An intelligent grading cooperative detection device for aquatic products and meat freshness in multiple scenes

Abstract

The application belongs to the technical field of food safety detection, and relates to a multi-scene-adaptive intelligent grading collaborative detection device for freshness of aquatic products and meat. The device integrates multiple detection technologies and builds a multi-modal detection collaborative model, and specifically comprises a power module, a multi-source detection module, a data processing module, a display and early warning module and a storage module. The multi-source detection module integrates at least two types of detection units, which can be flexibly configured according to the detection scene. Through the collaborative work of multiple detection methods, multi-modal feature parameters are obtained, and a dynamic weighted collaborative algorithm is used to realize the intelligent grading of the freshness of aquatic products and meat. The application solves the technical problem that the freshness of aquatic products and meat cannot be quickly and accurately detected in the processing, circulation and retail links, makes up for the defects of traditional single detection means, is compatible with multi-category detection and can improve the detection accuracy, provides a new technical solution for food safety guarantee, and has high application value.

Description

Technical Field

[0001] This invention belongs to the field of food safety testing technology, specifically relating to an intelligent grading and collaborative testing device for the freshness of aquatic products and meat, adaptable to multiple scenarios, and applied to the freshness testing of aquatic and meat products in processing, distribution, and retail. This invention integrates multi-source testing technologies to construct a multimodal, dynamically weighted, and collaborative testing system, enabling rapid assessment and intelligent detection of the freshness of aquatic and meat products. Addressing the shortcomings of traditional freshness testing methods that rely heavily on single-sensor technologies, resulting in limited detection capabilities, susceptibility to environmental interference, and difficulty in covering the testing needs of multiple product categories, this invention provides an intelligent grading and testing technology solution adaptable to multiple scenarios. Background Technology

[0002] With increasing awareness of food safety, evaluating the freshness of high-protein aquatic and meat products has become a crucial requirement in cold chain logistics, supermarket sales, and food processing. During spoilage, fresh produce typically undergoes microbial growth, protein decomposition, and fat oxidation, leading to significant changes in its sensory properties and chemical composition, and potentially causing food safety issues. Therefore, developing a rapid and accurate device for detecting the freshness of aquatic and meat products is of great importance for ensuring public food safety.

[0003] Currently, the main technologies for detecting the freshness of meat and seafood include the following categories: First, traditional physicochemical detection methods, which measure volatile basic nitrogen (TVB-N), hydrogen sulfide, pH, and peroxide value with high accuracy (Meat Research, 2025, 39, 64-71). However, these methods typically require laboratory equipment, reagents, and professional personnel, making them complex, time-consuming, and unable to provide real-time detection. Second, gas sensor technology, which quantitatively evaluates the freshness of food by detecting volatile substances such as ammonia and hydrogen sulfide released from the food. This method has the advantages of fast response and low cost, but it is easily affected by humidity and temperature factors, and a single gas sensor signal cannot comprehensively reflect the overall spoilage status of fresh food (Food Analytical Methods, 2018, 11, 2916-2924). Third, computer vision and image analysis methods assess changes on the surface of fresh food based on color, texture, and gloss features. This is a non-contact detection method, but image information is easily affected by light and surface contamination. Differences in parameters may introduce noise, interfering with feature extraction (FoodChemistry, 2025, 495, 146430). Fourth, emerging methods such as near-infrared spectroscopy, electrochemical sensing, and colorimetry each have unique advantages in certain dimensions. However, if used alone, the feature information is limited, making it difficult to accurately identify the complex changes in fresh food at different stages of spoilage (Food Packaging and Shelf Life, 2026, 37, 101089).

[0004] To address the limitations of single detection technologies, existing research has attempted to improve detection performance using multimodal fusion detection schemes. However, most of these schemes employ fixed-weight fusion strategies, failing to dynamically adjust weights based on food type, spoilage stage, and detection scenario. This fixed fusion strategy limits the model's robustness and practical application value. Therefore, this invention proposes an intelligent collaborative detection device for the freshness of aquatic products and meat, adaptable to multiple scenarios. Through the flexible configuration of modular multi-source detection units, combined with a multi-factor dynamic weighted collaborative fusion algorithm, it overcomes the limitations of traditional single technologies and fixed fusion detection strategies. This enables rapid, accurate, and multi-scenario-adaptive detection of the freshness of aquatic products and meat, providing a reliable technical solution for the entire fresh food supply chain quality control. Summary of the Invention

[0005] To address the limitations of existing food freshness testing technologies, the present invention aims to provide an intelligent grading and collaborative testing device for the freshness of aquatic products and meat that is adaptable to multiple scenarios. This device utilizes a multi-source detection module to extract data on various freshness indicators and combines a dynamic weighted collaborative intelligent algorithm to grade and classify the data for evaluation, thereby solving the problems existing in the aforementioned background technology.

[0006] To achieve the above objectives, the present invention provides an intelligent grading and collaborative detection device for the freshness of aquatic products and meat that is adaptable to multiple scenarios, including a power module, a multi-source detection module, a data processing module, a display and early warning module, and a storage module.

[0007] The power module includes a built-in rechargeable lithium battery and an external DC power interface, which are electrically connected to the multi-source detection module, data processing module, display and early warning module and storage module, respectively, to provide a stable operating voltage for each module, and have built-in overcharge, over-discharge and short-circuit protection circuits.

[0008] The multi-source detection module includes at least two of the following: an image acquisition unit, a gas-sensitive detection unit, a spectral detection unit, an electrochemical detection unit, and a colorimetric reaction unit. It is electrically connected to the input of the data processing module and is used to collect quantitative index data of freshness in the aquatic products and meat to be tested and transmit them to the data processing module.

[0009] The data processing module includes at least one industrial-grade microprocessor. The data processing module is bidirectionally electrically connected to the storage module and electrically connected to the input terminal of the display and warning module. It is used to receive freshness measurement index data output by the multi-source detection module, call the dynamic weighted cooperative algorithm preset in the storage module to perform fusion processing on the at least two types of freshness measurement index data, generate the freshness score and grading result of the sample to be tested, and output the corresponding control signal and data information.

[0010] The display and early warning module includes a display and an early warning unit, which is electrically connected to the data processing module. It is used to receive and display the sample category, freshness score, grade, key characteristic parameters and dynamic weight allocation results output by the data processing module, and to trigger an early warning when the freshness score reaches a preset early warning threshold.

[0011] The storage module includes at least one industrial-grade SD memory card for storing the processing algorithms, detection parameter thresholds, dynamic weight base database required for device operation, as well as storing all data of the detection process; the dynamic weight base database contains a library of basic weight parameters and weight self-learning iteration rules corresponding to different sample categories, detection scenarios, and decay stages.

[0012] Optionally, the multi-source detection module includes at least two of the following: an image acquisition unit, a gas-sensitive detection unit, a spectral detection unit, an electrochemical detection unit, and a colorimetric reaction unit.

[0013] Optionally, the data processing module includes at least one industrial-grade microprocessor. The industrial-grade microprocessor has built-in standard interfaces, including digital interfaces, analog interfaces, bus interfaces, and network interfaces, which support bidirectional data communication with external devices or external systems. The industrial-grade microprocessor integrates a weight adaptive calibration subunit, which communicates bidirectionally with the dynamic weight base database in the storage module to retrieve base weight parameters and output updated dynamic weight coefficients to the dynamic weighted cooperative algorithm execution unit.

[0014] Optionally, the storage module includes at least one industrial-grade SD memory card.

[0015] Optionally, the processing algorithms in the storage module include deep learning-based image processing algorithms and dynamic weighted cooperative algorithms.

[0016] Optionally, the display and warning module includes a display and a warning unit; the display includes at least one of an LCD display module and an OLED display module; the warning unit includes at least one of a tri-color LED light and a buzzer.

[0017] Optionally, the aquatic and meat samples to be tested include any one of freshwater fish, saltwater fish, octopus, crab, shrimp, clams, chicken, pork, beef, and mutton.

[0018] The image acquisition unit includes at least one high-definition camera, used to acquire RGB color channel data, texture distribution features and morphological parameters of the surface of the sample to be tested, and output visual freshness measurement index data A.

[0019] The gas-sensitive detection unit includes a temperature control module, a gas pump module, a gas sensor array, and an analog-to-digital conversion module, used to collect quantitative index data B based on volatile odors. The temperature control module includes a heating element and a thermocouple closed-loop temperature control structure, used to preheat and maintain the temperature of the gas sensor array, ensuring stable operation of different types of gas sensors at their optimal operating temperature. The gas pump module includes a miniature diaphragm pump, an electromagnetic reversing valve, a silicone gas guide tube, and a semi-enclosed chamber. The silicone gas guide tube is connected to the semi-enclosed chamber for dynamic gas sampling and exhaust of gases volatilized from the sample under test.

[0020] Optionally, the gas sensor array includes at least two of a hydrogen sulfide sensor, an ammonia sensor, and a trimethylamine sensor, used to acquire the concentration response signal of the volatile gas of the sample to be tested.

[0021] The spectral detection unit includes a near-infrared-visible spectral detector and an optical fiber probe, used to acquire the intensity of spectral characteristic peaks and wavenumber shift information of changes in the internal composition of the sample under test, and output freshness quantification data C based on the spectrum.

[0022] The electrochemical detection unit includes a three-electrode system and a micro electrochemical workstation, used to measure the redox potential, impedance value and current response signal in the extract of the sample to be tested, and output freshness quantitative index data D based on electrochemistry.

[0023] The colorimetric reaction unit includes a miniature cuvette, an LED light source, and a photodetector. The LED light source and the photodetector are respectively located on opposite sides of the miniature cuvette to form a transmission detection light path, which is used to collect information on the absorbance change after the volatile acids and aldehydes generated by the sample react with the colorimetric reagent, and output freshness quantification index data E based on colorimetry.

[0024] The intelligent grading and collaborative detection device for the freshness of aquatic products and meat, which is adaptable to multiple scenarios, includes the following steps during operation: S1. Sample preparation and device startup: Place the aquatic and meat samples to be tested in the corresponding detection positions, start the device, configure at least two detection units in the multi-source detection module according to the detection scenario, and complete the initialization and parameter calibration; identify the category of the sample to be tested through the image acquisition unit, retrieve the basic weight parameters of the corresponding category from the dynamic weight basic database in the storage module, and complete the weight initialization configuration. S2. Multi-source data synchronous acquisition: Start the configured multi-source detection module, synchronously acquire at least two types of fresh quantitative index data of the sample to be tested according to preset parameters, and transmit them to the data processing module. S3. Data Standardization Processing: The data processing module performs noise reduction, quantization, and normalization on the received fresh quantization index data in sequence to construct a standardized multi-source feature matrix; at the same time, it removes and completes abnormal data, completes feature validity verification, and marks the confidence level of each detection unit's data. S4. Dynamic weighted collaborative fusion and grading: The data processing module calls the pre-set weighted collaborative algorithm in the storage module to perform weighted fusion calculation on the standardized multi-source feature matrix and output the freshness score and corresponding freshness level of the sample to be tested. S5. Display, Early Warning and Data Storage: The display and early warning module displays the sample information, freshness score, freshness level, key characteristic parameters and dynamic weight allocation results. If the freshness score reaches the preset early warning threshold, an early warning is triggered. At the same time, all data of this test is stored in the storage module and the online self-learning update of the dynamic weight basic database is completed. After the test is completed, the device enters a low-power standby state.

[0025] In step S4, the execution process of the dynamic weighted cooperation algorithm includes: S41. Basic Weight Matching: Based on the identified sample category, retrieve the basic weights ω0 of each detection unit for the corresponding category from the dynamic weight database. x , where x is the enabled detection unit indicator; S42. Multi-dimensional dynamic factor calculation: The weight adaptive calibration subunit calculates the category adaptation factor α based on the collected real-time data. x Corruption stage discrimination factor β x Data confidence factor δ x Dynamic adjustment coefficients in three dimensions; S43. Dynamic Weight Coefficient Calculation: Using a pre-defined weight calculation model, the real-time dynamic weight coefficient ω of each detection unit is calculated. The formula is: ω = ω0 x × α x × β x × δ x All weight coefficients are normalized to ensure Σω = 1; S44. Freshness score calculation: By using a weighted summation method, the standardized multi-source feature matrix and the corresponding dynamic weight coefficients are fused into a comprehensive freshness evaluation vector to calculate the freshness score of the sample to be tested. S45. Freshness rating determination: Based on the freshness score, the rating is determined and the rating results are output.

[0026] Optionally, x corresponds to the enabled detection unit indicators A, B, C, D, and E.

[0027] Optionally, the freshness score ranges from 0 to 100 points, and the freshness level is divided into three levels: fresh, slightly fresh, and rotten; wherein, 81-100 points is considered fresh, 41-80 points is considered slightly fresh, and 0-40 points is considered rotten.

[0028] Optionally, the preset warning threshold is: a spoilage warning is triggered when the freshness score is ≤ 40, and a near-freshness warning is triggered when the score is 41-80.

[0029] Optionally, the category fit factor α x Based on the physicochemical properties and spoilage patterns of the samples to be tested, a weight priority system is established to match the core spoilage characteristics of different sample categories. Among these, the category adaptation factor α is used for the gas-sensitive detection unit and the colorimetric reaction unit. x ≥ 1.2, category adaptation factor α for image acquisition unit, spectral detection unit, and electrochemical detection unit x ≥ 1.0.

[0030] Optionally, the corruption stage discriminant factor β x Based on the feature distribution of a standardized multi-source feature matrix, a pre-trained deep learning model is used to determine the current spoilage stage of a sample and dynamically adjust the weights of each detection unit; among them, the discriminant factor β for the freshness stage... x ≥ 1.2, discriminant factor β for the near-freshness stage x ≥ 1.15, the discriminant factor β for the decay stage x ≥ 1.3.

[0031] Optionally, the data confidence factor δ x Based on the signal-to-noise ratio and repeatability deviation of the data collected by each detection unit, δ is calculated. The higher the signal-to-noise ratio and the smaller the repeatability deviation, the higher the δ. x The closer to 1; when the relative deviation of repeated measurements of a single set of data is ≥10%, the corresponding δ x ≤ 0.6, when the signal-to-noise ratio of a single data set is ≤ 20dB, corresponding to δ x ≤ 0.5.

[0032] Compared with existing technologies, the technical solution provided by this invention has the following characteristics: This invention provides an intelligent collaborative detection device for freshness grading of aquatic products and meat, adaptable to multiple scenarios. It constructs a multi-factor dynamic weighted collaborative fusion model, overcoming the limitations of existing fixed-weight fusion strategies. It integrates multi-source detection modules and combines data from at least two sensors to quantify freshness indicators such as surface morphology changes, volatile gas release, and internal component changes in aquatic and meat samples. A multi-modal detection collaborative model is constructed, and signals from various detection modules are fused through a dynamic weighted collaborative algorithm to more comprehensively and accurately reflect the freshness level, ultimately achieving intelligent grading detection of freshness for aquatic and meat products. Furthermore, the testing device of this invention does not require sample destruction or complex processing, is easy to operate, and has a fast detection speed. Multi-source indicator data mutually verify each other, effectively reducing environmental interference and improving the accuracy and stability of detection. This invention is applicable to multiple stages of food processing, distribution, and retail, with a wider range of application scenarios, providing strong support for freshness grading detection of aquatic products and meat in different scenarios. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0034] Figure 1 A flowchart of the apparatus provided in an embodiment of the present invention; Figure 2 This is a device system block diagram provided in an embodiment of the present invention; Figure 3 This is a structural unit composition diagram of the device provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the device structure provided in an embodiment of the present invention; Figure 5 This is a diagram illustrating the second type of device structure unit composition provided in an embodiment of the present invention; Figure 6 This is a diagram illustrating the third type of device structure unit composition provided in this embodiment of the invention; The diagram is marked as follows: 1. Image acquisition unit; 2. Gas-sensitive detection unit; 3. Spectroscopic detection unit; 4. Electrochemical detection unit; 5. Colorimetric reaction unit; 6. Data processing module; 7. Display and early warning module; 8. Power supply module; 9. Storage module; 10. Air pump module. Detailed Implementation

[0035] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described below are only for explaining the present invention and are not intended to limit the present invention. Example 1

[0036] Please see Figure 1 , Figure 1 This is a flowchart of an intelligent grading and collaborative detection device for the freshness of aquatic products and meat that is adapted to multiple scenarios, provided by an embodiment of the present invention, including the following steps: S1. Sample Preparation and Device Start-up: Place 50g of the pork sample to be tested into the semi-enclosed sample chamber; start the device, configure the image acquisition unit, gas-sensitive detection unit, spectral detection unit, and colorimetric reaction unit to start operation, complete system initialization, sensor preheating, and parameter calibration; simultaneously, identify the sample to be tested as pork through the image acquisition unit, and retrieve the basic weight parameters corresponding to pork from the dynamic weight database in the storage module: Image Acquisition Unit ω0 A = 0.2, Gas-sensitive detection unit ω0 B = 0.35, spectral detection unit ω0 C = 0.25, colorimetric reaction unit ω0 E = 0.2, weight initialization configuration complete; S2. Multi-source data synchronous acquisition: Start the configured multi-source detection module, synchronously acquire the freshness quantitative index data of the sample to be tested according to the preset parameters, and synchronously transmit it to the data processing module. S3. Data Standardization Processing: The data processing module sequentially performs denoising, quantization, and normalization on the received fresh metric data to construct a standardized multi-source feature matrix. Simultaneously, it removes and completes outlier data, verifies feature validity, calculates the signal-to-noise ratio and repeatability bias of each detection unit's data, and labels the confidence factor δ of each detection unit's data. A = 0.98, δ B = 0.95, δ C = 0.99, δ E = 0.96; S4. Dynamic weighted collaborative fusion and grading: The data processing module calls the pre-set dynamic weighted collaborative algorithm in the storage module, calculates the real-time dynamic weight coefficients of the corresponding indicators of each detection unit through the weight adaptive calibration subunit, performs weighted fusion calculation on the standardized multi-source feature matrix, and outputs the freshness score and corresponding freshness level of the sample to be tested. In this step, the specific execution process of the dynamic weighted cooperation algorithm is as follows: S41. Complete the basic weight matching and determine the basic weight ω0 for each detection unit corresponding to the salmon sample. A = 0.2、ω0B = 0.35、ω0 C = 0.25、ω0 E = 0.2; S42. Multi-dimensional dynamic factor calculation: Category adaptation factor: Salmon is a high-protein marine fish, and its core spoilage characteristic is the production of volatile amines from protein decomposition. Therefore, α is calculated. A = 1.0, α B = 1.25, α C = 1.0, α E = 1.2; Spoilage stage discrimination factor: Based on the standardized multi-source feature matrix, a pre-trained deep learning model is used to determine whether a sample is in the fresh stage, thus β is calculated. A = 1.2, β B = 1.0, β C = 1.1、β E = 1.0; Data confidence factor: based on the calculation results of step S3, δ A = 0.98, δ B = 0.95, δ C = 0.99, δ E = 0.96; S43. Calculation of dynamic weighting coefficients: using the formula ω = ω0 x × α x × β x × δ x The initial dynamic weights of each detection unit are calculated as follows: ω A = 0.2 × 1.0 × 1.2 × 1.0 × 0.98 = 0.2352 ω B = 0.35 × 1.25 × 1.0 × 1.0 × 0.95 = 0.4156 ω C = 0.25 × 1.0 × 1.1 × 1.0 × 0.99 = 0.2723 ω E = 0.2 × 1.2 × 1.0 × 1.0 × 0.96 = 0.2304 The above weights are normalized to obtain the real-time dynamic weight coefficients: ω A = 0.204、ω B =0.360、ω C = 0.236、ω E = 0.200, satisfying Σω = 1; S44. Freshness score calculation: By using a weighted summation method, the standardized multi-source feature matrix is ​​fused with the corresponding dynamic weight coefficients to calculate the freshness score of the pork sample to be tested as 88 points. S45. Freshness rating: 88 points falls within the 81-100 point range and is rated as fresh. S5. Display, Warning and Data Storage: The OLED display of the display and warning module shows the sample information, freshness score of 88, freshness level, key characteristic parameters and dynamic weight allocation results. If the warning threshold is not reached, no warning is triggered. At the same time, the full data of this test and the dynamic weight coefficient are stored in the storage module. After the test is completed, the air pump system in the gas-sensitive detection unit ventilates the cavity, and the data processing module controls each module to enter a low-power standby state. Example 2

[0037] Optionally, please see Figure 2 , Figure 2 This is a system block diagram of an intelligent grading and collaborative detection device for the freshness of aquatic products and meat, adapted to multiple scenarios, provided by an embodiment of the present invention. The system mainly consists of a power supply module, a multi-source detection module, a data processing module, a display and early warning module, and a storage module, wherein: The multi-source detection module consists of multiple detection units, including an image acquisition unit, a gas-sensitive detection unit, a spectral detection unit, an electrochemical detection unit, and a colorimetric reaction unit, each used to acquire various types of detection information. The system's main operating logic is as follows: After startup, each sensing module operates synchronously. The image acquisition unit identifies the sample type, captures sample images, and obtains surface color and texture features; the gas-sensitive detection unit acquires gas-sensitive response features through a gas sensor array; the spectral detection unit captures the near-infrared and visible spectra of the sample, identifying internal component features; the electrochemical detection unit and the colorimetric reaction unit acquire redox potential signal features and color change features, respectively; then, the data processing module calls the intelligent grading algorithm preset in the storage module to generate a freshness score and grade, and outputs the detection results through the display and warning module; the storage module saves the detection information and results, ensuring long-term traceability; after detection is completed, the air pump system in the gas-sensitive detection unit ventilates the cavity, and the data processing module controls each module to enter a low-power standby state. Example 3

[0038] Optionally, for the structural unit composition diagram of the first type of intelligent grading and collaborative detection device for freshness of aquatic products and meat adapted to multiple scenarios provided in this embodiment of the invention, please refer to... Figure 3 Specifically: The device is powered by a power module, and the multi-source detection module operates synchronously according to preset parameters. The camera in the image acquisition unit captures images of the samples, acquiring RGB color channel data, texture distribution characteristics, and morphological parameters of the food surface. This data is electrically connected to the input of the data processing module to determine the category of the target sample and outputs visual-based freshness quantification data A. The gas-sensitive detection unit acquires gas-sensitive response characteristics through a gas-sensitive sensor array, detects the concentration information of characteristic gases released from the sample, and is electrically connected to the input of the data processing module, outputting odor-based freshness quantification data B. The spectral detection unit acquires the spectral signal reflected from the surface of the target sample through a fiber optic probe and is electrically connected to the input of the data processing module, outputting spectral-based freshness quantification data C. The electrochemical detection unit is electrically connected to the input of the data processing module via a shielded cable, measuring the redox potential, impedance value, and current response signal in the food acquisition liquid, and outputting electrochemical-based freshness quantification. The indicator data is D; the colorimetric reaction unit collects the color change information of the volatile acids and aldehydes generated by the target sample and the colorimetric reagent, and is electrically connected to the input of the data processing module to output the colorimetric freshness quantitative indicator data E; after receiving the raw indicator data A, B, C, D, and E from each detection module, the data processing module executes the processing algorithm preset by the storage module to generate a freshness score and grade, and generates control commands to output control signals and data information to the display and early warning module; after receiving the commands, the display and early warning module displays the sample category, freshness score, grade, and key detection information on the OLED display, and displays the freshness grade through three-color LEDs: fresh is green, slightly fresh is yellow, and spoiled is red. When the spoilage threshold is reached, a buzzer is triggered simultaneously to provide an early warning; the storage module records and stores the detection information and results; after the detection is completed, the air pump system in the gas-sensitive detection unit ventilates the cavity, and the data processing module controls each module to enter a low-power standby state. Example 4

[0039] Optionally, please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an intelligent collaborative detection device for freshness of aquatic products and meat adapted to multiple scenarios, provided in an embodiment of the present invention. Specifically, it is used in the first type of intelligent collaborative detection device for freshness of aquatic products and meat adapted to multiple scenarios described in Embodiment 3. The system comprises an image acquisition module 1, a gas-sensitive detection unit 2, a spectral detection unit 3, an electrochemical detection unit 4, a colorimetric reaction unit 5, a data processing module 6, a display and early warning module 7, a power supply module 8, a storage module 9, and an air pump module 10. Specifically, the image acquisition module 1 includes a camera, electrically connected to the data processing module 6; the display and early warning module 7 includes an OLED display and a tri-color LED, where the display outputs detection results and status data, and the tri-color LED is used to mark early warning status. Controlled by the data processing module 6, it can output three colors: green, yellow, and red, where green indicates freshness, yellow indicates near-freshness, and red indicates spoilage; the storage module 9 stores preset processing algorithms and is electrically connected to the data processing module 6; the air pump module 10 includes an inlet pipe and an outlet pipe, the inlet pipe being connected to air, and the outlet pipe being connected to the semi-enclosed cavity outside the gas-sensitive detection unit 2. During measurement, it can be mounted on a handheld terminal or mobile testing vehicle, and the freshness of the dispersed target samples can be graded and tested manually. The test results are fed back on-site through the display and early warning module 7. Example 5

[0040] Optionally, please see Figure 5 , Figure 5 This is a structural unit composition diagram of the second type of intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios provided in this embodiment of the invention, specifically as follows: The device is powered by a power module, and the multi-source detection module operates synchronously according to preset parameters. The camera in the image acquisition unit captures images of the sample, acquiring RGB color channel data, texture distribution characteristics, and morphological parameters of the food surface. This data is electrically connected to the input of the data processing module to determine the category of the target sample and outputs visual-based freshness quantification data A. The gas-sensitive detection unit acquires gas-sensitive response characteristics through a gas-sensitive sensor array, detects the concentration information of characteristic gases released by the sample, and is electrically connected to the input of the data processing module, outputting odor-based freshness quantification data B. The spectral detection unit acquires the spectral signal reflected from the surface of the target sample through a fiber optic probe and is electrically connected to the input of the data processing module, outputting spectral-based freshness quantification data C. The colorimetric reaction unit collects volatile acids, aldehydes, and colorimetric components produced by the target sample. The color change information of the reagent's colorimetric reaction is electrically connected to the input of the data processing module, outputting colorimetric freshness quantification index data E. After receiving the raw index data A, B, C, and E from each detection module, the data processing module executes the processing algorithm preset in the storage module to generate a freshness score and grade, and generates control commands to output control signals and data information to the display and early warning module. The OLED display screen in the display and early warning module outputs the sample category, freshness score, grade, and key detection information, and displays the freshness grade through three-color LEDs: green for fresh, yellow for slightly fresh, and red for spoiled. When the spoilage threshold is reached, a buzzer is triggered simultaneously to provide an early warning. The storage module records and stores the detection information and results. After the detection is completed, the air pump system in the gas-sensitive detection unit ventilates the cavity, and the data processing module controls each module to enter a low-power standby state. Example 6

[0041] Optionally, please see Figure 6 , Figure 6 This is a structural unit composition diagram of the third type of intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios provided in this embodiment of the invention, specifically as follows: The device is powered by a power module, and the multi-source detection module operates synchronously according to preset parameters. The camera in the image acquisition unit captures images of the sample, acquiring RGB color channel data, texture distribution characteristics, and morphological parameters of the food surface. This data is electrically connected to the input of the data processing module to determine the category of the target sample and outputs visual-based freshness quantification data A. The gas-sensitive detection unit acquires gas-sensitive response characteristics through a gas-sensitive sensor array, detects the concentration information of characteristic gases released by the sample, and is electrically connected to the input of the data processing module, outputting odor-based freshness quantification data B. The spectral detection unit acquires the spectral signal reflected from the surface of the target sample through a fiber optic probe and is electrically connected to the input of the data processing module, outputting spectral-based freshness quantification data C. After receiving the raw index data A, B, and C from each detection module, the data processing module executes the processing algorithm preset in the storage module to generate a freshness score and grade, and generates control commands to output control signals and data information to the display and warning module. After receiving the commands, the display and warning module displays the sample category, freshness score, grade, and key detection information on the OLED screen. When the corrosion threshold is reached, a buzzer is triggered to issue an early warning; the storage module records and stores the detection information and results; after the detection is completed, the air pump system in the gas-sensitive detection unit ventilates the cavity, and the data processing module controls each module to enter a low-power standby state.

[0042] The above description is only a preferred embodiment 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.

Claims

1. A smart collaborative detection device for freshness grading of aquatic products and meat, adaptable to multiple scenarios, characterized in that, It includes a power supply module, a multi-source detection module, a data processing module, a display and early warning module, and a storage module, among which: The power module includes a built-in rechargeable lithium battery and an external DC power interface, which are electrically connected to the multi-source detection module, data processing module, display and early warning module and storage module, respectively, to provide a stable operating voltage for each module. The power module has built-in overcharge protection, over-discharge protection and short circuit protection circuits. The multi-source detection module includes at least two of the following: an image acquisition unit, a gas-sensitive detection unit, a spectral detection unit, an electrochemical detection unit, and a colorimetric reaction unit. It is electrically connected to the input terminal of the data processing module and is used to collect quantitative index data of freshness in the aquatic products and meat to be tested and transmit them to the data processing module. The data processing module includes at least one industrial-grade microprocessor. The data processing module is bidirectionally electrically connected to the storage module and electrically connected to the input terminal of the display and warning module. It is used to receive freshness measurement index data output by the multi-source detection module, call the dynamic weighted cooperative algorithm preset in the storage module to perform fusion processing on the at least two types of freshness measurement index data, generate the freshness score and grading result of the sample to be tested, and output the corresponding control signal and data information. The display and warning module includes a display and a warning unit, which is electrically connected to the data processing module. It is used to receive and display the sample category, freshness score, grade, key characteristic parameters and dynamic weight allocation results output by the data processing module, and to trigger a warning when the freshness score reaches a preset warning threshold. The storage module includes at least one industrial-grade SD memory card for storing the processing algorithms, detection parameter thresholds, dynamic weight base database required for device operation, as well as storing all data of the detection process; the dynamic weight base database contains a library of basic weight parameters and weight self-learning iteration rules corresponding to different sample categories, detection scenarios, and decay stages.

2. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that: The image acquisition unit includes at least one high-definition camera, used to acquire RGB color channel data, texture distribution features and morphological parameters of the surface of the sample to be tested, and output visual freshness measurement index data A. The gas-sensitive detection unit includes a temperature control module, a gas pump module, a gas sensor array, and an analog-to-digital conversion module, used to acquire quantitative index data B based on volatile odors. The temperature control module includes a heating element and a thermocouple closed-loop temperature control structure, used to preheat and maintain the temperature of the gas sensor array, ensuring stable operation of different types of gas sensors at their optimal operating temperatures. The gas pump module includes a miniature diaphragm pump, an electromagnetic reversing valve, a silicone gas guide tube, and a semi-enclosed chamber. The silicone gas guide tube is connected to the semi-enclosed chamber for dynamic gas sampling and exhaust from the sample. The gas sensor array includes at least two of a hydrogen sulfide sensor, an ammonia sensor, and a trimethylamine sensor, used to acquire the concentration response signal of the volatile gases from the sample. The spectral detection unit includes a near-infrared-visible spectral detector and an optical fiber probe, used to acquire the intensity of spectral characteristic peaks and wavenumber shift information of changes in the internal composition of the sample under test, and output freshness quantification index data C based on the spectrum. The electrochemical detection unit includes a three-electrode system and a micro electrochemical workstation, used to measure the redox potential, impedance value and current response signal in the extract of the sample to be tested, and output freshness quantitative index data D based on electrochemistry; The colorimetric reaction unit includes a miniature cuvette, an LED light source, and a photodetector. The LED light source and the photodetector are respectively located on opposite sides of the miniature cuvette to form a transmission detection light path, which is used to collect information on the absorbance change after the volatile acids and aldehydes generated by the sample react with the colorimetric reagent, and output freshness quantification index data E based on colorimetry.

3. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that, The industrial-grade microprocessor has built-in standard interfaces, including digital interfaces, analog interfaces, bus interfaces, and network interfaces, which support bidirectional data communication with external devices or systems. The industrial-grade microprocessor integrates a weight adaptive calibration subunit, which communicates bidirectionally with the dynamic weight base database in the storage module to retrieve the base weight parameters and output the updated dynamic weight coefficients to the dynamic weighted cooperative algorithm execution unit.

4. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that, The display includes at least one of an LCD display module and an OLED display module; the warning unit includes at least one of a tri-color LED light and a buzzer.

5. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that, The processing algorithms in the storage module include deep learning-based image processing algorithms and dynamic weighted cooperative algorithms.

6. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that, The samples to be tested include any one of the following: freshwater fish, saltwater fish, octopus, crab, shrimp, clams, chicken, pork, beef, and mutton.

7. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that, This method integrates multiple detection modules to collaboratively detect the freshness of aquatic products and meat, constructs a multimodal detection collaboration model, and assigns adaptive dynamic weights to the signals of each detection module through a dynamic weighted collaboration algorithm. Ultimately, it achieves intelligent grading and detection of the freshness of aquatic products and meat, specifically including the following detection steps: S1. Sample preparation and device startup: Place the aquatic and meat samples to be tested in the corresponding detection positions, start the device, configure at least two detection units in the multi-source detection module according to the detection scenario, and complete the initialization and parameter calibration; identify the category of the sample to be tested through the image acquisition unit, retrieve the basic weight parameters of the corresponding category from the dynamic weight basic database in the storage module, and complete the weight initialization configuration. S2. Multi-source data synchronous acquisition: Start the configured multi-source detection module, synchronously acquire at least two types of fresh quantitative index data of the sample to be tested according to preset parameters, and transmit them to the data processing module. S3. Data Standardization Processing: The data processing module performs noise reduction, quantization, and normalization on the received fresh quantization index data in sequence to construct a standardized multi-source feature matrix; at the same time, it removes and completes abnormal data, completes feature validity verification, and marks the confidence level of each detection unit's data. S4. Dynamic weighted collaborative fusion and grading: The data processing module calls the pre-set dynamic weighted collaborative algorithm in the storage module to perform weighted fusion calculation on the standardized multi-source feature matrix and output the freshness score and corresponding freshness level of the sample to be tested. S5. Display, Early Warning and Data Storage: The display and early warning module displays the sample information, freshness score, freshness level, key characteristic parameters and dynamic weight allocation results; if the freshness score reaches the preset early warning threshold, an early warning is triggered; at the same time, all data of this test is stored in the storage module and the online self-learning update of the dynamic weight basic database is completed. After the test is completed, the device enters a low-power standby state.

8. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that, In step S4, the execution process of the dynamic weighted cooperation algorithm includes: S41. Basic Weight Matching: Based on the identified sample category, retrieve the basic weights ω0 of each detection unit for the corresponding category from the dynamic weight database. x , x is the enabled detection unit indicator, which can be A, B, C, D, or E; S42. Multi-dimensional dynamic factor calculation: The weight adaptive calibration subunit calculates the category adaptation factor α based on the collected real-time data. x Corruption stage discriminant factor β x Data confidence factor δ x Dynamic adjustment coefficients in three dimensions; S43. Dynamic Weight Coefficient Calculation: Using a pre-defined weight calculation model, the real-time dynamic weight coefficient ω of each detection unit is calculated. The formula is: ω = ω0 x × α x × β x × δ x All dynamic weight coefficients are normalized to ensure Σω = 1; S44. Freshness score calculation: By using a weighted summation method, the standardized multi-source feature matrix and the corresponding dynamic weight coefficients are fused into a comprehensive freshness evaluation vector to calculate the freshness score of the sample to be tested. S45. Freshness rating determination: Based on the freshness score, the rating is determined and the rating results are output.

9. The intelligent grading and collaborative detection device for the freshness of aquatic products and meat adapted to multiple scenarios as described in claim 1, characterized in that: The category compatibility factor α x Based on the physicochemical properties and spoilage patterns of the samples to be tested, a weight priority system is established to match the core spoilage characteristics of different sample categories. Among these, the category adaptation factor α is used for the gas-sensitive detection unit and the colorimetric reaction unit. x ≥ 1.2, category adaptation factor α for image acquisition unit, spectral detection unit, and electrochemical detection unit x ≥1.0; the corruption stage discriminant factor β x Based on the feature distribution of a standardized multi-source feature matrix, a pre-trained deep learning model is used to determine the current spoilage stage of a sample and dynamically adjust the weights of each detection unit; among them, the discriminant factor β for the freshness stage... x ≥ 1.2, discriminant factor β for the near-freshness stage x ≥ 1.15, the discriminant factor β for the decay stage x ≥ 1.3; the data confidence factor δ x Based on the signal-to-noise ratio and repeatability deviation of the data collected by each detection unit, δ is calculated. The higher the signal-to-noise ratio and the smaller the repeatability deviation, the higher the δ. x The closer to 1; when the relative deviation of repeated measurements of a single set of data is ≥ 10%, the corresponding δ x ≤ 0.6, when the signal-to-noise ratio of a single data set is ≤ 20dB, corresponding to δ x ≤ 0.5.

Images