A multi-parameter rapid screening food detection system

Abstract

The application discloses a kind of multi-parameter rapid screening food detection system, it is related to food detection technical field, comprising: the raw material place information of screening food is obtained, the pollution risk confidence index of screening food is evaluated;Establish known pollutant knowledge database;Collect the original signal parameter of screening food, analyze the signal fingerprint corresponding to original signal parameter, the pollution risk confidence index of screening food is used as interference factor and known pollutant knowledge database is associated and matched, and the pollutant list of screening food is generated;According to the pollutant list of screening food, the corresponding detection mode of the pollutant list of screening food is determined, the characteristic parameter data of each pollutant list is collected, and the comprehensive pollutant probability of screening food is evaluated;Determine whether the comprehensive pollutant probability of screening food exceeds predetermined food safety tolerance threshold, if yes, it is determined that it is not allowed to ship out.The application has the advantages that: detection efficiency is significantly improved, and detection cost is optimized.

Description

Technical Field

[0001] This invention relates to the field of food testing technology, specifically to a multi-parameter rapid screening food testing system. Background Technology

[0002] Food testing refers to the process of using physical, chemical, biological, or instrumental analysis techniques to qualitatively or quantitatively analyze the components, contaminants, additives, microorganisms, and other indicators in raw materials, processed products, and finished products of food in order to assess their safety, nutritional value, authenticity, and compliance.

[0003] Current food testing methods are insufficient to simultaneously screen multiple contaminants in complex food matrices. Furthermore, rapid food screening relies on step-by-step operations and manual interpretation, with individual sample testing taking several hours. This results in low efficiency and an inability to meet the rapid screening needs of large-scale food distribution. Summary of the Invention

[0004] To address the aforementioned technical problems, a multi-parameter rapid screening food detection system is provided. This technical solution solves the problem that existing food detection methods, when detecting contaminants, are unable to meet the need for simultaneous screening of multiple contaminants in complex matrix foods; and that rapid food screening relies on step-by-step operations and manual interpretation, with each sample taking several hours to test, resulting in low efficiency and an inability to meet the rapid screening needs of large-scale food distribution.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A rapid multi-parameter screening method for food detection includes:

[0007] Obtain information on the origin of raw materials for screened foods, mark potential contamination factors in the origin of raw materials, and assess the confidence index of contamination risk for screened foods.

[0008] Establish a knowledge database of known contaminants based on global historical food safety events;

[0009] The raw signal parameters of the screened food are collected, and the signal fingerprints corresponding to the raw signal parameters are analyzed. The contamination risk confidence index of the screened food is used as an interference factor and matched with the knowledge database of known pollutants to generate a list of pollutants in the screened food.

[0010] Based on the contaminant list of the screened food, determine the corresponding detection methods for the contaminants in the screened food, collect characteristic parameter data for each contaminant in the list, and assess the overall contaminant probability of the screened food.

[0011] Determine whether the overall contaminant probability of the screened food exceeds the predetermined food safety tolerance threshold. If not, allow shipment; if so, disallow shipment.

[0012] Furthermore, based on the raw material origin supply chain data of the screened food, historical contamination events in the raw material origin supply chain data are marked to determine the characteristic parameters of potential contamination factors in the screened food.

[0013] Obtain the geographic coordinate parameters of potential contaminants in the screened food, associate and bind them with the characteristic parameters of potential contaminants in the screened food, and construct a historical contamination event distribution matrix;

[0014] Using historical pollution events as label data, the mutual information value between each element in the historical pollution event distribution matrix and the label data is calculated. Elements with positive correlation in mutual information value are selected to obtain the set A of pollution factors positively correlated with historical pollution events.

[0015] Furthermore, the set of polluting factors positively correlated with historical pollution events is normalized.

[0016] Calculate the standard deviation and correlation coefficient among the characteristic parameters of each pollution factor in the set of positively correlated pollution factors of historical pollution events;

[0017] Using the CRITIC objective weighting method, weights are assigned to each pollution factor characteristic parameter in the set of positively correlated pollution factors of historical pollution events based on the standard deviation and correlation coefficient between the characteristic parameters of each pollution factor. This results in a weighted set B of positively correlated pollution factors of historical pollution events.

[0018] Furthermore, based on Bayesian networks, the weighted set of pollution factors in historical pollution events is used as input, and the pollution probability of pollution factors in historical pollution events is used as output.

[0019] Based on the pollution probability of pollution factors in historical pollution events and the weight of characteristic parameters of each pollution factor in historical pollution events, the confidence index of pollution risk of screened food is calculated.

[0020] 5. The multi-parameter rapid screening food detection method according to claim 4, characterized in that:

[0021] Based on microfluidic chips, raw signal parameter data of food is collected for screening.

[0022] Principal component analysis (PCA) was used to reduce the dimensionality of the original signal parameters of the screened food, resulting in the dimensionality-reduced parameter data of the original signal of the screened food.

[0023] Based on the original signal dimensionality reduction parameter data of the screened food, the time-domain and frequency-domain feature values ​​of the original signal dimensionality reduction parameter data are extracted using the wavelet transform method to obtain the original signal fingerprint set of the screened food.

[0024] The original signal fingerprint set of the screened food was corrected by using the confidence index of the contamination risk of the screened food as an interference factor, thus obtaining the corrected signal fingerprint set of the screened food.

[0025] Furthermore, based on the known pollutant knowledge database, a standardized signal fingerprint corresponding to each pollutant is determined, and a set of standardized signal fingerprints for known pollutants is constructed.

[0026] Using linear mapping, a vector transformation is performed on the modified signal fingerprint set of screened foods and the standardized signal fingerprint set of known contaminants to obtain the modified signal fingerprint vector set of screened foods and the standardized signal fingerprint vector set of known contaminants.

[0027] Based on the cosine similarity formula, the contamination risk confidence index of the screened food is used as an interference factor to calculate the matching degree between the modified signal fingerprint vector set of the screened food and the standardized signal fingerprint vector set of known contaminants, thereby generating a contaminant list for the screened food.

[0028] Furthermore, based on the contaminant list of the screened food, the corresponding detection methods for the contaminant list of the screened food are used to obtain the characteristic parameter data of the contaminant list of the screened food;

[0029] Standardize the characteristic parameter data of the contaminant inventory for screened foods;

[0030] Based on the toxicological data of the contaminants corresponding to the characteristic parameter data of the contaminant inventory of the screened food, determine the characteristic toxicity weights of the contaminant inventory of the screened food.

[0031] The original sample dataset of the screened food is packaged based on the matching degree between the modified signal fingerprint vector set of the screened food and the standardized signal fingerprint vector set of the known contaminants, the characteristic parameter data of each contaminant list of the screened food, and the characteristic toxicity weight of each contaminant list of the screened food.

[0032] Furthermore, based on logistic regression, a single contaminant assessment model for screening food was established;

[0033] Using the original sample dataset of screened foods, a single contaminant assessment model for screened foods is trained to quantify the risk probability of each contaminant in the screened foods.

[0034] By correcting using the probability multiplication formula, the risk probability of each contaminant in the screened food is integrated to obtain the comprehensive contaminant probability of the screened food.

[0035] Furthermore, a multi-parameter rapid screening food detection system includes:

[0036] Confidence module, knowledge base, pollutant inventory module, pollution rate assessment module, screening and judgment module;

[0037] The confidence module is used to obtain the origin information of the raw materials of the screened food, mark the potential contamination factors of the raw material origin information, and assess the confidence index of the contamination risk of the screened food, including:

[0038] include:

[0039] Based on the raw material origin supply chain data of the screened food, historical contamination events in the raw material origin supply chain data are marked to determine the characteristic parameters of potential contamination factors of the screened food.

[0040] Obtain the geographic coordinate parameters of potential contaminants in the screened food, associate and bind them with the characteristic parameters of potential contaminants in the screened food, and construct a historical contamination event distribution matrix;

[0041] Using historical pollution events as label data, the mutual information value between each element in the historical pollution event distribution matrix and the label data is calculated. Elements with positive correlation in mutual information value are selected to obtain the set A of pollution factors positively correlated with historical pollution events.

[0042] Normalization was performed on the set of polluting factors positively correlated with historical pollution events;

[0043] Calculate the standard deviation and correlation coefficient among the characteristic parameters of each pollution factor in the set of positively correlated pollution factors of historical pollution events;

[0044] Using the CRITIC objective weighting method, weights are assigned to each pollution factor characteristic parameter in the set of positively correlated pollution factors of historical pollution events based on the standard deviation and correlation coefficient between the characteristic parameters of each pollution factor in the set of positively correlated pollution factors of historical pollution events, thus obtaining the weighted set B of positively correlated pollution factors of historical pollution events;

[0045] Based on Bayesian networks, the weighted set of pollution factors in historical pollution events is used as input, and the pollution probability of pollution factors in historical pollution events is used as output.

[0046] Based on the pollution probability of pollution factors in historical pollution events and the weight of characteristic parameters of each pollution factor in historical pollution events, the confidence index of pollution risk of screened food is calculated.

[0047] The knowledge base is used to establish a database of known contaminants based on global historical food safety events;

[0048] The contaminant inventory module is electrically connected to the knowledge base and the confidence module. The contaminant inventory module is used to collect the original signal parameters of the screened food, analyze the signal fingerprints corresponding to the original signal parameters, use the contamination risk confidence index of the screened food as an interference factor, and match it with the known contaminant knowledge database to generate a contaminant inventory of the screened food.

[0049] The contamination rate assessment module is electrically connected to the contaminant inventory module. The contamination rate assessment module is used to determine the corresponding detection method for each contaminant in the screened food based on the contaminant inventory, collect characteristic parameter data for each contaminant in the inventory, and assess the overall contamination probability of the screened food, including:

[0050] Based on microfluidic chips, raw signal parameter data of food is collected for screening.

[0051] Principal component analysis (PCA) was used to reduce the dimensionality of the original signal parameters of the screened food, resulting in the dimensionality-reduced parameter data of the original signal of the screened food.

[0052] Based on the original signal dimensionality reduction parameter data of the screened food, the time-domain and frequency-domain feature values ​​of the original signal dimensionality reduction parameter data are extracted using the wavelet transform method to obtain the original signal fingerprint set of the screened food.

[0053] The original signal fingerprint set of the screened food was corrected by using the contamination risk confidence index of the screened food as an interference factor, resulting in the corrected signal fingerprint set of the screened food.

[0054] Based on the contaminant list of the screened food, determine the corresponding detection methods for the contaminants in the screened food, collect characteristic parameter data for each contaminant in the list, and assess the overall contaminant probability of the screened food.

[0055] The screening and judgment module is electrically connected to the contamination rate assessment module. The screening and judgment module is used to determine whether the overall contaminant probability of the screened food exceeds the predetermined food safety tolerance threshold. If not, it determines that shipment is allowed; if so, it determines that shipment is not allowed.

[0056] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0057] This invention proposes a multi-parameter rapid food screening detection scheme. It employs a pre-set dual food contaminant detection technology, using raw material origin information and a historical contamination database to predict potential risks. By matching signal fingerprints with a known contaminant database, a contaminant list for screening foods is rapidly generated. A comprehensive detection of this contaminant list is then performed, collecting characteristic data and calculating the overall contamination probability. This is compared with safety thresholds to achieve rapid decision-making. The advantages of this scheme are: significantly improved detection efficiency and optimized detection costs. Attached Figure Description

[0058] Figure 1 This is a flowchart of a multi-parameter rapid screening method for food testing.

[0059] Figure 2 This is a framework diagram of a multi-parameter rapid screening food testing system. Detailed Implementation

[0060] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0061] Reference Figure 1 As shown, a multi-parameter rapid screening method for food detection includes:

[0062] Step 1: Obtain the origin information of the raw materials of the screened food, mark the potential contamination factors of the raw material origin information, and assess the contamination risk confidence index of the screened food.

[0063] Step one includes the following:

[0064] Step 101: Based on the raw material origin supply chain data of the screened food, mark the historical contamination events in the raw material origin supply chain data, and determine the characteristic parameters of potential contamination factors of the screened food.

[0065] Obtain the geographic coordinate parameters of potential contaminants in the screened food, associate and bind them with the characteristic parameters of potential contaminants in the screened food, and construct a historical contamination event distribution matrix;

[0066] Using historical pollution events as label data, the mutual information value between each element in the historical pollution event distribution matrix and the label data is calculated. Elements with positive correlation in mutual information values ​​are selected to obtain the set of pollution factors positively correlated with historical pollution events. ;in, The characteristic parameter value of the j-th pollution factor that is positively correlated with the i-th pollution event in history;

[0067] Step 102: Normalize the set of polluting factors positively correlated with historical pollution events;

[0068] Calculate the standard deviation and correlation coefficient among the characteristic parameters of each pollution factor in the set of positively correlated pollution factors of historical pollution events;

[0069] Using the CRITIC objective weighting method, weights are assigned to each pollution factor characteristic parameter in the set of positively correlated pollution factors of historical pollution events based on the standard deviation and correlation coefficient between the characteristic parameters of each pollution factor. This yields the weighted set B of positively correlated pollution factors of historical pollution events, as follows: ,

[0070] in, The weight of the feature parameter of the j-th pollution factor that is positively correlated with the i-th pollution event in history. Let $\begin{pmatrix} \ ... The correlation coefficient between the characteristic parameters of the j-th pollution factor that is positively associated with the i-th pollution event in history;

[0071] Step 103: Based on Bayesian networks, take the weighted set of pollution factors of historical pollution events as input and the pollution probability of pollution factors of historical pollution events as output.

[0072] Based on the pollution probability of pollution factors in historical pollution events and the weight of characteristic parameters of each pollution factor in historical pollution events, the confidence index of pollution risk for screened foods is calculated as follows: ,

[0073] in, To screen for the risk of food contamination, a confidence level indicator is needed. Let Y be a probability function, and Y be a binary random variable.

[0074] When using it, refer to the content in sections 101 to 103.

[0075] As a further step, a precise assessment of food contamination risk is achieved by binding historical contamination events with corresponding geographic information, using mutual information filtering, CRITIC weighting, and Bayesian evaluation: First, a distribution matrix of historical contamination events is constructed based on the production area supply chain data, and a set of positively correlated contamination factors is obtained through mutual information filtering; then, the CRITIC method is used to calculate the objective weight of each factor to form a weighted feature set; finally, a Bayesian network is used to calculate the weighted contamination probability and generate a risk confidence index.

[0076] The beneficial effects are:

[0077] 1. Enhance the targeting of screening by linking geographic information with the spatiotemporal correlation of pollution factors;

[0078] 2. Employing a dual screening method of mutual information and CRITIC ensures the objectivity of key factor identification;

[0079] 3. The evaluation model that combines Bayesian networks and weights makes the risk quantification results applicable to large-scale real-time screening scenarios.

[0080] Step 2: Establish a knowledge database of known contaminants based on global historical food safety events.

[0081] Step 3: Collect the original signal parameters of the screened food, analyze the signal fingerprints corresponding to the original signal parameters, use the contamination risk confidence index of the screened food as an interference factor and match it with the known contaminant knowledge database to generate a contaminant list of the screened food.

[0082] Step three includes the following:

[0083] Step 301: Based on the microfluidic chip, collect the raw signal parameter data of the screened food;

[0084] Principal component analysis (PCA) was used to reduce the dimensionality of the original signal parameters of the screened food, resulting in the dimensionality-reduced parameter data of the original signal of the screened food.

[0085] Based on the original signal dimensionality reduction parameter data of the screened food, the time-domain and frequency-domain feature values ​​of the original signal dimensionality reduction parameter data are extracted using the wavelet transform method to obtain the original signal fingerprint set of the screened food.

[0086] The original signal fingerprint set of the screened food was corrected using the contamination risk confidence index as an interference factor, resulting in the corrected signal fingerprint set of the screened food. The method is as follows: ,

[0087] in, To screen the modified signal fingerprint set of food, To screen the original signal fingerprint set of food, For adjustment coefficients;

[0088] As further content, The range of adjustment coefficients is determined by fitting a curve of the impact of risk confidence on fingerprint signal matching error using the least squares method, in order to minimize the false positive rate. The value, this coefficient is essentially a coupling factor between risk confidence and fingerprint signal detection sensitivity;

[0089] Step 302: Based on the known pollutant knowledge database, determine the standardized signal fingerprint corresponding to each pollutant and construct a set of standardized signal fingerprints for known pollutants.

[0090] Using linear mapping, a vector transformation is performed on the modified signal fingerprint set of screened foods and the standardized signal fingerprint set of known contaminants to obtain the modified signal fingerprint vector set of screened foods and the standardized signal fingerprint vector set of known contaminants.

[0091] Based on the cosine similarity formula, the contamination risk confidence index of the screened food is used as an interference factor. The matching degree between the corrected signal fingerprint vector set of the screened food and the standardized signal fingerprint vector set of known contaminants is calculated to generate a contaminant list for the screened food. The method is as follows:

[0092] in, To screen for contaminants in food, Let j be the standardized signal fingerprint vector of the j-th pollutant in the set of standardized signal fingerprint vectors of known pollutants. To determine the matching degree between the set of corrected signal fingerprint vectors for screening food and the standardized signal fingerprint of the j-th contaminant in the set of standardized signal fingerprint vectors for known contaminants, Given a set of standardized signal fingerprint vectors for known pollutants, As the baseline threshold, The adjustment factor is m, which represents the total number of pollutant standardized signal fingerprint vectors.

[0093] When using this, please refer to the content in sections 301 to 302:

[0094] As a further step, a technical approach of microfluidic chip-PCA dimensionality reduction-wavelet feature extraction-risk weight correction-dynamic cosine matching is used to achieve accurate pollutant screening. First, multidimensional signal data is collected using a microfluidic chip, and time-frequency domain features are extracted through PCA dimensionality reduction and wavelet transform to form a signal fingerprint. Then, the signal fingerprint is weighted and corrected using the previously calculated pollution risk confidence index to enhance the sensitivity of high-risk samples. Finally, the corrected signal fingerprint is compared with a pollutant standard database to generate a list through linear mapping and dynamically adjusted cosine similarity matching.

[0095] The beneficial effects are:

[0096] 1. The combination of microfluidic chips and PCA-wavelet improves signal acquisition efficiency and feature representation capability, and reduces data redundancy compared with traditional methods;

[0097] 2. The dual interference mechanism of risk confidence (signal correction + dynamic threshold) improves the detection rate of low-concentration pollutants;

[0098] 3. Standardized vector mapping ensures the comparability of results from different test batches, making it suitable for rapid screening scenarios on production lines. It can accurately screen for potential contaminants in food, optimizing the accuracy and efficiency of food safety testing.

[0099] Step 4: Based on the list of contaminants in the screened food, determine the corresponding detection methods for the contaminants in the screened food, collect characteristic parameter data for each contaminant in the list, and assess the overall contaminant probability of the screened food.

[0100] Step four includes the following:

[0101] Step 401: Based on the contaminant list of the screened food, the corresponding detection methods for the contaminant list of the screened food are used to obtain the characteristic parameter data of the contaminant list of the screened food.

[0102] Standardize the characteristic parameter data of the contaminant inventory for screened foods;

[0103] Based on the toxicological data of the contaminants corresponding to the characteristic parameter data of the contaminant inventory of the screened food, determine the characteristic toxicity weights of the contaminant inventory of the screened food.

[0104] Based on the matching degree between the modified signal fingerprint vector set of the screened food and the standardized signal fingerprint vector set of known contaminants, the characteristic parameter data of each contaminant list of the screened food, and the characteristic toxicity weight of each contaminant list of the screened food, the original sample dataset of the screened food is packaged.

[0105] Step 402: Based on logistic regression, establish a single contaminant assessment model for screening food.

[0106] Using the original sample dataset of screened foods, a single contaminant assessment model for screened foods is trained to quantify the risk probability of each contaminant in the screened foods, as follows: ,

[0107] in, To screen for the risk probability of the j-th contaminant in food, For logical functions, To screen for the j-th contaminant in a food inventory, Let be the regression coefficient of the j-th pollutant inventory;

[0108] By correcting using the probability multiplication formula and integrating the risk probabilities of each contaminant in the screened food, the overall contaminant probability of the screened food is obtained, as follows: ,

[0109] in, To screen for the overall probability of contaminants in food, The characteristic toxicity weights for the j-th contaminant list in the screening of food.

[0110] When using this, refer to sections 401 to 402:

[0111] As a further development, a technical approach of multi-source data fusion, toxicity weight correction, and logistic regression modeling is used to achieve accurate quantification of pollutant risk: First, quantitative feature parameters are obtained based on the pollutant inventory matching detection method, and feature toxicity weights are assigned in combination with toxicological data. The matching degree, detection data, and toxicity weights are then packaged into a standardized dataset. Subsequently, a single pollutant risk probability model is constructed through logistic regression. The toxicity weights and detection concentrations are fused using the Sigmoid function to output the risk probability of a single pollutant. Finally, the risk probabilities of single pollutants are fused according to the probability multiplication formula to generate the comprehensive pollutant probability of the screened food.

[0112] The beneficial effects are:

[0113] 1. The coordinated correction of toxicity weights and detection data makes risk assessment conform to toxicological principles and avoids the limitations of relying solely on concentration determination;

[0114] 2. The Logistic regression model has strong interpretability, supports threshold-based decision-making on risk probabilities, and improves the accuracy of risk assessment;

[0115] 3. The standardized data packaging mechanism is compatible with results from multiple testing platforms and is suitable for integrating heterogeneous data from laboratory and field rapid screening.

[0116] Step 5: Determine whether the overall contaminant probability of the screened food exceeds the predetermined food safety tolerance threshold. If not, ship is allowed; if so, ship is not allowed.

[0117] Reference Figure 2 As shown, a multi-parameter rapid screening food detection system includes:

[0118] Confidence module, knowledge base, pollutant inventory module, pollution rate assessment module, screening and judgment module;

[0119] The confidence module is used to obtain the origin information of the raw materials of the screened food, mark the potential contamination factors of the raw material origin information, and assess the confidence index of the contamination risk of the screened food, including:

[0120] include:

[0121] Based on the raw material origin supply chain data of the screened food, historical contamination events in the raw material origin supply chain data are marked to determine the characteristic parameters of potential contamination factors of the screened food.

[0122] Obtain the geographic coordinate parameters of potential contaminants in the screened food, associate and bind them with the characteristic parameters of potential contaminants in the screened food, and construct a historical contamination event distribution matrix;

[0123] Using historical pollution events as label data, the mutual information value between each element in the historical pollution event distribution matrix and the label data is calculated. Elements with positive correlation in mutual information value are selected to obtain the set A of pollution factors positively correlated with historical pollution events.

[0124] Normalization was performed on the set of polluting factors positively correlated with historical pollution events;

[0125] Calculate the standard deviation and correlation coefficient among the characteristic parameters of each pollution factor in the set of positively correlated pollution factors of historical pollution events;

[0126] Using the CRITIC objective weighting method, weights are assigned to each pollution factor characteristic parameter in the set of positively correlated pollution factors of historical pollution events based on the standard deviation and correlation coefficient between the characteristic parameters of each pollution factor in the set of positively correlated pollution factors of historical pollution events, thus obtaining the weighted set B of positively correlated pollution factors of historical pollution events;

[0127] Based on Bayesian networks, the weighted set of pollution factors in historical pollution events is used as input, and the pollution probability of pollution factors in historical pollution events is used as output.

[0128] Based on the pollution probability of pollution factors in historical pollution events and the weight of characteristic parameters of each pollution factor in historical pollution events, the confidence index of pollution risk of screened food is calculated.

[0129] The knowledge base is used to establish a database of known contaminants based on global historical food safety events;

[0130] The contaminant inventory module is electrically connected to the knowledge base and the confidence module. The contaminant inventory module is used to collect the original signal parameters of the screened food, analyze the signal fingerprints corresponding to the original signal parameters, use the contamination risk confidence index of the screened food as an interference factor, and match it with the known contaminant knowledge database to generate a contaminant inventory of the screened food.

[0131] The contamination rate assessment module is electrically connected to the contaminant inventory module. The contamination rate assessment module is used to determine the corresponding detection method for each contaminant in the screened food based on the contaminant inventory, collect characteristic parameter data for each contaminant in the inventory, and assess the overall contamination probability of the screened food, including:

[0132] Based on microfluidic chips, raw signal parameter data of food is collected for screening.

[0133] Principal component analysis (PCA) was used to reduce the dimensionality of the original signal parameters of the screened food, resulting in the dimensionality-reduced parameter data of the original signal of the screened food.

[0134] Based on the original signal dimensionality reduction parameter data of the screened food, the time-domain and frequency-domain feature values ​​of the original signal dimensionality reduction parameter data are extracted using the wavelet transform method to obtain the original signal fingerprint set of the screened food.

[0135] The original signal fingerprint set of the screened food was corrected by using the contamination risk confidence index of the screened food as an interference factor, resulting in the corrected signal fingerprint set of the screened food.

[0136] Based on the contaminant list of the screened food, determine the corresponding detection methods for the contaminants in the screened food, collect characteristic parameter data for each contaminant in the list, and assess the overall contaminant probability of the screened food.

[0137] The screening and judgment module is electrically connected to the contamination rate assessment module. The screening and judgment module is used to determine whether the overall contaminant probability of the screened food exceeds the predetermined food safety tolerance threshold. If not, it determines that shipment is allowed; if so, it determines that shipment is not allowed.

[0138] In summary, the advantages of this invention are as follows:

[0139] The solution employs a pre-set dual food contaminant detection technology. It anticipates potential risks by leveraging raw material origin information and a historical contamination database. Then, it uses signal fingerprinting to match a known contaminant database, rapidly generating a contaminant list for screened foods. A comprehensive test is then conducted on this list, collecting characteristic data and calculating the overall contamination probability. This is compared to safety thresholds to enable rapid decision-making. The benefits of this solution include significantly improved detection efficiency and optimized detection costs.

[0140] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A multi-parameter rapid screening food detection method, characterized in that, The method comprises the following steps: Obtain the raw material origin information of the screening food, mark the potential pollution factors of the raw material origin information, evaluate the pollution risk confidence index of the screening food, including: Based on the raw material origin supply chain data of the screening food, mark the historical pollution events in the raw material origin supply chain data, determine the characteristic parameters of the potential pollution factors of the screening food; Obtain the geographic coordinate parameters of the potential pollution factors of the screening food, bind the geographic coordinate parameters with the characteristic parameters of the potential pollution factors of the screening food, and establish a historical pollution event distribution matrix; Take the historical pollution events as label data, calculate the mutual information value between each element in the historical pollution event distribution matrix and the label data, select the elements with positive correlation of mutual information value, and obtain a set A of positive correlation pollution factors of the historical pollution events; Normalization processing is performed on the set A of positive correlation pollution factors of the historical pollution events; Calculate the standard deviation and correlation coefficient between the characteristic parameters of each pollution factor in the set A of positive correlation pollution factors of the historical pollution events; Using CRITIC objective weighting method, according to the standard deviation and correlation coefficient between the characteristic parameters of each pollution factor in the set A of positive correlation pollution factors of the historical pollution events, a weight is assigned to each pollution factor characteristic parameter to obtain a weighted set B of positive correlation pollution factors of the historical pollution events; Based on the Bayesian network, taking the weighted set B of positive correlation pollution factors of the historical pollution events as input and the pollution probability of the pollution factors of the historical pollution events as output; According to the pollution probability of the pollution factors of the historical pollution events and the weight of each pollution factor characteristic parameter, the pollution risk confidence index of the screening food is calculated; Based on global historical food safety events, a known pollutant knowledge database is established; Collect the original signal parameters of the screening food, analyze the signal fingerprints corresponding to the original signal parameters, and associate and match the pollution risk confidence index of the screening food with the known pollutant knowledge database as an interference factor to generate a pollutant list of the screening food, including: Based on the microfluidic chip, the original signal parameter data of the screening food is collected; Using PCA principal component analysis method, the data dimensionality of the original signal parameters of the screening food is reduced to obtain the original signal reduced parameter data of the screening food; According to the original signal reduced parameter data of the screening food, the time domain and frequency domain characteristic values of the original signal reduced parameter data are extracted by using wavelet transform method to obtain the original signal fingerprint set of the screening food; The pollution risk confidence index of the screening food is taken as an interference factor to correct the original signal fingerprint set of the screening food to obtain a corrected signal fingerprint set of the screening food; According to the pollutant list of the screening food, the corresponding detection method of the pollutant list of the screening food is determined, the characteristic parameter data of each pollutant list is collected, and the comprehensive pollutant probability of the screening food is evaluated; Determine whether the comprehensive pollutant probability of the screening food exceeds the predetermined food safety tolerance threshold, if not, determine that the screening food is allowed to be shipped, if yes, determine that the screening food is not allowed to be shipped.

2. The multi-parameter rapid screening food detection method according to claim 1, wherein: Based on the known pollutant knowledge database, the standardized signal fingerprint corresponding to each pollutant is determined, and a standardized signal fingerprint set of known pollutants is established; Using linear mapping, the vector conversion is performed on the modified signal fingerprint set of the screened food and the standardized signal fingerprint set of the known pollutants, to obtain a modified signal fingerprint vector set of the screened food and a standardized signal fingerprint vector set of the known pollutants; Based on the cosine similarity formula, the matching degree between the modified signal fingerprint vector set of the screened food and the standardized signal fingerprint vector set of the known pollutants is calculated by taking the pollution risk confidence index of the screened food as an interference factor, and a pollutant list of the screened food is generated.

3. The multi-parameter rapid screening food detection method according to claim 2, characterized in that: Based on the pollutant list of the screened food, the characteristic parameter data of the pollutant list of the screened food is obtained according to the corresponding detection mode of the pollutant list of the screened food; The characteristic parameter data of the pollutant list of the screened food is standardized; According to the toxicological data of the pollutants corresponding to the characteristic parameter data of the pollutant list of the screened food, the characteristic toxicity weight of the pollutant list of the screened food is determined; According to the matching degree between the modified signal fingerprint vector set of the screened food and the standardized signal fingerprint vector set of the known pollutants, the characteristic parameter data of each pollutant list of the screened food, and the characteristic toxicity weight of each pollutant list of the screened food, the original sample data set of the screened food is packaged.

4. The multi-parameter rapid screening food detection method according to claim 3, characterized in that: Based on logistic regression, a single-pollutant evaluation model of the screened food is established; Using the original sample data set of the screened food, the single-pollutant evaluation model of the screened food is trained to quantify the risk probability of each pollutant of the screened food; Using the probability multiplication formula correction, the risk probability of each pollutant of the screened food is fused to obtain the comprehensive pollutant probability of the screened food.

5. A multi-parameter rapid screening food detection system, characterized in that, It includes: confidence module, knowledge base, pollutant list module, pollution rate evaluation module, screening judgment module; The confidence module is used to obtain the raw material origin information of the screened food, mark the potential pollution factors of the raw material origin information, and evaluate the pollution risk confidence index of the screened food, including: Based on the raw material origin supply chain data of the screened food, the historical pollution events in the raw material origin supply chain data are marked, and the characteristic parameters of the potential pollution factors of the screened food are determined; The geographic coordinate parameters of the potential pollution factors of the screened food are obtained and associated with the characteristic parameters of the potential pollution factors of the screened food to establish a historical pollution event distribution matrix; The historical pollution events are taken as label data, the mutual information value between each element in the historical pollution event distribution matrix and the label data is calculated, the elements with positive correlation of mutual information value are selected, and a historical pollution event positively correlated pollution factor set A is obtained; The historical pollution event positively correlated pollution factor set is normalized; The standard deviation and correlation coefficient between the characteristic parameters of each pollution factor in the historical pollution event positively correlated pollution factor set are calculated; ​ The CRITIC objective weighting method is used, the standard deviation and the correlation coefficient between each pollution factor characteristic parameter in the positive correlation pollution factor set of the historical pollution event are used to weight each pollution factor characteristic parameter, and a weighted set B of the positive correlation pollution factor of the historical pollution event is obtained; Based on the Bayesian network, the weighted set of the pollution factor of the historical pollution event is used as the input, and the pollution probability of the pollution factor of the historical pollution event is used as the output; According to the pollution probability of the pollution factor of the historical pollution event and the weight of each pollution factor characteristic parameter of the historical pollution event, the pollution risk confidence index of the screened food is calculated; The knowledge base is used to establish a known pollutant knowledge database based on global historical food safety events; The pollutant list module is electrically connected with the knowledge base and the confidence module, and the pollutant list module is used to collect the original signal parameters of the screened food, analyze the signal fingerprint corresponding to the original signal parameters, and associate and match the pollution risk confidence index of the screened food with the known pollutant knowledge database as an interference factor to generate a pollutant list of the screened food; The pollution rate evaluation module is electrically connected with the pollutant list module, and the pollution rate evaluation module is used to determine the corresponding detection mode of the pollutant list of the screened food according to the pollutant list of the screened food, collect the characteristic parameter data of each pollutant list, and evaluate the comprehensive pollutant probability of the screened food, including: Based on the microfluidic chip, the original signal parameter data of the screened food is collected; PCA principal component analysis method is used to reduce the dimension of the original signal parameter data of the screened food, and the original signal dimension reduction parameter data of the screened food is obtained; According to the original signal dimension reduction parameter data of the screened food, the wavelet transform method is used to extract the time domain frequency domain characteristic value of the original signal dimension reduction parameter data, and the original signal fingerprint set of the screened food is obtained; The pollution risk confidence index of the screened food is used as an interference factor to correct the original signal fingerprint set of the screened food, and the corrected signal fingerprint set of the screened food is obtained; According to the pollutant list of the screened food, the corresponding detection mode of the pollutant list of the screened food is determined, the characteristic parameter data of each pollutant list is collected, and the comprehensive pollutant probability of the screened food is evaluated; The screening judgment module is electrically connected with the pollution rate evaluation module, and the screening judgment module is used to judge whether the comprehensive pollutant probability of the screened food exceeds the predetermined food safety tolerance threshold, if not, it is determined that the goods are allowed to be shipped, if yes, it is determined that the goods are not allowed to be shipped.

Images