Food detection sample intelligent management system and method

Abstract

The invention discloses an intelligent management system and method for a food detection sample, and relates to the technical field of food management, and the method achieves the real-time collection and dynamic perception of multi-source smell diffusion data such as smell diffusion data through setting a plurality of data collection points in a target area of a food detection storage environment. And after the sample management center platform receives the data transmitted by each data acquisition point, high-precision alignment and dimensionless integration of the data are realized through preprocessing. The processing flow effectively eliminates the influence of sensing error, time drift and unit dimension difference, and ensures the comparability and consistency of smell data between samples. Therefore, compared with a traditional monitoring mode only depending on a single-point gas sensor, the storage smell dynamic monitoring system has the advantages that the'smell basic data set 'of the sample can be constructed in a software algorithm mode, and the storage smell dynamic monitoring system with low hardware dependence, uniform data standard and higher calculation precision is realized.

Description

Technical Field

[0001] This invention relates to the field of food management technology, specifically to an intelligent management system and method for food testing samples. Background Technology

[0002] With the rapid development of food safety testing technology and intelligent laboratory management systems, the food testing field is gradually shifting from traditional manual recording and timed testing methods to data-driven intelligent management. In the food testing process, a large number of samples need to be stored, numbered, preserved, and sent for testing in batches in closed or semi-closed storage environments. During this process, the samples continuously release volatile chemical components, forming unique odor characteristics. Because the volatile chemical components of different food samples vary and change over time, effectively identifying the odor diffusion characteristics of samples during the testing and storage stage, establishing calculable odor digital fingerprints, and intelligently identifying and dynamically controlling potential cross-contamination risks have become core issues that urgently need to be addressed in the intelligent management system of food testing.

[0003] Currently, food testing and storage management systems primarily rely on manual inspections, fixed-point gas sensor monitoring, or static concentration threshold alarm mechanisms to determine whether there is cross-influence of odors between samples. However, these methods generally have several shortcomings: First, the limited number of sensor deployment points and low spatial coverage make it difficult to accurately reflect the overall odor diffusion status of the storage area; second, the monitoring frequency is limited by the hardware refresh rate, resulting in insufficient real-time performance, often triggering alarms only after contamination has already spread; third, the judgment models rely solely on gas concentration or fixed thresholds, failing to quantify the similarity and dynamic evolution trends of odor characteristics between different samples; and fourth, data analysis mainly relies on single-point sampling and manual comparison, lacking global modeling and temporal correlation analysis capabilities. Due to the lack of systematic data description and mathematical modeling of "odor diffusion behavior," existing technologies struggle to predict potential odor diffusion and contamination trends between samples in advance, and cannot achieve data-driven dynamic adjustments and risk control.

[0004] Current food testing and storage management systems primarily rely on manual inspections, fixed-point gas sensor monitoring, or static concentration threshold alarm mechanisms to determine whether cross-influence of odors exists between samples. However, these methods generally have several shortcomings: First, the limited number of sensor deployment points and low spatial coverage make it difficult to accurately reflect the overall odor diffusion status of the storage area; second, the monitoring frequency is limited by the hardware refresh rate, resulting in insufficient real-time performance, often triggering alarms only after contamination has already spread; third, the judgment models rely solely on gas concentration or fixed thresholds, failing to quantify the similarity and dynamic evolution trends of odor characteristics between different samples; and fourth, data analysis mainly relies on single-point sampling and manual comparison, lacking global modeling and temporal correlation analysis capabilities. Due to the lack of systematic data description and mathematical modeling of "odor diffusion behavior," existing technologies struggle to predict potential odor diffusion and contamination trends between samples in advance, and cannot achieve data-driven dynamic adjustments and risk control. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent management system and method for food testing samples, solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution, comprising the following steps:

[0007] S1. Set up multiple data collection points in the target area of ​​the food testing and storage environment. The data collection points are connected to the sample management center platform through a wireless communication module to collect the odor diffusion data of the samples in real time, process it, and establish a basic sample dataset.

[0008] S2. Based on the sample basic dataset, calculate the odor diffusion potential energy intensity Eod for each sample, perform similarity calculation on the odor diffusion potential energy intensity Eod, output the similarity value Simod, and perform similarity evaluation.

[0009] S3. Based on the similarity assessment results, trigger the odor migration trend analysis mechanism. The odor migration trend analysis mechanism calculates the odor migration rate index Rmig between samples by using the odor diffusion potential energy intensity Eod of each sample, sets the migration threshold Rth to determine the time-varying migration trend, and then executes the protection control mechanism based on the determination result.

[0010] Preferably, S1 includes S11;

[0011] S11. By setting up acquisition devices at data acquisition points, odor diffusion data of samples in the food testing and storage environment are collected in real time; and a network connection is established between the wireless communication module built into each acquisition device and the receiving terminal of the sample management center platform to transmit the real-time collected odor diffusion data to the sample management center platform.

[0012] The data acquisition points include a first data acquisition point D1, a second data acquisition point D2, and a third data acquisition point D3;

[0013] The data acquisition device includes an embedded gas phase detection interface, a wireless temperature and humidity sensor, a radio frequency identification module, and a spatial positioning module;

[0014] The odor diffusion data includes component concentration C, ambient temperature Tenv, ambient humidity Henv, and the center-to-center distance Δx between sample n and sample m in the food testing storage space. nm .

[0015] Preferably, S1 further includes S12;

[0016] S12. After receiving the odor diffusion data transmitted from each data collection point, the sample management center platform preprocesses the odor diffusion data to establish a sample basic dataset. The preprocessing includes data synchronization and time alignment, signal denoising and outlier detection, and standardization.

[0017] The data synchronization and time alignment are achieved by calling the built-in time series synchronization module through the sample management center platform. The timestamps of each data collection point are calibrated through the NTP network time synchronization protocol. When the time difference exceeds the preset threshold of 2 seconds, automatic linear interpolation correction is performed.

[0018] The signal denoising and outlier detection are performed by using the moving average method and outlier detection algorithm. Based on the three-standard-deviation principle, noise filtering and outlier removal are performed on the component concentration C. The ambient temperature Tenv and ambient humidity Henv are processed by a joint algorithm of median filtering and Kalman filtering for steady-state smoothing.

[0019] The standardization process involves scaling the odor diffusion data by range scaling after signal denoising and outlier detection, converging all parameters in the odor diffusion data to between 0 and 1, eliminating the influence of unit dimensions, and then summarizing the standardized odor diffusion data to establish a basic sample dataset.

[0020] Preferably, S2 includes S21;

[0021] S21. Based on the obtained sample dataset, calculate and output the odor diffusion potential energy intensity Eod;

[0022] The odor diffusion potential energy intensity Eod is calculated and output using the following algorithm formula:

[0023] In the formula, I represents the total number of volatile components in the sample that are being modeled and analyzed, and D... i Chs represents the diffusion coefficient of the i-th specific chemical component.i Cbg represents the headspace concentration of the i-th specific chemical component. i Let represent the environmental background concentration of the i-th specific chemical component, and e represent the exponential function. This represents the temperature correction factor, and Tref represents the reference temperature.

[0024] Preferably, S2 further includes S22;

[0025] S21. Calculate and output the odor diffusion potential energy intensity Eod based on traversing all food samples, store all odor diffusion potential energy intensities Eod in the sample database of the sample management center platform, and calculate the similarity of the odor diffusion potential energy intensity Eod of all food samples, output the similarity value Simod, and perform quantitative analysis of odor overlap.

[0026] The similarity value Simod is calculated and output using the following algorithm formula: In the formula, Eod n Eod represents the odor diffusion potential intensity of the nth sample. m This represents the odor diffusion potential intensity of the m-th sample.

[0027] Preferably, S2 further includes S23;

[0028] S23. Based on the output of the similarity value Simod, a similarity assessment is performed to automatically determine the odor diffusion association between two food samples, and an odor migration trend analysis mechanism is automatically triggered based on the similarity assessment results; the specific assessment content is as follows:

[0029] When the similarity value Simod > 0.85, the odor fingerprint difference is not significant, and the food sample is marked as having abnormal odor similarity. At this time, the odor migration trend analysis mechanism is triggered.

[0030] When the similarity value Simod ≤ 0.85, the odor fingerprint difference is significant, and the food sample is marked as having normal odor similarity, at which point it enters the routine testing queue.

[0031] Preferably, S3 includes S31;

[0032] S31. When the similarity assessment triggers the odor migration trend analysis mechanism, the odor diffusion potential energy intensity Eod is combined with the center-to-center distance Δx between sample n and sample m in the food testing storage space. nm Time-varying migration trend analysis was performed to obtain the odor migration rate index Rmig(n,m) of sample n relative to sample m.

[0033] The odor migration rate index Rmig(n, m) of sample n relative to sample m is calculated and output using the following algorithm:

[0034] In the formula, d represents the integral function, dt represents the time integral function, and Eod represents the integral function. n (t) represents the odor diffusion potential energy intensity of the nth sample at time t, Eod m (t) represents the odor diffusion potential energy intensity of the m-th sample at time t.

[0035] Preferably, S3 further includes S32;

[0036] S32. Extract several sample pairs known to have and not have odor migration issues from historical storage batches. Record the odor migration rate index Rmig(n,m) sequence of sample n relative to sample m for these sample pairs within 24 hours before the problem occurs. Set the critical value of the odor migration rate index Rmig(n,m) of sample n relative to sample m as the migration threshold Rth when the odor migration problem exists or does not exist. Determine the time-varying migration trend between the migration threshold Rth and the real-time acquired odor migration rate index Rmig(n,m) of sample n relative to sample m, and implement a protective control mechanism based on the time-varying migration trend determination result. The specific determination content is as follows:

[0037] When the odor migration rate index Rmig(n,m) of sample n relative to sample m is greater than the migration threshold Rth, it indicates that the odor spillover poses a critical risk to neighboring samples, and a protective control mechanism is executed at this time.

[0038] When the odor migration rate index Rmig(n,m) of sample n relative to sample m is less than or equal to the migration threshold Rth, it means that odor leakage does not pose a risk to neighboring samples. In this case, the current configuration is maintained and monitoring is continued.

[0039] Preferably, S3 further includes S33;

[0040] S33. The protection and control mechanism includes software layer strategies, hardware layer coordination, and environmental adjustment strategies.

[0041] The software layer strategy marks food samples with critical risks, adjusts the priority of the marked food samples to the highest priority through the task scheduler, and initiates sample re-inspection within 15 minutes. At the same time, the status of risky food samples is highlighted on the user terminal interface, triggering a manual review prompt.

[0042] The hardware layer collaboration isolates video sample pairs marked with the highest priority through an automatic transfer device, maintaining a distance of greater than or equal to 1.8 meters from adjacent food samples, with the transfer time completed on average within 5 to 8 seconds.

[0043] The environmental control strategy adjusts the current food testing and storage environment to -0.5℃ to -1.0℃ by increasing the ventilation level by 15% to 25% and issuing a temperature control command.

[0044] An intelligent management system for food testing samples includes an odor acquisition module, an odor diffusion similarity analysis module, and an odor migration trend analysis module;

[0045] The odor collection module sets up multiple data collection points in the target area of ​​the food testing and storage environment. These data collection points are connected to the sample management center platform via a wireless communication module to collect and process the odor diffusion data of the samples in real time, thereby establishing a basic dataset of the samples.

[0046] The odor diffusion similarity analysis module calculates the odor diffusion potential energy intensity Eod for each sample based on the sample dataset, performs similarity calculations on the odor diffusion potential energy intensity Eod, outputs a similarity value Simod, and conducts a similarity evaluation.

[0047] The odor migration trend analysis module triggers the odor migration trend analysis mechanism based on the similarity assessment results. The odor migration trend analysis mechanism calculates the odor migration rate index Rmig between samples based on the odor diffusion potential energy intensity Eod of each sample, sets a migration threshold Rth to determine the time-varying migration trend, and then executes the protection control mechanism based on the determination results.

[0048] This invention provides an intelligent management system and method for food testing samples. It has the following beneficial effects:

[0049] (1) This method sets up multiple data collection points within the target area of ​​the food testing and storage environment, and deploys embedded gas phase detection interfaces, wireless temperature and humidity sensors, radio frequency identification modules, and spatial positioning modules at each data collection point. This enables real-time acquisition and dynamic sensing of multi-source odor diffusion data, such as component concentration C, ambient temperature Tenv, ambient humidity Henv, and sample center-to-center distance Δxnm. After receiving the data transmitted from each data collection point, the sample management center platform performs multi-level preprocessing, including time series synchronization, signal denoising, outlier removal, and range standardization, to achieve high-precision alignment and dimensionless integration of the data. This processing effectively eliminates the influence of sensing errors, time drift, and differences in unit dimensions, ensuring the comparability and consistency of odor data between samples. Therefore, compared with the traditional method that relies solely on single-point gas sensor monitoring, this invention can construct a "basic odor dataset" for samples using software algorithms, realizing a dynamic monitoring system for storage odors with low hardware dependence, unified data standards, and higher computational accuracy, providing a more stable foundation for subsequent diffusion potential calculation and pollution risk modeling.

[0050] (2) This method calculates the odor diffusion potential energy intensity based on an improved Fick diffusion law by executing an odor diffusion potential energy intensity calculation model in the sample management center platform. It combines the diffusion coefficient Di of each specific chemical component, the headspace concentration Chsi, and the environmental background concentration Cbgi to form the odor diffusion potential energy intensity Eod. A temperature correction coefficient β is introduced to achieve exponential correction of the diffusion capacity by environmental temperature. Based on this, the similarity of the odor diffusion potential energy intensity Eod of different food samples is calculated, and the odor diffusion similarity value Simod is output. When Simod is greater than 0.85, the system automatically determines that the odor fingerprint difference is not significant and triggers the odor migration trend analysis mechanism; when Simod is less than or equal to 0.85, the system automatically determines that the odor fingerprint difference is significant and maintains the conventional detection mode. Through this similarity assessment mechanism, potential cross-contamination risks can be identified and dynamically responded to in advance before contamination manifests. Compared to the traditional method relying on concentration threshold alarms, this invention utilizes the quantitative modeling of odor diffusion potential energy and fingerprint similarity to achieve a shift from static detection to dynamic prediction, effectively improving the foresight and intelligence level of food testing and storage management.

[0051] (3) This method, after the similarity assessment result triggers the odor migration trend analysis mechanism, calculates the odor migration rate index Rmig based on the odor diffusion potential energy intensity Eod and the spatial center distance Δxnm between samples, and compares it with the migration threshold Rth in real time. When Rmig exceeds the migration threshold Rth, the sample management center platform automatically executes the protection control mechanism, realizing a fully automatic closed loop from risk identification to prevention and control execution. The protection control mechanism includes three parts: software layer strategy, hardware layer coordination, and environmental adjustment strategy. The software layer marks and prioritizes food samples with critical risks through the scheduling system, ensuring that re-inspection is completed within 15 minutes and prompting manual review. The hardware layer isolates samples to a position with an adjacent distance of not less than 1.8 meters through an automatic racking device, and the moving average time is controlled within 5-8 seconds. The environmental layer increases the local ventilation volume by 15%-25% and reduces the storage temperature by 0.5℃-1.0℃ by adjusting the ventilation system and temperature control system, thereby weakening the odor overflow ability. The aforementioned three-layer linkage strategy enables the system to respond instantly, adjust automatically, and take proactive measures after detecting a contamination trend. This transforms risk response from passive alarm to proactive intervention, significantly reducing the probability of odor migration and cross-contamination between food samples and ensuring the reliability of test data and the stability of the storage environment. Attached Figure Description

[0052] Figure 1 This is a schematic diagram illustrating the steps of an intelligent management method for food testing samples according to the present invention.

[0053] Figure 2 This is a schematic diagram of a smart management system module for food testing samples according to the present invention.

[0054] Figure 3 This is a graph showing the trend of odor migration. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1

[0057] Please see Figure 1 This invention provides an intelligent management method for food testing samples. To achieve the above objectives, this invention is implemented through the following technical solution, including the following steps:

[0058] S1. Set up multiple data collection points in the target area of ​​the food testing and storage environment. The data collection points are connected to the sample management center platform through a wireless communication module to collect the odor diffusion data of the samples in real time, process it, and establish a basic dataset of the samples.

[0059] S2. Based on the sample dataset, calculate the odor diffusion potential energy intensity Eod for each sample, perform similarity calculation on the odor diffusion potential energy intensity Eod, output the similarity value Simod, and evaluate the similarity.

[0060] S3. Based on the similarity assessment results, the odor migration trend analysis mechanism is triggered. The odor migration trend analysis mechanism calculates the odor migration rate index Rmig between samples by using the odor diffusion potential energy intensity Eod of each sample, and sets a migration threshold Rth to determine the time-varying migration trend. Then, the protection control mechanism is executed based on the determination results.

[0061] In this embodiment, the method monitors the spatial distribution of odor diffusion characteristics by deploying multiple data collection points in the target area of ​​the food testing storage environment. A distributed acquisition structure is adopted because the odor diffusion of food samples in the storage space exhibits a significant spatial gradient effect, and single-point detection cannot reflect the overall diffusion state. The collected component concentration, ambient temperature and humidity, and sample spacing data are transmitted in real-time to the sample management center platform via a wireless communication module, enabling continuous tracking and dynamic sensing of the odor diffusion state. This real-time acquisition not only reduces delays caused by human intervention but also captures weak change signals in the early stages of odor anomalies, thus providing a highly timely data foundation for subsequent judgment. When calculating the odor diffusion potential energy intensity Eod based on the sample's basic dataset, an improved diffusion model incorporating a temperature correction coefficient β is used, reflecting the sample's odor release capacity at different temperatures. Its physical meaning is: the higher the temperature, the stronger the molecular kinetic energy, the greater the diffusion potential energy of the odor components, and the higher the Eod value; conversely, when the ambient temperature decreases, the diffusion rate slows down, and the Eod value decreases. Through this dynamic modeling, a direct correspondence can be established between physical diffusion behavior and changes in data parameters, thereby achieving accurate quantification of the sample's odor release trend. Furthermore, the similarity of the odor diffusion potential energy intensity Eod between different samples is calculated and the similarity value Simod is output, which can effectively distinguish odor fingerprint differences. When Simod is higher than 0.85, it indicates that the odor diffusion characteristics of the two samples are similar, and there is a potential risk of cross-contamination. At this time, the system will automatically trigger the migration trend analysis module. In the odor migration trend analysis mechanism, the system evaluates the direction and rate of odor diffusion between samples by calculating the odor migration rate index Rmig. The purpose of introducing the migration threshold Rth is to establish a dynamic critical judgment benchmark. When Rmig exceeds Rth, it means that the diffusion rate between adjacent samples has reached a level that may cause contamination. For example, when two alcohol samples are too close (e.g., less than 1.5 meters) and the ventilation rate is low, the odor spillover components may form a "superposition zone" locally, causing the odor concentration to be falsely high and interfering with the detection of adjacent samples. Through the threshold judgment mechanism, the system can intervene in advance before the detection value is abnormal and implement protective control mechanisms, including automatic sample isolation, environmental temperature control and ventilation adjustment, thereby physically blocking the odor migration path.

[0062] Example 2

[0063] Please see Figure 1 Specifically: S1 includes S11;

[0064] S11. By setting up acquisition devices at data acquisition points, odor diffusion data of samples in the food testing and storage environment are collected in real time; and a network connection is established between the wireless communication module built into each acquisition device and the receiving terminal of the sample management center platform to transmit the real-time collected odor diffusion data to the sample management center platform.

[0065] The data acquisition points include the first data acquisition point D1, the second data acquisition point D2, and the third data acquisition point D3;

[0066] The data acquisition equipment includes an embedded gas phase detection interface, a wireless temperature and humidity sensor, an RFID module, and a spatial positioning module;

[0067] Odor diffusion data include component concentration C, ambient temperature Tenv, ambient humidity Henv, and the center-to-center distance Δx between sample n and sample m in the food testing storage space. nm ;

[0068] Wherein: the first data acquisition point D1 is set near the detection position of each sample storage rack, and is used to read the concentration parameter C of each component in the sample detection database through the embedded gas phase detection interface, specifically representing the concentration C of the i-th specific chemical component. i , refers to the mass concentration of the i-th specific chemical component in a food test sample or its headspace, specifically including: ethanol, lactic acid, ethyl acetate and thiol;

[0069] The second data acquisition point D2 is located in the ventilation duct above the storage environment and is equipped with a wireless temperature and humidity sensor to collect ambient temperature Tenv and ambient humidity Henv in real time.

[0070] The third data acquisition point D3 is set in the warehouse video positioning system. The RFID module and the spatial positioning module obtain the warehouse spatial coordinate system and collect the short distance between samples. Specifically, by attaching RFID positioning tags to the bottom of each sample tray, the RFID module identifies the samples, and then the spatial positioning module locates the identified samples in the warehouse spatial coordinate system. After positioning, the distance is calculated using the Euclidean distance formula.

[0071] S1 also includes S12;

[0072] S12. After receiving the odor diffusion data transmitted from each data collection point, the sample management center platform preprocesses the odor diffusion data to establish a basic sample dataset. The preprocessing includes data synchronization and time alignment, signal denoising and outlier detection, and standardization.

[0073] Data synchronization and time alignment are achieved by calling the built-in time series synchronization module through the sample management center platform. The timestamps of each data acquisition point are calibrated through the NTP network time synchronization protocol. When the time difference exceeds the preset threshold of 2 seconds, automatic linear interpolation correction is performed to ensure the accuracy of data time alignment.

[0074] Signal denoising and outlier detection employ a moving average method and an outlier detection algorithm. Based on the principle of three standard deviations, noise filtering and outlier removal are performed on component concentration C. For ambient temperature Tenv and ambient humidity Henv, a joint algorithm of median filtering and Kalman filtering is used for steady-state smoothing to eliminate instantaneous communication interference and equipment errors.

[0075] Standardization processing involves range scaling of the odor diffusion data after signal denoising and outlier detection, converging all parameters in the odor diffusion data to between 0 and 1, eliminating the influence of unit dimensions, and then summarizing the standardized odor diffusion data to establish a basic sample dataset.

[0076] In this embodiment, the intelligent management system for food testing samples first deploys multiple data acquisition points in the storage area. The first data acquisition point D1, located near the sample storage rack detection position, is used to collect the concentration C of specific chemical components in the sample head space in real time via an embedded gas phase detection interface, capturing the most authentic odor release state of the food sample. The second data acquisition point D2, located above the ventilation channel, is used to monitor the ambient temperature (Tenv) and ambient humidity (Henv) in real time, as these parameters directly affect the evaporation rate and diffusion behavior. The third data acquisition point D3 integrates a radio frequency identification module and a spatial positioning module, used to accurately calculate the center-to-center distance Δx between samples. nmThis allows the system to obtain the spatial distribution relationship between samples. The reason for using three-point collaborative acquisition is that the diffusion of food odors exhibits significant spatial non-uniformity. Relying solely on single-point detection can easily lead to misjudgments such as "locally high concentrations and distorted overall trends." Through this spatially distributed acquisition layout, the system can simultaneously acquire the odor source release intensity, environmental disturbance factors, and geometric position changes, forming complete physical diffusion field data. After the data is transmitted in real-time to the sample management center platform via the wireless communication module, the system immediately executes a preprocessing procedure. First, the timestamps of each acquisition point are aligned using the NTP network time synchronization protocol. When a time difference exceeds 2 seconds, automatic linear interpolation correction is performed to ensure that data from different acquisition points are calculated on the same time reference. This avoids temporal misalignment in the calculation of the odor diffusion potential energy intensity Eod. Subsequently, the system uses a moving average method and outlier detection algorithm to perform noise filtering and anomaly removal on the component concentration C. Simultaneously, a combination of mean filtering and Kalman filtering algorithms is applied to the temperature (Tenv) and humidity (Henv) to achieve steady-state smoothing, preventing false concentration mutations caused by instantaneous communication jitter or equipment drift. For example, if the temperature probe is impacted by airflow during a ventilation switch, the unfiltered wave will cause an instantaneous increase in Eod, thus falsely triggering the migration trend determination. Through this noise reduction process, the system ensures that each data point represents the true physical state. After data purification, the system performs range scaling on the odor diffusion data, uniformly mapping all parameters to a standardized range of 0–1, achieving dimensionless processing. The physical significance of this is that it makes the component concentration C, temperature and humidity parameters Tenv and Henv correlated with the spatial distance Δx. nm By using the same numerical scale for model calculations, computational bias caused by unit differences is avoided. This process makes the calculation results of odor diffusion potential energy intensity Eod, odor similarity Simod, and odor migration rate index Rmig more stable and comparable. Especially in scenarios with small sample spacing (e.g., less than 1.5 meters) and rising temperatures, neglecting distance normalization can lead to adjacent samples being misclassified as having "similar odors" due to spatial gas superposition. However, by introducing Δxnm and unifying the dimensions, the system can accurately identify the true diffusion coupling relationship.

[0077] Example 3

[0078] Please see Figure 1 Specifically: S2 includes S21;

[0079] S21. Based on the obtained sample dataset, calculate and output the odor diffusion potential energy intensity Eod;

[0080] The odor diffusion potential energy intensity Eod is calculated and output using the following algorithm formula:

[0081] In the formula, I represents the total number of volatile components in the sample that are being modeled and analyzed, and D... i Chs represents the diffusion coefficient of the i-th specific chemical component, and its value is dimensionless. i Cbg represents the headspace concentration of the i-th specific chemical component. i Let represent the environmental background concentration of the i-th specific chemical component, and e represent the exponential function. This represents the temperature correction factor, a dimensionless value ranging from 0.02 to 0.4, which is set by the user according to the food testing and storage environment. Tref represents the reference temperature.

[0082] It should be noted that the headspace concentration Chs of the i-th specific chemical component i It represents the concentration of the i-th volatile chemical component in the gas phase within a fixed space above the surface of a food sample in a food testing and storage environment. In other words, it reflects the concentration level of the component released from the sample matrix and represents the intensity of the sample's own "odor".

[0083] The environmental background concentration of the i-th specific chemical component (Cbg) i This represents the average concentration of the i-th type of chemical component in the air of the storage environment where the sample is located. It reflects the "background presence" of this gas in the ambient air. In food storage environments, this typically comes from: the volatilization of other samples in the same storage area, residual odors in the environment (such as alcohol, acid, sulfur, etc.), and residues from the ambient gas circulation system; the difference between the two (Chs) i -Cbg i () indicates that the concentration gradient of the i-th component from the interior of the sample to the external environment is the "potential energy difference" or "diffusion motive force" driving diffusion.

[0084] The diffusion coefficient D of the i-th specific chemical component i This reflects the velocity at which a certain gas is uniformly distributed in air due to molecular thermal motion. The diffusion coefficients for ethanol, lactic acid, ethyl acetate, and thiols are respectively: 1.20 × 10⁻⁶. -5 0.85×10 -5 1.05×10 -5 and 1.30×10 -5 ; After being dimensionless, the values ​​are 0.923, 0.808, 0.654 and 1.000;

[0085] This formula is derived from a combination of Fick's first law and the temperature effect. Fick flux is proportional to the diffusion coefficient multiplied by the concentration difference. Increased temperature enhances molecular kinetic energy, therefore an exponential temperature correction term is added. This formula, considering the characteristics of headspace detection for food samples, proposes the following two improvements: It no longer uses spatial gradient to represent concentration differences, but instead uses the concentration difference between the headspace and the ambient background (Chs). i-Cbg i This simplifies parameter acquisition; the diffusion coefficient correction term is incorporated into the overall model, and an overall summation framework is introduced to form a model of the overall escape potential of the sample, resulting in the final improved formula.

[0086] S2 also includes S22;

[0087] S21. Calculate and output the odor diffusion potential energy intensity Eod based on traversing all food samples, store all odor diffusion potential energy intensities Eod in the sample database of the sample management center platform, and calculate the similarity of the odor diffusion potential energy intensity Eod of all food samples, output the similarity value Simod, and perform quantitative analysis of odor overlap.

[0088] The similarity value Simod is calculated and output using the following algorithm formula: In the formula, Eod n Eod represents the odor diffusion potential intensity of the nth sample. m This represents the odor diffusion potential energy intensity of the m-th sample; this quantification value defines the odor diffusion fingerprint spectrum similarity, used to measure the similarity of odor characteristics between two samples; the formula originates from: cosine similarity (a classic algorithm in mathematics used to measure the similarity of feature vectors), and in this application, it means: used to measure the angle similarity between the odor diffusion feature vectors (i.e., Eod vectors) of sample n and sample m, reflecting the similarity between the two in the odor diffusion potential energy distribution, thereby used to determine the potential risk of cross-contamination of odors between samples.

[0089] S2 also includes S23;

[0090] S23. Based on the output of the similarity value Simod, a similarity assessment is performed to automatically determine the odor diffusion association between two food samples, and an odor migration trend analysis mechanism is automatically triggered based on the similarity assessment results; the specific assessment content is as follows:

[0091] When the similarity value Simod > 0.85, the odor fingerprint difference is not significant, and the food sample is marked as having abnormal odor similarity. At this time, the odor migration trend analysis mechanism is triggered.

[0092] When the similarity value Simod ≤ 0.85, the odor fingerprint difference is significant, and the food sample is marked as having normal odor similarity, at which point it enters the routine testing queue.

[0093] In this embodiment, the odor diffusion potential energy intensity Eod of each food sample is calculated based on the established sample dataset. This parameter is chosen as the core indicator because Eod can quantitatively characterize the diffusion capacity of odor components from the inside out; its physical meaning is equivalent to "odor escape potential energy." The Eod calculation formula introduces a temperature correction term based on Fick's first law, utilizing the headspace concentration Chs. i Compared with the background concentration of Cbg i The difference in odor values ​​is used instead of the traditional spatial gradient to simplify parameter acquisition and improve real-time computation. For example, when the ambient temperature (Tenv) increases or ventilation efficiency decreases, the kinetic energy of odor molecules increases, the diffusion potential energy rises accordingly, and the Eod value increases significantly; conversely, if the ambient temperature is stable and the airflow is uniform, the Eod remains at a low level. This design allows the system to capture changes in the odor diffusion trend of food samples in real time when there are slight temperature fluctuations in the storage environment. After calculating the Eod of each sample, the system calculates a similarity value (Simod) using a cosine similarity algorithm to assess the degree of similarity of odor characteristics between different samples. Cosine similarity is chosen instead of Euclidean distance because it avoids bias caused by differences in the absolute value of sample Eod and can more accurately reflect the "shape similarity" of odor distribution patterns. In real-world scenarios, when two samples (such as fermented dairy products and fruit juice samples) volatilize some common components (such as ethyl acetate and ethanol) under similar conditions, their Simod will increase; when Simod is greater than 0.85, it indicates that the odor diffusion characteristics of the two have converged, and the system automatically determines "odor similarity anomaly" and triggers the odor migration trend analysis mechanism. At this point, the platform will further analyze the spatial distribution between samples and the change in odor diffusion potential energy over time to determine whether there is a potential risk of cross-contamination. The real physical significance of this implementation method lies in transforming odor diffusion from "single-point gas detection" to "multi-dimensional spatiotemporal modeling": Eod characterizes the odor release intensity, Simod quantifies the proximity of diffusion characteristics between samples, and the evaluation threshold (0.85) serves as the physical boundary for determining diffusion coupling between two samples. For example, in a storage space, if sample n and sample m are close to each other (e.g., Δx... nm Even if the distance between two food items is less than 1.2 meters and the ambient temperature rises by 2°C, the detection signals may appear to overlap due to the superposition of odor components, leading to misjudgment as having the same odor. This algorithm allows the system to quickly identify such "false similarity" and trigger protective control strategies in advance. Therefore, this implementation process achieves real-time quantification, dynamic assessment, and intelligent judgment of odor diffusion in food detection. Its beneficial effects include significantly improving the sensitivity and stability of odor monitoring, reducing misjudgments caused by environmental fluctuations or close sample proximity, and achieving automated identification and early intervention of odor migration risks, thereby effectively preventing cross-contamination of odors in the storage environment.

[0094] Example 4

[0095] Please see Figure 1 and Figure 3 Specifically: S3 includes S31;

[0096] S31. When the similarity assessment triggers the odor migration trend analysis mechanism, the odor diffusion potential energy intensity Eod is combined with the center-to-center distance Δx between sample n and sample m in the food testing storage space. nm Time-varying migration trend analysis was performed to obtain the odor migration rate index Rmig(n,m) of sample n relative to sample m.

[0097] The odor migration rate index Rmig(n, m) of sample n relative to sample m is calculated and output using the following algorithm:

[0098] In the formula, d represents the integral function, dt represents the time integral function, and Eod represents the integral function. n (t) represents the odor diffusion potential energy intensity of the nth sample at time t, Eod m (t) represents the odor diffusion potential energy intensity of the m-th sample at time t;

[0099] This formula treats the odor diffusion potential energy intensity Eod as a time signal of the "sample odor escape potential," and uses Eod to represent the odor diffusion potential between two samples n and m. n (t)−Eod m (t) represents the "time difference" of the outflow potential of the two; the time derivative d / dt measures the "rate of change of the difference over time" (which is getting stronger and which is getting weaker); and the center-to-center distance Δx between the samples is used as the metric. nm By performing spatial normalization, we obtain the "time-varying velocity per unit distance", which is the odor migration rate index Rmig(n, m).

[0100] S3 also includes S32;

[0101] S32. Extract several sample pairs known to have and not have odor migration issues from historical storage batches. Record the odor migration rate index Rmig(n,m) sequence of sample n relative to sample m for these sample pairs within 24 hours before the problem occurs. Set the critical value of the odor migration rate index Rmig(n,m) of sample n relative to sample m as the migration threshold Rth when the odor migration problem exists or does not exist. Determine the time-varying migration trend between the migration threshold Rth and the real-time acquired odor migration rate index Rmig(n,m) of sample n relative to sample m, and implement a protective control mechanism based on the time-varying migration trend determination result. The specific determination content is as follows:

[0102] When the odor migration rate index Rmig(n,m) of sample n relative to sample m is greater than the migration threshold Rth, it indicates that the odor spillover poses a critical risk to neighboring samples, and a protective control mechanism is executed at this time.

[0103] When the odor migration rate index Rmig(n,m) of sample n relative to sample m is less than or equal to the migration threshold Rth, it means that odor leakage does not pose a risk to neighboring samples. In this case, the current configuration is maintained and monitoring is continued.

[0104] S3 also includes S33;

[0105] S33. The protection and control mechanism includes software layer strategies, hardware layer coordination, and environmental adjustment strategies.

[0106] The software-layer strategy involves marking food samples with critical risks, adjusting the priority of the marked food samples to the highest priority through the task scheduler, initiating sample re-inspection within 15 minutes, highlighting the status of risky food samples on the user terminal interface, and triggering a manual review prompt.

[0107] Hardware layer collaboration isolates video sample pairs marked with the highest priority using an automatic transfer device, maintaining a distance of at least 1.8 meters between them and adjacent food samples, with the transfer time averaging within 5 to 8 seconds.

[0108] The environmental control strategy involves adjusting the ventilation level of the current food testing and storage environment to -0.5℃ to -1.0℃ by increasing it by 15% to 25% and issuing temperature control instructions.

[0109] In this embodiment, after similarity assessment is triggered, the method uses the temporal difference of the odor diffusion potential energy intensity Eod as a measure of "source strength change", and combines it with the center-to-center distance Δx between sample n and sample m. nmThe odor migration rate index Rmig(n,m) is calculated, which is physically defined as "the rate of change of the outward potential energy difference per unit distance". It is used to characterize who exerts odor influence on whom faster. The purpose of setting the migration threshold Rth is to provide an executable critical criterion, so that the judgment is upgraded from "whether they are similar" to "whether they are accelerating migration". When Rmig(n, m) exceeds Rth, it indicates an accelerated diffusion trend towards neighboring samples under the current Δxnm and environmental conditions. For example, when Δxnm decreases from 1.8 meters to 1.2 meters and ventilation temporarily decreases, local accumulation of alcoholic and ester-containing samples can lead to falsely high headspace readings. In this case, the system immediately marks the risky sample pair, prioritizes retesting, and initiates verification within 15 minutes. Simultaneously, it instructs the automatic rack-shifting device to isolate the two samples to a distance ≥1.8 meters and coordinates with the environmental layer to increase ventilation by 15%–25% and lower the temperature by 0.5℃–1.0℃ to reduce the superposition effect of molecular kinetic energy and convection. If Rmig(n, m) does not exceed Rth, the existing storage location and routine monitoring are maintained to avoid unnecessary mechanical actions and energy consumption. This is achieved through "Eod time-varying + Δx". nm The real-time closed loop of "normalization + Rmig thresholding" enables the pre-judgment and interpretable intervention of cross-contamination. The direct effect is to reduce misjudgment and re-inspection caused by close proximity and short-term temperature rise, shorten the time from risk emergence to disposal completion, and significantly improve the stability, response efficiency and data reliability of warehouse testing without changing the hardware architecture.

[0110] Example 5

[0111] Please see Figure 1 and Figure 2 An intelligent management system for food testing samples includes an odor acquisition module, an odor diffusion similarity analysis module, and an odor migration trend analysis module.

[0112] The odor acquisition module sets up multiple data acquisition points in the target area of ​​the food testing and storage environment. The data acquisition points are connected to the sample management center platform through a wireless communication module to collect odor diffusion data of the samples in real time, process the data, and establish a basic dataset of the samples.

[0113] The odor diffusion similarity analysis module calculates the odor diffusion potential energy intensity Eod for each sample based on the sample dataset, performs similarity calculations on the odor diffusion potential energy intensity Eod, outputs the similarity value Simod, and evaluates the similarity.

[0114] The odor migration trend analysis module triggers the odor migration trend analysis mechanism based on the similarity assessment results. The odor migration trend analysis mechanism calculates the odor migration rate index Rmig between samples based on the odor diffusion potential energy intensity Eod of each sample, sets a migration threshold Rth to determine the time-varying migration trend, and then executes the protection control mechanism based on the determination results.

[0115] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A food detection sample intelligent management method, characterized in that: The method comprises the following steps: S1, setting multiple data collection points in a target area of a food detection storage environment, connecting the data collection points with a sample management center platform through a wireless communication module, collecting odor diffusion data of samples in real time, processing the odor diffusion data, and establishing a sample basic data set; S2, based on the sample basic data set, calculating the odor diffusion potential intensity Eod of each sample, performing similarity calculation on the odor diffusion potential intensity Eod, outputting a similarity value Simod, and performing similarity evaluation; S3, based on the similarity evaluation result, triggering an odor migration trend analysis mechanism, the odor migration trend analysis mechanism calculating the odor migration rate index Rmig between samples through the odor diffusion potential intensity Eod of each sample, setting a migration threshold Rth to determine the time-varying migration trend, and executing a protection control mechanism based on the determination result. 2.The food detection sample intelligent management method according to claim 1, characterized in that: The S1 comprises S11; S11, collecting odor diffusion data of samples in a food detection storage environment in real time by setting collection equipment in the data collection points; and based on the network connection between the wireless communication module in each collection equipment and the receiving terminal of the sample management center platform, transmitting the real-time collected odor diffusion data to the sample management center platform; The data collection points comprise a first data collection point D1, a second data collection point D2, and a third data collection point D3; The collection equipment comprises an embedded gas phase detection interface, a wireless temperature and humidity sensor, a radio frequency identification module, and a spatial positioning module; The odor diffusion data includes component concentration C, ambient temperature Tenv, ambient humidity Henv, and the center-to-center distance Δx of sample n and sample m in the food detection warehouse space nm . 3.The food detection sample intelligent management method according to claim 2, characterized in that: The S1 further comprises S12; S12, after receiving the odor diffusion data transmitted by each data collection point, the sample management center platform pre-processes the odor diffusion data, establishes a sample basic data set, and the pre-processing comprises data synchronization and time alignment, signal denoising and outlier detection, and standardization processing; The data synchronization and time alignment calibrate the time stamps of each data collection point through the NTP network time synchronization protocol by calling the built-in time sequence synchronization module of the sample management center platform, and automatically linearly interpolate and correct when detecting that the time difference exceeds the preset threshold of 2 seconds; The signal denoising and outlier detection perform noise filtering and outlier rejection on the component concentration C based on the three-sigma principle by using the moving average method and the outlier detection algorithm, and perform steady-state smoothing processing on the environmental temperature Tenv and the environmental humidity Henv by using the median filter and Kalman filter joint algorithm; The standardization processing converges all parameters in the odor diffusion data to 0-1 after completing the signal denoising and outlier detection, eliminates the influence of unit dimension, and then aggregates the odor diffusion data after standardization processing to establish a sample basic data set.

4. The food detection sample intelligent management method according to claim 3, characterized in that: The S2 comprises S21; S21, based on the obtained sample basic data set, calculating and outputting the odor diffusion potential intensity Eod; The odor diffusion potential intensity Eod is calculated and outputted through the following algorithm formula: ; where I represents the total number of volatile components in the sample being modeled, D i represents the diffusion coefficient of the i-th specific chemical component, Chs i represents the headspace concentration of the i-th specific chemical component, Cbg i represents the environmental background concentration of the i-th specific chemical component, and e represents the exponential function, represents the temperature correction factor, and Tref represents the reference temperature.

5. The food detection sample intelligent management method according to claim 4, characterized in that: The S2 further comprises S22; S21, calculate and output odor diffusion potential energy intensity Eod based on traversing all food samples, store all odor diffusion potential energy intensity Eod in the sample database of the sample management center platform, and perform similarity calculation on the odor diffusion potential energy intensity Eod of all food samples, output similarity value Simod, and perform quantitative analysis on odor overlap; The similarity value Simod is calculated and output by the following algorithm formula: ; wherein, Eod n represents the odor diffusion potential energy intensity of the nth sample, Eod m represents the odor diffusion potential energy intensity of the mth sample.

6. The food detection sample intelligent management method according to claim 5, characterized in that: The S2 further includes S23; S23, based on the output result of the similarity value Simod, automatically determine the odor diffusion correlation between the two food samples through similarity evaluation, and automatically trigger the odor migration trend analysis mechanism based on the similarity evaluation result; the specific evaluation content is as follows: When the similarity value Simod is greater than 0.85, the odor fingerprint difference is not significant, and the food sample is marked as odor similarity abnormality, at which time the odor migration trend analysis mechanism is triggered; When the similarity value Simod is less than or equal to 0.85, the odor fingerprint difference is significant, and the food sample is marked as odor similarity normal, at which time it enters the regular detection queue.

7. The food detection sample intelligent management method according to claim 6, characterized in that: The S3 includes S31; S31、When the similarity evaluation triggers the smell migration trend analysis mechanism, based on the smell diffusion potential energy intensity Eod, the center distance △x between sample n and sample m in the food detection storage space is combined to perform time-varying migration trend analysis, and the smell migration rate index Rmig(n,m) of sample n relative to sample m is obtained. nm , perform time-varying migration trend analysis, and obtain the smell migration rate index Rmig(n,m) of sample n relative to sample m. The odor migration rate index Rmig(n,m) of sample n relative to sample m is calculated and output by the following algorithm formula: wherein d represents an integral function, dt represents a time integral function, Eod n Eod (t) represents the odor diffusion potential energy intensity of the n-th sample at the time t, Eod m Eod (t) represents the odor diffusion potential energy intensity of the m-th sample at the time t. 8.The food detection sample intelligent management method of claim 6, wherein: The S3 further includes S32; S32, from the historical warehouse batch, extract a number of sample pairs known to have odor migration problems or not, record the odor migration rate index Rmig(n,m) sequence of sample n relative to sample m within 24 hours before the problem occurs, set the critical value of the odor migration rate index Rmig(n,m) of sample n relative to sample m as the migration threshold Rth when the value exists or does not exist odor migration problem, and perform time-varying migration trend judgment on the migration threshold Rth and the real-time acquired odor migration rate index Rmig(n,m) of sample n relative to sample m, and execute the protection control mechanism based on the time-varying migration trend judgment result; the specific judgment content is as follows: When the odor migration rate index Rmig(n,m) of sample n relative to sample m is greater than the migration threshold Rth, it indicates that odor overflow constitutes a critical risk to adjacent samples, at which time the protection control mechanism is executed; When the odor migration rate index Rmig(n,m) of sample n relative to sample m is less than or equal to the migration threshold Rth, it indicates that odor overflow does not constitute a risk to adjacent samples, at which time the current configuration is maintained and only continuous monitoring is performed.

9. The food detection sample intelligent management method according to claim 8, characterized in that: The S3 further includes S33; S33, the protection control mechanism includes software layer strategy, hardware layer cooperation and environment adjustment strategy; The software layer strategy marks the food sample pair with critical risk, adjusts the priority of the marked food sample to the highest priority through the task scheduler, starts sample re-inspection within less than or equal to 15 minutes, and highlights the risk food sample state on the user terminal interface to trigger manual review prompt; The hardware layer cooperation performs sample isolation through automatic rack moving equipment for the video sample pair marked as the highest priority, keeps the distance between adjacent food samples greater than or equal to 1.8 meters, and moves the time to complete within an average of 5-8 seconds; The environment adjusting strategy prompts 15%-25% of the ventilation component of the current food detection storage environment, and issues a temperature control instruction, so as to adjust the current food detection storage environment to-0.5℃ to-1.0℃.

10. A food detection sample intelligent management system applied to the food detection sample intelligent management method of any one of claims 1-9, characterized in that: The system comprises an odor collection module, an odor diffusion similarity analysis module and an odor migration trend analysis module. The odor collection module sets multiple data collection points in a target area of the food detection storage environment, and connects the data collection points with the sample management center platform through a wireless communication module, so as to collect odor diffusion data of samples in real time, process the data, and establish a sample basic data set. The odor diffusion similarity analysis module calculates the odor diffusion potential intensity Eod of each sample based on the sample basic data set, performs similarity calculation on the odor diffusion potential intensity Eod, outputs a similarity value Simod, and performs similarity evaluation. The odor migration trend analysis module triggers an odor migration trend analysis mechanism based on the similarity evaluation result, the odor migration trend analysis mechanism calculates the odor migration rate index Rmig between samples based on the odor diffusion potential intensity Eod of each sample, sets a migration threshold Rth to perform time-varying migration trend judgment, and executes a protection control mechanism based on the judgment result.

Images