A fresh pork freshness rapid determination method

Abstract

The application relates to the technical field of food quality detection, and discloses a fresh pork freshness rapid determination method; the method is characterized in that: firstly, the to-be-detected fresh pork is placed into a detection bin for temperature and humidity pre-equilibration, and gas response signals, fat oxidation coloration signals, visible near-infrared spectrum signals, bin environment signals and sample surface temperature signals are synchronously collected; after the signals are corrected, compensated and abnormally removed, microbial corruption characteristics, fat oxidation characteristics and sample color, luster and moisture state characteristics are extracted; temperature change influence is determined according to the deviation relationship among the three types of characteristics, the microbial corruption characteristics and the fat oxidation characteristics are corrected, the corrected characteristics are matched with a state reference library, and the freshness grade of the to-be-detected fresh pork is output; the freshness of the fresh pork in a cold chain circulation scene is rapidly, objectively and reliably determined.

Description

Technical Field

[0001] This invention relates to the field of food quality testing technology, specifically a method for rapidly determining the freshness of fresh pork. Background Technology

[0002] Fresh pork is susceptible to quality degradation during slaughtering, processing, cold chain transportation, wholesale and retail, and terminal sales due to factors such as temperature fluctuations, microbial growth, protein decomposition, and fat oxidation. Therefore, the ability to quickly and accurately determine the freshness of fresh pork is directly related to food safety control and handling in the distribution process. Existing technologies include, for example, the authorized Chinese invention patent application CN114805875B, which discloses a low-temperature molded intelligent indicator film for visualizing pork freshness, its preparation method, and its application. This film uses a shikonin indicator film to generate a color response to volatile basic nitrogen in the packaging and combines this color parameter to determine the freshness of the pork. Another example is the authorized Chinese invention patent application CN112465094B, which discloses a dynamic QR code intelligent sensor label system for real-time monitoring of food freshness. This system uses a pH indicator to form a QR code label and provides freshness indications based on color changes caused by volatile alkaline gases produced during food spoilage.

[0003] The methods described above can achieve visual detection of meat freshness to a certain extent, but their judgment is mainly based on volatile alkaline substances or color changes. Fresh pork often experiences temperature fluctuations during actual circulation due to changes in loading, unloading, transfer, transportation, and display conditions. Temperature fluctuations simultaneously affect the rate of microbial reproduction, the degree of protein decomposition, and the degree of fat oxidation. Since the responses of microbial spoilage and fat oxidation to temperature changes are not entirely consistent, when the sample under test experiences unknown temperature changes, or when fat oxidation has occurred significantly but volatile basic nitrogen has not yet increased significantly, relying solely on volatile alkaline gases or color changes for judgment can easily lead to inconsistencies between the test results and the actual quality of the sample. This is especially true in the critical state between fresh and near-fresh, where misjudgments are likely to occur, and samples with food safety risks may even be judged as being suitable for sale. Therefore, there is an urgent need to provide a rapid freshness determination method for fresh pork that can simultaneously reflect microbial spoilage and fat oxidation, and combine the effects of temperature changes for comprehensive judgment. This would solve the problem that existing single-indicator detection methods are difficult to adapt to real cold chain circulation scenarios, thereby improving the accuracy of freshness determination for samples in critical states, reducing the risk of misjudgment, and achieving rapid, objective, and reliable on-site determination of fresh pork freshness. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for rapidly determining the freshness of fresh pork, thus solving the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for rapidly determining the freshness of fresh pork, characterized by comprising:

[0007] The fresh pork to be tested was placed in the testing chamber, and the temperature and humidity of the testing chamber were pre-balanced. At the same time reference, gas response signals, fat oxidation color development signals, visible and near-infrared spectral signals, chamber environment signals and sample surface temperature signals were collected.

[0008] Baseline correction, temperature compensation, channel response correction, and outlier sampling removal are performed on the acquired signals.

[0009] Based on the corrected signals, microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics were extracted respectively.

[0010] Based on the correspondence between microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics, the influence of temperature changes experienced by the fresh pork under test was determined, and the microbial spoilage characteristics and lipid oxidation characteristics were corrected.

[0011] The corrected microbial spoilage characteristics, corrected lipid oxidation characteristics, and sample color and moisture state characteristics are matched with a preset state reference library to output the freshness grade of the fresh pork to be tested.

[0012] Further, the fresh pork to be tested is placed into the testing chamber, and the temperature and humidity of the testing chamber are pre-equilibrated, including:

[0013] First, check the sealing status of the detection chamber, the stability of the light source, the zero point of the gas response channel, the colorimetric acquisition area, and the spectral background;

[0014] Record the sample part and sample morphology of the fresh pork to be tested, and remove blood, frost and packaging condensation from the surface of the fresh pork to be tested;

[0015] When there is a water film on the surface of the fresh pork to be tested, the fresh pork to be tested is left to stand, and then the testing chamber is closed and the temperature and humidity inside the testing chamber are adjusted.

[0016] If the preset environmental range is not met, the pre-balancing process will be extended, the detection will be suspended, or the degradation decision will be made.

[0017] Furthermore, gas response signals, aliphatic oxidation colorimetric signals, visible and near-infrared spectral signals, chamber environment signals, and sample surface temperature signals were collected at the same time reference, including:

[0018] When collecting data at the same time reference, hardware synchronization triggering, software time stamp alignment, or group synchronization methods are used to collect gas response, colorimetric images, visible and near-infrared spectra, chamber environment, and sample surface temperature, and generate multi-channel data records.

[0019] The multi-channel data acquisition record should include at least the acquisition time identifier, signal type, acquisition channel number, original signal value, chamber environmental parameters, sample surface temperature, acquisition method identifier, and abnormal status identifier.

[0020] Furthermore, baseline correction, temperature compensation, channel response correction, and outlier removal are performed on each acquired signal, including:

[0021] First, verify the validity of the gas response signal, the colorimetric signal of fat oxidation, and the visible and near-infrared spectral signal;

[0022] Based on the signal types of gas response signal, fat oxidation color development signal and visible and near-infrared spectral signal, empty chamber baseline, initial color data, reference color data, reference plate data and dark background data are selected respectively;

[0023] Based on the detection chamber environment, sample surface temperature, and channel calibration data, each signal is corrected, and burrs, missing, saturated, and time-misaligned data are removed, generating a calibration data record with baseline source, compensation identifier, correction coefficient, and anomaly identifier.

[0024] Furthermore, based on the corrected signals, microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics were extracted, including:

[0025] Verify the availability of the gas response sequence, colorimetric image sequence, and spectral reflectance sequence;

[0026] Select the effective response segment from the gas response sequence and extract the response amplitude, rise rate, settling time, settling segment fluctuation amplitude, and channel response sequence;

[0027] The effective color-developing area is determined from the color-developing image sequence, and the color change amplitude, color change direction, color change speed, and color stability after reaction are extracted.

[0028] The band ratio, band differences, and local variation trends were extracted from the spectral reflectance sequence according to the color-related band and the moisture scattering-related band.

[0029] Generate feature data records that include validity flags, anomaly flags, missing item flags, and inconsistency flags.

[0030] Furthermore, the availability of the gas response sequence, colorimetric image sequence, and spectral reflectance sequence is verified, including:

[0031] When determining the available state of each sequence, the gas response sequence has a continuous rising segment or a stable changing segment;

[0032] The colorimetric image sequence has a recognizable effective colorimetric region;

[0033] The spectral reflectance sequence has been corrected to the reference.

[0034] When only two types of sequences among the gas response sequence, colorimetric image sequence, and spectral reflectance sequence meet the usability criteria, extract the features corresponding to the sequences that meet the usability criteria, and record the missing items and the reasons for the missing items.

[0035] Furthermore, based on the correspondence between microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics, the influence of temperature changes experienced by the tested fresh pork was determined, including:

[0036] It calls upon the microbial spoilage characteristic group, lipid oxidation characteristic group, sample color and moisture state characteristic group, environmental state records during the detection period, sample surface temperature records, sample type information, and isothermal reference data and temperature fluctuation reference data in the state reference library;

[0037] The three types of features are respectively located in the freshness range. Based on the mutual deviation relationship between the three types of features, a temperature change influence label is generated by comparing the grade difference, matching the reference trajectory, and combining the deviation direction. The temperature change influence label is one of the following values: low influence, medium influence, high influence, isothermal proximity, short-term temperature rise, repeated fluctuations, and suspected chain breakage.

[0038] Furthermore, the characteristics of microbial spoilage and lipid oxidation are modified, including:

[0039] Process according to the rule of prioritizing feature consistency and higher risk;

[0040] When the characteristics of microbial spoilage or lipid oxidation and the characteristics of sample color and moisture state point to the same freshness range, the consistent characteristics are used as the matching criteria.

[0041] When the gas response feature or fat oxidation feature deviates from other features, the matching weight of the corresponding feature is adjusted according to the temperature change influence indicator, the direction of deviation, and the current temperature status indicator.

[0042] When the three types of features are completely inconsistent or there are missing features, an anomaly judgment mark, a low confidence mark, or a re-inspection prompt mark is generated accordingly, and a corrected data record is generated.

[0043] Furthermore, the corrected microbial spoilage characteristics, corrected lipid oxidation characteristics, and sample color and moisture state characteristics are matched with a preset state reference library, including:

[0044] Select the status reference library index based on sample type, testing scenario, temperature change impact indicator, and anomaly judgment indicator;

[0045] If sample location information exists, retrieve the corresponding location reference data;

[0046] If no sample location information is available, general reference data will be retrieved based on the color and moisture status reflected in the spectrum and the sample appearance information.

[0047] The matching priorities for fat oxidation characteristics, gas response characteristics, and sample color and moisture state characteristics are set according to the sample location. Then, hierarchical boundary matching, similar sample matching, or trajectory position matching are used to compare the three types of features with reference boundaries, combinations of known sample features, or state trajectories.

[0048] Furthermore, the freshness grade of the tested pork is output, including:

[0049] The grade is determined based on the matching consistency of the three types of features, the influence of temperature changes, the priority of sample location, and the feature correction results;

[0050] The freshness grade is one of the following: Fresh, Nearly Fresh, Borderline, and Unacceptable.

[0051] If an anomaly detection flag exists, output the anomaly detection information;

[0052] A verification prompt message is generated for critical grade samples, and a stop-process prompt message is generated for non-conforming grade samples.

[0053] Generate a judgment result record containing sample identifier, status reference library version number, matching method identifier, reliability identifier, and output time, and update the status reference library version or revert to the previous version based on the review results.

[0054] Compared with the prior art, the present invention provides a method for rapidly determining the freshness of fresh pork, which has the following beneficial effects:

[0055] 1. This invention pre-balances the temperature and humidity of fresh pork under test in a detection chamber, and simultaneously collects gas response signals, fat oxidation color development signals, visible and near-infrared spectral signals, chamber environment signals, and sample surface temperature signals. After correcting, compensating, and removing anomalies from each of the collected signals, microbial spoilage characteristics, fat oxidation characteristics, and sample color and moisture state characteristics are obtained respectively. Then, based on the deviation relationship between the three types of characteristics, the influence of temperature changes such as short-term temperature rise, repeated fluctuations, or suspected chain breakage is identified. After correcting the microbial spoilage characteristics and fat oxidation characteristics, they are matched with a state reference library. This can adapt to the detection scenario of temperature fluctuation and asynchronous microbial spoilage and fat oxidation in real cold chain circulation, reduce the risk of misjudging freshness level by relying solely on volatile alkaline gases or color changes, and improve the accuracy of critical state sample determination and the reliability of on-site rapid test results.

[0056] 2. This invention establishes corresponding indexes in the state reference library according to sample location, detection scenario, and reference sample characteristics. During level matching, it calls the corresponding reference data or general reference data based on the sample type. Simultaneously, it sets the matching priority of lipid oxidation characteristics, gas response characteristics, and sample color and moisture state characteristics according to the sample location. After the judgment is completed, a judgment result record is generated, which includes the sample identifier, state reference library version number, matching method identifier, reliability identifier, and output time. The state reference library is updated or rolled back according to the review results. This allows different segmentation locations, different detection scenarios, and different batches of samples to use matching reference ranges, improving the traceability of the judgment process, the controllability of state reference library updates, and the scenario adaptability of the on-site rapid testing equipment during long-term use. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the process for a rapid method for determining the freshness of fresh pork according to the present invention.

[0058] Figure 2 This is a flowchart of the sample loading pre-balancing and synchronous acquisition process of the present invention;

[0059] Figure 3 This is a schematic diagram of the multi-channel signal correction and feature extraction of the present invention;

[0060] Figure 4 This is a schematic diagram illustrating the influence of temperature changes on the three types of feature deviations in this invention.

[0061] Figure 5 This is a schematic diagram of the feature correction and state reference library matching output of the present invention. Detailed Implementation

[0062] 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.

[0063] Example 1: Figure 1 - Figure 5 A rapid method for determining the freshness of pork is presented, including:

[0064] The fresh pork to be tested was placed in the testing chamber, and the temperature and humidity of the testing chamber were pre-balanced. At the same time reference, gas response signals, fat oxidation color development signals, visible and near-infrared spectral signals, chamber environment signals and sample surface temperature signals were collected.

[0065] Baseline correction, temperature compensation, channel response correction, and outlier sampling removal are performed on the acquired signals.

[0066] Based on the corrected signals, microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics were extracted respectively.

[0067] Based on the correspondence between microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics, the influence of temperature changes experienced by the fresh pork under test was determined, and the microbial spoilage characteristics and lipid oxidation characteristics were corrected.

[0068] The corrected microbial spoilage characteristics, corrected lipid oxidation characteristics, and sample color and moisture state characteristics are matched with a preset state reference library to output the freshness grade of the fresh pork to be tested.

[0069] This method can be applied to various scenarios, including factory sampling inspection after slaughter and processing, sampling inspection during cold chain transportation, rapid on-site inspection at wholesale markets, rapid inspection at supermarket fresh food counters, rapid inspection of e-commerce fresh food warehousing and sorting, and regulatory sampling inspection. This method can be executed by electronic devices with data processing capabilities, signal acquisition capabilities, and information interaction capabilities. The electronic devices can be portable rapid inspection terminals, edge computing controllers, vehicle-mounted cold chain inspection terminals, industrial controllers, or quality inspection desktop terminals.

[0070] The electronic device includes at least a detection chamber, a gas response acquisition unit, a fat oxidation colorimetric acquisition unit, a visible and near-infrared spectroscopy acquisition unit, an environmental status acquisition unit, a sample surface temperature acquisition unit, a data processing unit, a storage unit, and a result output unit. The detection chamber contains the fresh pork to be tested and creates a relatively stable detection environment. The gas response acquisition unit acquires the response signals caused by volatile substances in the fresh pork. The fat oxidation colorimetric acquisition unit acquires the color changes of the colorimetric film or colorimetric area. The visible and near-infrared spectroscopy acquisition unit acquires the reflectance spectrum of the surface of the fresh pork. The environmental status acquisition unit acquires the detection chamber temperature, humidity, and air pressure. The sample surface temperature acquisition unit acquires the current surface temperature of the fresh pork. The data processing unit performs signal correction, feature extraction, temperature change influence judgment, feature correction, and grade matching. The storage unit stores reference sample data, grade boundaries, detection records, and a status reference library. The result output unit outputs the freshness grade, anomaly determination results, and re-inspection prompts.

[0071] The test subject is fresh pork. The input information collected by the electronic device includes at least gas response signals, fat oxidation color development signals, visible and near-infrared spectral signals, detection chamber environmental signals, and sample surface temperature signals. The detection chamber environmental signals include at least the detection chamber temperature, detection chamber humidity, and detection chamber air pressure. The electronic device identifies the influence of temperature changes based on the correspondence between microbial spoilage, fat oxidation, and sample color and moisture status, and makes an on-site determination of the freshness level of the pork based on the identification results. After the determination is completed, the electronic device outputs the freshness level of the pork, and if necessary, simultaneously outputs temperature change risk indicators, abnormality indicators, and re-inspection prompts. The above data, parameters, thresholds, rules, status names, and processing methods are used to illustrate the technical solution of this invention. Those skilled in the art can make adaptive adjustments according to the sample part, detection equipment, detection cycle, and application scenario.

[0072] Before the method is executed, the electronic device can access a pre-established state reference library. This library is built using fresh pork samples with known freshness states as reference samples. The reference samples cover common cuts such as tenderloin, hind leg, pork belly, shoulder, and foreleg, and cover storage or distribution conditions including stable refrigeration, short-term temperature increases, repeated temperature fluctuations, and suspected cold chain disruptions. After collecting gas response signals, fat oxidation colorimetric signals, and visible-near-infrared spectral signals, the freshness state of the reference samples is determined by combining laboratory re-inspection results, existing quality inspection conclusions, or regulatory sampling conclusions. The state reference library includes at least the sample cut and storage conditions. The system includes fields such as sample type, type of testing equipment, range of microbial spoilage characteristics, range of fat oxidation characteristics, range of sample color and moisture state characteristics, grade boundaries, characteristic combination relationships under temperature fluctuation conditions, and abnormal sample identification. The boundaries of each grade are determined by the characteristic distribution of reference samples and the corresponding retest results, and indexes are established according to sample location, type of testing equipment, and storage conditions. By calling this state reference library, the sample to be tested can preferentially use the reference range of the same or similar location and similar testing conditions during subsequent matching, reducing the impact of differences in fat content, moisture state, and surface structure of different locations on the judgment results.

[0073] Specifically, such as Figure 2As shown: After the electronic equipment completes its power-on self-test and enters the test-ready state, it initiates sample loading, pre-balancing of the detection chamber, and synchronous acquisition of multiple signals. The power-on self-test includes at least the following: detection chamber sealing check, light source stability check, gas response channel zero-point check, colorimetric acquisition area check, and spectral background check. The fresh pork to be tested can be taken from the tenderloin, hind leg, belly, shoulder, or foreleg after slaughter and cutting. The sample form can be bare meat pieces, tray-packaged meat pieces, or representative sampled meat pieces after opening. Before the sample enters the detection chamber, obvious blood, frost, and condensation on the outer surface of the packaging can be removed. When there is an obvious water film on the sample surface, it can be left to stand for 30 to 120 seconds before being placed in the detection chamber. This time range is determined according to the on-site rapid testing cycle and the time required for the water film on the sample surface to naturally disappear. If the time is too short, the water film may easily interfere with the spectral reflection signal and colorimetric image acquisition. If the time is too long, the sample surface temperature and evaporation state may change, which is not conducive to reflecting the current state of the sample.

[0074] After the sample is placed in the detection chamber, the chamber is closed and a temperature and humidity pre-equilibration process is initiated. The internal temperature of the detection chamber can be selected from 20°C to 24°C, the relative humidity from 60% to 70%, and the pre-equilibration time from 1 minute to 3 minutes. The above temperature range refers to the detection conditions, not storage conditions. This range is set to balance gas release rate, colorimetric response stability, and to prevent further changes in the sample state during detection. When the temperature is below 20°C, the release of volatile substances in the sample is slower, which can easily prolong the detection time. When the temperature is above 24°C, the sample may undergo new state changes during detection. The above humidity range is based on the anti-condensation properties of the sensing surface. The requirements for clarity of the optical window and normal response of the color development area are determined. When the humidity is too low, the stability of the color development response may decrease. When the humidity is too high, the gas response channel and optical acquisition window are easily affected by water vapor. The pre-equilibration time is controlled between 1 and 3 minutes to balance the stability of the chamber environment and the efficiency of on-site rapid testing. If the test chamber does not reach the target range within 3 minutes, the pre-equilibration can be extended once, with the extension time not exceeding 2 minutes, to eliminate short-term environmental disturbances and avoid the sample staying in the test chamber for too long and changing the current quality status. If the target range is still not reached after the extension, the electronic equipment generates an environmental abnormality mark, suspends the current test, or enters the downgrade judgment process.

[0075] After pre-balancing, the electronic equipment triggers multi-channel signal acquisition at the same time reference; the gas response signal is acquired by the gas response acquisition unit to reflect the changes in volatile substances during the spoilage of fresh pork; the gas response acquisition unit can use no less than 4 detection channels, preferably 6 to 8 detection channels; the number of channels is determined according to the coverage requirements of common spoilage volatiles, equipment size, and power consumption limitations of portable equipment; fewer than 4 channels are not conducive to distinguishing different volatile response sources, and more than 8 channels have limited improvement on the accuracy of on-site rapid detection and will increase equipment cost and power consumption; the sampling interval of the gas response signal The sampling interval can be selected from 0.5 seconds to 2 seconds. This range is determined based on the time required for gas to diffuse into the detection channel and form a recognizable response. Too short a sampling interval will increase the amount of data and power consumption, while too long a sampling interval will reduce the continuity of the response curve. The acquisition duration can be selected from 3 minutes to 10 minutes. Specifically, 3 minutes to 5 minutes can be selected for offline sampling inspections at slaughterhouses, while 6 minutes to 10 minutes can be selected for regulatory sampling inspections or pre-screening before dispute review. This time range is determined based on the detection pattern that it usually takes several minutes for the gas response to rise to a stable state. Too short an acquisition time will result in insufficient response, while too long an acquisition time will affect the efficiency of rapid detection.

[0076] The colorimetric signal from fat oxidation is acquired by the colorimetric acquisition unit, used to record the color change of the colorimetric film or area over time. The colorimetric image acquisition interval can be selected from 5 to 30 seconds. This range is determined based on the characteristic that color changes typically develop gradually over seconds to minutes, balancing the continuity of color changes with data volume control. The visible and near-infrared spectral signal is acquired by the spectral acquisition unit, covering a detection band from 400 nm to 1100 nm. This range is commonly used in portable visible and near-infrared detection, capable of simultaneously covering the surface color, moisture state, and composition of fresh pork. Textile scattering information; spectral acquisition can focus on collecting reflection signals near 540 nm, 580 nm, 630 nm, 730 nm, 760 nm and 860 nm. The above bands are selected based on the commonly used detection characteristics of fresh pork color changes, moisture state changes and muscle fiber scattering changes in the visible and near-infrared range; the environmental signals inside the chamber include at least the fields of chamber temperature, humidity and air pressure, and the sample surface temperature signal is collected by a non-contact temperature detection unit. Both are used to record the environmental state and the current temperature state of the sample during the detection process, and are not directly equivalent to the historical temperature of the sample;

[0077] Multi-channel acquisition can employ hardware synchronous triggering, software time stamp alignment, or group synchronization. Hardware synchronous triggering uses the same trigger signal to simultaneously initiate gas response acquisition, colorimetric image acquisition, spectral acquisition, and environmental acquisition, suitable for desktop quality inspection equipment or fixed production line equipment. Software time stamp alignment involves each acquisition unit working independently and writing a unified time stamp when generating data; the data processing unit aligns the signals according to the time stamp, suitable for portable rapid testing terminals. Group synchronization treats gas response and environmental signals as a continuous acquisition group, and colorimetric images and spectral signals as an interval acquisition group, aligning them based on the start time of each acquisition cycle, suitable for low-power devices. The synchronization error is preferably no more than 100 milliseconds; in high-precision equipment requiring fine correlation between spectral and colorimetric changes, the synchronization error is preferably no more than 10 milliseconds. These error ranges are determined based on the fact that gas response, colorimetric response, and spectral state changes typically occur on the order of seconds to minutes. An error within 100 milliseconds does not affect the overall timing judgment of on-site rapid testing; an error within 10 milliseconds meets the requirements of high-precision equipment for consistency in multi-channel acquisition time.

[0078] After acquisition, the electronic device generates multiple acquisition data records. These records include at least the following fields: acquisition time identifier, signal type, acquisition channel number, original signal value, detection chamber temperature, detection chamber humidity, detection chamber air pressure, sample surface temperature, acquisition method identifier, and abnormal status identifier. Among these, gas response data is recorded in time series format, colorimetric image data is recorded in image series format, spectral data is recorded in reflectance spectrum series format, and environmental state data and sample surface temperature data are associated and saved according to the same time reference. These data records serve as the data source for subsequent signal correction, feature extraction, and freshness determination.

[0079] Specifically, such as Figure 3 As shown: The electronic device takes multiple raw data streams as input and preprocesses the gas response signal, aliphatic oxidation colorimetric signal, visible and near-infrared spectral signal, and environmental records. Before preprocessing, at least two of the gas response signal, colorimetric signal, and spectral signal should be valid. The input data includes at least the acquisition time identifier, multiple raw signals, empty chamber reference data, standard color chart data, reference plate data, detection channel calibration data, detection chamber temperature, detection chamber humidity, detection chamber pressure, and sample surface temperature. The above data is used to reduce the impact of zero-point drift, environmental fluctuations, differences in response of different detection channels, and transient interference on subsequent feature extraction, so that the signals are comparable.

[0080] Baseline correction is performed separately for different types of signals. For gas response signals, the baseline can be selected as the stable average value of each gas response channel in an empty chamber state within 30 to 60 seconds, or as the data in the first 10 to 30 seconds after the sample enters the chamber before a significant response is formed. The range of 30 to 60 seconds is determined based on the time required for the gas response channel to form a stable zero point in an empty chamber state. The range of the first 10 to 30 seconds is determined based on the detection law that volatile substances have not fully diffused when the sample is first entered into the chamber. The empty chamber baseline is suitable for scenarios where the equipment is stable and the detection cycle is fixed. The field baseline is suitable for scenarios with large environmental changes during transportation, wholesale sampling, or retail rapid testing. For aliphatic oxidation color development signals, the baseline can be selected as the initial color data of the color development membrane before contact with the volatile substances in the sample, or as the reference color data of unreacted membranes in the same batch. For visible and near-infrared spectral signals, the baseline can be selected as the white reference plate reflection signal and the dark background signal to subtract the influence of light source intensity changes and dark current. After baseline correction is completed, the electronic equipment retains the original time sequence and records the baseline source identifier for easy subsequent anomaly tracing.

[0081] Temperature compensation is based on the detection chamber temperature, detection chamber humidity, and sample surface temperature as inputs. For gas response signals, compensation is initiated when the detection chamber temperature deviates from the target temperature by 1 to 2 degrees Celsius. When the deviation is less than 1 degree Celsius, only the environmental conditions are recorded, and the gas response signal remains unchanged. This range is determined based on the common temperature control accuracy of portable rapid testing equipment and small detection chambers used in the field. Compensation is less meaningful when the temperature deviation is too small, and it will have a significant impact on the release of volatile substances and the response of the detection channel when the temperature deviation exceeds 2 degrees Celsius. For the fat oxidation colorimetric image signal, if the light source intensity fluctuates by more than 5% within 5 minutes after preheating, the reference color data is reacquired. The 5% range is based on common LED light sources. Short-term operational stability is determined; beyond this range, color acquisition is easily affected by changes in illumination. If the humidity in the detection chamber exceeds 75% and persists for more than 30 seconds, the color development signal is marked as having a risk of moisture interference. This condition is determined based on the fact that moisture easily adheres to the optical window and color development area in high humidity environments; a duration of more than 30 seconds indicates that the interference is not an instantaneous fluctuation. For visible and near-infrared spectral signals, when the change in the reflectance intensity of the reference plate exceeds 3% to 5%, the spectral baseline is recalculated using the most recent reference plate data. This range is determined based on the repeatability fluctuations of portable spectral modules and fixed light sources in short-term on-site detection; fluctuations below this range can be considered normal equipment fluctuations, while fluctuations exceeding this range require recalibration.

[0082] Channel response correction is used to reduce response differences between different detection channels. For gas response channels, the response correction coefficient can be set based on the standard reference response obtained during the power-on self-test phase. For colorimetric channels, color correction can be performed based on the difference between the measured color and the calibrated color of the standard color chart. For spectral channels, intensity correction can be performed based on the white reference plate and dark background data. This process does not change the true state of the sample, but rather allows different channels to participate in subsequent judgments under the same comparison standard. If the correction coefficient of a certain channel is lower than 70% of the historical average or higher than 130% of the historical average, it is determined that the channel has response attenuation or abnormal enhancement, and the participation weight of the channel is reduced in subsequent processing, and the channel is removed if necessary. The range of 70% to 130% is used as the warning range for abnormal channel status, which is determined based on the response drift, on-site assembly errors, and fluctuations in repeated detection of the standard reference material during long-term use of the sensor. When it is lower than 70% or higher than 130%, the channel output has significantly deviated from the normal operating state, and continued equal weighting may easily introduce misjudgment.

[0083] Anomaly removal targets transient spikes, missing data, acquisition saturation, and time misalignment. For continuous time series, a sliding window approach can be used, with a window length of 5 to 15 sampling points. This range is determined based on the 0.5-2 second sampling interval of the gas response signal, covering short-term changes from several seconds to tens of seconds, identifying transient anomalies without mistakenly deleting normal upward trends. When a single sampling point experiences a sudden increase or decrease relative to adjacent data within the window, and this change does not continue within the next 2 to 3 sampling points, it can be identified as a transient spike and replaced with adjacent valid data. For color images... It can remove image frames that are obviously overexposed, underexposed, have color areas blocked, or have severe reflections. When the number of consecutive invalid frames exceeds 20% of the total number of frames, the color signal is marked as low confidence, and this mark is passed to subsequent processing along with the color features. The 20% range is determined based on the requirement that the color image sequence needs to retain enough consecutive frames to reflect the color change trend. When this proportion is exceeded, the color change trajectory is no longer stable. For spectral signals, if the reflection intensity is close to the upper limit of saturation or lower than the lower limit of effectiveness for a long time, one resampling is triggered. If it is still abnormal after resampling, the gas response signal and color signal are retained, and the spectral state features are marked as missing.

[0084] After preprocessing, the electronic device generates a calibration data record. The calibration data record includes at least the following fields: acquisition time identifier, signal type, channel number, baseline source identifier, temperature compensation identifier, channel response correction coefficient, abnormal sampling identifier, calibrated gas response sequence, calibrated colorimetric image sequence, calibrated spectral reflectance sequence, and valid environmental record. The calibration data record serves as the data source for subsequent extraction of microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics.

[0085] Specifically, such as Figure 3 As shown: The electronic device takes the corrected gas response sequence, colorimetric image sequence, and spectral reflectance sequence as inputs to extract microbial spoilage characteristics, fat oxidation characteristics, and sample color and moisture state characteristics for determining the freshness of fresh pork, respectively. Before feature extraction, the gas response sequence should have a continuous rising segment or a stable changing segment, the colorimetric image sequence should have an identifiable effective colorimetric region, and the spectral reflectance sequence should have a usable reference correction result. If only two types of signals are valid among the gas response signal, colorimetric signal, and spectral signal, the features corresponding to the valid signals are extracted first, and the missing items and reasons for the missing items are recorded for subsequent downgrade processing or re-examination prompts.

[0086] Microbial spoilage characteristics are extracted from the corrected gas response sequence. The electronic device aligns the signals according to the time signature of each gas response channel and selects an effective response segment from the start of acquisition to the stabilization of the response. The effective response segment can be a fixed-duration response segment or a stable response segment. When using a fixed-duration response segment, data from 3 to 5 minutes after the start of acquisition can be collected. This range is suitable for scenarios with high requirements for detection cycle, such as offline sampling inspections at slaughterhouses. Its setting is determined based on the detection pattern that it usually takes several minutes for the gas response to rise from its initial state to form a identifiable change. When using a stable response segment, when the response change amplitude within 30 consecutive seconds is less than 5% to 10% of the previous maximum change amplitude, it can be determined that the gas response has entered a stable stage. This method is suitable for regulatory sampling inspections or pre-screening scenarios before benchtop verification. The 30-second continuous observation time is used to exclude single-point fluctuations and short-term disturbances, and the 5% to 10% change amplitude range is used to distinguish between a normal stabilizing state and a state that is still in a rapid change state.

[0087] After determining the effective response segment, the electronic device extracts the response amplitude, response rise rate, time required to reach stability, stability segment fluctuation amplitude, and multi-channel response sequence from each gas response channel. The response amplitude reflects the degree of volatile substance accumulation, the response rise rate reflects the volatile substance release rate, the time required to reach stability reflects the diffusion state of volatile substances in the sample, the stability segment fluctuation amplitude identifies the airtightness of the detection chamber or the instability of the sample state, and the multi-channel response sequence distinguishes the influence of different volatile substance sources on the detection results. Based on the above data, the electronic device forms a microbial spoilage characteristic group, which includes at least the following fields: channel number, response amplitude, response rise rate, stabilization time, stability segment fluctuation amplitude, channel response sequence, and channel valid identifier.

[0088] Fat oxidation features are extracted from the corrected colorimetric image sequence. The electronic device first determines the position of the colorimetric membrane or colorimetric region. The position of the colorimetric region can be determined by fixed area reading, edge positioning, or color segmentation. Fixed area reading is suitable for standardized detection chambers where the colorimetric region is installed in a fixed position. Edge positioning is suitable for portable terminals where the colorimetric region has a slight offset. Color segmentation is suitable for detection accessories where the colorimetric region and the background color are significantly different. After determining the effective colorimetric region, the electronic device extracts the color change amplitude, color change direction, color change rate, and post-reaction color stability of the colorimetric region. The color change amplitude is used to reflect the influence of fat oxidation-related substances on the colorimetric region. The color change direction is used to exclude abnormal color changes caused by non-target interference. The color change rate is used to identify the release state of related substances. The post-reaction color stability is used to determine whether the colorimetric response has reached a definite state.

[0089] When local contamination, wrinkles, occlusion, or strong reflections occur in the color rendering area, only image frames with an unoccluded area greater than 70% of the effective area are retained. If the area is less than this proportion, the color features of the frame are not extracted, and the low-confidence marker formed in the previous processing is used. The 70% proportion is determined based on the statistical stability of the image area. When the effective area is too small, the color mean is easily dominated by local defects and cannot represent the overall color rendering change. The electronic device forms a fat oxidation feature group based on the effective image frames. The fat oxidation feature group includes at least the following fields: image time marker, color rendering area location, effective area ratio, color change amplitude, color change direction, color change speed, color stability after reaction, and low-confidence marker.

[0090] The color and moisture state characteristics of the sample were extracted from the corrected visible and near-infrared spectral reflectance sequence. The electronic device divided the spectrum into color-related bands and moisture scattering-related bands according to their intended use. The color-related bands primarily utilize reflectance signals near 540 nm, 580 nm, and 630 nm to describe the surface color state of fresh pork and changes such as darkening, graying, and browning. The moisture scattering-related bands primarily utilize reflectance signals near 730 nm, 760 nm, and 860 nm to describe the surface moisture state and muscle fiber scattering changes. These bands were selected based on commonly used detection characteristics of fresh pork color, moisture, and tissue scattering information within the visible and near-infrared range, and can be completed using conventional visible and near-infrared spectral acquisition equipment.

[0091] Spectral features can be obtained through band ratio extraction, band difference extraction, or local trend extraction. Band ratio extraction obtains state features by the relative changes in reflection intensity of different bands, which is suitable for reducing the influence of overall light intensity fluctuations. Band difference extraction highlights local absorption changes by the difference between the target band and the adjacent background bands, which is suitable for identifying samples with obvious color changes. Local trend extraction judges the changes in surface moisture and tissue state of the sample by the trend of reflection intensity changes in adjacent bands, which is suitable for scenarios where the sample surface state is uneven. The electronic device forms a sample color and moisture state feature group based on the above results. The sample color and moisture state feature group includes at least the following fields: spectral acquisition time identifier, target band reflection intensity, band ratio, band difference, local change trend, reference correction identifier, and spectral validity identifier.

[0092] After feature extraction is completed, the electronic device generates feature data records. The feature data records include at least the following fields: sample identifier, collection time identifier, signal source identifier, microbial spoilage feature group, lipid oxidation feature group, sample color and moisture state feature group, validity identifier, anomaly identifier, and missing item identifier. To facilitate subsequent comparison, the electronic device can convert each feature group into a unified comparison level or feature position, such as locating intervals according to the reference range of fresh samples, the reference range of slightly fresh samples, the reference range of critical samples, and the reference range of unqualified samples. If a certain type of feature is significantly inconsistent with the other two types of features, the feature is not directly deleted. Instead, the inconsistency identifier is retained along with the feature data record for subsequent judgment based on the influence of temperature changes, avoiding premature removal of abnormal information with discriminative value.

[0093] Specifically, such as Figure 4 and Figure 5 As shown: The electronic device takes as input the microbial spoilage feature group, lipid oxidation feature group, sample color and moisture state feature group, environmental state record during detection, sample surface temperature record, sample type information, and isothermal reference data and temperature fluctuation reference data stored in the state reference library. It identifies the impact of temperature changes on the spoilage process of fresh pork and corrects the microbial spoilage feature and lipid oxidation feature accordingly. This process does not require the restoration of the complete historical temperature curve of the sample, but rather determines whether the fresh pork to be tested is close to a stable cold chain state, whether it has experienced a short-term temperature rise, whether there are repeated temperature fluctuations, or whether there is a suspected cold chain break.

[0094] The impact of temperature changes is determined based on the deviations between microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics. Under stable cold chain conditions, these three characteristics typically change in a relatively consistent direction: the gas response gradually increases, the color change gradually increases, and the states related to color and moisture in the spectrum gradually shift towards quality decline. When the fresh pork being tested experiences a short-term temperature increase or repeated temperature fluctuations, the degree of change in these three characteristics may be inconsistent. For example, a significant increase in microbial spoilage characteristics while relatively small changes in lipid oxidation characteristics usually corresponds to a rapid increase in volatile substances after short-term exposure to higher temperatures; a significantly higher lipid oxidation characteristic while the gas response remains at a low level usually corresponds to the gradual accumulation of lipid oxidation after long-term storage at low temperatures or temperature fluctuations; a significant deviation in sample color and moisture state while microbial spoilage and lipid oxidation characteristics are not completely synchronized usually indicates that the sample may have undergone freezing, thawing, surface dehydration, or short-term cold chain disruption. The electronic device generates a temperature change impact indicator based on the above deviations. The temperature change impact indicator can be selected from three levels: low impact, medium impact, and high impact, and can also be further classified into states such as near-isothermal conditions, short-term temperature increases, repeated fluctuations, and suspected cold chain disruption.

[0095] The impact of temperature changes can be determined through grade difference comparison, reference trajectory matching, or channel deviation direction assessment. When using grade difference comparison, the electronic device first positions the three types of features near the four grades of fresh, slightly fresh, critical, and unqualified, and then compares the grade differences between the three types of features. When the three types of features are at the same grade, or there is only a difference of no more than one grade, the impact of temperature changes is considered low. When any spoilage feature is more than two grades higher than the other features, or when the three types of features point to different grades, the impact of temperature changes is considered high. Other situations are considered to have a medium impact. A difference of one grade may usually be caused by differences in sample location, sampling position, or detection fluctuations; differences of two grades or more are usually... The fluctuations exceeding the normal range of on-site rapid testing can therefore be used as a criterion for judging the enhanced influence of temperature changes. When using reference trajectory matching, the electronic device compares the three types of features with the isothermal reference trajectory and temperature fluctuation reference trajectory in the state reference library. If the feature combination of the sample to be tested is closer to the temperature fluctuation reference trajectory, a temperature fluctuation identifier is generated. When using channel deviation direction judgment, the electronic device determines the type of temperature change based on the combination relationship between high microbial putrefaction features, high lipid oxidation features, and deviations in sample color and moisture state, and uses the deviation direction as the priority source for subsequent feature correction. The above methods can be used individually or in combination depending on the processing capacity of the detection equipment.

[0096] When correcting for microbial putrefaction and lipid oxidation characteristics, the electronic device follows a rule of prioritizing consistency and risk. When one putrefaction characteristic points to the same level as the sample's color and moisture state characteristics, while another putrefaction characteristic deviates from that level, the two consistent characteristics are prioritized as the level matching basis. When the impact of temperature change is low, the microbial putrefaction and lipid oxidation characteristics maintain their original positions. When the impact of temperature change is moderate, putrefaction characteristics with large deviations are slightly corrected to bring them back to the range consistent with the other two characteristics. When the impact of temperature change is high and the gas response is significantly higher than other characteristics, it is determined that the increase in temperature has a negative impact on the gas response. In terms of amplification, the individual influence of gas response characteristics is reduced, and the level is matched in combination with the sample color and moisture state characteristics; when the lipid oxidation characteristic is more than 1 level higher than the microbial putrefaction characteristic, the higher risk indication of the lipid oxidation characteristic is retained; when the microbial putrefaction characteristic is more than 2 levels higher than the lipid oxidation characteristic and the influence of temperature change is high, the individual influence of the microbial putrefaction characteristic is reduced, and the level is matched in combination with the sample color and moisture state characteristics; the reason for the above 1 and 2 levels is that in on-site rapid testing, the difference of a single level may come from individual sample differences, sampling location differences or channel fluctuations, while differences of more than 2 levels usually need to be conservatively handled in combination with temperature changes or abnormal scenarios.

[0097] When the three types of features are completely inconsistent and the difference reaches more than two levels, the electronic device sets the temperature change impact indicator to high and generates an anomaly judgment indicator. If the difference between the sample surface temperature and the ambient temperature inside the detection chamber exceeds 5 degrees Celsius and lasts for more than 1 minute, it indicates that the sample may have just left the refrigerated environment or just experienced external temperature exposure. The electronic device can extend the pre-equilibrium period or add a current temperature status indicator during feature correction. The 5-degree Celsius setting is determined based on the common temperature difference between refrigerated samples and the on-site detection environment. When the difference is small, it has a limited impact on the release of volatile substances and spectral reflectance. When the difference exceeds 5 degrees Celsius and lasts for more than 1 minute, the sample surface condition is usually sufficient to affect the on-site detection signal. If a certain type of feature is missing, the electronic device does not generate a substitute result for that type of feature, but instead performs conservative correction based on the remaining two types of features and retains a low confidence indicator or a re-examination prompt.

[0098] After feature correction is completed, the electronic device generates a correction data record. The correction data record includes at least the following fields: sample identifier, sample type, level of microbial spoilage feature before correction, level of microbial spoilage feature after correction, level of lipid oxidation feature before correction, level of lipid oxidation feature after correction, sample color and moisture status, temperature change influence indicator, current temperature status indicator, anomaly judgment indicator, low confidence indicator, and re-inspection prompt indicator. The above correction data record serves as the data source for subsequent status reference library matching and freshness level output.

[0099] Specifically, such as Figure 5As shown: The electronic device takes as input the corrected microbial spoilage characteristics, corrected fat oxidation characteristics, sample color and moisture state characteristics, temperature change influence indicators, anomaly judgment indicators, and sample type information generated by previous processing. It matches these characteristics with a state reference library to determine the freshness grade of the pork to be tested. Before matching, the electronic device selects a state reference library index based on the sample type and testing scenario. When the operator inputs the sample part, the reference data for that part is prioritized. When no sample part is input, general reference data can be retrieved based on spectral color, moisture state, and sample appearance information. For parts with high fat content, such as pork belly, the priority of fat oxidation characteristics in matching is increased. For parts mainly composed of lean meat, such as tenderloin, the consistency requirement between sample color and moisture state characteristics and gas response characteristics is increased. The above priority rules are determined based on the differences in fat content, moisture state, and surface tissue structure of different cuts, which can avoid judgment bias caused by using the same reference boundary to process samples from different parts.

[0100] The state reference library matching can employ hierarchical boundary matching, similar sample matching, or trajectory position matching. When using hierarchical boundary matching, the electronic device compares the microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics with the fresh, slightly fresh, critical, and unqualified reference boundaries in the state reference library, respectively, to obtain their corresponding freshness levels. This is then combined with sample location priority, temperature change influence indicators, and feature correction results to output a comprehensive level. When using similar sample matching, the electronic device compares the feature combinations of the sample to be tested with the feature combinations of known samples in the state reference library, selecting the closest set of one or more reference samples. The freshness level of the reference sample is used as the basis for judgment. When using trajectory position matching, the electronic device locates the characteristic combination of the sample to be tested on the microbial putrefaction trajectory, lipid oxidation trajectory, and spectral state trajectory, compares whether the positions of the three trajectories are consistent, and determines the freshness level and credibility label based on the degree of consistency. The above matching methods can be selected and used according to the computing power of the equipment, the capacity of the state reference library, and the on-site testing cycle. Among them, graded boundary matching is suitable for rapid testing terminals with limited computing resources, similar sample matching is suitable for wholesale market quality inspection or regulatory sampling equipment with a rich sample library, and trajectory position matching is suitable for scenarios that require special attention to samples with temperature changes.

[0101] When outputting the freshness level, if the matching results of the three types of features are consistent, the electronic device directly outputs the corresponding freshness level; if the difference between the three types of features is no more than one level, the electronic device outputs a comprehensive result based on the influence of temperature change, the priority of sample location, and the feature correction result; when the influence of temperature change is high, the output level is adjusted to the side with higher risk; a difference of one level may usually be caused by differences in sample location, sampling position, or fluctuations in the detection channel, and a stable result can still be obtained through comprehensive judgment of multiple features; for samples that have generated an anomaly judgment mark, the electronic device does not force the output of the conventional level, but outputs the anomaly judgment result and prompts for resampling, manual review, or laboratory retesting;

[0102] Freshness grades can be divided into fresh, near-fresh, critical, and unqualified grades. Fresh grade corresponds to multiple characteristics falling within the fresh reference range of the status reference library. Near-fresh grade corresponds to at least one characteristic deviating from the fresh reference range but not reaching the critical boundary. Critical grade corresponds to microbial spoilage or lipid oxidation characteristics approaching the safety boundary, requiring careful handling. Unqualified grade corresponds to at least one key characteristic exceeding the unqualified reference boundary in the status reference library, or multiple characteristics pointing to significant quality degradation. For critical grade samples, the electronic device can simultaneously output a verification prompt. For unqualified grade samples, the electronic device can output a stop-processing prompt or a re-inspection prompt. The above grade boundaries are determined by the characteristic distribution of reference samples in the status reference library and their re-inspection conclusions. Corresponding reference boundaries can be established for different locations, different testing equipment, and different storage conditions.

[0103] To ensure the long-term applicability of the status reference library, the electronic device can record the judgment results, feature combinations, temperature change impact indicators, and retest results after testing. When subsequent laboratory test results or regulatory review results are obtained, the confirmed samples can be written into the status reference library update buffer. Before writing, the electronic device performs a consistency check on the new sample. If the new sample deviates significantly from the reference range of similar samples and there are no explanations of temperature anomalies, sample location differences, or testing equipment differences, it will not be directly written into the formal status reference library. If the new sample can supplement new sample locations, temperature fluctuation scenarios, or seasonal batches, incremental updates can be completed when the device is idle. Each update generates a version identifier and retains the previous version for rollback. If the judgment difference ratio of historical verification samples after the update exceeds 5% to 10%, it will roll back to the previous version. The 5% to 10% range is determined based on the balance between the long-term consistency of the rapid testing equipment and the sample expansion capability. If the ratio is too low, the status reference library will have difficulty absorbing new batches of samples, and if the ratio is too high, it will affect the consistency of existing judgment results.

[0104] After matching and grade output are completed, the electronic device generates a judgment result record. The judgment result record includes at least the following fields: sample identifier, sample type, status reference library version number, microbial spoilage characteristic grade, lipid oxidation characteristic grade, sample color and moisture status grade, temperature change influence indicator, matching method identifier, freshness grade, reliability indicator, anomaly judgment indicator, review prompt indicator, and output time. The above record can be used for on-site display, batch traceability, subsequent re-inspection, and status reference library updates.

[0105] Through the above processing, the fresh pork under test forms a continuous data chain from entry testing to grade output, sequentially completing multi-channel signal synchronous acquisition, signal correction, feature extraction, temperature change impact correction, and state reference library matching. This method can comprehensively judge problems such as temperature fluctuations, insufficient single indicators, and asynchronous microbial spoilage and fat oxidation in real cold chain circulation, so that the detection process has clear data sources, processing rules, anomaly handling, and rollback mechanisms, thereby reducing the risk of misjudgment caused by temperature changes or fat oxidation as the main factors, and improving the stability, synchronicity, and reliability of on-site rapid test results.

[0106] Example 2: Based on Example 1, the specific application process of a method for rapidly determining the freshness of fresh pork is further explained:

[0107] In a specific application scenario, the quality inspection station at a wholesale market conducts on-site rapid testing on a batch of fresh pork transported via cold chain. The sample to be tested is a pork belly sample taken from the transport vehicle upon arrival. The transport document for this sample indicates refrigerated transport, but it is impossible to confirm whether there were any brief openings, loading / unloading delays, or temperature rises during transport. After opening the outer packaging, the quality inspectors select pieces of meat with intact surfaces, no obvious contamination, and no obvious frost coverage as the sample to be tested. The electronic device reads or receives the sample type information. When the sample type is pork belly, it prioritizes calling the reference data corresponding to the pork belly part in the status reference library. When no sample type is entered, it can call general reference data based on the sample appearance image, initial spectral state, and manually selected information. Since the fat ratio in pork belly is relatively high, the electronic device increases the priority of fat oxidation characteristics in subsequent matching and uses gas response characteristics and sample color and moisture state characteristics for joint verification.

[0108] Before testing begins, the electronic equipment performs a self-check on the sealing of the testing chamber, the stability of the light source, the zero point of the gas response channel, the colorimetric acquisition area, and the spectral background. After passing the self-check, quality inspectors remove any obvious blood and packaging condensation from the surface of the sample to be tested. If a significant water film still exists on the sample surface, the sample can be left to stand on the testing platform for 30 to 120 seconds. This time range is determined based on the on-site rapid testing cycle and the time required for the water film on the sample surface to naturally dissipate. If the standing time is too short, the water film may cause abnormal spectral reflection and a brighter colorimetric image. If the standing time is too long, the sample surface temperature and evaporation state may undergo new changes. After completing the surface treatment, the sample to be tested is placed into the testing chamber, the testing chamber is closed, and temperature and humidity are pre-equilibrated.

[0109] The temperature inside the testing chamber can be selected from 20°C to 24°C, the relative humidity can be selected from 60% to 70%, and the pre-equilibration time can be selected from 1 minute to 3 minutes. The above temperature range is for testing conditions and is not used as the storage temperature for fresh pork. The reason for setting it is that this range can take into account the release of volatile substances, the stability of color response, and the prevention of further accelerated changes in the sample during testing. If the testing chamber does not reach the target range within 3 minutes, the electronic equipment can extend the pre-equilibration once, with the extension time not exceeding 2 minutes. If the target range is still not reached after the extension, the electronic equipment generates an environmental anomaly indicator and prompts for retesting or entry into the downgrade judgment process.

[0110] After pre-equilibration, the electronic equipment acquires gas response signals, lipid oxidation color development signals, visible and near-infrared spectral signals, detection chamber temperature, detection chamber humidity, detection chamber pressure, and sample surface temperature at the same time reference. The gas response acquisition unit can use 6 detection channels to cover different response sources of common putrefactive volatiles. The sampling interval of the gas response signal can be selected as 1 second, and the acquisition duration can be selected as 5 minutes. This sampling interval is within the range of 0.5 seconds to 2 seconds, which can take into account both the continuity of the response curve and the control of data volume. The 5-minute acquisition duration is suitable for the on-site testing rhythm of the wholesale market quality inspection station and can also cover the process of gas response from rising to stabilizing. The color development image can be acquired every 10 seconds, with an interval between 5 seconds and 30 seconds, which can record the color change trend of the color development area. The visible and near-infrared spectrum can cover 400 nm to 1100 nm, and focuses on reading the reflection signals near 540 nm, 580 nm, 630 nm, 730 nm, 760 nm, and 860 nm to reflect the sample surface color, moisture state, and tissue scattering state.

[0111] After acquisition, the electronic device generates multiple acquisition data records. These records include at least the following fields: sample identifier, acquisition time identifier, signal type, acquisition channel number, original signal value, detection chamber temperature, detection chamber humidity, detection chamber air pressure, sample surface temperature, acquisition method identifier, and abnormal status identifier. Among these, gas response data is saved in time series format, colorimetric image data is saved in image series format, spectral data is saved in reflectance spectrum series format, and environmental state data and sample surface temperature data are associated and saved according to the same time reference.

[0112] The electronic equipment preprocesses the collected data. For gas response signals, the stable average value within 30 to 60 seconds under empty chamber conditions can be selected as the baseline. If the environment at the wholesale market changes significantly, the data within the first 10 to 30 seconds after the sample enters the chamber, before a significant gas response has formed, can also be selected as the baseline. The above two time ranges are determined based on the zero-point stabilization time of the empty chamber and the detection law that the volatile substances in the sample have not yet fully diffused. For the colorimetric image signal of fat oxidation, the electronic equipment uses the initial color of the colorimetric membrane before it comes into contact with the volatile substances in the sample or the reference color of the unreacted membrane in the same batch as the baseline. For visible and near-infrared spectral signals, the electronic equipment uses the reflected signal from the white reference plate and the dark background signal to subtract the influence of light source intensity changes and dark current.

[0113] Temperature compensation is performed based on the detection chamber temperature, detection chamber humidity, and sample surface temperature. When the detection chamber temperature deviates from the target temperature by 1 to 2 degrees Celsius, compensation is applied to the gas response signal. When the deviation is less than 1 degree Celsius, only the environmental conditions are recorded, and the gas response signal remains unchanged. This range is determined based on the common temperature control accuracy of portable rapid testing equipment and small detection chambers used in the field. If the intensity fluctuation of the colorimetric acquisition light source exceeds 5% within 5 minutes after preheating, the reference color data is reacquired. If the detection chamber humidity exceeds 75% and remains above 30 seconds, the colorimetric signal is marked as having a risk of moisture interference. For visible and near-infrared spectral signals, when the reference plate... When the reflectance intensity changes by more than 3% to 5%, the spectral baseline is recalculated using the most recent reference plate data. If the correction coefficient of a certain detection channel is lower than 70% of the historical average or higher than 130% of the historical average, the electronic device determines that the channel has response attenuation or abnormal enhancement and reduces the participation weight of the channel. For continuous time series, a sliding window of 5 to 15 sampling points can be used to identify transient spikes. For color images, if the number of consecutive invalid frames exceeds 20% of the total number of frames, the color signal is marked as low confidence. After preprocessing, the electronic device generates the corrected gas response sequence, color image sequence, spectral reflectance sequence, and effective environmental record.

[0114] The electronic device extracts microbial spoilage characteristics from the calibrated gas response sequence. For rapid testing in wholesale markets, the effective response segment can be determined using a stable response segment. When the change in gas response amplitude is less than 5% to 10% of the previous maximum change amplitude within 30 consecutive seconds, the gas response is considered to have entered a stable phase. The electronic device extracts the response amplitude, response rise rate, time required to reach stability, stability segment fluctuation amplitude, and the sequence of responses from the effective response segment, forming a microbial spoilage characteristic group. If multiple gas response channels all show increased response amplitude, faster rise rate, and shorter stabilization time, it indicates that the release of volatile substances related to microbial spoilage in the sample is significant.

[0115] Electronic devices extract fat oxidation features from colorimetric image sequences. For pork belly samples, the colorimetric area may be affected by reflections from the fat layer or grease on the sample surface. Therefore, the effective area of ​​the colorimetric membrane is first identified, and contaminated, wrinkled, obstructed, or strongly reflective image frames are removed. When the effective unobstructed area is greater than 70% of the colorimetric area, the image frame is retained, and the color change amplitude, color change direction, color change speed, and color stability after the reaction are extracted. When the effective area is less than this proportion, the color features of the image frame are not extracted, and the low confidence label is used. The 70% proportion is determined based on the statistical stability of the image region. When the effective area is too small, the color mean is easily dominated by local defects and cannot represent the overall color change. If the colorimetric area continues to change towards the fat oxidation reference direction and the color change speed is fast, it indicates a high risk of fat oxidation.

[0116] Electronic devices extract sample color and moisture state characteristics from visible and near-infrared spectral reflectance sequences. For pork belly samples, reflectance signals near 540 nm, 580 nm, and 630 nm are used to describe surface color changes, while reflectance signals near 730 nm, 760 nm, and 860 nm are used to describe moisture state and tissue scattering changes. If the spectrum shows that the sample surface has darkened color, localized graying, or moisture state deviates from the normal range for refrigerated samples, the sample color and moisture state characteristics are located in a higher risk range than the fresh reference range.

[0117] After feature extraction, the electronic device locates the microbial spoilage features, lipid oxidation features, and sample color and moisture state features into reference ranges in the state reference library. For example, in a single test, the gas response feature is close to the near-freshness reference range, the lipid oxidation feature is close to the critical reference range, and the sample color and moisture state features are between the near-freshness reference range and the critical reference range. In this case, the three types of features are not completely consistent, and the risk of the lipid oxidation feature is higher than that of the gas response feature. The electronic device makes a judgment based on the influence of temperature changes. Since the lipid oxidation feature is significantly higher than the gas response has not yet reached the same risk level, this combination usually corresponds to the situation of long-term storage at low temperature or gradual accumulation of lipid oxidation after temperature fluctuations. Therefore, it is not appropriate to judge the sample as fresh simply because the gas response has not reached the critical level.

[0118] Electronic devices can use grade difference comparison and reference trajectory matching to jointly determine the impact of temperature changes. If the difference between the three types of features is no more than one grade, and their combination is close to the stable refrigerated reference trajectory in the state reference library, then the impact of temperature changes can be judged as low. If any putrefaction feature is more than two grades higher than the other features, or the feature combination is closer to the temperature fluctuation reference trajectory, then the impact of temperature changes can be judged as high. In the above detection scenario, the lipid oxidation feature is higher than the gas response feature, but the difference is not more than two grades, and the spectral state also shifts to the higher risk side. The electronic device can judge the impact of temperature changes as medium and make coordinated corrections to the gas response feature and lipid oxidation feature.

[0119] During corrections, the electronic equipment follows a rule of prioritizing consistency and risk. When both lipid oxidation characteristics and sample color and moisture state characteristics point to a higher risk, while microbial spoilage characteristics are relatively low, the risk indication of lipid oxidation characteristics is retained first, and the impact of gas response characteristics alone reducing the risk level is reduced. If the temperature difference between the sample surface and the detection chamber exceeds 5 degrees Celsius and lasts for more than 1 minute, the electronic equipment adds a current temperature status indicator, indicating that the sample may have just left the refrigerated environment or experienced external temperature exposure. The 5-degree Celsius difference is determined based on the common temperature difference between refrigerated samples and the on-site detection environment. When the difference exceeds this range and lasts for more than 1 minute, the sample surface condition is usually sufficient to affect the release of volatile substances and spectral reflectance.

[0120] After correction, the electronic device matches the corrected microbial spoilage characteristics, corrected fat oxidation characteristics, and sample color and moisture state characteristics with the state reference library. Since the sample is pork belly, the reference range corresponding to the pork belly part is used first during matching, and the participation priority of fat oxidation characteristics is increased. If the matching results show that the corrected fat oxidation characteristics and sample color and moisture state characteristics are close to the critical reference boundary, and the microbial spoilage characteristics are in the sub-freshness reference range, the electronic device can output a freshness level of sub-freshness to critical or critical level, and simultaneously output a verification prompt. If the matching results of the three types of characteristics in the state reference library are completely inconsistent, and the difference reaches more than 2 levels, the electronic device will not force the output of the normal level, but will output the abnormal judgment result and prompt resampling, manual verification, or laboratory re-examination.

[0121] After the determination is completed, the electronic device generates a determination result record. The determination result record includes at least the following fields: sample identifier, sample type, status reference library version number, microbial spoilage characteristic level, lipid oxidation characteristic level, sample color and moisture status level, temperature change influence indicator, matching method indicator, freshness level, reliability indicator, anomaly determination indicator, re-verification prompt indicator, and output time. This record can be used for on-site display, batch traceability, subsequent re-testing, and status reference library updates. If the subsequent laboratory re-testing results are consistent with the current rapid test results, the electronic device can write the sample's characteristic combination and re-testing results into the status reference library update buffer. If the determination difference ratio of historical verification samples after the update exceeds 5% to 10%, it will revert to the previous version. This range is determined based on the balance between the long-term consistency of the rapid testing equipment and the sample expansion capability, avoiding both an overly conservative status reference library that is difficult to absorb new batches of samples and an overly rapid update that leads to instability of existing determination results.

[0122] Through the above processing, the electronic device can simultaneously utilize gas response, fat oxidation color development, and visible and near-infrared spectral information in a single on-site rapid test to comprehensively determine the freshness of the pork to be tested. This method does not rely solely on volatile alkaline gases or color changes, but rather combines the sample location, the influence of temperature changes, and the consistency relationship between multiple characteristics to make a judgment. It can adapt to real circulation scenarios such as cold chain transportation, wholesale sampling, and terminal sales, reducing the risk of misjudging between fresh and less fresh, and between less fresh and critical state.

[0123] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0124] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above embodiments can be implemented in whole or in part by a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions of the embodiments of this application are implemented in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted wirelessly or wiredly from one website, computer, server, or data center to another website, computer, server, or data center. Wired methods include optical fiber, twisted pair, coaxial cable, etc. Wireless methods include infrared, microwave, etc. Available media include any available media that can be accessed by a computer or data storage devices such as servers and data centers that contain one or more sets of available media. Available media can be magnetic media (floppy disks, hard disks, magnetic tapes), optical media (DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0125] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0126] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for rapidly determining the freshness of fresh pork, characterized in that, include: The fresh pork to be tested was placed in the testing chamber, and the temperature and humidity of the testing chamber were pre-balanced. At the same time reference, gas response signals, fat oxidation color development signals, visible and near-infrared spectral signals, chamber environment signals and sample surface temperature signals were collected. Baseline correction, temperature compensation, channel response correction, and outlier sampling removal are performed on the acquired signals. Based on the corrected signals, microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics were extracted respectively. Based on the correspondence between microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics, the impact of temperature changes experienced by the tested fresh pork was determined. This involved accessing data on microbial spoilage characteristics, lipid oxidation characteristics, sample color and moisture state characteristics, environmental records during the testing period, sample surface temperature records, sample type information, and isothermal and temperature fluctuation reference data from a state reference library. These three types of characteristics were then assigned to freshness ranges. Based on the deviation relationships between these three types of characteristics, a temperature change impact indicator was generated through grade difference comparison, reference trajectory matching, and deviation direction combination. The temperature change impact indicator values ​​were categorized as low impact, medium impact, high impact, near-isothermal, and short-term temperature rise. The system identifies one of the following: repeated fluctuations and suspected chain breaks. It corrects for microbial spoilage and lipid oxidation characteristics and processes them according to the rules of characteristic consistency and priority of higher risk. When the microbial spoilage or lipid oxidation characteristics and the sample color and moisture status characteristics point to the same freshness range, the consistent characteristics are used as the matching basis. When the microbial spoilage or lipid oxidation characteristics deviate from other characteristics, the matching weight of the corresponding characteristics is adjusted according to the temperature change influence indicator, the direction of deviation, and the current temperature status indicator. When the three types of characteristics are completely inconsistent or there are missing characteristics, an anomaly judgment indicator, a low confidence indicator, or a re-inspection prompt indicator is generated, and a corrected data record is generated. The corrected microbial spoilage characteristics, corrected lipid oxidation characteristics, and sample color and moisture state characteristics are matched with a preset state reference library to output the freshness grade of the fresh pork to be tested.

2. The method for rapidly determining the freshness of fresh pork according to claim 1, characterized in that, The fresh pork to be tested is placed in the testing chamber, and the temperature and humidity of the testing chamber are pre-balanced, including: First, check the sealing status of the detection chamber, the stability of the light source, the zero point of the gas response channel, the colorimetric acquisition area, and the spectral background; Record the sample part and sample morphology of the fresh pork to be tested, and remove blood, frost and packaging condensation from the surface of the fresh pork to be tested; When there is a water film on the surface of the fresh pork to be tested, the fresh pork to be tested is left to stand, and then the testing chamber is closed and the temperature and humidity inside the testing chamber are adjusted. If the preset environmental range is not met, the pre-balancing process will be extended, the detection will be suspended, or the degradation decision will be made.

3. The method for rapidly determining the freshness of fresh pork according to claim 1, characterized in that, Gas response signals, fat oxidation color development signals, visible and near-infrared spectral signals, chamber environment signals, and sample surface temperature signals were collected at the same time reference, including: When collecting data at the same time reference, hardware synchronization triggering, software time stamp alignment, or group synchronization methods are used to collect gas response, colorimetric images, visible and near-infrared spectra, chamber environment, and sample surface temperature, and generate multi-channel data records. The multi-channel data acquisition record should include at least the acquisition time identifier, signal type, acquisition channel number, original signal value, chamber environmental parameters, sample surface temperature, acquisition method identifier, and abnormal status identifier.

4. The method for rapidly determining the freshness of fresh pork according to claim 1, characterized in that, The acquired signals undergo baseline correction, temperature compensation, channel response correction, and outlier removal, including: First, verify the validity of the gas response signal, the colorimetric signal of fat oxidation, and the visible and near-infrared spectral signal; Based on the signal types of gas response signal, fat oxidation color development signal and visible and near-infrared spectral signal, empty chamber baseline, initial color data, reference color data, reference plate data and dark background data are selected respectively; Based on the detection chamber environment, sample surface temperature, and channel calibration data, each signal is corrected, and burrs, missing, saturated, and time-misaligned data are removed, generating a calibration data record with baseline source, compensation identifier, correction coefficient, and anomaly identifier.

5. The method for rapidly determining the freshness of fresh pork according to claim 1, characterized in that, Based on the corrected signals, microbial spoilage characteristics, lipid oxidation characteristics, and sample color and moisture state characteristics were extracted, including: Verify the availability of the gas response sequence, colorimetric image sequence, and spectral reflectance sequence; Select the effective response segment from the gas response sequence and extract the response amplitude, rise rate, settling time, settling segment fluctuation amplitude, and channel response sequence; The effective color-developing area is determined from the color-developing image sequence, and the color change amplitude, color change direction, color change speed, and color stability after reaction are extracted. The band ratio, band differences, and local variation trends were extracted from the spectral reflectance sequence according to the color-related band and the moisture scattering-related band. Generate feature data records that include validity flags, anomaly flags, missing item flags, and inconsistency flags.

6. The method for rapidly determining the freshness of fresh pork according to claim 5, characterized in that, Verify the availability of the gas response sequence, colorimetric image sequence, and spectral reflectance sequence, including: When determining the available state of each sequence, the gas response sequence has a continuous rising segment or a stable changing segment; The colorimetric image sequence has a recognizable effective colorimetric region; The spectral reflectance sequence has been corrected to the reference. When only two types of sequences among the gas response sequence, colorimetric image sequence, and spectral reflectance sequence meet the usability criteria, extract the features corresponding to the sequences that meet the usability criteria, and record the missing items and the reasons for the missing items.

7. The method for rapidly determining the freshness of fresh pork according to claim 1, characterized in that, The corrected microbial spoilage characteristics, corrected lipid oxidation characteristics, and sample color and moisture state characteristics were matched with a preset state reference library, including: Select the status reference library index based on sample type, testing scenario, temperature change impact indicator, and anomaly judgment indicator; If sample location information exists, retrieve the corresponding location reference data; If no sample location information is available, general reference data will be retrieved based on the color and moisture status reflected in the spectrum and the sample appearance information. The matching priorities for lipid oxidation characteristics, microbial spoilage characteristics, and sample color and moisture state characteristics are set according to the sample location. Then, hierarchical boundary matching, similar sample matching, or trajectory position matching are used to compare the three types of features with reference boundaries, combinations of known sample features, or state trajectories.

8. The method for rapidly determining the freshness of fresh pork according to claim 1, characterized in that, Output the freshness grade of the pork to be tested, including: The grade is determined based on the matching consistency of the three types of features, the influence of temperature changes, the priority of sample location, and the feature correction results; The freshness grade is one of the following: Fresh, Nearly Fresh, Borderline, and Unacceptable. If an anomaly detection flag exists, output the anomaly detection information; A verification prompt message is generated for critical grade samples, and a stop-process prompt message is generated for non-conforming grade samples. Generate a judgment result record containing sample identifier, status reference library version number, matching method identifier, reliability identifier, and output time, and update the status reference library version or revert to the previous version based on the review results.

Images