A rapid detection system for the degree of rancidity of edible oil based on an electrochemical sensor

Abstract

The application relates to the technical field of electrochemical detection and food quality and safety, and particularly discloses a rapid detection system for the degree of rancidity of edible oil based on an electrochemical sensor, which comprises a data acquisition module, which receives original electrochemical response signals, and performs format standardization, abnormal data elimination and integrity verification; a data preprocessing module, which adopts a statistical optimization algorithm to reduce noise and correct baseline drift, generates a unified dimension data set and completes stability verification; a feature analysis module, which extracts rancidity-related characteristic parameters through statistical correlation analysis, fuses and calculates characteristic values representing rancidity after significance verification; and an evaluation and judgment module, which inputs the characteristic values into a pre-trained model, outputs quantitative indexes through multivariate statistical calculation, dynamically sets a safety threshold according to the type of edible oil, and outputs a rancidity judgment conclusion after comparison. The application relies on statistical data processing and analysis, improves detection reliability and accuracy, and is suitable for rapid screening and quality monitoring of the quality and safety of edible oil.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrochemical detection and food quality and safety, more specifically, the present application relates to a rapid detection system for the degree of rancidity of edible oil based on an electrochemical sensor. BACKGROUND

[0002] As an indispensable food raw material in daily life, the quality and safety of edible oil is directly related to human health. During storage, processing and use, oil is susceptible to oxidation and hydrolysis due to factors such as light, oxygen, temperature and moisture, leading to rancidity. Rancid edible oil not only produces an unpleasant smell, reducing food quality, but also generates harmful substances such as peroxides, aldehydes and ketones, which can increase the risk of cardiovascular disease and cancer if consumed for a long time. Therefore, it is crucial to quickly and accurately detect the degree of rancidity of edible oil.

[0003] Currently, the detection of the degree of rancidity of edible oil mainly relies on chromatographic and spectroscopic analysis methods and electrochemical detection methods. Chromatographic and spectroscopic analysis methods include high-performance liquid chromatography, gas chromatography and near-infrared spectroscopy. Chromatography can accurately separate and quantify specific rancidity products, has high sensitivity, and near-infrared spectroscopy can achieve non-destructive rapid detection. Electrochemical detection method is based on the response signal of electrochemical sensor to the electroactive substances in edible oil. The typical implementation of this method is to apply a specific electrical stimulus to the oil sample through the working electrode and detect the current or impedance response signal generated, and then establish a correlation model between the signal characteristic value and the acid value or peroxide value index. The advantage of this method is that it has fast detection speed, small equipment size and good portability, and is expected to realize rapid on-site screening.

[0004] However, these methods still have some shortcomings in actual use:

[0005] 1. Significant detection condition limitations: Although chromatographic and spectroscopic analysis methods have high detection accuracy, they are heavily dependent on large and precise instruments and professional laboratory environments, with high equipment purchase and maintenance costs, making them difficult to be deployed in on-site environments such as production lines and kitchens, and unable to achieve true on-site rapid detection.

[0006] 2. Insufficient model robustness: Existing electrochemical detection methods usually use a simple mapping model of a single electrical parameter and rancidity index. Due to the large difference in matrix composition of different types of edible oil and the diversity of electroactive substances generated during the rancidity process, this type of model has poor generalization ability when facing unknown or complex oil samples, resulting in a significant decrease in prediction accuracy.

[0007] 3. Lack of anti-interference ability: Electrochemical signals are easily disturbed by environmental factors such as temperature fluctuations, electrode surface contamination and water content in the sample. Existing technologies lack effective signal preprocessing and feature optimization algorithms, resulting in poor stability and repeatability of detection results, affecting the reliability of the method.

[0008] 4. Limited intelligence: Existing electrochemical solutions usually only output a single indicator such as acid value or peroxide value, and cannot achieve multi-indicator collaborative analysis and comprehensive decision-making. When the detection results of each indicator are contradictory, the system cannot automatically give a clear final judgment conclusion, and still needs to rely on manual interpretation, and cannot realize the automation and intelligence of the whole detection process. SUMMARY

[0009] In order to overcome the above-mentioned defects of the prior art, the present application provides a kind of based on electrochemical sensor's edible oil rancidity degree rapid detection system, by the following scheme, to solve the problems raised in the above background art.

[0010] To achieve the above object, the present application provides the following technical scheme: a kind of based on electrochemical sensor's edible oil rancidity degree rapid detection system, comprising:

[0011] Data acquisition module is used to receive the original electrochemical response signal of edible oil sample output by electrochemical sensor, and the signal is processed by format standardization, and the integrity is verified by removing abnormal data through statistical rules;

[0012] Data preprocessing module is used to adopt statistical optimization algorithm to denoise and correct baseline drift of received data, generate data set of uniform dimension after standardization processing, and transmit to feature analysis module after stability verification;

[0013] Feature analysis module is used to extract characteristic parameters related to rancidity from data set by statistical correlation analysis, and after significance verification, the parameters are fused by numerical calculation method to obtain characteristic value representing rancidity degree;

[0014] Evaluation and judgment module is used to input characteristic value into pre-trained rancidity index prediction model, and the rancidity index prediction model is obtained by machine learning algorithm on training data set containing multiple characteristic values and corresponding standard acid value and peroxide value labels, and the quantitative indicators of acid value and peroxide value are output by multivariate statistical calculation, and the safety threshold is dynamically set according to the type of edible oil, and the rancidity judgment conclusion is output after comparison.

[0015] Preferably, the original electrochemical response signal received by the data acquisition module is one of voltammetry curve, impedance spectrum or current time sequence, and each signal needs to meet the statistical sampling requirement.

[0016] Preferably, the voltammetry curve is differential pulse voltammetry curve or square wave pulse voltammetry curve, and the signal quality of the two curves needs to be verified by statistical method.

[0017] Preferably, the preprocessing includes the following steps:

[0018] S1: Adopting filtering algorithm based on signal statistical characteristics to reduce noise;

[0019] S2: Using statistical correction algorithm for baseline drift rule to eliminate drift;

[0020] S3: Converting the processed data into a unified dimension data set through a standardization algorithm.

[0021] Preferably, the feature analysis module extracts the feature parameters related to rancidity, which needs to be confirmed by statistical correlation analysis and verified by statistical significance test.

[0022] Preferably, the numerical calculation method used by the feature analysis module is one of principal component analysis, partial least squares regression or support vector regression, and each method needs to be combined with statistical optimization logic.

[0023] Preferably, the rancidity index prediction model of the evaluation and judgment module is obtained by machine learning algorithm on the training data set with statistical representation, and the training data set contains multiple sets of feature values and corresponding standard acid value and peroxide value labels.

[0024] Preferably, the quantitative indicators output by the evaluation and judgment module are acid value and peroxide value, and the accuracy of the indicators needs to be verified by statistical method.

[0025] Preferably, the preset safety threshold of the evaluation and judgment module needs to be dynamically set in combination with the national standard limit value of the type of edible oil and the statistical distribution of the historical detection data of the type of oil sample.

[0026] Preferably, the relationship between the quantitative indicators and the threshold is analyzed by a statistical model, and if the acid value or the peroxide value exceeds the statistical determination standard of the corresponding threshold, rancidity is output; if both indicators do not exceed, no rancidity is output.

[0027] The technical effects and advantages of the present application are:

[0028] 1. Realize on-site rapid and accurate detection: The present application adopts a portable electrochemical sensor and an intelligent processing system, which overcomes the problem of on-site deployment of large instruments. Through the control of the signal acquisition module on the sensor scanning, the detection is completed within a few minutes without complex pretreatment and toxic reagents, realizing the rapid and accurate detection of the on-site environment such as production line and kitchen.

[0029] 2. Improve the robustness and accuracy of the model: Multi-dimensional feature extraction and machine learning algorithm are combined. The feature extraction module extracts multiple types of feature parameters from the data set, and the parameter calculation module uses multivariate statistical method to establish a prediction model, which effectively overcomes the influence of different edible oil matrix differences.

[0030] 3. Enhanced anti-interference ability and reliability: The signal pre-processing module solves the problem of signal susceptibility to interference. Digital filtering is used for noise reduction, baseline correction is used to eliminate drift, and data is standardized, which significantly improves signal quality and stability and ensures the reproducibility of detection results.

[0031] 4. Intelligent comprehensive decision-making: The evaluation output module realizes multi-index collaborative analysis and automatic decision-making. The system synchronously outputs acid value and peroxide value, automatically compares according to dynamic safety threshold, directly outputs clear conclusion, avoids manual error, and realizes full-process automation and intelligence. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The figure is a schematic diagram of the module structure of the system of the present application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0034] Reference Figure 1 The figure shows a rapid detection system for the degree of rancidity of edible oil based on an electrochemical sensor, comprising:

[0035] A data acquisition module for receiving the original electrochemical response signal of the edible oil sample output by the electrochemical sensor, performing format standardization processing on the signal, removing abnormal data and checking integrity through statistical rules;

[0036] A data pre-processing module for using a statistical optimization algorithm to denoise and correct baseline drift of the received data, generating a unified dimension data set after standardization processing, and transmitting to a feature analysis module after stability verification;

[0037] A feature analysis module for extracting feature parameters related to rancidity from the data set through statistical correlation analysis, fusing parameters through numerical calculation method after significance verification, and obtaining feature values representing the degree of rancidity;

[0038] An evaluation and judgment module for inputting the feature values into a pre-trained rancidity index prediction model, the rancidity index prediction model being obtained by machine learning algorithm on a training data set containing multiple groups of feature values and corresponding standard acid value and peroxide value labels, outputting quantitative indexes of acid value and peroxide value through multivariate statistical calculation, setting dynamic safety threshold according to the type of edible oil, and outputting rancidity judgment conclusion after comparison.

[0039] The data acquisition module is used for receiving the original electrochemical response signal of the edible oil sample output by the electrochemical sensor, performing format standardization processing on the signal, removing abnormal data through a statistical rule, and checking integrity.

[0040] The data acquisition module is a direct interface between the system and the electrochemical sensor hardware, responsible for acquiring original signals and ensuring that the quality meets the requirements of subsequent statistical analysis. The physical composition of the data acquisition module includes a signal input interface, an embedded processor, a cache memory, and a communication unit. The data acquisition module realizes data standardization, abnormality removal, and integrity checking through algorithms. The specific steps are as follows:

[0041] 1. Signal reception and format standardization

[0042] The voltammetry curve signal output by the electrochemical sensor is converted into a digital signal by the multi-channel analog-to-digital converter of the signal input interface at a sampling rate of not less than 100 kS / s. Preferably, the voltammetry curve is a differential pulse voltammetry curve; alternatively, the voltammetry curve is a square wave voltammetry curve.

[0043] The embedded processor processes the digital signal, extracts the voltage-current data pair, and converts it into a standardized data format.

[0044] The standardized data is temporarily stored in the cache memory through direct memory access for subsequent statistical processing.

[0045] 2. Abnormal data removal based on statistical rules

[0046] 2.1 Abnormality detection algorithm: The embedded processor calculates the arithmetic mean and standard deviation of the data segment , and removes abnormal points outside the range using the criterion. A data segment refers to a short time period or a small set of data points extracted by the embedded processor from continuous voltammetry curve digital signals for independent analysis. For voltammetry curves, the peak detection algorithm running on the embedded processor verifies the number and position of redox peaks, and removes abnormal curves without peaks or multiple peaks. For differential pulse voltammetry curves, a single characteristic peak should be present.

[0047] 2.2 Statistical sampling requirement guarantee: To ensure statistical significance, the voltage scanning range of the voltammetry curve must cover the typical redox window, with a voltage scanning range of -0.5 V to +1.0 V and a scanning point number of not less than 500 points. All intermediate data in the processing process are stored in the cache memory.

[0048] 3. Data integrity checking

[0049] ​Data integrity check is performed by embedded processor, including:

[0050] 3.1 Data length verification: check if the number of voltammetry curve data points reaches the minimum threshold of 500 points, if not, send instructions to the electrochemical sensor through the communication unit to trigger re-collection;

[0051] 3.2 Time stamp continuity check: ensure uniform sampling interval and avoid time stamp breakage leading to sequence analysis failure.

[0052] Data passing the check is tagged with metadata by the embedded processor, including signal quality score, check status code, etc., and is transferred to the output queue area of the cache memory.

[0053] 4. Data transmission to data preprocessing module

[0054] The communication unit reads data packets from the output queue of the cache memory and sends them to the data preprocessing module.

[0055] The data acquisition module significantly improves the quality of the original signal through statistical rule-driven anomaly rejection and integrity check, providing a high-reliability data foundation for subsequent rancidity feature analysis.

[0056] The data preprocessing module is used to adopt statistical optimization algorithm to denoise and correct baseline drift of the received data, generate a unified dimension data set after standardization processing, and transmit to the feature analysis module after stability check.

[0057] The data preprocessing module is the core processing unit of the system, responsible for deep processing of the original data transmitted by the acquisition module, and provides a high-quality data foundation for subsequent feature extraction. The physical composition of the data preprocessing includes: data receiving interface, digital signal processor, algorithm processing unit and data storage unit. The data preprocessing module realizes data denoising, baseline drift correction and generation of unified dimension data set through multi-level processing algorithm. The specific steps are as follows:

[0058] 1. Filter denoising processing based on statistical characteristics

[0059] The data receiving interface obtains the standardized data packet from the data acquisition module and transmits it to the input buffer of the digital signal processor through direct memory access. The digital signal processor uses wavelet threshold denoising method to perform multi-scale decomposition on the voltammetry curve signal: 5-layer decomposition is performed using Symlets wavelet basis function; adaptive threshold processing is applied to each layer of detail coefficients, and the threshold calculation formula is: , where is the noise standard deviation, which is the value obtained by statistical estimation of the first layer high-frequency detail coefficient after wavelet decomposition, The data length is the total number of sampling points for the volt-ampere curve signal; the reconstructed signal yields the denoised data. The processed data is temporarily stored in a dedicated buffer area of ​​the data storage unit and marked with a processing status flag.

[0060] 2. Baseline drift statistical correction

[0061] The algorithm processing unit uses an adaptive iterative reweighted penalized least squares method to perform baseline correction on the volt-ampere curve signal: the baseline drift pattern is estimated through statistical methods, and an objective function is constructed. ,in, For the first The original current values ​​of each data point For the first Baseline current estimates for each data point No. The weight of each data point For regularization parameters, for The second reciprocal, the formula is: The baseline curvature is used to solve for the optimal baseline current estimate using an iterative algorithm. The baseline component is subtracted from the original signal to obtain the corrected signal. During the correction process, the algorithm processing unit calculates the baseline parameters of each data segment in parallel, and the processing results are stored in real time to the data storage unit.

[0062] 3. Data standardization and unit unification

[0063] The digital signal processor performs normalization processing: using the Z-score normalization algorithm. ,in, This is the arithmetic mean of all current values ​​along the entire volt-ampere curve. The standard deviation of all current values ​​on the entire volt-ampere curve. This represents a single original current value on the current-voltage curve. The standardized new current value is a dimensionless numerical value; for the volt-ampere curve, maximum-minimum normalization is used to transform each feature to the [0,1] interval; the current signal is uniformly converted to μA units, and the voltage signal is uniformly converted to V units. After generating the dataset with uniform dimensions, a stability check is performed: the coefficient of variation (CV) of the dataset is calculated. ,in, It is the standard deviation of the dataset. It is the arithmetic mean of the dataset. To verify the stability of the data, the CV should be less than 0.05.

[0064] 4. Data is transmitted to the feature analysis module.

[0065] After preprocessing, the data receiving interface transmits the processed volt-ampere curve dataset to the feature analysis module.

[0066] The preprocessing module realizes efficient purification and standardization of data through a statistical optimization algorithm, thereby providing a reliable data basis for subsequent rancidity feature analysis.

[0067] The feature analysis module is configured to extract feature parameters related to rancidity from the data set through statistical correlation analysis, and after significance verification, fuse the parameters by using a numerical calculation method to obtain a feature value representing the degree of rancidity.

[0068] The feature analysis module is the intelligent analysis core of the system, and the physical composition of the feature analysis module includes a data interface unit, a feature extraction unit, a statistical analysis unit and a feature database, which is responsible for extracting feature parameters related to rancidity from the preprocessed voltammogram data, verifying the effectiveness thereof through a statistical method, and finally generating a feature value representing the degree of rancidity by using a numerical calculation method. The specific steps are as follows:

[0069] 1. Voltammogram feature parameter extraction

[0070] The data interface unit receives the standardized voltammogram data set from the data preprocessing module and transmits it to the memory buffer of the feature extraction unit through a direct memory access mode. Then, the voltammogram feature parameters are extracted, including electrochemical feature parameters such as oxidation peak current value, reduction peak current value, peak potential difference and half-peak width; morphological features such as curve slope, curvature change and peak symmetry index; and integral features such as oxidation peak area, reduction peak area and peak area ratio. All the feature parameters are temporarily stored in the temporary area of the feature database, waiting for further verification.

[0071] 2. Statistical correlation analysis and significance verification

[0072] The statistical analysis unit performs the following verification process:

[0073] 2.1 Statistical correlation analysis: Pearson correlation coefficient method is used to analyze the correlation between each feature parameter and the standard acid value and peroxide value, and the calculation formula is as follows: wherein, N is the total number of samples, i.e. the number of edible oil samples detected which have both feature parameters and standard chemical values; is the value of a certain feature parameter of the i-th edible oil sample, for example, the oxidation peak current value or the peak area of the sample; is the arithmetic mean value of the same feature parameter of all N samples; is the true acid value or peroxide value of the i-th edible oil sample measured by a standard chemical method; is the arithmetic mean value of the same feature parameter of all N samples; is the true acid value or peroxide value of the i-th edible oil sample measured by a standard chemical method; is the arithmetic mean value of the same feature parameter of all N samples; is the true acid value or peroxide value of the i-th edible oil sample measured by a standard chemical method; Arithmetic mean of acid value or peroxide value of samples; after calculation, keep the characteristic parameters with |r|>0.7.

[0074] 2.2 Statistical significance test: t-test is used to verify the significance of characteristic parameters: ,

[0075] The requirement is p<0.05, which ensures that the characteristic parameters can effectively reflect the change of rancidity.

[0076] The verified characteristic parameters are marked by the statistical analysis unit and transferred to the verified area of the characteristic database.

[0077] 3. Feature fusion based on partial least squares regression method

[0078] The feature extraction unit uses the partial least squares regression method for feature fusion:

[0079] 3.1 Model establishment: Establish the regression model between the characteristic parameter matrix of voltammetry curve and the rancidity index matrix : ,

[0080] wherein, and are score matrices, and are loading matrices, and are residual matrices.

[0081] 3.2 Latent variable extraction: Extract latent variables through nonlinear iterative partial least squares algorithm to maximize the covariance between and .

[0082] 3.3 Statistical optimization: Use leave-one-out cross-validation to determine the optimal number of latent variables to avoid overfitting.

[0083] 3.4 Prediction model: Establish the regression equation: ,

[0084] wherein, is the regression coefficient matrix, is the intercept term.

[0085] Through statistical optimization logic, the reliability and stability of the characteristic value calculation are ensured. All model parameters and intermediate calculation results are stored in the characteristic database.

[0086] 4. Characteristic value output and transmission

[0087] The fused characteristic values are transmitted to the evaluation and judgment module through the data interface unit.

[0088] The characteristic analysis module uses a partial least squares regression method for feature fusion for voltammetry curve signal characteristics, which not only considers the correlation between the characteristic parameters and the rancidity index, but also improves the reliability of the characteristic value calculation through statistical optimization logic. This method is particularly suitable for processing the case where multiple collinearity exists in the voltammetry curve data, and can effectively extract the most predictive feature combination.

[0089] The evaluation and judgment module inputs the characteristic value into a pre-trained rancidity index prediction model, the rancidity index prediction model is learned by a machine learning algorithm on a training data set containing multiple groups of characteristic values and corresponding standard acid value and peroxide value labels, and outputs quantitative indicators of acid value and peroxide value through multivariate statistical calculation. Combine the dynamic safety threshold of the type of edible oil, and output the rancidity judgment conclusion after comparison.

[0090] The evaluation and judgment module is the decision center of the system, and the physical composition of the evaluation and judgment module includes a data receiving unit, a model calculation unit, a threshold management unit and a result display unit, which is responsible for converting the characteristic value output by the characteristic analysis module into specific rancidity index, and making a final judgment according to the dynamic safety threshold. The specific steps are as follows:

[0091] 1. Acid value and peroxide value prediction model loading and calculation

[0092] The data receiving unit receives the fused characteristic value data packet from the characteristic analysis module, and transmits it to the model calculation unit after decryption and verification.

[0093] The model calculation unit loads the pre-trained rancidity index prediction model: , wherein, is the input feature vector, is the predicted index: acid value and peroxide value .

[0094] A multivariate statistical calculation method is used to simultaneously calculate the acid value and peroxide value of two indicators: ,

[0095] wherein, is the characteristic parameter value of the first edible oil sample used in the model after significance and correlation, is the regression coefficient of the first characteristic parameter in the acid value prediction model, is the regression coefficient of the first characteristic parameter in the peroxide value prediction model, and are the intercept terms of the acid value prediction model and the peroxide value prediction model, respectively.

[0096] 2. Dynamic safety threshold setting

[0097] The threshold management unit dynamically sets the safety threshold according to the type of edible oil: calls the built-in national standard database to obtain the standard limit value of this type of oil; refers to the statistical distribution of historical detection data of this type of oil sample: arithmetic mean and standard deviation , dynamically adjust the threshold value according to the principle: ,

[0098] Where k is the adjustment coefficient, which is dynamically valued between 2.0-3.0 according to the characteristics of the oil.

[0099] 3. Index comparison and result judgment

[0100] Statistical comparison of the calculated acid value and peroxide value with the corresponding dynamic threshold value:

[0101] Statistical determination is made by using hypothesis testing method:

[0102] , where is the arithmetic mean of the acid value or peroxide value measurement result, is the safety threshold of the standard acid value or peroxide value, is the standard deviation of the acid value or peroxide value measurement result, is the number of repeated measurements. Judgment rule:

[0103] If the acid value or peroxide value of any index exceeds the statistical determination standard of its corresponding threshold value, meets

[0104] , then it is determined to be rancid; If both the acid value and the peroxide value do not exceed the threshold determination standard, meets

[0105] , then it is determined to be not rancid.

[0106] 4. Result output and display

[0107] The result display unit outputs the detection results, including the acid value, the peroxide value, the safety threshold range and the judgment conclusion.

[0108] The evaluation and judgment module combines multivariate statistical calculation with dynamic threshold management to ensure the accuracy and adaptability of the detection results. The pre-trained machine learning model can effectively process the feature parameters extracted from the voltammetry curve, realizing fast and accurate rancidity judgment.

[0109] ​Secondly: the embodiment of the present application discloses only the structure related to the embodiment of the present application, other structures can refer to the general design, and in the case of no conflict, the same embodiment and different embodiments of the present application can be combined with each other;

[0110] Finally: the above only for the preferred embodiment of the present application, and does not limit the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A rapid detection system for the degree of rancidity of edible oil based on an electrochemical sensor, characterized by, The application relates to a method for evaluating the rancidity of edible oil, which comprises the following steps: a data acquisition module is used to receive the original electrochemical response signals of edible oil samples output by an electrochemical sensor, the signals are subjected to format standardization treatment, abnormal data are removed through a statistical rule, and the integrity is verified; a data preprocessing module is used to adopt a statistical optimization algorithm to denoise and correct the baseline drift of the received data, a unified dimension data set is generated through standardization treatment, and the data set is transmitted to a feature analysis module after stability verification; the feature analysis module is used to extract feature parameters related to rancidity from the data set through statistical correlation analysis, the parameters are fused through a numerical calculation method after significance verification, and a feature value representing the rancidity degree is obtained; and an evaluation and judgment module is used to input the feature value into a pre-trained rancidity index prediction model, the rancidity index prediction model is obtained through machine learning algorithm learning on a training data set containing multiple groups of feature values and corresponding standard acid value and peroxide value labels, quantitative indexes of the acid value and the peroxide value are output through multivariate statistical calculation, a safety threshold is dynamically set in combination with the type of edible oil, and a rancidity judgment conclusion is output after comparison. The original electrochemical response signals received by the data acquisition module are one of a voltammetry curve, an impedance spectrum or a current time sequence, and each signal needs to meet statistical sampling requirements. The voltammetry curve is a differential pulse voltammetry curve or a square wave pulse voltammetry curve, and the signal quality of the two curves needs to be verified through a statistical method. The preprocessing module comprises the following steps: S1: a filtering algorithm based on signal statistical characteristics is adopted to denoise; 2. The rapid detection system for the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, S2: a statistical correction algorithm for baseline drift rules is used to eliminate drift; 3. The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 2, characterized in that, S3: a standardization algorithm is used to convert the processed data into a unified dimension data set.

4. The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, The feature analysis module extracts feature parameters related to rancidity, which need to be confirmed for relevance through statistical correlation analysis and verified for effectiveness through statistical significance test. The numerical calculation method adopted by the feature analysis module is one of a principal component analysis method, a partial least squares regression method or a support vector regression method, and each method needs to be combined with statistical optimization logic. The rancidity index prediction model of the evaluation and judgment module is obtained through machine learning algorithm learning on a training data set with statistical representation, and the training data set contains multiple groups of feature values and corresponding standard acid value and peroxide value labels. The quantitative indexes output by the evaluation and judgment module are the acid value and the peroxide value, and the accuracy of the indexes needs to be verified through a statistical method. The preset safety threshold of the evaluation and judgment module needs to be dynamically set in combination with the national standard limit value of the type of edible oil and in reference to the statistical distribution of historical detection data of the type of oil sample.

5. The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, The relationship between the quantitative indexes and the threshold value is analyzed through a statistical model, if the acid value or the peroxide value exceeds the statistical determination standard of the corresponding threshold value, rancidity is output; if both indexes do not exceed, no rancidity is output.

6. The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, ​ 7. The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, ​ 8.The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, ​ 9. The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, ​ 10. The rapid detection system of the degree of rancidity of edible oil based on an electrochemical sensor according to claim 1, characterized in that, ​

Images