Food safety detection method and system based on intelligent spectrophotometry

Abstract

The invention provides a food safety detection method based on intelligent spectrophotometry, and the method comprises the steps: carrying out the spectrum scanning of a food sample through an intelligent spectrophotometer, obtaining the high-resolution absorption spectrum data of the sample, carrying out the mode recognition of the high-resolution absorption spectrum data through a deep learning model, and carrying out the recognition of the high-resolution absorption spectrum data. According to the method, natural components in food and illegal additives existing outside a preset target are distinguished, meanwhile, a component concentration distribution diagram is generated, interaction among the components in the food and potential influences of the components are evaluated by adopting a multivariable statistical analysis method in combination with a food component spectrum database, and a health risk evaluation report is generated. And mapping the component concentration distribution diagram and the health risk assessment report into a preset safety risk model, and generating a comprehensive safety assessment report. According to the technical scheme provided by the invention, the detection time is shortened, the accuracy and reliability of detection are improved, more comprehensive assessment is provided for food safety, and the safety of food is improved. And a one-stop solution from detection to evaluation is realized.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of intelligent spectrophotometry, and in particular to a food safety detection method based on intelligent spectrophotometry. Background Art

[0002] With the improvement of living standards, consumers are paying more and more attention to food safety. Traditional food safety detection methods mainly rely on chemical analysis and laboratory tests, which are often time-consuming, costly, and difficult to achieve large-scale rapid screening. In recent years, intelligent spectrophotometry technology, as a non-destructive rapid detection method, has shown great potential in the field of food safety. This technology can quickly obtain food ingredient information by analyzing the absorption spectrum of food samples in the visible to near-infrared range, making it possible to identify natural ingredients and illegal additives.

[0003] However, current food safety testing methods based on spectrophotometry still have some limitations. On the one hand, traditional methods have difficulty in accurately distinguishing natural ingredients from illegal additives when dealing with complex food samples, especially when faced with trace or trace additives, and their detection accuracy and reliability are insufficient. On the other hand, existing risk assessments mostly remain at the qualitative description stage, lacking quantitative analysis of the interactions between various ingredients in food and their potential health effects, resulting in the generated health risk assessment reports being not comprehensive and specific enough. Summary of the invention

[0004] The present invention provides a food safety detection method and system based on intelligent spectrophotometry

[0005] , which is used to solve the problem that traditional methods in the existing technology are difficult to accurately distinguish natural ingredients from illegal additives when processing complex food samples, especially when facing trace or trace additives, the detection accuracy and reliability are insufficient, and there is a lack of quantitative analysis of the interactions between various ingredients in food and their potential health effects, resulting in the generated health risk assessment report being not comprehensive and specific enough.

[0006] In a first aspect, an embodiment of the present invention provides a food safety detection method based on intelligent spectrophotometry, comprising:

[0007] The food sample is scanned by a smart spectrophotometer for a preset number of times to obtain high-resolution absorption spectrum data of the sample in the range from visible light to near infrared;

[0008] Using a deep learning model to perform pattern recognition on the high-resolution absorption spectrum data, distinguishing natural ingredients in food from illegal additives outside of preset targets, and generating a component concentration distribution map;

[0009] Based on the component concentration distribution map, combined with a dynamically updated food component spectral database, a multivariate statistical analysis method is used to evaluate the interaction between the components in the food and the potential impact of each component, and a health risk assessment report is generated;

[0010] The ingredient concentration distribution map and the health risk assessment report are mapped to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake.

[0011] Optionally, a deep learning model is used to perform pattern recognition on the high-resolution absorption spectrum data to distinguish natural ingredients in food from illegal additives outside the preset target, and a component concentration distribution map is generated, including:

[0012] Inputting the high-resolution absorption spectrum data into a deep convolutional neural network trained with a preset target component spectrum, and using the deep convolutional neural network to perform feature recognition processing on the high-resolution absorption spectrum data to obtain the spectral features of natural ingredients and illegal additives;

[0013] Based on the spectral features, an adaptive threshold segmentation algorithm is used to perform optimization processing to obtain optimized results for distinguishing natural ingredients from illegal additives;

[0014] Combining the optimized differentiation results of natural ingredients and illegal additives, a concentration estimation process is performed using a nonlinear regression analysis method to obtain a preliminary component concentration distribution map;

[0015] The preliminary component concentration distribution map is corrected by introducing a chemometric method to obtain a target component concentration distribution map.

[0016] Optionally, the high-resolution absorption spectrum data is input into a deep convolutional neural network trained with a preset target component spectrum, and the deep convolutional neural network is used to perform feature recognition processing on the high-resolution absorption spectrum data to obtain the spectral features of natural ingredients and illegal additives, including:

[0017] A multi-channel spectrum acquisition system is used to collect high-resolution absorption spectrum data of food samples in a wide bandwidth, and high-resolution absorption spectrum data covering characteristic absorption peaks of all natural ingredients and illegal additives are obtained;

[0018] Preprocessing the high-resolution absorption spectrum data using a baseline correction method to obtain absorption spectrum data;

[0019] Constructing and using a deep convolutional neural network model that has been pre-trained with a large number of spectral data sets of known ingredients to perform feature recognition on the absorption spectral data to obtain preliminary spectral features containing natural ingredients and illegal additives;

[0020] The preliminary spectral features are optimized using an adaptive threshold segmentation algorithm to obtain optimized target spectral features of natural ingredients and illegal additives.

[0021] Optionally, the preliminary spectral features are optimized using an adaptive threshold segmentation algorithm to obtain optimized target spectral features of natural ingredients and illegal additives, including:

[0022] Using an adaptive threshold segmentation algorithm combined with a local contrast enhancement technique, the preliminary spectral features are subjected to detail enhancement processing to obtain enhanced preliminary spectral features;

[0023] Based on the enhanced preliminary spectral features, a multi-scale morphological analysis method is used to extract structural features to obtain spectral features with multi-scale structural features;

[0024] The spectral features with multi-scale structural features are compared with a preset standard spectral library, and a similarity matching algorithm based on deep learning is used for accurate calibration to obtain calibrated spectral features;

[0025] Combined with the calibrated spectral features, an iterative optimization process is performed using a genetic algorithm or a particle swarm optimization algorithm to obtain optimized target spectral features of natural ingredients and illegal additives.

[0026] Optionally, based on the component concentration distribution map, combined with a dynamically updated food component spectrum database, a multivariate statistical analysis method is used to evaluate the interaction between the components in the food and the potential impact of each component to generate a health risk assessment report, including:

[0027] Using advanced spectral analysis technology to extract feature vectors from the component concentration distribution diagram to obtain feature vectors of key components;

[0028] Comparing the characteristic vector of the key component with the standard spectral feature in the dynamically updated food component spectral database to obtain a comparison result;

[0029] Based on the comparison results, a method combining principal component analysis and partial least squares regression is used to perform dimensionality reduction processing on each component in the food to obtain a component data set after dimensionality reduction;

[0030] Inputting the reduced-dimensional component data set into a multivariate statistical model based on a deep neural network, calculating the correlation coefficient matrix between the components, and obtaining an interaction evaluation between the components;

[0031] Collecting the user's personal health record data, the personal health record data includes: age, gender, weight, health status;

[0032] The personal health record data is combined with the interaction assessment between the components, and the potential impact of each component is quantified using a Bayesian reasoning algorithm to generate a health risk assessment report.

[0033] Optionally, the personal health record data is combined with the interaction assessment between the components, and the potential impact of each component is quantified using a Bayesian inference algorithm to generate a health risk assessment report, including:

[0034] Combined with the interaction evaluation among the components, a Bayesian inference algorithm is used to perform probability modeling on the potential impact of each component based on prior knowledge and observation data to obtain the probability distribution of the potential impact of each component;

[0035] Based on the probability distribution of the potential impact of each component, uncertainty analysis is introduced, and the health risks of each component under different scenarios are evaluated by Monte Carlo simulation method to obtain scenario analysis results;

[0036] Combining the scenario analysis results with personal health record data, using a personalized risk assessment model to calculate the impact of individual differences on health risks, and generating a personalized health risk assessment report;

[0037] The personalized health risk assessment report is associated with the source tracking information of the food ingredients, and the source tracking information is made transparent and tamper-proof through blockchain technology to form a target health risk assessment report.

[0038] Optionally, the ingredient concentration distribution map and the health risk assessment report are mapped to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake, including:

[0039] Inputting the ingredient concentration distribution map and the evaluation results into a preset safety risk model constructed based on historical food safety data and epidemiological research results, performing preliminary mapping processing, and obtaining preliminary mapping results;

[0040] The preliminary mapping results are processed by a multi-factor risk assessment algorithm to calculate a comprehensive risk index to obtain a food safety score;

[0041] Based on the food safety score, health risk warning information is generated and processed using preset health risk warning rules to obtain health risk warning information, wherein the health risk warning information includes: short-term consumption risk, long-term consumption risk, and consumption risk for special groups;

[0042] In combination with the health risk warning information, a recommended intake amount is formulated through a comprehensive analysis of nutrition and toxicology to obtain a specific recommended intake amount;

[0043] The food safety score, health risk warning information and specific recommended intake are integrated, and natural language generation technology is used to convert text descriptions and chart displays to obtain a comprehensive safety assessment report.

[0044] In a second aspect, the present application provides a food safety detection system based on intelligent spectrophotometry, including:

[0045] A scanning module is used to perform a preset number of spectral scans on food samples through an intelligent spectrophotometer to obtain high-resolution absorption spectrum data of the samples in the range of visible light to near infrared;

[0046] An identification module, for performing pattern recognition on the high-resolution absorption spectrum data using a deep learning model, distinguishing natural ingredients in food from illegal additives outside of preset targets, and generating a component concentration distribution map;

[0047] An evaluation module, for evaluating the interaction between ingredients in food and the potential impact of each ingredient based on the ingredient concentration distribution map and in combination with a dynamically updated food ingredient spectrum database, and generating a health risk assessment report by using a multivariate statistical analysis method;

[0048] A generation module is used to map the ingredient concentration distribution map and the health risk assessment report to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake.

[0049] In a third aspect, an embodiment of the present invention provides a computing device, comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute a food safety detection method based on intelligent spectrophotometry as described in any one of the first aspects.

[0050] In a fourth aspect, an embodiment of the present invention provides a computer storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement a food safety detection method based on intelligent spectrophotometry as described in any one of the first aspects.

[0051] In an embodiment of the present invention, a food sample is spectrally scanned a preset number of times by an intelligent spectrophotometer to obtain high-resolution absorption spectrum data of the sample in the range of visible light to near infrared, and a deep learning model is used to perform pattern recognition on the high-resolution absorption spectrum data to distinguish natural ingredients in food from illegal additives outside the preset target, and a component concentration distribution map is generated at the same time. Based on the component concentration distribution map, combined with a dynamically updated food component spectrum database, a multivariate statistical analysis method is used to evaluate the interaction between the components in the food and the potential impact of each component, and a health risk assessment report is generated. The component concentration distribution map and the health risk assessment report are mapped to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake. The technical solution provided by the present invention greatly shortens the detection time, improves the accuracy and reliability of the detection, provides a more comprehensive and specific assessment for food safety, and realizes a one-stop solution from detection to assessment, helping consumers make wise choices.

[0052] Furthermore, the clarity of the preliminary spectral features can be significantly improved by combining the adaptive threshold segmentation algorithm with the local contrast enhancement technology, making subsequent processing more accurate and effective; the multi-scale morphological analysis method can be used to extract structural features from different scales, enhancing the understanding and analysis of the spectral features of complex components; the spectral features with multi-scale structural features are compared with the standard spectral library, and accurately calibrated using a similarity matching algorithm based on deep learning to ensure the accuracy of spectral feature recognition; combined with the calibrated spectral features, iterative optimization processing is performed through a genetic algorithm or a particle swarm optimization algorithm, and finally the optimized target spectral features of natural ingredients and illegal additives are obtained, further improving the accuracy and reliability of component identification; a variety of advanced algorithms and technologies are introduced in the entire optimization process, such as adaptive threshold segmentation, multi-scale morphological analysis, deep learning, and genetic algorithms. The application of these technologies enables this method to continuously improve itself with the development of new data and technologies and maintain its leading edge.

[0053] These and other aspects of the present invention will become more apparent from the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1A flowchart of a food safety detection method based on intelligent spectrophotometry provided by an embodiment of the present invention;

[0056] Figure 2 A schematic diagram of the structure of a food safety detection system based on intelligent spectrophotometry provided by an embodiment of the present invention;

[0057] Figure 3 A schematic diagram of the structure of a computing device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0059] In some of the processes described in the specification and claims of the present invention and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0061] When dealing with complex food samples, traditional methods have difficulty in accurately distinguishing natural ingredients from illegal additives, especially when faced with trace or trace additives. The detection accuracy and reliability are insufficient, and the existing risk assessments mostly remain at the qualitative description stage, lacking quantitative analysis of the interactions between various ingredients in food and their potential health effects, resulting in the generated health risk assessment reports being not comprehensive and specific. Based on this, the present invention provides a food safety detection method based on intelligent spectrophotometry, such as Figure 1 ,include:

[0062] Step 101: Scanning the food sample by a smart spectrophotometer for a preset number of times to obtain high-resolution absorption spectrum data of the sample in the range from visible light to near infrared;

[0063] In this step, a smart spectrophotometer, an instrument that can measure the degree to which different wavelengths of light are absorbed by a substance, is used to perform multiple spectral scans of the food sample to obtain information on how the sample absorbs different wavelengths of light in the range from visible light (approximately 380-700 nanometers) to near-infrared (approximately 700-2500 nanometers). This information is presented in the form of high resolution, meaning that very detailed and precise data can be obtained.

[0064] Step 102: using a deep learning model to perform pattern recognition on the high-resolution absorption spectrum data, distinguishing natural ingredients in food from illegal additives outside the preset target, and generating a component concentration distribution map;

[0065] In this step, the deep learning model refers to a machine learning algorithm based on artificial neural networks, which can automatically learn features from a large amount of data and perform classification or prediction. The high-resolution absorption spectrum data is input into the trained deep learning model, and the natural ingredients in food and possible illegal additives are distinguished through pattern recognition technology. In addition, the model will also generate a spatial distribution image describing the concentration of each component based on the absorption spectrum data, that is, the component concentration distribution map.

[0066] Step 103: Based on the component concentration distribution map and in combination with a dynamically updated food component spectrum database, a multivariate statistical analysis method is used to evaluate the interaction between components in the food and the potential impact of each component, and a health risk assessment report is generated;

[0067] In this step, the component concentration distribution map refers to an image showing the changes in the concentration of various components in food. The dynamically updated food component spectral database is a continuously maintained and updated database that contains known food components and their corresponding spectral characteristics. It uses multivariate statistical analysis methods, a statistical technique that considers the relationship between multiple variables, to evaluate the interactions between components in food and their potential effects on human health, and generate a health risk assessment report based on this.

[0068] Step 104: Mapping the ingredient concentration distribution map and the health risk assessment report to a preset safety risk model to generate a comprehensive safety assessment report, wherein the comprehensive safety assessment report includes: food safety score, health risk warning, and recommended intake;

[0069] In this step, the safety risk model is a pre-established framework for evaluating and quantifying the risks posed by different factors. The ingredient concentration distribution map obtained in the previous step and the information from the health risk assessment report are integrated into this model to generate a comprehensive safety assessment report. This report not only contains an overall evaluation (score) of food safety, but also provides specific warnings (early warnings) about health risks and recommended intake guidance for specific ingredients.

[0070] Based on this, the present invention provides a specific embodiment, wherein the step 102 uses a deep learning model to perform pattern recognition on the high-resolution absorption spectrum data, distinguishes natural ingredients in food from illegal additives outside the preset target, and generates a component concentration distribution map, which specifically includes the following steps:

[0071] Step 201: inputting the high-resolution absorption spectrum data into a deep convolutional neural network trained with a preset target component spectrum, and using the deep convolutional neural network to perform feature recognition processing on the high-resolution absorption spectrum data to obtain the spectral features of natural ingredients and illegal additives;

[0072] In this step, the high-resolution absorption spectrum data obtained from the food samples is input into a pre-trained deep convolutional neural network (DCNN), which has been trained with a large amount of spectral data of known ingredients, especially for the spectral characteristics of natural ingredients and potential illegal additives. Through this deep learning model, the characteristic patterns in the high-resolution absorption spectrum can be automatically extracted and identified, thereby distinguishing natural ingredients and illegal additives and generating their respective spectral characteristics.

[0073] Step 202: Based on the spectral features, an adaptive threshold segmentation algorithm is used to perform optimization processing to obtain optimized results for distinguishing natural ingredients from illegal additives;

[0074] In this step, an adaptive threshold segmentation algorithm is used to further optimize the spectral features obtained in the previous step. The algorithm can dynamically adjust the threshold according to the distribution of spectral features to better separate the signals of natural ingredients and illegal additives. In this way, the influence of noise and background interference can be reduced, the accuracy of ingredient differentiation can be improved, and finally the optimized natural ingredients and illegal additives differentiation results can be obtained.

[0075] Step 203: combining the optimized natural ingredients and illegal additives differentiation results, using nonlinear regression analysis to perform concentration estimation processing to obtain a preliminary ingredient concentration distribution map;

[0076] In this step, combined with the optimized distinction between natural ingredients and illegal additives, nonlinear regression analysis is used to estimate the relative concentration of each ingredient in the food sample. Nonlinear regression analysis is a statistical method that can establish a mathematical model based on the relationship between the spectral characteristics of known ingredients and their actual concentrations, thereby predicting the concentration of each ingredient in an unknown sample. The result of this step is a preliminary component concentration distribution map that shows the spatial distribution of different ingredients in the sample and their concentration levels.

[0077] Step 204: Correcting the preliminary component concentration distribution map by introducing a chemometric method to obtain a target component concentration distribution map;

[0078] In this step, the chemometric method is introduced to correct the preliminary component concentration distribution map. Chemometrics is a methodology that combines chemistry, mathematics and statistics to solve complex problems in chemical measurement. It can help correct concentration deviations caused by sample preparation or instrument errors, ensuring that the target component concentration distribution map generated in the end more accurately reflects the actual concentration distribution of each component in the food sample. After this correction process, more accurate and reliable component concentration information can be obtained.

[0079] Based on this, the present invention provides a specific embodiment, wherein the step 201 inputs the high-resolution absorption spectrum data into a deep convolutional neural network trained with a preset target component spectrum, and uses the deep convolutional neural network to perform feature recognition processing on the high-resolution absorption spectrum data to obtain the spectral features of natural ingredients and illegal additives, specifically comprising the following steps:

[0080] Step 301: using a multi-channel spectrum acquisition system to collect high-resolution absorption spectrum data of food samples under a wide bandwidth, and obtaining high-resolution absorption spectrum data covering characteristic absorption peaks of all natural ingredients and illegal additives;

[0081] In this step, a multi-channel spectral acquisition system is used to perform spectral scanning on the food sample. The system is capable of simultaneously acquiring high-resolution absorption spectral data at multiple wavelengths in a wide band (e.g., from visible light to near-infrared regions). In this way, it can be ensured that the collected data covers all possible natural ingredients and all characteristic absorption peaks of potential illegal additives in the food, thereby providing detailed and accurate basic information for subsequent analysis.

[0082] Step 302: preprocessing the high-resolution absorption spectrum data using a baseline correction method to obtain absorption spectrum data;

[0083] In this step, the high-resolution absorption spectrum data obtained in the previous step are preprocessed using the baseline correction method. Baseline correction refers to correcting the baseline offset caused by instrument drift, background signal or other non-specific absorption to eliminate the influence of these factors on the spectrum data. After baseline correction, purer and smoother absorption spectrum data are obtained, which helps to improve the accuracy and reliability of subsequent analysis.

[0084] Step 303: construct and use a deep convolutional neural network model that has been pre-trained with a large number of spectral data sets of known components to perform feature recognition on the absorption spectrum data to obtain preliminary spectral features containing natural ingredients and illegal additives;

[0085] In this step, a pre-trained deep convolutional neural network (DCNN) model is applied to process the pre-processed absorption spectrum data. This DCNN model has been trained with a large amount of known component spectrum data and can automatically learn and identify the spectral characteristics of different components. By inputting the absorption spectrum data into this model, natural ingredients and illegal additives can be distinguished and their respective preliminary spectral characteristics can be generated, providing a basis for further analysis and evaluation.

[0086] Step 304: optimizing the preliminary spectral features using an adaptive threshold segmentation algorithm to obtain optimized target spectral features of natural ingredients and illegal additives;

[0087] In this step, the preliminary spectral features are optimized using an adaptive threshold segmentation algorithm. The algorithm dynamically adjusts the threshold according to the specific distribution of the spectral features to more accurately distinguish natural ingredients from illegal additives. In this way, the influence of noise and background interference can be reduced, and the accuracy of component differentiation can be improved. Ultimately, the optimized target spectral features of natural ingredients and illegal additives are obtained. These features are clearer and more accurate, and are suitable for subsequent concentration estimation and other quantitative analysis.

[0088] Based on this, the present invention provides a specific embodiment, wherein the step 301 uses an adaptive threshold segmentation algorithm to optimize the preliminary spectral features to obtain optimized target spectral features of natural ingredients and illegal additives, specifically comprising the following steps:

[0089] Step 401: using an adaptive threshold segmentation algorithm in combination with a local contrast enhancement technique, performing detail enhancement processing on the preliminary spectral feature to obtain an enhanced preliminary spectral feature;

[0090] In this step, an adaptive threshold segmentation algorithm is used in combination with local contrast enhancement technology to process preliminary spectral features. The adaptive threshold segmentation algorithm can dynamically adjust the threshold according to the specific distribution of the spectral features, while the local contrast enhancement technology can highlight subtle differences and important features in the spectrum. Through the combination of these two, the details of the preliminary spectral features can be effectively enhanced, making the difference between natural ingredients and illegal additives more obvious, thereby obtaining enhanced preliminary spectral features.

[0091] Step 402: Based on the enhanced preliminary spectral features, a multi-scale morphological analysis method is used to extract structural features to obtain spectral features with multi-scale structural features;

[0092] In this step, the enhanced preliminary spectral features are processed using a multiscale morphological analysis method. Multiscale morphological analysis is an image processing technique that can extract spatial structural information of spectral features at different scales. Through this method, different hierarchical structures of spectral features can be captured from multiple angles to generate spectral features with multiscale structural features. These structural features are crucial for subsequent precise calibration and optimization because they provide richer information about component distribution.

[0093] Step 403: comparing the spectral features with multi-scale structural features with a preset standard spectral library, and performing precise calibration using a similarity matching algorithm based on deep learning to obtain calibrated spectral features;

[0094] In this step, the spectral features with multi-scale structural features are compared with a preset standard spectral library. The standard spectral library contains known components and their corresponding spectral feature data. It is a strictly verified and widely recognized data set. In order to achieve accurate calibration, a similarity matching algorithm based on deep learning is used here. This algorithm can compare the multi-scale structural features with the features in the standard spectral library and automatically find the closest match, thereby calibrating the spectral features to ensure its accuracy and reliability.

[0095] Step 404: combining the calibrated spectral features, performing iterative optimization processing through a genetic algorithm or a particle swarm optimization algorithm to obtain optimized target spectral features of natural ingredients and illegal additives;

[0096] In this step, the calibrated spectral features are iteratively optimized using genetic algorithms or particle swarm optimization algorithms. Both algorithms are global optimization methods that can search for optimal solutions in complex spaces. Through multiple iterations, the spectral features are gradually adjusted and optimized to maximize the ability to distinguish natural ingredients from illegal additives. Ultimately, the target spectral features of natural ingredients and illegal additives that have been optimized are obtained. These features are not only clearer and more accurate, but also more suitable for further quantitative analysis and risk assessment.

[0097] Based on this, the present invention provides a specific embodiment, wherein step 103, based on the component concentration distribution diagram, combined with a dynamically updated food component spectrum database, uses a multivariate statistical analysis method to evaluate the interaction between the components in the food and the potential impact of each component, and generates a health risk assessment report, specifically comprising the following steps:

[0098] Step 501: extracting feature vectors from the component concentration distribution diagram using advanced spectral analysis technology to obtain feature vectors of key components;

[0099] In this step, advanced spectral analysis technology is used to process the component concentration distribution map. This technology can extract key features representing different components from complex spectral data and convert these features into mathematical feature vectors. Feature vectors are numerical representations that describe spectral characteristics and can be used for subsequent analysis and comparison. Through this process, the main components in food samples can be identified and quantified, providing a basis for the next step of analysis.

[0100] Step 502: comparing the feature vector of the key component with the standard spectral feature in the dynamically updated food component spectral database to obtain a comparison result;

[0101] In this step, the key component feature vectors extracted in the previous step are compared with the standard spectral features in a dynamically updated food component spectral database, which contains a large amount of standard spectral information of known components and is constantly updated as new data is added. Through the comparison, it can be determined whether the components in the sample exist in the database and their specific matching degree, so as to obtain detailed comparison results, which helps to further confirm the identity of the components and their existence in the food.

[0102] Step 503: Based on the comparison results, a method combining principal component analysis and partial least squares regression is used to perform dimensionality reduction processing on each component in the food to obtain a component data set after dimensionality reduction;

[0103] In this step, based on the comparison results, the principal component analysis (PCA) and partial least squares regression (PLSR) method are combined to reduce the dimensionality of the data of each component in the food. PCA is a statistical method that aims to reduce the data dimension while retaining as much information as possible; PLSR is a regression analysis method that can simultaneously consider the relationship between multiple independent variables and dependent variables. Through the combination of these two methods, the information integrity of the interaction between components can be maintained while reducing the complexity of the data, and finally a component data set with reduced dimensionality is obtained.

[0104] Step 504: input the reduced-dimensional component data set into a multivariate statistical model based on a deep neural network, calculate the correlation coefficient matrix between the components, and obtain an interaction evaluation between the components;

[0105] In this step, the reduced dimension ingredient dataset is input into a multivariate statistical model based on a deep neural network (DNN). The model is trained to automatically learn and adapt to the characteristics of different food samples. By calculating the correlation coefficient matrix between the ingredients, the direct and indirect interactions between the ingredients can be evaluated. This evaluation not only reveals the possible synergistic or antagonistic effects between the ingredients, but also provides an important basis for subsequent risk assessment.

[0106] Step 505: Collect the user's personal health record data, the personal health record data includes: age, gender, weight, health status;

[0107] In this step, the user's personal health record data is collected, including but not limited to information such as age, gender, weight, and health status. The personal health record data provides important background information about individual differences among users, which is crucial for personalized risk assessment. This data will be combined with the interaction assessment between components in subsequent steps to generate a more customized health risk assessment report.

[0108] Step 506: combining the personal health record data with the interaction assessment between the components, using the Bayesian inference algorithm to quantify the potential impact of each component, and generating a health risk assessment report;

[0109] In this step, the personal health record data is combined with the interaction assessment between ingredients, and the potential health effects of each ingredient are quantitatively analyzed using the Bayesian inference algorithm. Bayesian inference is a probabilistic statistical method that can infer the probability distribution of unknown parameters based on existing data and prior knowledge. This method can more accurately estimate the potential impact of different ingredients on the health of specific individuals, and ultimately generate a health risk assessment report that includes food safety scores, health risk warnings, and recommended intakes. This report not only takes into account the scientific properties of food ingredients, but also combines the user's individualized health information to provide users with more personalized guidance.

[0110] Based on this, the present invention provides a specific embodiment, wherein the step 506 combines the personal health record data with the interaction evaluation between the components, uses the Bayesian reasoning algorithm to quantify the potential impact of each component, and generates a health risk assessment report, which specifically includes the following steps:

[0111] Step 601: combining the interaction evaluation between the components, using the Bayesian inference algorithm, probabilistically modeling the potential impact of each component based on prior knowledge and observation data, and obtaining the probability distribution of the potential impact of each component;

[0112] In this step, the results of the interaction assessment between ingredients are used as input information, and the Bayesian inference algorithm is used to integrate prior knowledge (such as known health effects of ingredients) and new observational data (such as ingredient concentrations, interactions between ingredients). By constructing a probability model, the probability distribution of each ingredient's possible impact on human health can be estimated. The key to this step is to quantify uncertainty and provide a statistical basis for subsequent risk assessment.

[0113] Step 602: Based on the probability distribution of the potential impact of each component, uncertainty analysis is introduced, and the health risk of each component under different scenarios is evaluated by Monte Carlo simulation method to obtain scenario analysis results;

[0114] In this step, the probability distribution obtained in the previous step is used to simulate the health risks under different conditions or hypothetical scenarios using the Monte Carlo simulation method. Monte Carlo simulation is a powerful tool for understanding uncertainty and risk. It simulates various possible outcomes through a large number of random samplings. This can evaluate the health risks of each component under different scenarios (such as different exposure levels, different individual sensitivities, etc.), thereby gaining a more comprehensive understanding of the potential impact of the components.

[0115] Step 603: combining the scenario analysis results with the personal health record data, using a personalized risk assessment model to calculate the impact of individual differences on health risks, and generating a personalized health risk assessment report;

[0116] In this step, the results of the scenario analysis and personal health record data (such as age, gender, weight, health status, etc.) are combined to apply a personalized risk assessment model to calculate how individual differences affect health risks. This model takes into account each person's unique physiological characteristics and lifestyle habits, making the health risk assessment more in line with the individual's actual situation. The final output is a customized health risk assessment report that can provide specific guidance and suggestions on individual dietary choices.

[0117] Step 604: Associating the personalized health risk assessment report with the source tracking information of the food ingredients, making the source tracking information transparent and tamper-proof through blockchain technology, and forming a target health risk assessment report;

[0118] In this step, the personalized health risk assessment report will be connected with the source tracking information of food ingredients. By adopting blockchain technology, it is ensured that the source tracking information is transparent and cannot be tampered with. As a distributed ledger technology, blockchain provides a high degree of security and reliability, ensuring the authenticity and integrity of all transaction records (including information on food production, processing, transportation, etc.). Therefore, consumers can obtain reliable information about the source and quality of food and enhance their trust in food safety. The final result is a target health risk assessment report that includes detailed ingredient information, health risk assessment and supply chain transparency, providing users with a comprehensive decision-making support tool.

[0119] Based on this, the present invention provides a specific embodiment, wherein the step 104 maps the ingredient concentration distribution map and the health risk assessment report to a preset safety risk model to generate a comprehensive safety assessment report, wherein the comprehensive safety assessment report includes: a food safety score, a health risk warning, and a recommended intake, and specifically includes the following steps:

[0120] Step 701: input the component concentration distribution map and the evaluation results into a preset safety risk model constructed based on historical food safety data and epidemiological research results, perform preliminary mapping processing, and obtain preliminary mapping results;

[0121] In this step, the ingredient concentration distribution map and health risk assessment results are input into a pre-built safety risk model. This model is based on historical food safety data (such as past test records, recall events, etc.) and epidemiological research results (such as the relationship between different ingredients and health effects). It can comprehensively consider the impact of multiple factors on food safety. Through preliminary mapping processing, the model converts these input information into preliminary mapping results, providing a basis for subsequent risk assessment.

[0122] Step 702: Calculate the comprehensive risk index of the preliminary mapping result through a multi-factor risk assessment algorithm to obtain a food safety score;

[0123] In this step, the preliminary mapping results are further analyzed using a multi-factor risk assessment algorithm. The multi-factor risk assessment algorithm is a statistical method that comprehensively considers multiple risk factors. It can quantify the impact of various risk factors on food safety. Through this process, a comprehensive risk index can be calculated, and finally a food safety score is generated. This score reflects the overall safety level of the food sample and is an important basis for subsequent decision-making.

[0124] Step 703: Based on the food safety score, health risk warning information is generated and processed using preset health risk warning rules to obtain health risk warning information, wherein the health risk warning information includes short-term consumption risk, long-term consumption risk, and consumption risk for special groups;

[0125] In this step, based on the food safety score, the preset health risk warning rules are applied to generate specific health risk warning information. This warning information covers short-term consumption risks (such as acute poisoning), long-term consumption risks (such as chronic disease risks) and consumption risks for special groups (such as pregnant women, children or the elderly). Through this classification method, the possible health impacts of food in different situations can be conveyed more accurately, helping consumers make more informed choices.

[0126] Step 704: combining the health risk warning information, formulating a recommended intake through comprehensive analysis of nutrition and toxicology to obtain a specific recommended intake;

[0127] In this step, a comprehensive analysis is conducted using knowledge from nutrition and toxicology in combination with health risk warning information. Nutrition focuses on the positive effects of ingredients on human health, while toxicology focuses on the potential harmful effects of ingredients. Through the combination of these two disciplines, specific recommended intake amounts can be scientifically and rationally formulated. These recommendations not only take into account the safety and health benefits of the ingredients, but also take into account individual differences and the needs of specific populations, ensuring the scientificity and practicality of the recommendations.

[0128] Step 705: Integrate the food safety score, health risk warning information and specific recommended intake, and use natural language generation technology to convert text description and chart display to obtain a comprehensive safety assessment report;

[0129] In this step, the food safety score, health risk warning information and specific recommended intake are integrated to form a comprehensive integrated safety assessment report. In order to make the report easy to understand, natural language generation technology is used to convert complex scientific data and technical terms into easy-to-understand text descriptions, accompanied by intuitive charts. This method not only improves the readability of the report, but also enhances consumers' understanding and trust in food safety information, helping them make better dietary choices based on the evaluation results.

[0130] Figure 2 A structural diagram of a food safety detection system based on intelligent spectrophotometry is provided for the embodiment of the present application, such as Figure 2 As shown, the system includes:

[0131] The scanning module 21 is used to perform a preset number of spectral scans on the food sample through an intelligent spectrophotometer to obtain high-resolution absorption spectrum data of the sample in the range of visible light to near infrared;

[0132] An identification module 22, for performing pattern recognition on the high-resolution absorption spectrum data using a deep learning model, distinguishing natural ingredients in food from illegal additives outside of a preset target, and generating a component concentration distribution map;

[0133] An evaluation module 23 is used to evaluate the interaction between the ingredients in the food and the potential impact of each ingredient based on the ingredient concentration distribution map and in combination with a dynamically updated food ingredient spectrum database, and generate a health risk assessment report by using a multivariate statistical analysis method;

[0134] The generation module 24 is used to map the ingredient concentration distribution map and the health risk assessment report to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake.

[0135] Figure 2 The food safety detection system based on intelligent spectrophotometry can perform Figure 1 The implementation principle and technical effect of the food safety detection method based on intelligent spectrophotometry described in the embodiment shown will not be repeated. The specific way in which each module and unit performs operations in the food safety detection system based on intelligent spectrophotometry in the above embodiment has been described in detail in the embodiment of the method, and will not be elaborated here.

[0136] Figure 2 The food safety detection system based on intelligent spectrophotometry in the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0137] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0138] The processing component 32 is used to perform a preset number of spectral scans on the food sample through an intelligent spectrophotometer to obtain high-resolution absorption spectrum data of the sample in the range of visible light to near infrared;

[0139] Using a deep learning model to perform pattern recognition on the high-resolution absorption spectrum data, distinguishing natural ingredients in food from illegal additives outside of preset targets, and generating a component concentration distribution map;

[0140] Based on the component concentration distribution map, combined with a dynamically updated food component spectral database, a multivariate statistical analysis method is used to evaluate the interaction between the components in the food and the potential impact of each component, and a health risk assessment report is generated;

[0141] The ingredient concentration distribution map and the health risk assessment report are mapped to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake.

[0142] The processing component 32 includes one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be implemented by one or more application-specific integrated circuits (AICs), digital signal processors (DPs), digital signal processing devices (DPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above method.

[0143] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (RAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0144] Computing devices also include other components, such as input / output interfaces, display components, and communication components.

[0145] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above-mentioned peripheral interface module can be an output device or an input device.

[0146] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0147] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0148] The embodiment of the present invention further provides a computer storage medium storing a computer program, which can achieve the above-mentioned Figure 1 The illustrated embodiment provides a food safety detection method and system based on intelligent spectrophotometry.

[0149] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0150] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0151] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0152] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A food safety detection method based on intelligent spectrophotometry, characterized in that: include: The food sample is scanned by a smart spectrophotometer for a preset number of times to obtain high-resolution absorption spectrum data of the sample in the range from visible light to near infrared; Using a deep learning model to perform pattern recognition on the high-resolution absorption spectrum data, distinguishing natural ingredients in food from illegal additives outside of preset targets, and generating a component concentration distribution map; Based on the component concentration distribution map, combined with a dynamically updated food component spectral database, a multivariate statistical analysis method is used to evaluate the interaction between the components in the food and the potential impact of each component, and a health risk assessment report is generated; The ingredient concentration distribution map and the health risk assessment report are mapped to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake.

2. The method according to claim 1, characterized in that The deep learning model is used to perform pattern recognition on the high-resolution absorption spectrum data to distinguish natural ingredients in food from illegal additives outside the preset target, and generate a component concentration distribution map, including: Inputting the high-resolution absorption spectrum data into a deep convolutional neural network trained with a preset target component spectrum, and using the deep convolutional neural network to perform feature recognition processing on the high-resolution absorption spectrum data to obtain the spectral features of natural ingredients and illegal additives; Based on the spectral features, an adaptive threshold segmentation algorithm is used to perform optimization processing to obtain optimized results for distinguishing natural ingredients from illegal additives; Combining the optimized differentiation results of natural ingredients and illegal additives, a concentration estimation process is performed using a nonlinear regression analysis method to obtain a preliminary component concentration distribution map; The preliminary component concentration distribution map is corrected by introducing a chemometric method to obtain a target component concentration distribution map.

3. The method according to claim 2, characterized in that The high-resolution absorption spectrum data is input into a deep convolutional neural network trained with a preset target component spectrum, and the deep convolutional neural network is used to perform feature recognition processing on the high-resolution absorption spectrum data to obtain the spectral features of natural ingredients and illegal additives, including: A multi-channel spectrum acquisition system is used to collect high-resolution absorption spectrum data of food samples in a wide bandwidth, and high-resolution absorption spectrum data covering characteristic absorption peaks of all natural ingredients and illegal additives are obtained; Preprocessing the high-resolution absorption spectrum data using a baseline correction method to obtain absorption spectrum data; Constructing and using a deep convolutional neural network model that has been pre-trained with a large number of spectral data sets of known ingredients to perform feature recognition on the absorption spectral data to obtain preliminary spectral features containing natural ingredients and illegal additives; The preliminary spectral features are optimized using an adaptive threshold segmentation algorithm to obtain optimized target spectral features of natural ingredients and illegal additives.

4. The method according to claim 3, characterized in that The preliminary spectral features are optimized using an adaptive threshold segmentation algorithm to obtain optimized target spectral features of natural ingredients and illegal additives, including: Using an adaptive threshold segmentation algorithm combined with a local contrast enhancement technique, the preliminary spectral features are subjected to detail enhancement processing to obtain enhanced preliminary spectral features; Based on the enhanced preliminary spectral features, a multi-scale morphological analysis method is used to extract structural features to obtain spectral features with multi-scale structural features; The spectral features with multi-scale structural features are compared with a preset standard spectral library, and a similarity matching algorithm based on deep learning is used for accurate calibration to obtain calibrated spectral features; Combined with the calibrated spectral features, an iterative optimization process is performed using a genetic algorithm or a particle swarm optimization algorithm to obtain optimized target spectral features of natural ingredients and illegal additives.

5. The method according to claim 1, characterized in that Based on the component concentration distribution map, combined with the dynamically updated food component spectral database, a multivariate statistical analysis method is used to evaluate the interactions between the components in the food and the potential impact of each component, and a health risk assessment report is generated, including: Using advanced spectral analysis technology to extract feature vectors from the component concentration distribution diagram to obtain feature vectors of key components; Comparing the characteristic vector of the key component with the standard spectral feature in the dynamically updated food component spectral database to obtain a comparison result; Based on the comparison results, a method combining principal component analysis and partial least squares regression is used to perform dimensionality reduction processing on each component in the food to obtain a component data set after dimensionality reduction; Inputting the reduced-dimensional component data set into a multivariate statistical model based on a deep neural network, calculating the correlation coefficient matrix between the components, and obtaining an interaction evaluation between the components; Collecting the user's personal health record data, the personal health record data includes: age, gender, weight, health status; The personal health record data is combined with the interaction assessment between the components, and the potential impact of each component is quantified using a Bayesian reasoning algorithm to generate a health risk assessment report.

6. The method according to claim 5, characterized in that The personal health record data is combined with the interaction assessment between the ingredients, and the potential impact of each ingredient is quantified using a Bayesian inference algorithm to generate a health risk assessment report, including: Combined with the interaction evaluation among the components, a Bayesian inference algorithm is used to perform probability modeling on the potential impact of each component based on prior knowledge and observation data to obtain the probability distribution of the potential impact of each component; Based on the probability distribution of the potential impact of each component, uncertainty analysis is introduced, and the health risks of each component under different scenarios are evaluated by Monte Carlo simulation method to obtain scenario analysis results; Combining the scenario analysis results with personal health record data, using a personalized risk assessment model to calculate the impact of individual differences on health risks, and generating a personalized health risk assessment report; The personalized health risk assessment report is associated with the source tracking information of the food ingredients, and the source tracking information is made transparent and tamper-proof through blockchain technology to form a target health risk assessment report.

7. The method according to claim 1, characterized in that The component concentration distribution map and health risk assessment report are mapped to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake, including: Inputting the ingredient concentration distribution map and the evaluation results into a preset safety risk model constructed based on historical food safety data and epidemiological research results, performing preliminary mapping processing, and obtaining preliminary mapping results; The preliminary mapping results are processed by a multi-factor risk assessment algorithm to calculate a comprehensive risk index to obtain a food safety score; Based on the food safety score, health risk warning information is generated and processed using preset health risk warning rules to obtain health risk warning information, wherein the health risk warning information includes: short-term consumption risk, long-term consumption risk, and consumption risk for special groups; In combination with the health risk warning information, a recommended intake amount is formulated through a comprehensive analysis of nutrition and toxicology to obtain a specific recommended intake amount; The food safety score, health risk warning information and specific recommended intake are integrated, and natural language generation technology is used to convert text descriptions and chart displays to obtain a comprehensive safety assessment report.

8. A food safety detection method and system based on intelligent spectrophotometry, characterized in that: include: A scanning module is used to perform a preset number of spectral scans on food samples through an intelligent spectrophotometer to obtain high-resolution absorption spectrum data of the samples in the range from visible light to near infrared; An identification module, for performing pattern recognition on the high-resolution absorption spectrum data using a deep learning model, distinguishing natural ingredients in food from illegal additives outside of preset targets, and generating a component concentration distribution map; An evaluation module, for evaluating the interaction between ingredients in food and the potential impact of each ingredient based on the ingredient concentration distribution map and in combination with a dynamically updated food ingredient spectrum database, and generating a health risk assessment report by using a multivariate statistical analysis method; A generation module is used to map the ingredient concentration distribution map and the health risk assessment report to a preset safety risk model to generate a comprehensive safety assessment report, which includes: food safety score, health risk warning and recommended intake.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a food safety detection method based on intelligent spectrophotometry as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, a food safety detection method based on intelligent spectrophotometry as described in any one of claims 1 to 7 is implemented.

Images