Food safety toxic and harmful substance inspection and detection big data structuring system

Abstract

The invention discloses a food safety toxic and harmful substance inspection and detection big data structuring system, and relates to the technical field of food safety detection, and the system comprises the following components: data acquisition, neural mimicry calculation, data analysis and classification, result output and early warning, and system maintenance and optimization. Through the prediction model and a data analysis algorithm, the intelligent level of detection is improved, the prediction model can be established and continuously optimized based on historical detection data and known toxic and harmful substance characteristics, early warning and preferential detection of potential harmful substances are realized, in addition, the system also has self-learning and optimization capabilities, and the detection efficiency is improved. The intelligent food safety detection method can continuously adapt to new detection requirements and scenes, the detection accuracy and efficiency are further improved, and the intelligent detection mode not only improves the scientificity and precision of food safety detection, but also provides a more powerful technical means for food safety supervision.

Description

Technical Field

[0001] The present invention relates to the technical field of food safety detection, and specifically to a system for structuring big data on the inspection and detection of toxic and harmful substances in food safety. Background Art

[0002] With the rapid development of the food industry and the increasing requirements of consumers for food safety, food safety detection has become an important link in ensuring public health. However, traditional food safety detection methods have problems such as long detection cycles, high costs, and low efficiency, making it difficult to meet the large-scale, rapid, and accurate detection needs. At the same time, there are a wide variety of toxic and harmful substances in food, and their contents are often extremely low, posing higher requirements for the sensitivity and accuracy of detection technologies.

[0003] In traditional technologies, there are many deficiencies in neuromorphic computing in the field of food safety detection. When processing food safety detection data, the feature extraction algorithm cannot accurately capture the features of toxic and harmful substances, resulting in missed detections and false detections. Traditional methods are difficult to predict the types of toxic and harmful substances that may exist in food samples, and based on the prediction results, provide a priority detection direction for actual detection. Therefore, it is particularly important to develop a system for structuring big data on the inspection and detection of toxic and harmful substances in food safety. Summary of the Invention

[0004] The purpose of the present invention is to make up for the deficiencies of the prior art, and provide a system for structuring big data on the inspection and detection of toxic and harmful substances in food safety. This system realizes the rapid and accurate detection of toxic and harmful substances in food samples by integrating multiple functional modules including data acquisition, neuromorphic computing, data analysis and classification, result output and warning, and system maintenance and optimization. At the same time, this system makes full use of the advantages of big data analysis and neuromorphic computing, improves the accuracy and efficiency of detection, and provides a new solution for the field of food safety detection.

[0005] To solve the above technical problems, the present invention provides the following technical solution: A system for structuring big data on the inspection and detection of toxic and harmful substances in food safety, which includes data acquisition, neuromorphic computing, data analysis and classification, result output and warning, and system maintenance and optimization;

[0006] In the data acquisition link: First, evaluate the detection requirements, select and purchase detection equipment, complete installation, debugging, and calibration verification. After preparing food samples, perform multimodal data acquisition with the help of a spectral analyzer, image acquisition equipment, and chemical sensors, and do a good job in storage management. At the same time, plan the network, install and configure network equipment, test data transmission, select a transmission protocol and develop transmission software, and configure a data storage and management center to achieve real-time transmission and monitoring of the acquired data;

[0007] The neuromorphic computing part: Deeply study the performance of different neuromorphic chips, select a chip according to the detection requirements. For example, choose a low-power chip for portable devices. Install the chip into the computing device to complete the hardware configuration. Install the development tools and software libraries, configure the development environment, preprocess the collected data, select a feature extraction algorithm to obtain the feature vector and store it. Select a pattern recognition algorithm to judge the presence and types of toxic and harmful substances in food samples. Use historical data and substance characteristics to establish a prediction model and continuously optimize it to provide a priority direction for the detection of new samples;

[0008] The data analysis and classification module: Establish a data interface with the neuromorphic computing module, clarify the transmission format and protocol, store and classify the output results, perform data cleaning, feature extraction and analysis, select a classification algorithm to train the model, predict the types of toxic and harmful substances, and use big data analysis technology to mine potential food safety problems;

[0009] The result output and warning function: Determine the result presentation form according to user needs and scenarios, organize and format the data, design a simple user interface, combine with standards to set safety thresholds, establish a real-time monitoring mechanism. Once the detection result exceeds the threshold, issue a warning through various methods. After relevant personnel receive it, take measures such as suspending production and sales, investigating and rectifying;

[0010] The system maintenance and optimization aspect: Regularly maintain, check the hardware, update the software, clean and back up the database, continuously collect new data, control the quality and store and manage it, evaluate and analyze the model and optimize and improve it, introduce new technologies to improve the detection ability.

[0011] Furthermore, for the data acquisition and installation of the detection equipment, it is necessary to first conduct a demand assessment, equipment selection and procurement, then carry out installation, debugging and calibration verification. After preparing the food samples, use a spectral analyzer, image acquisition equipment, and chemical sensors to collect multi-modal data and store and manage it. Then conduct network planning, install and configure network equipment, test data transmission, select a data transmission protocol, develop data transmission software, configure the data storage and management center, start the equipment and software to transmit the collected data to the center in real time and monitor it.

[0012] Furthermore, the pattern recognition algorithm selected by the neuromorphic computing performs pattern recognition and classification processing on the feature vector to determine whether there are toxic and harmful substances in the food sample and the specific types: Among them, represents the feature vector of the food sample, represents the weight vector, w i f i represents multiplying each feature value f i by the corresponding weight value w i and denotes the sum of the products of all eigenvalues and weight values, b represents the bias term, and for the presence or absence of toxic and harmful substances in food samples, a threshold T is set. Then it is preliminarily judged that there may be toxic and harmful substances in the food sample, E Then it is judged that there may be no toxic and harmful substances in the food sample.

[0013] Furthermore, the neuromorphic computing uses historical detection data and known characteristics of toxic and harmful substances to establish a prediction model: Among them, represents the input feature vector of the food sample, where each f i represents the i-th eigenvalue, θ represents the set of model parameters, σ is the activation function, u j i is the weight connecting the first-layer feature f i and the intermediate-layer node j, c j is the bias of the intermediate-layer node j, v j is the weight connecting the intermediate layer and the output layer, d is the bias of the output layer, and g is the activation function or transformation function of the output layer.

[0014] Furthermore, the neuromorphic computing inputs the collected data into the prediction model to predict the types of possible toxic and harmful substances: Among them, Φ is the set of model parameters after training and optimization, σ is the activation function, u ji the value after the fusion of the first-layer connected features and the weight of the intermediate-layer node j, v j the weight connecting the intermediate layer and the output layer, c j the bias of the intermediate-layer node j, d is the bias of the output layer, and g is the activation function or transformation function of the output layer.

[0015] Furthermore, the data analysis and classification establish a data interface with the neuromorphic computing module, determine the transmission format and protocol, store its output results in a dedicated data storage system and classify them, perform data cleaning, feature extraction and analysis, select a classification algorithm to train the model and make predictions to determine the types of toxic and harmful substances in the food sample, and use big data analysis techniques for association analysis and clustering analysis to discover potential association rules and food safety problems.

[0016] Furthermore, the result output and early warning analysis determine the intuitive form of presenting the detection results according to the user requirements and usage scenarios, conduct data collation and formatting, have a simple and clear user interface and add annotation descriptions, and can also output results through multiple channels. Determine the safety thresholds of toxic and harmful substances in combination with standards and store them in the system, establish a real-time monitoring mechanism to obtain the latest data, compare the detection results with the thresholds for judgment, and issue early warnings through system pop-ups, email and SMS notifications, and audible and visual alarms when the thresholds are exceeded. After receiving the early warnings, relevant departments and personnel should suspend production or sales, investigate the reasons and take rectification measures, and at the same time strengthen detection and supervision to prevent the problem from expanding.

[0017] Furthermore, the system maintenance and optimization determine the maintenance frequency and list the task list for regular maintenance, covering hardware inspection, software update, database cleaning and backup restoration, continuously establish data collection channels, control data quality and conduct storage management, regularly evaluate the analysis model, optimize and improve according to the results, establish a continuous improvement mechanism after verification and confirmation, and continuously collect new data and introduce new technologies to improve the system detection ability.

[0018] Compared with the prior art, the big data structured system for the inspection and detection of toxic and harmful substances in food safety has the following beneficial effects:

[0019] First, through the prediction model and data analysis algorithm, the system improves the intelligent level of detection. Based on historical detection data and known characteristics of toxic and harmful substances, it can establish and continuously optimize the prediction model to achieve early warning and priority detection of potential harmful substances. In addition, it also has the ability of self-learning and optimization, and can continuously adapt to new detection requirements and scenarios, further improving the accuracy and efficiency of detection. This intelligent detection method not only improves the scientificity and accuracy of food safety detection, but also provides a more powerful technical means for food safety supervision.

[0020] Second, by integrating multiple functional modules such as data acquisition, neuromorphic computing, data analysis and classification, result output and early warning, and system maintenance and optimization, the system significantly improves the efficiency and accuracy of food safety detection. Compared with traditional detection methods, the system can complete the multi-modal data acquisition of food samples more quickly, and through the neuromorphic computing module for efficient processing, it can achieve accurate identification and classification of toxic and harmful substances, which not only shortens the detection cycle, but also reduces the risks of missed detection and false detection, providing a solid technical support for ensuring food safety. Description of the Drawings

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0022] Figure 1 System operation flowchart for structuring big data of food safety toxic and harmful substance inspection and testing. Detailed implementation manners

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0024] Embodiment 1

[0025] Determine the types of toxic and harmful substances to be detected, understand the characteristics and detection method requirements of different substances, analyze the types, forms and characteristics of food samples, determine the appropriate types of detection equipment, consider factors such as detection accuracy, speed and cost, formulate an equipment procurement plan, select the appropriate spectral range and resolution according to the detection requirements. Raman spectrometers can be used to detect the chemical structure and composition of substances in food. Select a high-resolution camera that can take clear images of the appearance of food for detecting the color, shape, size characteristics of food, as well as surface contaminants and damages;

[0026] Select corresponding sensors according to the toxic and harmful substances to be detected. Electrochemical sensors can be used to detect heavy metal ions and pesticide residues in food. According to actual needs, other detection equipment can also be selected. Select a suitable installation location to ensure that the equipment can operate stably and is convenient for operation and maintenance. Install and debug according to the equipment instruction manual to ensure that the various parameter settings of the equipment are correct and it can work properly. Calibrate and verify the equipment to ensure the accuracy and reliability of the detection results;

[0027] Select representative food samples to ensure that the samples can reflect the overall quality and safety status of the food. Pretreat the samples to facilitate the operation of the detection equipment and data collection. Number and label the samples, and record the source, production date, and batch information of the samples for subsequent data management and analysis. Spectral data collection: Use a spectral analyzer to scan the food samples to obtain the spectral data of the samples. Set appropriate scanning parameters according to different types of spectral analyzers and detection requirements. Use an image acquisition device to take pictures of the food samples to obtain the appearance images of the samples. Set appropriate shooting parameters to ensure that the images clearly and accurately reflect the appearance characteristics of the food. Install chemical sensors near the food samples or place the samples in the sensor detection chamber to obtain the chemical sensor data of the samples. Set appropriate detection parameters according to different sensor types and detection requirements. If necessary, other detection equipment can also be used to collect data from the food samples. Timely store the collected data in a data storage device to ensure the security and integrity of the data, and avoid data loss and damage. Classify and label the data for subsequent data management and analysis;

[0028] Determine the requirements and scope of data transmission, understand the distribution of data collection equipment and the size of the data volume, select appropriate network technologies and equipment, consider factors such as network stability, reliability, security, and scalability, formulate a network topology and layout plan to ensure the efficiency and convenience of data transmission. Install network equipment, install and debug it according to the equipment manual to ensure that the equipment can work properly. Configure network parameters to ensure that network devices can communicate with each other and can be connected to the data storage and management center;

[0029] After the installation and configuration of network equipment are completed, conduct data transmission tests. Use data collection equipment to collect data from some food samples and transmit the data to the data storage and management center. Check the speed, stability, and accuracy of data transmission to ensure that the network can meet the requirements of data transmission;

[0030] Select an appropriate data transmission protocol, consider factors such as data transmission reliability, real-time performance, and security. According to the operating systems and software environments of the data collection equipment and the data storage and management center, select a data transmission protocol with good compatibility. Develop data transmission software according to the selected data transmission protocol. The software should have functions such as data collection, data compression, data encryption, and data transmission to ensure that the data can be securely and quickly transmitted to the data storage and management center;

[0031] Test and optimize the data transmission software to ensure the stability and reliability of the software. Install a database management system and data storage devices in the data storage and management center to ensure that data can be stored and managed securely and reliably. Configure the database management system, create data tables and data fields to facilitate the storage and management of the collected data. Set appropriate database access permissions to ensure the security and confidentiality of the data. Start the data collection device and data transmission software, and transmit the collected data to the data storage and management center in real time. Set up a data monitoring system in the data storage and management center to monitor the status and progress of data transmission in real time. If data transmission anomalies are found, handle and repair them in a timely manner.

[0032] Study the performance characteristics of different neuromorphic chips. According to the requirements of food safety toxic and harmful substance inspection and detection, select a neuromorphic chip. For portable detection devices, select a chip with lower power consumption. Install the selected neuromorphic chip into the computing device to ensure the stable and reliable connection between the chip and other hardware components, and perform necessary hardware configuration;

[0033] Install the development tools and software libraries of the neuromorphic chip to provide necessary software support for the development of the neuromorphic computing module. Configure the development environment. According to the specific requirements of food safety detection, adopt the architecture and functions of the neuromorphic computing module, and use the development tools of the neuromorphic chip to implement each functional module of the neuromorphic computing module;

[0034] Preprocess the collected multimodal data, including but not limited to data cleaning, denoising, and normalization operations. Convert the preprocessed data into a format that can be processed by the neuromorphic computing module. Study the feature extraction algorithms for food safety toxic and harmful substance detection. According to the characteristics of the data and the detection requirements, select a feature extraction algorithm, extract features from the preprocessed data to obtain a feature vector that can reflect the characteristics of toxic and harmful substances, and store the extracted feature vector;

[0035] Study the pattern recognition algorithms for food safety toxic and harmful substance detection. According to the characteristics of the feature vector and the detection requirements, select a pattern recognition algorithm, perform pattern recognition and classification processing on the feature vector to determine whether there are toxic and harmful substances in the food sample and the specific types, Among them, represents the feature vector of the food sample, represents the weight vector, w i f i represents multiplying each feature value f i by the corresponding weight value w i and represents adding up the products of all feature values and weight values. b represents the bias term. For the presence or absence of toxic and harmful substances in the food sample, set a threshold T, Then it is preliminarily judged that there may be toxic and harmful substances in the food sample. Then it is judged that there may be no toxic and harmful substances in the food sample, and the results of classification processing are stored.

[0036] Predict the types of existing toxic and harmful substances and provide a priority detection direction for actual detection. Using historical detection data and the characteristics of known toxic and harmful substances, establish a prediction model. Among them, represents the feature vector of the input food sample, where each f i represents the i-th eigenvalue, θ represents the set of model parameters, σ is the activation function, u j i is the weight connecting the first-layer feature f i and the middle-layer node j, c j is the bias of the middle-layer node j, v j is the weight connecting the middle layer and the output layer, d is the bias of the output layer, and g is the activation function or transformation function of the output layer. Train and model the data, continuously optimize the prediction model, and improve the accuracy and reliability of the prediction. When a new food sample enters the detection process, input the collected data into the prediction model to predict the types of possible toxic and harmful substances. Among them, Φ is the set of model parameters after training and optimization, σ is the activation function, u ji The value after the fusion of the first-layer connection features and the weight of the middle-layer node j, v j The weight connecting the middle layer and the output layer, c j The bias of the middle-layer node j, d is the bias of the output layer, and g is the activation function or transformation function of the output layer. According to the prediction results, provide a priority detection direction for actual detection.

[0037] Design a data transmission interface with the neuromorphic computing module to ensure that its output results can be received stably and efficiently. Determine the format and protocol of data transmission for subsequent data processing. Store the received output results in a dedicated data storage system, and classify and store the data for subsequent retrieval and analysis.

[0038] Remove the noise and outliers in the output results to improve the data quality. Check the integrity and consistency of the data to ensure that there is no missing or incorrect data. Extract the key features from the output results that can reflect the characteristics of the food sample and the presence of toxic and harmful substances. Principal component analysis and independent component analysis methods can be used for feature extraction. Conduct statistical analysis on the extracted features and perform correlation analysis to determine the relationships between different features.

[0039] Research classification algorithms suitable for food safety issues, select the most appropriate classification algorithm based on data characteristics and classification requirements, use known food sample data and corresponding toxic and hazardous substance category labels to train the classification algorithm, adjust the algorithm parameters to improve the accuracy and stability of classification, input the characteristics of the food samples to be classified into the trained classification model, perform classification prediction, and obtain the category to which the food samples belong, that is, determine the types of toxic and hazardous substances that may exist in them;

[0040] Using known food sample data and the corresponding toxic and harmful substance content, a content prediction model is established. Regression analysis and neural network methods can be used to predict the content. The content prediction model is trained using historical data, and model parameters are continuously adjusted to improve the accuracy of the prediction. Cross-validation and error analysis are performed to ensure the reliability of the model. The characteristics of the food sample to be analyzed are input into the content prediction model to obtain the predicted value of the toxic and harmful substance content.

[0041] Analyze the correlation between the characteristics of different food samples and the presence of toxic and hazardous substances, use association rule mining algorithms to discover potential association rules, cluster food samples according to their characteristics, and find sample groups with similar characteristics. Use clustering algorithms to analyze the characteristics of food samples in each cluster and discover potential food safety issues.

[0042] Analyze user needs and usage scenarios, determine the most intuitive and easy-to-understand presentation format for test results, consider charts, reports, and visual interfaces, organize test result data, convert complex data into a format suitable for presentation, classify and label text data, and design a user interface based on the determined presentation format. The interface should be concise and clear, highlighting the key information of the test results, and use clear fonts, appropriate color matching and layout to ensure that users can quickly and accurately obtain information;

[0043] For charts and visualization interfaces, add appropriate annotations and instructions to help users understand the meaning of the data. In addition to presenting the test results on a specific software interface, you can also consider outputting the results through multiple channels to meet the needs of different users, send the test results to relevant personnel via email or text messages, or provide an API interface for other systems to call the test result data;

[0044] Combined with national food safety standards and industry specifications, the safety thresholds of different toxic and hazardous substances can be determined. They can be segmented according to food types, uses, and consumer groups, and more stringent or loose threshold standards can be formulated. The safety thresholds can be stored in the system for comparison during the detection process, and a real-time monitoring mechanism can be established to continuously obtain the latest test result data. Real-time data updates can be achieved through data interfaces with detection equipment and database queries;

[0045] Compare the detection results with the safety threshold to determine whether there are toxic and harmful substances exceeding the safety threshold. If the detected value exceeds the threshold, trigger the warning mechanism. If the detected value is within the safe range, continue monitoring. When it is detected that the toxic and harmful substances exceed the safety threshold, the system immediately issues a warning notice, pops up an obvious warning window on the software interface that the user is using, displays the information of the toxic and harmful substances exceeding the threshold and the key content of the food sample source, and sends an email or text message to relevant departments and personnel to notify them of the situation of toxic and harmful substances exceeding the threshold, and provide detailed detection results and recommended measures to be taken;

[0046] Set up an audible and visual alarm at the detection site or the office premises of relevant departments to emit an obvious alarm signal to remind the staff to handle it in a timely manner. After receiving the warning notice, relevant departments and personnel should immediately take emergency response measures. If it is detected that the toxic and harmful substances exceed the threshold during the production process, production should be immediately suspended, and the production equipment and raw materials should be inspected and processed. If problems are found during the sales process, relevant foods should be promptly taken off the shelves to prevent the further circulation of problem foods. Organize professional personnel to investigate the reasons for the toxic and harmful substances exceeding the threshold, which may involve the source of raw materials, production processes, transportation and storage links. After determining the root cause of the problem, take corresponding rectification measures to prevent similar problems from occurring again. Before the problem is solved, strengthen the detection frequency and supervision intensity of relevant foods to ensure food safety. At the same time, conduct a screening of other foods that may be affected to prevent the problem from expanding.

[0047] Determine the system maintenance frequency, such as conducting a comprehensive maintenance once a month, quarter, or half-year. List the maintenance task list, check the operating status of the server, storage device, and network device hardware to ensure the normal operation of the devices. Clean the dust on the hardware devices to keep the devices well cooled. Check the connection lines of the hardware devices to ensure stable and reliable connections. Check the update status of the operating system, database management system, and application software used by the system, and install the latest security patches and function updates in a timely manner. Upgrade the driver programs and firmware of the neuromorphic computing chips used by the system to improve system performance and stability;

[0048] Clean up the expired and useless data in the database to free up storage space. Optimize the indexing and query of the database to improve the data retrieval speed. Regularly back up the system data to prevent data loss. A combination of full backup and incremental backup can be used to ensure the integrity and availability of the backup data. Test the recovery function of the system backup to ensure that data can be quickly restored in case of system failures;

[0049] Establish cooperation relationships with food testing institutions, enterprises, and regulatory departments to obtain new testing data. Use sensor networks and Internet of Things technologies to automatically collect data in the food production, circulation, and sales links. Conduct quality inspections on the collected data, remove noise and outliers to ensure the accuracy and reliability of the data. Standardize the data, unify the data format and units for subsequent analysis and processing. Timely store the newly collected data in the database to ensure data security and accessibility. Classify and store the data for subsequent retrieval and analysis;

[0050] Regularly evaluate the analysis model of the system, analyze the accuracy, efficiency, and stability of the model. Use the newly collected data to test the model, identify problems and deficiencies in the model. According to the results of the model evaluation, optimize and improve the model. Adjust model parameters, add new features, and improve algorithm methods to enhance the performance of the model. Use machine learning technologies to further improve the accuracy and efficiency of the model;

[0051] Verify and validate the optimized model to ensure that the performance of the model has been effectively improved. Use an independent test data set to test the model and verify the generalization ability of the model. Establish a mechanism for continuous improvement of the model, continuously collect new data, optimize and improve the model. Pay attention to the latest technologies and research results in the field of food safety, and timely introduce new methods and technologies to improve the detection ability of the system.

[0052] Example 2

[0053] Install high-resolution spectrometers, high-definition image acquisition devices, and high-precision chemical sensors to achieve all-round and multi-angle data collection of food samples. Integrate automated sample injection, sample pretreatment, and analysis pretreatment modules to reduce manual intervention and improve detection efficiency. Establish a high-speed and stable data transmission network, adopt advanced network protocols and encryption technologies to ensure the real-time and security of data transmission.

[0054] Based on historical detection data and known characteristics of toxic and harmful substances, construct a prediction model, where, represents the input food sample feature vector, where each f i represents the i-th eigenvalue, θ represents the set of model parameters, σ is the activation function, u j i is the weight connecting the feature f i and the intermediate layer node j, c j is the bias of the intermediate layer node j, v jis the weight connecting the middle layer and the output layer, d is the bias of the output layer, g is the activation function or transformation function of the output layer. Utilizing the advantages of neuromorphic computing, simulating the working principle of human brain neurons, performing deep learning and optimization on the model, researching the feature extraction algorithm for detecting toxic and harmful substances in food safety, extracting features from the preprocessed data to obtain a feature vector that can reflect the characteristics of toxic and harmful substances, and using pattern recognition algorithms to perform pattern recognition and classification processing on the feature vector to determine whether there are toxic and harmful substances in the food sample and the specific types. Among them, represents the feature vector of the food sample, represents the weight vector, w i f i represents multiplying each feature value f i by the corresponding weight value w i and represents summing up the products of all feature values and weight values. b represents the bias term. For the presence or absence of toxic and harmful substances in the food sample, a threshold T is set. Then it is preliminarily judged that there may be toxic and harmful substances in the food sample. Then it is judged that there may be no toxic and harmful substances in the food sample. Using the continuously optimized and trained prediction model, quickly predicting the food samples newly entering the detection process and judging the possible types of toxic and harmful substances. Among them, Φ is the set of model parameters after training and optimization, σ is the activation function, u ji is the value after the first-layer connection feature fusion and the weight connecting to the middle layer node j, v j is the weight connecting the middle layer and the output layer, c j is the bias of the middle layer node j, d is the bias of the output layer, g is the activation function or transformation function of the output layer, providing a priority detection direction for actual detection.

[0055] Using big data analysis technology, deeply mining the massive detection data, discovering potential safety hazards and association rules, combining machine learning algorithms to automatically classify food samples, accurately identifying the types and contents of toxic and harmful substances, and providing rich data visualization tools to help users intuitively understand the detection results and data analysis results.

[0056] Design a simple and intuitive user interface, support multi-channel output of results, establish a real-time monitoring mechanism to monitor the detection results in real time. Once an over-standard situation is found, immediately trigger the warning system and notify the relevant departments and personnel in a timely manner through multiple methods such as pop-up windows, sound and light alarms, and text message notifications.

[0057] Formulate a detailed system maintenance plan, including hardware inspection, software updates, database cleaning and backup, to ensure the stable operation of the system. Continuously collect new detection data, iteratively optimize the analysis model, introduce new technologies to improve detection capabilities, and provide users with system operation training and technical support to ensure that users can efficiently utilize system resources.

[0058] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be construed as limiting the claimed invention.

Claims

1. A big data structured system for food safety toxic and hazardous substance inspection and detection, characterized by: The system includes data collection, neuromorphic computing, data analysis and classification, result output and early warning, and system maintenance and optimization; The data collection stage: first evaluate the testing needs, select and purchase testing equipment, complete installation, commissioning and calibration verification, prepare food samples, use spectrometers, image acquisition equipment, and chemical sensors to collect multimodal data, and do a good job of storage management. At the same time, plan the network, install and configure network equipment, test data transmission, select the transmission protocol and develop transmission software, configure the data storage and management center, and realize real-time transmission and monitoring of collected data; The neuromorphic computing part: in-depth study of the performance of different neuromorphic chips, selection of chips based on detection requirements, such as low-power chips for portable devices, installation of chips on computing devices, completion of hardware configuration, installation of development tools and software libraries, configuration of development environment, pre-processing of collected data, selection of feature extraction algorithms to obtain feature vectors and storage, selection of pattern recognition algorithms to determine the presence and types of toxic and harmful substances in food samples, establishment of prediction models using historical data and material characteristics, continuous optimization, and providing priority directions for new sample detection; The data analysis and classification module: establishes a data interface with the neuromorphic computing module, clarifies the transmission format and protocol, stores and classifies the output results, performs data cleaning, feature extraction and analysis, selects the classification algorithm training model, predicts the types of toxic and hazardous substances, and uses big data analysis technology to explore potential food safety issues; The result output and warning function: determine the result presentation format according to user needs and scenarios, organize and format data, design a simple user interface, set safety thresholds in combination with standards, and establish a real-time monitoring mechanism. Once the test results exceed the threshold, a warning is issued in various ways, and relevant personnel will take measures such as suspending production and sales, investigating and rectifying after receiving the warning; The system maintenance and optimization aspects include: regular maintenance, hardware inspection, software update, cleanup and backup database, continuous collection of new data, quality control and storage management, evaluation and optimization of analysis models, and introduction of new technologies to enhance detection capabilities.

2. The system for structuring big data of food safety toxic and hazardous substances inspection and detection according to claim 1 is characterized in that: The data acquisition, installation and testing equipment must first undergo demand assessment, equipment selection and procurement, and then installation, debugging and calibration verification. After preparing the food samples, multimodal data acquisition and storage management are performed using spectrometers, image acquisition equipment, and chemical sensors. Next, network planning is performed, network equipment is installed and configured, data transmission is tested, data transmission protocols are selected, data transmission software is developed, a data storage and management center is configured, and the equipment and software are started to transmit the collected data to the center in real time for monitoring.

3. The system for structuring big data of food safety toxic and hazardous substance inspection and detection according to claim 1 is characterized in that: The pattern recognition algorithm selected by the neuromorphic computing performs pattern recognition and classification processing on the feature vector to determine whether there are toxic and harmful substances in the food sample and the specific types: in, represents the feature vector of the food sample, represents the weight vector, w i f i It means that each eigenvalue f i And the corresponding weight value w i Multiply, It means adding the product of all feature values ​​and weight values, b means bias term, and a threshold T is set for the presence or absence of toxic and harmful substances in food samples. It is preliminarily judged that there may be toxic and harmful substances in the food sample. , it can be judged that there may be no toxic or harmful substances in the food sample.

4. The system for structuring big data of food safety toxic and hazardous substances inspection and detection according to claim 1 is characterized in that: The neuromorphic computing uses historical detection data and known characteristics of toxic and hazardous substances to establish a prediction model: in, Represents the input food sample feature vector, where each f i represents the i-th eigenvalue, θ represents the parameter set of the model, σ is the activation function, and u j i is the first layer connection feature f i and the weight of the middle layer node j, c j is the bias of the middle layer node j, v j is the weight connecting the intermediate layer and the output layer, d is the bias of the output layer, and g is the activation function or transformation function of the output layer.

5. The system for structuring big data of food safety toxic and hazardous substances inspection and detection according to claim 1 is characterized in that: The neuromorphic computing inputs the collected data into the prediction model to predict the types of toxic and hazardous substances that may exist: Among them, Φ is the set of model parameters after training and optimization, σ is the activation function, u ji The value after the first layer connection feature fusion and the weight of the middle layer node j, v j The weights connecting the middle layer and the output layer, c j The bias of the intermediate layer node j, d the bias of the output layer, g the activation function or transformation function of the output layer.

6. The system for structuring big data of food safety toxic and hazardous substances inspection and detection according to claim 1 is characterized in that: The data analysis and classification establishes a data interface with the neuromorphic computing module, determines the transmission format and protocol, stores and classifies the output results in a dedicated data storage system, performs data cleaning, feature extraction and analysis, selects a classification algorithm training model and performs prediction to determine the types of toxic and harmful substances in food samples, uses big data analysis technology to perform association analysis and cluster analysis, and discovers potential association rules and food safety issues.

7. The system for structuring big data of food safety toxic and hazardous substances inspection and detection according to claim 1 is characterized in that: The result output and early warning analysis determines the intuitive presentation form of the test results based on user needs and usage scenarios, organizes and formats data, and provides a concise and clear user interface with annotations. The results can also be output through multiple channels, and the safety thresholds of toxic and hazardous substances are determined in combination with standards and stored in the system. A real-time monitoring mechanism is established to obtain the latest data, and the test results are compared and judged with the thresholds. When the thresholds are exceeded, early warnings are issued through system pop-ups, email and SMS notifications, and sound and light alarms. After receiving the early warning, relevant departments and personnel should suspend production or sales, investigate the causes and take corrective measures, while strengthening detection and supervision to prevent the problem from escalating.

8. The system for structuring big data of food safety toxic and hazardous substance inspection and detection according to claim 1 is characterized in that: The system maintenance and optimization described herein determines the maintenance frequency and lists a task list for regular maintenance, covering hardware inspection, software update, database cleanup and backup recovery, continuously establishing data collection channels, controlling data quality and performing storage management, regularly evaluating analysis models, optimizing and improving based on the results, establishing a continuous improvement mechanism after verification and confirmation, constantly collecting new data and introducing new technologies to improve system detection capabilities.

Images