrix interference, resulting in insufficient accuracy and repeatability of taste rating.

[0004] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a method for evaluating the digital taste information of cured meat based on biosensing, the method comprising:

[0005] In response to the excitation response signal acquisition operation of the marinated meat sample to be tested, the current response matrix and impedance spectrum matrix of the marinated meat sample under a preset potential excitation sequence are obtained, wherein the preset potential excitation sequence includes multiple frequency segments and multiple potential amplitude levels. Preferably, the marinated meat sample to be tested is placed in a three-electrode electrolytic cell, and each potential excitation signal in the potential excitation sequence is applied to the working electrode. The potential excitation sequence is arranged from low to high frequency, and each frequency segment contains at least three potential amplitude levels. During the application of each potential excitation signal, the response current signal between the working electrode and the counter electrode is synchronously acquired at a predetermined sampling frequency, and the current amplitude at each sampling time is arranged in chronological order to form a current response matrix. After the application of each potential excitation signal, an AC perturbation signal with a constant amplitude is applied to the working electrode, and the scanning frequency range covers the characteristic frequency bands corresponding to each taste attribute. The impedance amplitude and impedance phase angle of the working electrode are acquired to construct the impedance spectrum matrix. By combining multi-frequency segmentation and multi-potential amplitude levels, the excitation mode can fully stimulate different types of taste-active substances in the test sample to produce differentiated electrochemical responses. At the same time, the obtained current response matrix and impedance spectrum matrix provide a comprehensive signal foundation for subsequent multi-dimensional information fusion.

[0006] Based on the current amplitude jump characteristics at each sampling time point in the current response matrix, the characteristic response peak group corresponding to the taste-active substances in the marinated meat sample to be tested is identified. This characteristic response peak group includes peak potential, peak height current value, and peak width potential range. As a further improvement of the present invention, a second-order difference operation is performed on the current amplitude sequence of each row in the current response matrix to obtain the current amplitude change rate sequence, and the sampling time points that exceed the preset change rate threshold are marked as candidate peak points. Expanding to the left and right sides with each candidate peak point as the center, the left boundary potential and right boundary potential corresponding to the current amplitude decreasing to a predetermined proportion of the current amplitude at the candidate peak point are found respectively, and this interval is determined as the peak width potential range of a single characteristic response peak. Characteristic response peaks with peak height current values ​​greater than the preset peak height threshold are selected from all single characteristic response peaks, and the selected characteristic response peaks are arranged in ascending order of peak potential to form a characteristic response peak group. The above processing method can effectively eliminate the interference of false peaks caused by noise and baseline drift, accurately extract the characteristic current jump information closely related to the oxidation-reduction process of taste active substances, and make the identification results of characteristic peak groups have high chemical selectivity and reproducibility.

[0007] The characteristic response peaks are mapped onto a preset taste attribute coordinate map to generate an initial taste fingerprint vector for the marinated meat sample to be tested. The initial taste fingerprint vector contains quantified values ​​of the sour, sweet, salty, bitter, and umami response intensities. Specifically, the first potential coordinate interval corresponding to the sour attribute, the second potential coordinate interval corresponding to the sweet attribute, the third potential coordinate interval corresponding to the salty attribute, the fourth potential coordinate interval corresponding to the bitter attribute, and the fifth potential coordinate interval corresponding to the umami attribute are obtained from the preset taste attribute coordinate map. The peak potential of each characteristic response peak in the characteristic response peak group is assigned to each of the above potential coordinate intervals to determine the taste attribute category to which each characteristic response peak belongs. The peak height current values ​​of all characteristic response peaks belonging to the same taste attribute category are accumulated, and the accumulated value is used as the quantified value of the response intensity of that taste attribute. The quantified values ​​of the response intensity of all taste attributes are combined to form the initial taste fingerprint vector. By mapping peak potential space to taste attribute coordinate map, the physiological taste meaning of complex electrochemical response signals was analyzed, and the originally abstract current peak group was directly transformed into five basic taste intensity values ​​that match human tongue perception, thus initially establishing the "signal-taste" correspondence.

[0008] The impedance spectrum matrix is ​​subjected to segmented frequency domain integration to obtain the cumulative impedance of the marinated meat sample in the characteristic frequency bands corresponding to each taste attribute. The initial taste fingerprint vector is then corrected based on the cumulative impedance to generate a corrected taste fingerprint vector. In a preferred embodiment, impedance amplitude sequences are extracted from the impedance spectrum matrix for the first characteristic frequency band corresponding to sourness, the second characteristic frequency band corresponding to sweetness, the third characteristic frequency band corresponding to saltiness, the fourth characteristic frequency band corresponding to bitterness, and the fifth characteristic frequency band corresponding to umami. A trapezoidal integral is performed on the impedance amplitude sequence in each characteristic frequency band with a frequency step size to obtain the cumulative impedance corresponding to each taste attribute. Each cumulative impedance represents the impedance energy contribution value of the corresponding taste attribute in the frequency domain. Subsequently, the impedance accumulation corresponding to each taste attribute is compared with the preset impedance benchmark threshold of the corresponding taste attribute to obtain the impedance correction coefficient for each taste attribute. The quantized response intensity value of each taste attribute in the initial taste fingerprint vector is multiplied by the impedance correction coefficient of the corresponding taste attribute to obtain the corrected response intensity value. All corrected response intensity values ​​are combined to form the corrected taste fingerprint vector. The frequency domain impedance accumulation reflects the hindering effect of the sample bulk solution and matrix on charge transport. Introducing it into the correction process can effectively compensate for the current response deviation caused by matrix factors such as differences in sample conductivity and fat content. This makes the corrected taste fingerprint vector more faithfully reflect the actual content and flavor contribution of taste active substances, significantly improving the accuracy of evaluation and batch-to-batch comparability.

[0009] The modified taste fingerprint vector is matched with multiple standard fingerprint vectors in the standard taste fingerprint database for similarity, and the taste level identifier of the marinated meat sample to be tested is determined based on the matching results. The preferred similarity matching calculation process is as follows: the Euclidean distance between the modified taste fingerprint vector and each standard fingerprint vector in the standard taste fingerprint database is calculated, and the product of the magnitude of each standard fingerprint vector and the magnitude of the modified taste fingerprint vector is calculated. The Euclidean distance is divided by the product to obtain the initial similarity value. The initial similarity value is normalized so that the normalized similarity value is distributed within a predetermined interval. The normalized similarity value is used as the final similarity score between the modified taste fingerprint vector and each standard fingerprint vector. This similarity calculation method considers both spatial distance and directional differences between vectors, is sensitive to the overall contour and relative proportion of the taste fingerprint vector, and can more finely distinguish subtle differences.

[0010] When determining the taste level identifier, the highest similarity score is selected from all final similarity scores, and the preset taste level label associated with the standard fingerprint vector corresponding to the highest similarity score is determined. When the highest similarity score is greater than or equal to a first threshold, the preset taste level label is directly determined as the taste level identifier of the marinated meat sample to be tested. When the highest similarity score is less than the first threshold but greater than or equal to a second threshold, the preset taste level labels corresponding to the highest similarity score and its adjacent similarity scores are weighted and fused to generate a fused taste level identifier. Specifically, the preset taste level label corresponding to the highest similarity score can be assigned a first weight coefficient, and the preset taste level label corresponding to the largest value among the similarity scores adjacent to the highest similarity score can be assigned a second weight coefficient. The first weight coefficient is greater than the second weight coefficient. The level quantization value of each preset taste level label is multiplied by its corresponding weight coefficient and then added together to obtain the fused level quantization value. The preset taste level label with the closest fused level quantization value is determined as the fused taste level identifier. When the highest similarity score is less than the second threshold, the taste level identifier of the marinated meat sample to be tested is output as unrecognizable. This classification strategy balances the certainty of matching results with the flexibility of handling fuzzy boundary samples. It ensures rapid and accurate classification of high-confidence samples, provides reasonable and traceable fusion evaluation results for critical samples, and avoids misclassification of abnormal samples.

[0011] The construction of the standard taste fingerprint database is achieved simultaneously through the aforementioned methodology: Standard current response matrices and standard impedance spectrum matrices of multiple standard cured meat samples with known taste levels are collected under the preset potential excitation sequence; following the same processing method as the cured meat samples to be tested, standard feature response peak groups are extracted from the standard current response matrices and mapped to generate standard initial taste fingerprint vectors for each standard cured meat sample; then, the standard initial taste fingerprint vectors are corrected according to the standard impedance spectrum matrix in the same manner to obtain the standard fingerprint vectors corresponding to each standard cured meat sample, and these vectors are associated with known taste level labels and stored in the standard taste fingerprint database. The database established in this way adopts a completely consistent feature extraction and correction process with the detection method, ensuring the uniformity of the coordinate system for similarity matching from the source and effectively avoiding systematic biases introduced by differences in processing methods.

[0012] This invention organically integrates the current jump characteristic peak group obtained under potential excitation scanning with the impedance spectrum energy accumulation information, simulates the biological taste perception process from the electrochemical response mechanism level, and makes targeted compensation for matrix effects. Finally, it outputs an objective and quantitative taste level label through multi-level similarity matching, providing a stable, accurate and easily reproducible digital taste evaluation solution for cured meat products.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0014] By constructing a preset potential excitation sequence containing multiple frequency segments and multiple potential amplitude levels, the current response matrix of the marinated meat sample to be tested is obtained. Second-order difference operations and peak position identification are performed on this matrix to extract characteristic response peak groups caused by the redox reactions of gustatory active substances, including three types of parameters: peak potential, peak height current value, and peak width potential range. The characteristic response peak groups are mapped to a preset taste attribute coordinate map. The peak potential of each characteristic response peak is assigned to one of the five taste attributes (sour, sweet, salty, bitter, and umami) based on their corresponding potential coordinate ranges. The peak height current values ​​of characteristic response peaks belonging to the same taste attribute are accumulated to generate an initial taste fingerprint vector containing quantified values ​​of the five taste response intensities. This mapping method directly decouples the complex current signal into a digital indicator with a clear taste meaning, ensuring that each dimension of the initial fingerprint vector corresponds one-to-one with the sensory taste attribute, avoiding the taste category confusion caused by the lack of physical interpretability of features in conventional methods. Simultaneously, the segmented excitation of potential amplitude and frequency allows the responses of different taste-active substances to be fully developed within different potential ranges. Even substances with low content or similar redox potentials can form distinguishable characteristic response peaks, thereby improving the resolution of taste-active components. Based on the synchronously acquired impedance spectrum matrix, characteristic frequency bands are divided for each taste attribute. The impedance amplitude sequence within each frequency band is extracted and subjected to trapezoidal integration to obtain the impedance accumulation corresponding to each taste attribute. This impedance accumulation characterizes the energy contribution of the sample to charge transport and interfacial polarization processes within a specific frequency range. Its magnitude is comprehensively modulated by factors such as the electrolyte concentration, viscosity, and adsorption state of the electrode surface in the sample micro-region. The impedance accumulation is compared with a preset impedance reference threshold to obtain an impedance correction coefficient. This coefficient is then used to weight and correct the quantized response intensity values ​​of each taste attribute in the initial taste fingerprint vector, generating a corrected taste fingerprint vector. The introduction of impedance spectrum information is independent of the current response path and can independently reflect the overall influence of matrix differences on the electrode interface process, effectively correcting signal drift caused by changes in accompanying sample components, electrode surface contamination, or temperature fluctuations. The corrected taste fingerprint vector exhibits significantly enhanced consistency across different batches of samples, and the differences in fingerprint vectors between adjacent taste level samples are amplified, providing more reliable input features for subsequent similarity matching. Therefore, based on the matching results between the corrected vector and the standard taste fingerprint database, samples are assigned taste level labels. When the highest similarity score falls within the fuzzy range, a weighted fusion is used to generate a fused level label. This allows the taste evaluation results to distinguish clearly different quality categories while also reasonably handling samples at the level boundary, improving the accuracy and practicality of digital taste evaluation. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0016] Figure 1 This is a flowchart of a method for evaluating the digital taste information of cured meats based on biosensing.

[0017] Figure 2 This is a flowchart for obtaining the current response matrix and impedance spectrum matrix of a cured meat sample;

[0018] Figure 3 This is a flowchart of the standard taste fingerprint database construction process;

[0019] Figure 4 This is a spectrum of impedance amplitude and impedance phase angle for a sample of cured meat.

[0020] Figure 5 This is a distribution diagram of the peak potential and peak height current value of the characteristic response matrix;

[0021] Figure 6 This is a schematic diagram of the impedance spectrum and taste attribute frequency bands of a sample of cured meat. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] See Figure 1This invention provides a method for evaluating the digital taste information of cured meat based on biosensing. The method responds to the acquisition of excitation response signals from a sample of cured meat to be tested, obtaining the current response matrix and impedance spectrum matrix of the sample under a preset potential excitation sequence. The preset potential excitation sequence includes multiple frequency segments and multiple potential amplitude levels. Based on the current amplitude jump characteristics at each sampling time point in the current response matrix, characteristic response peak groups corresponding to the taste-active substances in the sample are identified. These peak groups include peak potential, peak height current value, and peak width potential range. The characteristic response peak groups are mapped to a preset taste attribute coordinate map to generate an initial taste fingerprint vector for the sample. The impedance spectrum matrix is ​​then subjected to segmented frequency domain integration to obtain the cumulative impedance of the sample within the characteristic frequency bands corresponding to each taste attribute. The initial taste fingerprint vector is then corrected based on the cumulative impedance to generate a corrected taste fingerprint vector. The modified taste fingerprint vector is matched with multiple standard fingerprint vectors in the standard taste fingerprint database for similarity, and the taste level label of the marinated meat sample to be tested is determined based on the matching results.

[0024] Example 1:

[0025] In specific implementation, please refer to Figure 2 The method for obtaining the current response matrix and impedance spectrum matrix of the marinated meat sample under a preset potential excitation sequence is described below:

[0026] The cured meat sample to be tested is placed in a three-electrode electrolytic cell, which includes a working electrode, a counter electrode, and a reference electrode. An inert metal electrode is used as the working electrode, a platinum wire electrode as the counter electrode, and a saturated calomel electrode as the reference electrode. After homogenization, the cured meat sample is formed into a paste, which is then filled into the sample compartment of the three-electrode electrolytic cell, ensuring contact between the paste and the surface of the working electrode to form a conductive path.

[0027] A preset potential excitation sequence is applied to the working electrode. The preset potential excitation sequence consists of multiple frequency segments, each containing at least three potential amplitude levels. There are five frequency segments: the first, second, third, fourth, and fifth frequency segments, with the frequency range gradually increasing. The application order is from low to high frequency. Each frequency segment contains three potential amplitude levels: the first, second, and third potential amplitudes, with the absolute values ​​of these amplitudes increasing sequentially. Each potential excitation signal is a step potential signal, and the duration of the step potential signal is set based on the period of the lowest frequency within the corresponding frequency segment.

[0028] During each application of a potential excitation signal, the response current signal between the working electrode and the counter electrode is synchronously acquired at a predetermined sampling frequency. The predetermined sampling frequency is set based on the highest frequency component contained in the potential excitation signal, and the predetermined sampling frequency is ten times the highest frequency component. The current amplitudes at each sampling moment are arranged in chronological order to form a row vector. The row vectors corresponding to all potential excitation signals are superimposed in the application order to generate a current response matrix. The element in the a-th row and b-th column of the current response matrix represents the current amplitude acquired at the b-th sampling moment during the application of the a-th potential excitation signal.

[0029] After each potential excitation signal is applied, an AC impedance measurement is performed. A constant-amplitude AC perturbation signal is applied to the working electrode, with the amplitude set to 7mV. The frequency scanning range of the AC perturbation signal covers the characteristic frequency bands corresponding to each taste attribute, including sour, sweet, salty, bitter, and umami. The characteristic frequency bands corresponding to each taste attribute are frequency ranges obtained in advance through standard taste substances. During the frequency scanning process, the impedance amplitude and impedance phase angle of the working electrode are simultaneously acquired, obtaining an impedance amplitude and an impedance phase angle for each scanning frequency point. The impedance amplitudes of all scanning frequency points are arranged in one column, and the impedance phase angles in another column, arranged in ascending order of scanning frequency to construct an impedance spectrum matrix. The element in the c-th row and 1-th column of the impedance spectrum matrix is ​​the impedance amplitude corresponding to the c-th scanning frequency point, and the element in the c-th row and 2-th column is the impedance phase angle corresponding to the c-th scanning frequency point.

[0030] When an AC disturbance signal is applied, the amplitude of the AC disturbance signal is maintained at 7mV. The basis for selecting 7mV is that, under a disturbance amplitude of 7mV, the current response and potential disturbance of the marinated meat sample to be tested in the three-electrode electrolytic cell system satisfy a linear relationship. The impedance measurement results will not produce harmonic distortion due to excessive disturbance, nor will the signal-to-noise ratio be insufficient due to insufficient disturbance.

[0031] In some embodiments, the sample cell volume of the three-electrode electrolytic cell is 5 ml, and the paste filling amount accounts for 80% of the sample cell volume to ensure that the paste completely covers the working electrode surface and forms a stable three-phase interface.

[0032] In constructing the impedance spectrum matrix, the frequency scanning step size adopts a logarithmic distribution, with a small scanning step size in the low-frequency band and a large scanning step size in the high-frequency band, thereby obtaining a uniformly distributed number of measurement points throughout the scanning range. The lower limit of the scanning frequency range is 0.1Hz, and the upper limit is 1MHz.

[0033] It can be understood that the number of rows in the current response matrix is ​​equal to the total number of potential excitation signals in the preset potential excitation sequence. The total number of potential excitation signals is determined by the product of the number of frequency segments and the number of potential amplitude levels within each frequency segment. For example, with five frequency segments and three potential amplitude levels within each frequency segment, the total number of potential excitation signals is fifteen, and the current response matrix has fifteen rows. The number of rows in the impedance spectrum matrix is ​​determined by the number of frequency scan points, with a typical value of ten points per octave.

[0034] See Figure 4 This figure shows the characteristic curves of impedance amplitude and impedance phase angle as a function of frequency during AC impedance measurement of the cured meat sample in the range of 0.1 Hz to 1 MHz after the application of a preset potential excitation sequence. The horizontal axis represents the scanning frequency (unit: Hz) on a logarithmic scale, the left vertical axis corresponds to the impedance amplitude (unit: Ω), and the right vertical axis corresponds to the impedance phase angle (unit: °). In the figure, the blue solid line represents the impedance amplitude, and the red dashed line represents the impedance phase angle.

[0035] As the impedance amplitude curves show, as the scanning frequency gradually increases from 0.1 Hz to approximately 100 Hz, the impedance amplitude rapidly decreases from approximately 1.6 × 10^6 Ω to a lower level, exhibiting a clear exponential decreasing trend. This indicates that the cured meat sample has a strong ability to impede AC disturbance signals in the low-frequency region. Above 100 Hz, the impedance amplitude approaches zero and remains stable, indicating that the impedance characteristics of the electrochemical interface tend to stabilize in the high-frequency range.

[0036] The impedance phase angle curve remained at approximately 90° in the low-frequency range (0.1 Hz to approximately 10 Hz), then gradually increased with increasing frequency, rapidly approaching 0° from a negative value around 1 kHz, and stabilizing at around 0° in the high-frequency range (>1 kHz). This phase angle change reflects the transition of the electrochemical interface of the cured meat sample from capacitive to resistive characteristics.

[0037] Based on the characteristic frequency bands of taste attributes defined in Example 3, the impedance amplitude range of 10Hz to 100Hz in the figure corresponds to the first characteristic frequency band of sour taste, 100Hz to 1kHz corresponds to sweet taste, 1kHz to 10kHz corresponds to salty taste, 10kHz to 100kHz corresponds to bitter taste, and 100kHz to 1MHz corresponds to umami taste. The impedance amplitude shows a significant decreasing trend within the characteristic frequency band of sour taste, indicating that the sample has a high impedance to AC signals in this frequency band, which is beneficial for obtaining the cumulative impedance related to sour taste through integration. The phase angle gradually approaches zero from negative in this frequency band, reflecting changes in electrochemical interface characteristics.

[0038] Example 2:

[0039] In practice, the process of identifying the characteristic response peaks of the taste-active substances in the marinated meat sample based on the current amplitude jump characteristics at each sampling time point in the current response matrix is ​​as follows.

[0040] Performing a second-order difference operation on each row of the current amplitude sequence in the current response matrix yields the current amplitude change rate sequence. The current response matrix contains the... Line number The current amplitude at each sampling time point is denoted as , for the The calculation method for the second-order difference operation of the row current amplitude sequence is as follows:

[0041]

[0042] in, Representing the Line number Second-order difference of current amplitude at each sampling time point Representing the Line number Current amplitude at each sampling time point Representing the Line number Current amplitude at each sampling time point Representing the Line number Current amplitude at each sampling time point The value ranges from 1 to the total number of rows in the current response matrix. The value range is from 2 to the 1st. Decrease the total number of sampling time points by 1. (The last part is incomplete and likely refers to a separate step.) All calculations obtained according to Arranged in ascending order, forming the [number]th [order]. The corresponding current amplitude change rate sequence is generated. Sampling time points in the current amplitude change rate sequence that exceed a preset change rate threshold are marked as candidate peak points. The preset change rate threshold is determined based on the global mean of the current amplitude in the current response matrix, and the preset change rate threshold is set to five percent of the global mean of the current amplitude.

[0043] Expanding outwards from each candidate peak point, the left and right boundary potentials are determined when the current amplitude drops to a predetermined percentage of the current amplitude at the candidate peak point. The predetermined percentage is set to 50%. Sampling time points are searched sequentially to the left of the candidate peak point. When the current amplitude at a sampling time point on the left is less than or equal to 50% of the current amplitude at the candidate peak point, the potential value corresponding to that sampling time point is recorded as the left boundary potential. Similarly, sampling time points are searched sequentially to the right of the candidate peak point. When the current amplitude at a sampling time point on the right is less than or equal to 50% of the current amplitude at the candidate peak point, the potential value corresponding to that sampling time point is recorded as the right boundary potential. The potential interval between the left and right boundary potentials is defined as the peak width potential interval of a single characteristic response peak. The potential value at the candidate peak point is recorded as the peak position potential of the characteristic response peak, and the current amplitude at the candidate peak point is recorded as the peak height current value of the characteristic response peak.

[0044] From all individual characteristic response peaks, select those with peak height current values ​​greater than a preset peak height threshold. The preset peak height threshold is set based on the background noise level of all current amplitudes in the current response matrix. The background noise level is set to five times the standard deviation of the baseline current before the application of the potential excitation signal. When the peak height current value of an individual characteristic response peak is greater than the preset peak height threshold, that individual characteristic response peak is retained; when the peak height current value of an individual characteristic response peak is less than or equal to the preset peak height threshold, that individual characteristic response peak is discarded. Arrange the selected characteristic response peaks in ascending order of peak potential to form a characteristic response peak group. If the number of selected characteristic response peaks is zero, the characteristic response peak group is an empty set.

[0045] In some embodiments, the process of obtaining the potential coordinate intervals corresponding to each taste attribute in a preset taste attribute coordinate map is as follows: Several standard taste substance solutions are selected and potential scanned. These standard taste substance solutions include citric acid solution, sucrose solution, sodium chloride solution, caffeine solution, and monosodium glutamate solution, corresponding to sour, sweet, salty, bitter, and umami attributes, respectively. The response currents of each standard taste substance solution under a preset potential excitation sequence are used to extract characteristic peaks, obtaining the peak potential distribution range of the characteristic response peaks for each standard taste substance solution. This distribution range is then expanded by ±10% to form the potential coordinate intervals corresponding to the taste attribute. The first potential coordinate interval corresponding to sourness, the second potential coordinate interval corresponding to sweetness, the third potential coordinate interval corresponding to saltiness, the fourth potential coordinate interval corresponding to bitterness, and the fifth potential coordinate interval corresponding to umami are calibrated in this manner.

[0046] Optionally, the method for determining the interval affiliation of each characteristic response peak in the characteristic response peak group with the first, second, third, fourth, and fifth potential coordinate intervals is as follows: when the peak potential of a characteristic response peak falls within the first potential coordinate interval, the characteristic response peak is determined to belong to the sour taste category; when the peak potential of a characteristic response peak falls within the second potential coordinate interval, the characteristic response peak is determined to belong to the sweet taste category; when the peak potential of a characteristic response peak falls within the third potential coordinate interval, the characteristic response peak is determined to belong to the salty taste category; when the peak potential of a characteristic response peak falls within the fourth potential coordinate interval, the characteristic response peak is determined to belong to the bitter taste category; and when the peak potential of a characteristic response peak falls within the fifth potential coordinate interval, the characteristic response peak is determined to belong to the umami taste category. When the peak potential of a characteristic response peak falls within the overlapping region of two or more potential coordinate intervals, the distance between the peak potential and the center value of each potential coordinate interval corresponding to the overlapping region is calculated, and the characteristic response peak is assigned to the taste attribute category corresponding to the potential coordinate interval with the smallest distance. When the peak potential of a characteristic response peak does not fall within any potential coordinate interval, the characteristic response peak is marked as an unassigned characteristic response peak, and unassigned characteristic response peaks are not included in subsequent accumulation calculations.

[0047] The peak height current values ​​of all feature response peaks belonging to the same taste attribute category are summed. Taking the sour taste attribute category as an example, assuming there are a total of feature response peaks belonging to the sour taste attribute category... The, the The peak height current value of the characteristic response peak belonging to the sour taste category is denoted as Sourness response intensity The calculation method is as follows:

[0048]

[0049] The sweet, salty, bitter, and umami response intensities are calculated using the same method as the sour response intensity, by summing the peak height current values ​​of all characteristic response peaks belonging to the corresponding taste attribute category. The combined sour, sweet, salty, bitter, and umami response intensities form a five-dimensional vector, which is the initial taste fingerprint vector of the marinated meat sample to be tested.

[0050] It can be understood that when the number of feature response peaks belonging to a certain taste attribute category is zero, the quantized value of the response intensity corresponding to that taste attribute is recorded as zero.

[0051] In some embodiments, if the feature response peak group is an empty set, then the response intensity quantization value of all taste attributes in the initial taste fingerprint vector is zero.

[0052] See Figure 5 In the figure, the horizontal axis represents the peak potential of the characteristic response peak, in volts (V), ranging from 0 to 1V; the vertical axis represents the peak current value of the characteristic response peak, in microamperes (μA), ranging from 0.5 to 3.0 μA. Each dot in the figure represents a single characteristic response peak extracted from the current response matrix. The horizontal and vertical error bars of the dots correspond to the uncertainty ranges of the peak potential and peak current values, respectively, reflecting the measurement fluctuations and processing errors in the characteristic response peak identification process. The "characteristic response peak" marked in the legend corresponds to the effective peak point obtained in Example 2 by performing second-order difference operations and threshold screening on the current amplitude sequence of the current response matrix.

[0053] A dashed line is drawn at the lower end of the peak potential axis in the figure, labeled "Preset Peak Height Threshold". Its vertical axis value is approximately 0.3 μA, representing the peak height current value threshold for screening characteristic response peaks according to Example 2. This threshold is set based on the background noise level of the current response matrix and is used to eliminate noise peaks smaller than this value, ensuring the accuracy of the characteristic response peaks.

[0054] From the overall distribution trend, as the peak potential gradually increases from 0.1V to 0.8V, the peak height current value shows a significant upward trend, increasing from approximately 0.85μA to approximately 2.8μA. This indicates that the response intensity of the taste-active substances in the tested marinated meat samples varies significantly under different potential excitations. The lateral error bars are generally less than 0.1V, and the longitudinal error bars are generally less than 0.2μA, indicating good measurement stability of the peak potential and peak height current values, and high repeatability of the characteristic response peak extraction.

[0055] The peak height current values ​​of the characteristic response peaks in the range of approximately 0.1V to 0.4V are relatively dispersed, with some peaks approaching the preset peak height threshold, reflecting a weak or diverse response of gustatory active substances within this potential range. The peak height current values ​​of the characteristic response peaks in the range of 0.45V to 0.8V are generally higher and more concentrated, and the peak height current values ​​show an increasing trend with increasing peak potential. This indicates that electrochemical excitation in this potential range is more sensitive to the response of gustatory active substances, facilitating subsequent mapping to a taste attribute coordinate map to generate a more representative initial taste fingerprint vector.

[0056] Example 3:

[0057] In practice, the impedance spectrum matrix is ​​subjected to segmented frequency domain integration to obtain the impedance accumulation of the marinated meat sample in the characteristic frequency bands corresponding to each taste attribute. The initial taste fingerprint vector is then corrected based on the impedance accumulation. The process of generating the corrected taste fingerprint vector is as follows.

[0058] The impedance amplitude sequences corresponding to the following five characteristic frequency bands are extracted from the impedance spectrum matrix: the first characteristic band for sourness, the second characteristic band for sweetness, the third characteristic band for saltiness, the fourth characteristic band for bitterness, and the fifth characteristic band for umami. Each row in the impedance spectrum matrix corresponds to a scanning frequency point, and the first column stores the impedance amplitude. The first, second, third, fourth, and fifth characteristic frequency bands are pre-calibrated frequency ranges: the first characteristic band for sourness is 10Hz to 100Hz, the second characteristic band for sweetness is 100Hz to 1kHz, the third characteristic band for saltiness is 1kHz to 10kHz, the fourth characteristic band for bitterness is 10kHz to 100kHz, and the fifth characteristic band for umami is 100kHz to 1MHz. When extracting the impedance amplitude sequence, all rows of the impedance spectrum matrix are traversed. When the scanning frequency point of a certain row falls into the first characteristic frequency band, the impedance amplitude of the first column of that row is included in the impedance amplitude sequence corresponding to the sour taste attribute. In this way, the impedance amplitude sequences corresponding to each of the five taste attributes are constructed respectively.

[0059] By performing a trapezoidal integral over a frequency step on the impedance amplitude sequence within each characteristic frequency band, the cumulative impedance corresponding to each taste attribute is obtained. Taking sourness as an example, the impedance amplitude sequence corresponding to sourness includes... The impedance amplitude data points are sorted in ascending order of frequency. The frequency value corresponding to each impedance amplitude data point is denoted as . , No. The impedance amplitude corresponding to each impedance amplitude data point is denoted as . ,in The value range is 1 to The cumulative impedance corresponding to the sour taste property. The calculation method is as follows:

[0060]

[0061] in, This represents the cumulative impedance corresponding to the sour taste attribute. This represents the total number of data points in the impedance amplitude sequence corresponding to the sour taste attribute. The first in the impedance amplitude sequence representing the sour taste attribute Each impedance magnitude, The first in the impedance amplitude sequence representing the sour taste attribute Each impedance magnitude, The first in the impedance amplitude sequence representing the sour taste attribute The frequency value corresponding to each impedance amplitude data point The first in the impedance amplitude sequence representing the sour taste attribute The frequency values ​​corresponding to each impedance amplitude data point. The trapezoidal integral approximates the change in impedance amplitude between two adjacent frequency points as a linear change, and the integration region between adjacent frequency points approximates a trapezoid. The area of ​​the trapezoid is obtained by multiplying the average of the two adjacent impedance amplitudes by the frequency interval. The cumulative impedance values ​​corresponding to sweet, salty, bitter, and umami attributes are calculated using the same method as those corresponding to sourness, obtained by performing trapezoidal integration within the characteristic frequency band corresponding to each taste attribute with a frequency step size. Each cumulative impedance value represents the impedance energy contribution value of the corresponding taste attribute in the frequency domain.

[0062] The impedance correction coefficient for each taste attribute is obtained by comparing the cumulative impedance value corresponding to that attribute with the preset impedance threshold value. The preset impedance threshold value for the sour taste attribute is denoted as... The impedance correction factor for the sour taste property The calculation method is as follows The preset impedance reference thresholds for sweetness, saltiness, bitterness, and umami are respectively denoted as follows: , , and The impedance correction coefficient for each taste attribute is obtained by dividing the cumulative impedance of the corresponding taste attribute by the preset impedance reference threshold for that taste attribute. The preset impedance reference threshold is obtained by collecting multiple standard cured meat samples with known taste levels, calculating the cumulative impedance of each standard cured meat sample in the characteristic frequency band corresponding to each taste attribute, and taking the average of the cumulative impedance of all standard cured meat samples on the same taste attribute.

[0063] The quantized response intensity values ​​of each taste attribute in the initial taste fingerprint vector are multiplied by the corresponding impedance correction coefficient to obtain the corrected response intensity values ​​for each taste attribute. The sour taste response intensity in the initial taste fingerprint vector is denoted as... The intensity of the sweetness response is denoted as The intensity of the salty response is denoted as The intensity of the bitterness response is denoted as The intensity of the umami response is denoted as The corrected response intensity value of the sourness attribute. The calculation method is as follows The modified response intensity value of the sweetness attribute. Corrected response intensity value of salty property The modified response intensity value of bitterness attribute And the modified response intensity value of umami attribute The calculation follows the same method: the quantized response intensity value of each taste attribute is multiplied by the corresponding impedance correction coefficient. The corrected response intensity values ​​of all taste attributes are then combined to form a corrected taste fingerprint vector, which contains... , , , and A five-dimensional vector.

[0064] In some embodiments, the ranges of the first, second, third, fourth, and fifth characteristic frequency bands are defined as follows: Several standard taste substance solutions with single components are selected, including citric acid solution, sucrose solution, sodium chloride solution, caffeine solution, and monosodium glutamate solution, corresponding to sour, sweet, salty, bitter, and umami attributes, respectively. For each standard taste substance solution, AC impedance spectroscopy measurements are performed at multiple concentration gradients to obtain a curve showing the impedance amplitude changing with frequency. In the curve showing the impedance amplitude changing with frequency, the segment where the impedance amplitude is most sensitive to frequency changes is identified, and the frequency range of this segment is taken as the characteristic frequency band corresponding to that taste attribute. The first characteristic frequency band corresponding to the sour taste attribute is calibrated from 10Hz to 100Hz, the second characteristic frequency band corresponding to the sweet taste attribute is calibrated from 100Hz to 1kHz, the third characteristic frequency band corresponding to the salty taste attribute is calibrated from 1kHz to 10kHz, the fourth characteristic frequency band corresponding to the bitter taste attribute is calibrated from 10kHz to 100kHz, and the fifth characteristic frequency band corresponding to the umami taste attribute is calibrated from 100kHz to 1MHz.

[0065] Optionally, when the cumulative impedance corresponding to a certain taste attribute is less than one-tenth of the preset impedance reference threshold corresponding to that taste attribute, the impedance correction coefficient for that taste attribute is limited to 0.1 to avoid excessive suppression of the quantized response intensity value of that taste attribute due to excessively small impedance correction coefficient caused by impedance measurement noise. When the cumulative impedance corresponding to a certain taste attribute is greater than ten times the preset impedance reference threshold corresponding to that taste attribute, the impedance correction coefficient for that taste attribute is limited to 10 to avoid excessive amplification of the quantized response intensity value of that taste attribute due to excessively large impedance correction coefficient caused by accidental factors.

[0066] It can be understood that the cumulative impedance reflects the overall ability of the tested cured meat sample to impede AC disturbance signals within the characteristic frequency bands corresponding to each taste attribute. A larger cumulative impedance indicates a denser electrochemical interface structure or a higher content of active substances related to that taste attribute. The impedance correction coefficient normalizes the cumulative impedance to a relative scale based on a preset impedance reference threshold through ratio calculation, so that the impedance characteristics of different taste attributes correct the initial taste fingerprint vector on the same order of magnitude.

[0067] See Figure 6 The horizontal axis of the graph represents frequency (Hz), ranging from 0.1Hz to 1MHz, and the vertical axis represents impedance amplitude (Ω), using a logarithmic scale, ranging from approximately 10^3 to 10^4 ohms. The blue solid line in the graph represents the curve of impedance amplitude of the tested cured meat sample as a function of frequency. Overall, it shows a trend of impedance amplitude gradually decreasing as frequency increases, but exhibits different variation characteristics in different frequency ranges.

[0068] The diagram uses colored shaded areas to indicate the characteristic frequency bands corresponding to five preset taste attributes: sourness (10Hz to 100Hz, red area), sweetness (100Hz to 1kHz, blue area), saltiness (1kHz to 10kHz, green area), bitterness (10kHz to 100kHz, purple area), and umami (100kHz to 1MHz, orange area). This segmentation is based on the characteristic frequency band ranges defined in Example 3, clearly indicating the impedance frequency domain characteristics corresponding to different taste attributes.

[0069] Within the 10Hz to 100Hz range corresponding to the sour taste attribute, the impedance amplitude shows a significant and rapid decreasing trend, indicating that the electrochemical interface has a high and significantly changing impedance to AC disturbance signals in this frequency band, reflecting the significant electrochemical response of the active substance with this taste attribute. In the sweet taste range of 100Hz to 1kHz, the impedance amplitude continues to decrease, but the rate of change slows down relatively, indicating that the electrochemical characteristics are different in this frequency band, and the impedance energy contribution is slightly lower than in the sour taste range. Subsequently, in the salty taste range of 1kHz to 10kHz, the impedance amplitude continues to decrease, and the curve shape is relatively flat, showing the characteristic frequency band of the electrochemically active substance corresponding to the taste attribute.

[0070] Within the 10kHz to 100kHz frequency band corresponding to the bitterness attribute, the impedance amplitude curve exhibits local fluctuations, which may be related to changes in the electrochemical interface structure or substance content associated with this taste attribute. In the 100kHz to 1MHz range corresponding to the umami attribute, the impedance amplitude further decreases to its lowest level, and the curve tends to stabilize, indicating that the impedance accumulation in this frequency band is relatively small, reflecting the electrochemical response characteristics of umami-related active substances.

[0071] Example 4:

[0072] In practice, the process of matching the modified taste fingerprint vector with multiple standard fingerprint vectors in the standard taste fingerprint database and determining the taste level identifier of the marinated meat sample to be tested based on the matching results is as follows.

[0073] The modified taste fingerprint vector is denoted as ,in The modified response strength value representing the sourness attribute. The value representing the modified response intensity of the sweetness attribute. The modified response intensity value representing the salty taste attribute. The modified response intensity value representing the bitterness attribute. The modified response intensity value representing the umami attribute. The standard taste fingerprint database stores... The first standard fingerprint vector, the... The standard fingerprint vector is denoted as ,in The value range is 1 to , Representing the The response intensity value of the acidity attribute in a standard fingerprint vector. Representing the The response intensity value of the sweetness attribute in a standard fingerprint vector. Representing the The response intensity value of the salty attribute in a standard fingerprint vector. Representing the The response intensity value of the bitterness attribute in a standard fingerprint vector. Representing the The response intensity value of the umami attribute in a standard fingerprint vector.

[0074] Calculate the modified taste fingerprint vector With the Standard fingerprint vectors Euclidean distance between The calculation method is as follows ,in Dimension index. Calculate the corrected taste fingerprint vector. Length of the module Calculate the first Standard fingerprint vectors Length of the module Euclidean distance Divide by the corrected taste fingerprint vector length With the A standard fingerprint vector magnitude The product of , we get the first The initial similarity value corresponding to each standard fingerprint vector The calculation method is as follows This operation is performed on each standard fingerprint vector to obtain... 1 initial similarity value.

[0075] The initial similarity values ​​are normalized so that the normalized similarity values ​​are distributed within the interval (0,1]. The final similarity score corresponding to each standard fingerprint vector The calculation method is as follows:

[0076]

[0077] in, Representing the The final similarity score corresponding to each standard fingerprint vector. Representing the The initial similarity value corresponding to the first standard fingerprint vector. When the corrected taste fingerprint vector is compared with the first... When all standard fingerprint vectors completely overlap =0, Take the maximum value of 1; when the corrected taste fingerprint vector is the same as the first... When the difference between the standard fingerprint vectors increases Increase Approaching 0. Final similarity score. The value range is (0,1).

[0078] from The highest final similarity score among all final similarity scores is recorded as the highest similarity score. Determine the highest similarity score. The corresponding standard fingerprint vector is obtained, and the preset taste level label associated with the standard fingerprint vector is acquired. Each standard fingerprint vector in the standard taste fingerprint database is pre-associated with a preset taste level label, which includes level 1, level 2, level 3, level 4 and level 5, representing the taste quality classification of cured meat from excellent to poor.

[0079] The highest similarity score Compared with the first threshold and the second threshold, the first threshold was set to 0.85, and the second threshold was set to 0.60. The first threshold was set to 0.85 because when the final similarity score reaches 0.85, the Gaussian distribution confidence interval corresponding to the degree of difference between the modified taste fingerprint vector and the standard fingerprint vector indicates that the probability of them belonging to the same taste level exceeds 95%. The second threshold was set to 0.60 because when the final similarity score is below 0.60, the difference between the modified taste fingerprint vector and the standard fingerprint vector exceeds the allowable deviation range for a single taste level, making reliable single-level determination impossible.

[0080] When the highest similarity score The highest similarity score is given when it is greater than or equal to the first threshold of 0.85. The preset taste level label associated with the corresponding standard fingerprint vector is directly determined as the taste level identifier of the marinated meat sample to be tested.

[0081] When the highest similarity score If the score is less than the first threshold of 0.85 and greater than or equal to the second threshold of 0.60, a weighted fusion operation is performed to generate a fused taste level label. The score with the highest similarity among the remaining final similarity scores is then selected. The final similarity score that is numerically closest is denoted as the adjacent similarity score. The highest similarity score will be used. The corresponding preset taste level label is assigned a first weight coefficient. Calculate the similarity score between adjacent pairs. The higher value among the corresponding preset taste level labels is assigned a second weighting coefficient. First weighting coefficient Set to 0.70, second weighting coefficient Set to 0.30, first weighting coefficient The setting of 0.70 is based on the fact that the taste level corresponding to the highest similarity score plays a dominant role in the fusion determination, and the second weighting coefficient... The setting of 0.30 is based on providing auxiliary reference for adjacent levels, with the dominant weight greater than the auxiliary weight. The level quantification value of the preset taste level label corresponding to the highest similarity score is used. Multiplied by the first weighting coefficient The quantification value of the preset taste level label corresponding to the adjacent similarity scores. Multiply by the second weighting factor The two are added together to obtain the fusion level quantification value. The calculation method is as follows The preset quantification values ​​for taste level labels correspond as follows: Level 1 is 1, Level 2 is 2, Level 3 is 3, Level 4 is 4, and Level 5 is 5. The quantification values ​​for the blending levels will be... The values ​​are compared with the preset taste level labels (1, 2, 3, 4, 5), and the numerical distance is selected. The preset taste level label corresponding to the most recent grading value is used as the identifier for the fusion taste level. If If the distance between the two level quantification values ​​is equal, the preset taste level label with the smaller level quantification value is selected as the integrated taste level identifier.

[0082] When the highest similarity score When the value is less than the second threshold of 0.60, the taste level of the marinated meat sample to be tested is output as unrecognizable, indicating that the taste fingerprint features of the marinated meat sample to be tested do not match any known level in the standard taste fingerprint database.

[0083] In some embodiments, when multiple standard fingerprint vectors and modified taste fingerprint vectors have the same highest final similarity score, the preset taste level label associated with the standard fingerprint vector with the smallest modulus is selected as the preset taste level label corresponding to the highest similarity score. When the highest similarity score... When there are multiple preset taste level labels in the corresponding standard fingerprint vector, the preset taste level label with the highest frequency of occurrence is selected first.

[0084] Optional, neighbor similarity score The determination method is as follows: except for the highest similarity score In addition, the remaining final similarity scores are sorted in descending order of value, and the final similarity score ranked first is selected as the adjacent similarity score. When adjacent similarity scores When there are multiple corresponding preset taste level labels, select the preset taste level label with the smallest difference in level quantification value with the preset taste level label corresponding to the highest similarity score.

[0085] It is understandable that the construction process of the standard taste fingerprint database ensures the accuracy of the correspondence between each standard fingerprint vector and the known taste level label. Therefore, the taste level determination is repeatable by comparing the highest similarity score with the threshold.

[0086] Example 5:

[0087] In specific implementation, please refer to Figure 3 The construction process of the standard taste fingerprint database is as follows:

[0088] Several standard cured meat samples with known taste grades were selected. Each standard cured meat sample was graded by professional sensory evaluators according to the taste evaluation standards for cured meat, and a corresponding known taste grade label was obtained. The known taste grade labels include five levels: Level 1, Level 2, Level 3, Level 4, and Level 5. The standard current response matrix and standard impedance spectrum matrix of each standard cured meat sample were collected under a preset potential excitation sequence. When collecting the standard current response matrix and standard impedance spectrum matrix, the standard cured meat sample was placed in the same three-electrode electrolytic cell as the cured meat sample to be measured. The three-electrode electrolytic cell includes a working electrode, a counter electrode, and a reference electrode. The preset potential excitation sequence consists of five frequency segments, each containing three potential amplitude levels. The frequency range of the five frequency segments gradually increases, and the application order is from low to high frequency. During each potential excitation signal application, the response current signal between the working electrode and the counter electrode is synchronously acquired at a predetermined sampling frequency. The current amplitudes at each sampling moment are arranged in chronological order to form row vectors of the standard current response matrix. All row vectors are stacked in the application order to form the standard current response matrix. After each potential excitation signal application is completed, an AC disturbance signal with a constant amplitude of 7mV is applied to the working electrode. The scanning frequency range covers the first characteristic frequency band corresponding to the sour taste attribute, the second characteristic frequency band corresponding to the sweet taste attribute, the third characteristic frequency band corresponding to the salty taste attribute, the fourth characteristic frequency band corresponding to the bitter taste attribute, and the fifth characteristic frequency band corresponding to the umami taste attribute. The impedance amplitude and impedance phase angle of the working electrode are acquired and arranged in ascending order of scanning frequency to construct a standard impedance spectrum matrix.

[0089] Following the same processing method as the marinated meat samples to be tested, standard characteristic response peak groups were extracted from the standard current response matrix. A second-order difference operation was performed on the current amplitude sequence of each row in the standard current response matrix to obtain the standard current amplitude change rate sequence. Sampling time points exceeding a preset change rate threshold in the standard current amplitude change rate sequence were marked as standard candidate peak sites. Expanding to the left and right sides from each standard candidate peak site, the left and right boundary potentials corresponding to the current amplitude decreasing to 50% of the current amplitude at the standard candidate peak site were found. The interval between the left and right boundary potentials was determined as the peak width potential interval of a single standard characteristic response peak. Standard characteristic response peaks with peak height current values ​​greater than a preset peak height threshold were selected from all single standard characteristic response peaks. The selected standard characteristic response peaks were arranged in ascending order of peak potential to form a standard characteristic response peak group.

[0090] Following the same mapping method as the marinated meat samples to be tested, the standard characteristic response peak groups are mapped to a preset taste attribute coordinate map to generate standard initial taste fingerprint vectors for each standard marinated meat sample. The taste attribute coordinate map includes a first potential coordinate interval corresponding to sourness, a second potential coordinate interval corresponding to sweetness, a third potential coordinate interval corresponding to saltiness, a fourth potential coordinate interval corresponding to bitterness, and a fifth potential coordinate interval corresponding to umami. The peak potential of each standard characteristic response peak in the standard characteristic response peak group is assigned to one of the first, second, third, fourth, and fifth potential coordinate intervals to determine the taste attribute category to which each standard characteristic response peak belongs. The peak height current values ​​of all standard characteristic response peaks belonging to the same taste attribute category are accumulated, and the accumulated value is used as the quantified value of the standard response intensity of that taste attribute. The quantified values ​​of sourness, sweetness, saltiness, bitterness, and umami response intensity are combined to form the standard initial taste fingerprint vector.

[0091] Following the same correction method as the tested cured meat samples, the initial standard taste fingerprint vector was corrected based on the standard impedance spectrum matrix. Impedance amplitude sequences were extracted from the standard impedance spectrum matrix for the following taste attributes: first characteristic band for sourness, second characteristic band for sweetness, third characteristic band for saltiness, fourth characteristic band for bitterness, and fifth characteristic band for umami. A trapezoidal integral was performed on the impedance amplitude sequence within each characteristic band over a frequency step to obtain the cumulative standard impedance for each taste attribute. The cumulative standard impedance for each taste attribute was then compared to a preset impedance threshold for that attribute to obtain the standard impedance correction coefficient for each taste attribute. The preset impedance threshold was obtained by averaging the cumulative impedance values ​​of all standard cured meat samples over the same taste attribute. The standard response intensity quantization value of each taste attribute in the standard initial taste fingerprint vector is multiplied by the standard impedance correction coefficient of the corresponding taste attribute to obtain the standard corrected response intensity value of each taste attribute. The standard corrected response intensity values ​​of all taste attributes are combined to form the standard fingerprint vector.

[0092] Each standard cured meat sample is associated with a standard fingerprint vector and its own known taste level label to form a database record. All database records for all standard cured meat samples are aggregated and stored in a standard taste fingerprint database. Each record in the standard taste fingerprint database contains five-dimensional data of a standard fingerprint vector and a predefined taste level label associated with that standard fingerprint vector. The standard taste fingerprint database is used to provide a set of standard fingerprint vectors for subsequent taste level determination of the cured meat samples to be tested.

[0093] In some embodiments, the number of standard cured meat samples is no less than fifty, and the number of standard cured meat samples corresponding to each known taste level label is no less than ten, in order to ensure the statistical representativeness of each level of samples in the standard taste fingerprint database.

[0094] Optionally, the known taste rating label of the standard cured meat sample shall be independently evaluated by no fewer than five professional sensory evaluators according to the taste evaluation standard for cured meat. When the consistency rate of the evaluation results reaches more than 80%, the rating label shall be used as the known taste rating label of the standard cured meat sample.

[0095] It is understandable that the construction process of the standard taste fingerprint database is completely consistent with the testing process of the marinated meat samples to be tested. From the acquisition of potential excitation sequences, current response matrices and impedance spectrum matrices to the extraction of feature response peak groups, generation and correction of taste fingerprint vectors, the same operating parameters and algorithms are used to ensure that the standard fingerprint vector and the corrected taste fingerprint vector are matched for similarity on the same scale.

[0096] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for evaluating the digital taste information of cured meats based on biosensing, characterized in that, The method includes: In response to the excitation response signal acquisition operation of the marinated meat sample to be tested, the current response matrix and impedance spectrum matrix of the marinated meat sample to be tested under a preset potential excitation sequence are obtained. The preset potential excitation sequence includes multiple frequency segments and multiple potential amplitude levels. Based on the current amplitude jump characteristics at each sampling time point in the current response matrix, the characteristic response peak group corresponding to the taste active substances of the marinated meat sample to be tested is identified. The characteristic response peak group includes peak position potential, peak height current value and peak width potential range. The feature response peaks are mapped onto a preset taste attribute coordinate map to generate the initial taste fingerprint vector of the marinated meat sample to be tested. The impedance spectrum matrix is ​​subjected to frequency domain segmented integration to obtain the impedance accumulation of the marinated meat sample under test in the characteristic frequency bands corresponding to each taste attribute, and the initial taste fingerprint vector is corrected according to the impedance accumulation to generate a corrected taste fingerprint vector. The modified taste fingerprint vector is matched with multiple standard fingerprint vectors in the standard taste fingerprint database for similarity, and the taste level identifier of the marinated meat sample to be tested is determined based on the matching results.

2. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 1, characterized in that, The process of obtaining the current response matrix and impedance spectrum matrix of the marinated meat sample under a preset potential excitation sequence includes: The sample of marinated meat to be tested is placed in a three-electrode electrolytic cell, and the preset potential excitation signals in the preset potential excitation sequence are applied to the working electrode. The preset potential excitation sequence is arranged from low to high frequency and each frequency segment contains at least three potential amplitude levels. During each application of the potential excitation signal, the response current signal between the working electrode and the counter electrode is synchronously acquired at a predetermined sampling frequency, and the current amplitude at each sampling moment is arranged in chronological order to form the current response matrix; After each of the aforementioned potential excitation signals is applied, an AC perturbation signal with a constant amplitude is applied to the working electrode, the scanning frequency range covers the characteristic frequency bands corresponding to each taste attribute, the impedance amplitude and impedance phase angle of the working electrode are collected, and the impedance spectrum matrix is ​​constructed.

3. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 2, characterized in that, The step of identifying the characteristic response peaks corresponding to the taste-active substances in the marinated meat sample based on the current amplitude jump characteristics at each sampling time point in the current response matrix includes: A second-order difference operation is performed on each row of the current amplitude sequence in the current response matrix to obtain the current amplitude change rate sequence, and the sampling time points in the current amplitude change rate sequence that exceed the preset change rate threshold are marked as candidate peak points; Expanding to the left and right sides from each candidate peak point, find the left and right boundary potentials corresponding to the current amplitude dropping to a predetermined proportion of the current amplitude at the candidate peak point, and determine the interval between the left and right boundary potentials as the peak width potential interval of a single characteristic response peak. Select characteristic response peaks whose peak height current value is greater than a preset peak height threshold from all individual characteristic response peaks, and arrange the selected characteristic response peaks in ascending order of peak potential to form the characteristic response peak group.

4. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 3, characterized in that, The step of mapping the feature response peaks to a preset taste attribute coordinate map to generate the initial taste fingerprint vector of the marinated meat sample to be tested includes: The initial taste fingerprint vector includes quantified values ​​of sour taste response intensity, sweet taste response intensity, salty taste response intensity, bitter taste response intensity, and umami taste response intensity; Obtain the first potential coordinate interval corresponding to the sour taste attribute, the second potential coordinate interval corresponding to the sweet taste attribute, the third potential coordinate interval corresponding to the salty taste attribute, the fourth potential coordinate interval corresponding to the bitter taste attribute, and the fifth potential coordinate interval corresponding to the umami taste attribute from the preset taste attribute coordinate map. The peak potential of each characteristic response peak in the characteristic response peak group is assigned to the first potential coordinate interval, the second potential coordinate interval, the third potential coordinate interval, the fourth potential coordinate interval and the fifth potential coordinate interval respectively to determine the taste attribute category to which each characteristic response peak belongs. The peak height current values ​​of all feature response peaks belonging to the same taste attribute category are accumulated, and the accumulated value is used as the quantized value of the response intensity of that taste attribute. The quantized values ​​of the response intensity of all taste attributes are combined to form the initial taste fingerprint vector.

5. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 4, characterized in that, The step of performing frequency domain segmented integration on the impedance spectrum matrix to obtain the cumulative impedance of the marinated meat sample under test in the characteristic frequency bands corresponding to each taste attribute includes: The impedance amplitude sequences in the first characteristic frequency band corresponding to the sour taste attribute, the second characteristic frequency band corresponding to the sweet taste attribute, the third characteristic frequency band corresponding to the salty taste attribute, the fourth characteristic frequency band corresponding to the bitter taste attribute, and the fifth characteristic frequency band corresponding to the umami taste attribute are extracted from the impedance spectrum matrix respectively. The impedance amplitude sequence within each characteristic frequency band is integralized using a trapezoidal method with a frequency step size to obtain the impedance accumulation corresponding to each taste attribute. Each impedance accumulation represents the impedance energy contribution value of the corresponding taste attribute in the frequency domain.

6. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 5, characterized in that, The step of correcting the initial taste fingerprint vector based on the accumulated impedance to generate a corrected taste fingerprint vector includes: The impedance correction coefficient for each taste attribute is obtained by comparing the cumulative impedance of each taste attribute with the preset impedance reference threshold of the corresponding taste attribute. The quantized response intensity value of each taste attribute in the initial taste fingerprint vector is multiplied by the impedance correction coefficient of the corresponding taste attribute to obtain the corrected response intensity value of each taste attribute. The modified taste fingerprint vector is constructed by combining the modified response intensity values ​​of all taste attributes.

7. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 6, characterized in that, The step of performing similarity matching between the modified taste fingerprint vector and multiple standard fingerprint vectors in the standard taste fingerprint database includes: Calculate the Euclidean distance between the modified taste fingerprint vector and each standard fingerprint vector in the standard taste fingerprint database, and calculate the product of the magnitude of each standard fingerprint vector and the magnitude of the modified taste fingerprint vector. Divide the Euclidean distance by the product to obtain the initial similarity value. The initial similarity values ​​are normalized so that the normalized similarity values ​​are distributed within a predetermined interval. The normalized similarity values ​​are then used as the final similarity scores between the modified taste fingerprint vector and each standard fingerprint vector.

8. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 7, characterized in that, The step of determining the taste level identifier of the marinated meat sample to be tested based on the matching results includes: Select the highest similarity score from all final similarity scores, and determine the preset taste level label associated with the standard fingerprint vector corresponding to the highest similarity score; When the highest similarity score is greater than or equal to the first threshold, the preset taste level label is directly determined as the taste level identifier of the marinated meat sample to be tested. When the highest similarity score is less than the first threshold and greater than or equal to the second threshold, the preset taste level labels corresponding to the highest similarity score and its adjacent similarity scores are weighted and fused to generate a fused taste level label. When the highest similarity score is less than the second threshold, the taste level of the marinated meat sample to be tested is output as unrecognizable.

9. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 1, characterized in that, The steps for constructing the standard taste fingerprint database include: Standard current response matrices and standard impedance spectrum matrices of multiple standard cured meat samples with known taste levels were collected under the preset potential excitation sequence; Following the same processing method as the marinated meat samples to be tested, standard feature response peak groups are extracted from the standard current response matrix and mapped to generate standard initial taste fingerprint vectors for each standard marinated meat sample. Following the same correction method as the marinated meat sample to be tested, the standard initial taste fingerprint vector is corrected according to the standard impedance spectrum matrix to obtain the standard fingerprint vector corresponding to each standard marinated meat sample, and each standard fingerprint vector is associated with its known taste level label and stored in the standard taste fingerprint database.

10. The method for evaluating the digital taste information of cured meat based on biosensing according to claim 8, characterized in that, The step of weightedly fusing the preset taste level labels corresponding to the highest similarity score and its adjacent similarity scores to generate a fused taste level identifier includes: Assign a first weight coefficient to the preset taste level label corresponding to the highest similarity score, and assign a second weight coefficient to the preset taste level label corresponding to the largest value among the similarity scores adjacent to the highest similarity score. The first weight coefficient is greater than the second weight coefficient. The quantified value of each preset taste level label is multiplied by its corresponding weight coefficient and then summed to obtain the fusion level quantified value. The preset taste level label with the closest fusion level quantified value is determined as the fusion taste level identifier.

Images