Fresh food quality degradation pre-judgment method and system based on multi-modal dynamic coupling

Abstract

The invention provides a fresh food quality degradation pre-judgment method and system based on multi-modal dynamic coupling, and aims to solve the problem of prediction lag caused by data isolation and dynamic coupling deficiency in the traditional storage and transportation process. Specifically, sensory layer data, environment layer data and operation layer data are synchronously collected through a multi-source sensor, chromatic aberration quantification, chemical bond feature extraction and spatial chemical fusion processing are carried out on the sensory data, and a quality degradation feature tensor is generated; based on a graph network model and frequency domain analysis, constructing a dynamic coupling parameter set of quality degradation characteristics and environmental parameters; fusing the loading and unloading pulse trajectory tensor, the cold chain stability distribution and the coupling parameter set, and generating a degradation constraint distribution matrix through multivariable weight mapping; and finally, combining acoustic resonance mode coherence analysis and constraint matrix evolution, and outputting a four-stage quality label. According to the method, through multi-dimensional real-time sensing, dynamic coupling modeling and hierarchical management and control, the fresh food storage and transportation loss is remarkably reduced, and the cold chain management efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of cold chain logistics and food preservation, and in particular to a method and system for predicting the deterioration of fresh food quality based on multimodal dynamic coupling. Background Art

[0002] Due to their perishable nature, fresh produce is susceptible to environmental fluctuations, mechanical damage, and microbial activity during storage and transportation, leading to rapid quality deterioration. Globally, this spoilage results in hundreds of billions of dollars in economic losses each year, accompanied by significant resource waste and carbon emissions. As consumers increasingly demand fresh produce for its freshness and safety, early prediction and precise control of quality degradation have become key challenges in reducing losses and optimizing supply chain efficiency.

[0003] Current fresh produce quality monitoring relies primarily on single sensory indicators (such as surface color change) or environmental parameters (such as temperature and humidity threshold alarms), supplemented by manual spot checks and empirical judgment. Some technologies employ basic sensors (such as temperature recorders) or near-infrared spectroscopy equipment for localized detection. However, data collection is limited to static or isolated dimensions (such as focusing solely on temperature or humidity), lacking systematic integration of dynamic disturbances such as loading and unloading shock, stacking pressure, and transportation vibration.

[0004] Due to data isolation and static models, these methods struggle to capture the coupling effects of multimodal factors (sensory, environmental, and operational), resulting in delayed degradation predictions and high misjudgment rates. For example, relying solely on temperature monitoring cannot quantify the cumulative damage to the microstructure of goods caused by loading and unloading vibrations, while manual spot checks cannot respond in real time to temperature and humidity fluctuations caused by the frequent opening and closing of cold storage doors. These "data silos" and "lack of dynamic coupling" issues lead to extensive quality management decisions, making early warning and precise intervention difficult, ultimately exacerbating loss rates and operating costs. Summary of the Invention

[0005] In view of the above actual situation, this application proposes a method and system for predicting fresh food quality deterioration based on multimodal dynamic coupling to solve the problems of delayed quality deterioration prediction and rough decision-making caused by data isolation and lack of dynamic coupling effect in traditional fresh food storage and transportation.

[0006] A method for predicting fresh produce quality degradation based on multimodal dynamic coupling, the method comprising the following steps:

[0007] S1. Obtain data to be processed, including sensory layer data, environmental layer data, and operational layer data. The sensory layer data includes visible light images of fresh produce surfaces, near-infrared reflectance spectra, and acoustic resonance frequency response curves. The environmental layer data includes a temperature and humidity gradient matrix of the microenvironment within the cargo box, a heat map of the spatial distribution of stacking pressure, and a vibration spectrum of the transportation path. The operational layer data includes time and space stamp records of loading and unloading events and time series data on the opening and closing frequency of cold storage doors.

[0008] S2, extracting and fusing quality degradation features of the sensory layer data to obtain a quality degradation feature tensor;

[0009] S3, performing dynamic coupling modeling processing on the quality degradation feature tensor and the environmental layer data, thereby obtaining a quality degradation environment coupling parameter set;

[0010] S4, performing multi-dimensional dynamic constraint field modeling processing on the quality degradation environment coupling parameter set and the operation layer data, thereby obtaining a degradation constraint distribution matrix;

[0011] S5, performing adaptive degradation trajectory evolution processing on the acoustic wave resonance frequency response curve in the sensory layer data and the degradation constraint distribution matrix, thereby obtaining a four-level quality label.

[0012] Furthermore, the step S2 includes the following sub-steps:

[0013] S201, performing color difference quantization processing on the surface visible light image in the sensory layer data to obtain a surface color difference heat map, wherein the color difference quantization processing is to extract surface color abnormal areas through color space transformation;

[0014] S202, performing chemical bond feature extraction processing on the near-infrared reflectance spectrum in the sensory layer data to obtain a water loss feature vector, wherein the chemical bond feature extraction processing is performed by screening characteristic bands related to quality degradation through a PLS regression model;

[0015] S203, performing spatial chemical fusion processing on the surface color difference heat map and the water loss feature vector to obtain a quality degradation feature tensor, wherein the spatial chemical fusion processing maps chemical features to color difference abnormality areas through a spatial weighted matrix.

[0016] Furthermore, the S3 step includes the following sub-steps:

[0017] S301, performing diffusion path topology modeling processing on the quality degradation feature tensor and the microenvironment temperature and humidity gradient matrix in the environmental layer data to obtain a degradation diffusion rate field, wherein the diffusion path topology modeling processing simulates the propagation path of degradation along the direction of the maximum temperature and humidity gradient through a graph network model;

[0018] S302, performing mechanical path coupling processing on the degradation diffusion rate field and the stack pressure spatial distribution thermodynamic map in the environmental layer data to obtain a pressure path correlation tensor, wherein the mechanical path coupling processing calculates the local degradation acceleration effect through geometric superposition of the pressure distribution and the diffusion path;

[0019] S303, performing frequency domain stability analysis on the pressure path correlation tensor and the transport path vibration spectrum in the environmental layer data to obtain a quality degradation environment coupling parameter set. The frequency domain stability analysis quantifies the disturbance intensity of the vibration on the diffusion path through the frequency band energy distribution.

[0020] Furthermore, the S4 step includes the following sub-steps:

[0021] S401, performing pulse sequence trajectory tensor processing on the spatiotemporal stamp records of loading and unloading events in the operation layer data to obtain a loading and unloading pulse trajectory tensor, wherein the trajectory tensor processing is to reconstruct the loading and unloading energy impact distribution through the spatiotemporal trajectory of the pulse event;

[0022] S402: Performing temperature and humidity disturbance evolution modeling on the cold storage door opening and closing frequency time series data in the operation layer data to obtain a dynamic distribution of cold chain stability. The evolution modeling process is to construct a cold chain disturbance dynamic response model by coupling the time series disturbance curve with the humidity gradient;

[0023] S403, performing multi-dimensional dynamic constraint mapping processing on the loading and unloading pulse trajectory tensor, the dynamic distribution of cold chain stability and the quality degradation environment coupling parameter set, so as to obtain the degradation constraint distribution matrix. The mapping processing forms a constraint balance field through space-time-mechanical multivariate weight calculation.

[0024] Furthermore, the step S5 includes the following sub-steps:

[0025] S501, performing a resonance modal coherence analysis on the acoustic wave resonance frequency response curve in the sensory layer data to obtain a spectrum coherence drift factor, wherein the coherence analysis is to calculate the degree of acoustic wave mode shift through the temporal coherence of the resonance frequency band distribution;

[0026] S502, nonlinear degradation trajectory mapping processing is performed on the spectral coherence drift factor and the degradation constraint distribution matrix to obtain four-level quality labels. The trajectory mapping processing is to calculate the quality evolution critical point through multi-dimensional degradation space projection. The four-level quality labels include compliance label, warning label, abnormal label, and irreversible label.

[0027] Furthermore, the mechanical path coupling processing in step S302 is to calculate the local degradation acceleration effect through the geometric superposition of pressure distribution and diffusion path, including local stress coefficient extraction processing and pressure diffusion gain fusion processing; the frequency domain stability analysis processing in step S303 is to quantify the disturbance intensity of vibration on the diffusion path through frequency band energy distribution, including vibration energy spectrum extraction processing and frequency band coupling factor calculation processing.

[0028] Furthermore, the trajectory tensorization processing in step S401 is to reconstruct the loading and unloading energy impact distribution through the spatiotemporal trajectory of the pulse event, including spatiotemporal grid mapping processing and pulse intensity normalization processing.

[0029] Furthermore, the evolutionary modeling processing in step S402 is to construct a cold chain disturbance dynamic response model by coupling the time series disturbance curve with the humidity gradient, including time series disturbance curve decomposition processing and humidity gradient coupling processing; the mapping processing in step 403 is to form a constrained equilibrium field by space-time-mechanical multivariate weight calculation, including space-time dimension alignment processing and multivariate weight fusion processing.

[0030] Furthermore, the coherence analysis processing in the step S501 is to calculate the degree of acoustic wave mode deviation through the temporal coherence of the resonant frequency band distribution, including frequency band segmentation processing and temporal coherence calculation processing; the trajectory mapping processing in the step S502 is to calculate the critical point of quality evolution through multi-dimensional degradation space projection, including multi-dimensional feature space construction processing and critical point determination processing.

[0031] In addition, the present application discloses a fresh produce quality deterioration prediction system based on multimodal dynamic coupling, characterized in that the system includes:

[0032] An acquisition unit is configured to acquire data to be processed, the data to be processed including sensory layer data, environmental layer data, and operational layer data. The sensory layer data includes visible light images of fresh produce surfaces, near-infrared reflectance spectra, and acoustic resonance frequency response curves. The environmental layer data includes a temperature and humidity gradient matrix of the microenvironment within the cargo box, a thermal map of the spatial distribution of stacking pressure, and a vibration spectrum of the transport path. The operational layer data includes time and space stamp records of loading and unloading events and time series data on the opening and closing frequency of cold storage doors.

[0033] a feature fusion unit, configured to extract and fuse quality degradation features of the sensory layer data to obtain a quality degradation feature tensor;

[0034] A coupling modeling unit, configured to perform dynamic coupling modeling processing on the quality degradation feature tensor and the environmental layer data, thereby obtaining a quality degradation environment coupling parameter set;

[0035] a constraint modeling unit, configured to perform multi-dimensional dynamic constraint field modeling processing on the quality degradation environment coupling parameter set and the operation layer data, thereby obtaining a degradation constraint distribution matrix;

[0036] The hierarchical quality label unit is used to perform adaptive degradation trajectory evolution processing on the acoustic wave resonance frequency response curve in the sensory layer data and the degradation constraint distribution matrix, so as to obtain four-level quality labels.

[0037] The present application proposes a method and system for predicting the deterioration of fresh produce quality based on multimodal dynamic coupling, which realizes multi-dimensional real-time perception, dynamic coupling modeling and hierarchical control of the deterioration of fresh produce quality, significantly reduces storage and transportation losses and improves cold chain management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flow chart of a method for predicting fresh produce quality deterioration based on multimodal dynamic coupling proposed in this application;

[0039] Figure 2 This is a schematic diagram of the process of generating a quality degradation feature tensor for a fresh produce quality degradation prediction method based on multimodal dynamic coupling proposed in this application;

[0040] Figure 3 This is a flow chart of generating a quality degradation environment coupling parameter set for a fresh produce quality degradation prediction method based on multimodal dynamic coupling proposed in this application;

[0041] Figure 4 This is a flow chart of generating a degradation constraint distribution matrix for a method for predicting fresh produce quality degradation based on multimodal dynamic coupling proposed in this application;

[0042] Figure 5 This is a flow chart of obtaining a four-level quality label for a method for predicting fresh produce quality deterioration based on multimodal dynamic coupling proposed in this application;

[0043] Figure 6 A schematic diagram of the structure of a fresh produce quality deterioration prediction system based on multimodal dynamic coupling provided in an embodiment of the present application; DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the simulation technology route in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] The features and performance of the present invention are further described in detail below with reference to the embodiments. Figure 1 As shown, a method for predicting fresh food quality deterioration based on multimodal dynamic coupling includes the following steps:

[0047] S1. Obtain data to be processed, including sensory layer data, environmental layer data, and operational layer data. The sensory layer data includes visible light images of fresh produce surfaces, near-infrared reflectance spectra, and acoustic resonance frequency response curves. The environmental layer data includes a temperature and humidity gradient matrix of the microenvironment within the cargo box, a heat map of the spatial distribution of stacking pressure, and a vibration spectrum of the transportation path. The operational layer data includes time and space stamp records of loading and unloading events and time series data on the opening and closing frequency of cold storage doors.

[0048] In some embodiments, sensory layer data collection involves data from multiple different modalities. The collection of visible light images of the surface of fresh produce is achieved using high-resolution visible light imaging equipment to obtain color information of the product surface, which reflects the freshness and color changes of the product surface. The collection of near-infrared reflectance spectra is achieved using a near-infrared spectrometer. This instrument emits near-infrared light within a specific wavelength range and measures the intensity changes of the light reflected from the surface of fresh produce, thereby obtaining information about the internal chemical bonds of the product. In this embodiment, this is the vibration absorption information of the hydrogen-oxygen bonds of water molecules and the bonds between proteins and fats. The collection of acoustic wave resonance frequency response curves is achieved using an acoustic excitation-response sensing system, which uses acoustic waves within a specific frequency range to excite the surface of the product and detect changes in the product's own resonance response characteristics to reflect the integrity and degree of deterioration of its internal tissue structure.

[0049] In some embodiments, the collection of environmental layer data includes obtaining a temperature and humidity gradient matrix of the microenvironment inside the cargo box. The temperature and humidity gradient matrix reflects the temperature and humidity differences at different spatial locations inside the cargo box, and is collected in real time in a matrix manner through a temperature and humidity sensor array. In addition, the collection of a heat map of the spatial distribution of stacking pressure is achieved through a distributed pressure sensor array. The pressure sensor array measures and records the spatial distribution information of pressure caused by stacking between goods during transportation, and expresses the distribution characteristics of stacking pressure in the cargo space in the form of a heat map. In this embodiment, the collection of the vibration spectrum of the transportation path is completed by a high-sensitivity three-axis vibration sensor. The sensor detects and records the frequency distribution of vibration signals caused by road bumps or vehicle movement during transportation, thereby obtaining vibration spectrum data for subsequent evaluation of the disturbance effect of vibration on the quality deterioration process of the goods.

[0050] In some embodiments, the operational layer data includes time-space stamp records of loading and unloading events and time series data of the frequency of opening and closing of the cold storage door. The time-space stamp records of loading and unloading events are realized by installing a wireless radio frequency identification device and a time stamp recording device in the transport cargo box. The recording device automatically captures the time and space information of the time and location of the loading and unloading events to form clear time-space stamp data of the loading and unloading events. In some embodiments, the time series data of the opening and closing frequency of the cold storage door is collected by a magnetic induction or photoelectric sensor device installed on the cold storage door. The sensor records each opening and closing action of the cold storage door in real time and forms accurate time series data for evaluating the time series response characteristics of the temperature and humidity disturbances caused by the opening and closing of the cold storage door and its impact on the deterioration of the quality of the goods. In this embodiment, through the above-mentioned data collection process, the data set obtained is complete and clear, which provides an accurate and reliable data basis for subsequent multimodal fusion and dynamic coupling analysis, thereby laying the foundation for the refined prediction of quality deterioration.

[0051] S2, extracting and fusing quality degradation features of the sensory layer data to obtain a quality degradation feature tensor;

[0052] For details, please refer to the attached Figure 2 As shown, this step includes the following sub-steps:

[0053] S201, performing color difference quantization processing on the surface visible light image in the sensory layer data to obtain a surface color difference heat map, wherein the color difference quantization processing is to extract surface color abnormal areas through color space transformation;

[0054] In this embodiment, the core of the color difference quantization processing is to convert the original visible light image from the RGB color space to the Lab color space through a color space transformation method to achieve quantization of color information and efficient extraction of abnormal areas.

[0055] In this embodiment, after the original image data is acquired, the RGB color model is converted into the Lab color model using color space transformation with pixels as the basic unit. Specifically, let the RGB color vector of any pixel in the original image be (R, G, B), and convert it into a vector (L, a, b) of the Lab color space through nonlinear transformation. The specific mathematical expression of color space transformation is: X = 0.412453R + 0.57580G + 0.180423B; Y = 0.212671R + 0.715160G + 0.072169B; Z = 0.019334R + 0.119193G + 0.950227B, and the expression for converting the above XYZ space coordinates into Lab space is: L = 116f(Y / Y n )-16; a=500[f(X / X n )-f(Y / Y n)]; b = 200[f(Y / Y n )-f(Z / Z n )], where the reference white point is (X n ,Y n ,Z n ), the function f(t) is defined as follows: The constant In some embodiments, after converting the original RGB color space to Lab space through the above transformation, each pixel in the image is represented by a three-dimensional vector (L, a, b). The advantage of Lab space is that it more accurately reflects the human eye's perception of color differences, thereby facilitating color difference quantification.

[0056] In this embodiment, the color difference in Lab space is further quantified and the Euclidean distance between the chromaticity component of each pixel and the Lab vector of the reference color is used as the color difference metric. Specifically, the reference color (L0, a0, b0) is determined by the color vector of the normal area on the surface of the product. The color vector of any pixel (L i ,a i ,b i ) and the reference color is expressed as follows: Through the above calculation, in the process of color difference quantization, each pixel gets a corresponding color difference value ΔE i , and forming a two-dimensional spatial distribution matrix based on the color difference values. In some embodiments, the two-dimensional matrix is normalized and represented using heat map visualization technology to obtain a color difference heat map that intuitively expresses the spatial distribution of color abnormal areas on the surface of fresh produce.

[0057] In some embodiments, the surface color difference heat map obtained is spatial two-dimensional matrix data, in which the numerical value represents the degree of color abnormality, that is, the higher the numerical value, the more obvious the abnormality of the area relative to the baseline color, thereby providing a basic basis for the spatial-chemical feature fusion in subsequent steps.

[0058] S202, performing chemical bond feature extraction processing on the near-infrared reflectance spectrum in the sensory layer data to obtain a water loss feature vector, wherein the chemical bond feature extraction processing is performed by screening characteristic bands related to quality degradation through a PLS regression model;

[0059] In this embodiment, the near-infrared reflectance spectrum data is measured by a near-infrared spectrometer with a wavelength range of 780 to 2500 nm. The data is represented in the form of a one-dimensional numerical array of spectral reflectance intensity and corresponding wavelength. The absorption peak of each band corresponds to the vibration absorption signal of a specific chemical bond inside the product, including a characteristic band closely related to the OH chemical bond of water molecules.

[0060] In some embodiments, a partial least squares (PLS) regression model is applied to near-infrared reflectance spectral data to screen and extract characteristic bands. Specifically, the raw near-infrared spectral data is represented as a matrix X of size m × n, where m represents the number of spectral samples and n represents the number of wavelengths. Each row represents spectral data for a single sample, and each column corresponds to the reflectance intensity at a specific wavelength. A quality degradation indicator related to the degree of water loss is used as the model response variable, represented as Y, to form a corresponding response variable vector.

[0061] In this embodiment, the partial least squares regression method is used to construct a linear mapping between the spectral data matrix X and the response variable vector Y. By maximizing the covariance between the characteristic variables and the response variables, the screening and regression modeling of the characteristic bands are achieved. The specific mathematical modeling expression of PLS regression is as follows: Given a spectral data matrix X∈R n×m (where n is the number of samples and m is the number of spectral bands), the response variable vector Y∈R n×1 , a PLS regression model is established to achieve characteristic band selection and correlation coefficient calculation. The model aims to maximize the covariance between spectral data and response variables at the same time, extracting the latent variable matrix T and the loading matrix P. The specific mathematical expression is: X = TP T +E,Y=UQ T +F, where T and U are the extracted principal component score matrices, P and Q are the corresponding load matrices, and E and F are the residual matrices. By iterative solution, the principal component matrices T and U meet the maximum covariance condition, that is, the solution: Where ω and c are weight vectors, respectively used to generate the principal component score vector t = X ω 、u=Y c The weight vector is determined through iterative optimization to achieve dimensionality reduction of the original data and extract the set of characteristic bands with the greatest correlation with the quality degradation response variable.

[0062] In some embodiments, by performing a significance analysis on the regression coefficient of each band, the bands that contribute significantly to the response variable Y are selected and recorded as the chemical bond characteristic band set. After reconstructing the near-infrared spectral feature space using the selected characteristic bands, PLS regression modeling is performed again to obtain the score vector T of the water loss feature of quality deterioration. * , the score vector is the chemical bond feature vector to be extracted, which represents the quality change information caused by water loss in fresh products. In this embodiment, the water loss feature vector T * It reflects the changing characteristics of specific chemical bonds in fresh products under the state of water loss, and is one of the input data required for the spatial-chemical feature fusion processing in the subsequent steps.

[0063] S203, performing spatial chemical fusion processing on the surface color difference heat map and the water loss feature vector to obtain a quality degradation feature tensor, wherein the spatial chemical fusion processing maps chemical features to color difference abnormality areas through a spatial weighted matrix.

[0064] In this embodiment, the fusion processing is to project the water loss feature vector extracted from the near-infrared spectral data into the spatial anomaly area represented by the surface color difference heat map using spatial mapping, so as to realize the joint expression of quality degradation characteristics at the spatial and chemical information levels.

[0065] In this embodiment, the fusion process involves constructing a spatial weighted matrix to achieve the mapping of chemical features to abnormal surface color difference areas. In the specific implementation process, the surface color difference heat map is defined as a two-dimensional spatial matrix C∈R M×N , where each matrix element C(i,j) represents the abnormal intensity of color difference at position (i,j) in the image; at the same time, the water loss feature vector obtained by PLS regression is defined as T * ∈R d×1 , where d represents the feature dimension determined by PLS regression.

[0066] In some embodiments, to achieve spatial-chemical feature mapping, a spatial weighting matrix W∈R is defined. M ×N×d , where each element of the matrix represents the mapping strength between a specific spatial location (M×N represents the spatial resolution of the surface color difference heat map) and each chemical feature dimension. In this embodiment, the spatial weighting matrix is constructed by normalizing each pixel of the surface color difference heat map to generate a weight factor, which is recorded as: where w ij Indicates the color difference abnormality weight factor corresponding to the pixel at spatial position (i, j), ΔE ij is the quantized value of the color difference of pixel (i, j). The normalized weight factor reflects the contribution ratio of each pixel to the overall surface color difference abnormality.

[0067] In some embodiments, based on the above weight factors, the chemical feature vector T * Mapped to the spatial position of the color difference heat map, the spatial chemical fusion feature tensor F is constructed. The specific expression of the fusion feature tensor is: Where H and W represent the height and width of the surface color difference heat map respectively. represents the value of the k-th dimension feature in the water loss feature vector T*, and F(i,j,k) represents the fusion value of the k-th dimension feature at position (i,j). Through this fusion process, the quality degradation feature tensor F∈R H×W×dThat is, the precise spatial mapping expression between the chemical bond characteristics and the abnormal surface color difference areas is achieved.

[0068] In this embodiment, the quality degradation feature tensor obtained through the above fusion processing clearly reflects the spatial correlation between the abnormal surface color area and the internal chemical characteristics of water loss, providing the necessary multimodal input data for the next stage of dynamic coupling modeling of quality degradation.

[0069] S3, performing dynamic coupling modeling processing on the quality degradation feature tensor and the environmental layer data, thereby obtaining a quality degradation environment coupling parameter set;

[0070] For details, please refer to the attached Figure 3 As shown, this step includes the following sub-steps:

[0071] S301, performing diffusion path topology modeling processing on the quality degradation feature tensor and the microenvironment temperature and humidity gradient matrix in the environmental layer data to obtain a degradation diffusion rate field, wherein the diffusion path topology modeling processing simulates the propagation path of degradation along the direction of the maximum temperature and humidity gradient through a graph network model;

[0072] In some embodiments, the diffusion path topology modeling process is implemented through a graph network model. Specifically, a graph network method is used to simulate the spatial diffusion propagation process of the quality deterioration characteristics of fresh goods along the direction of the maximum temperature and humidity gradient of the microenvironment, so as to clarify the dynamic propagation trend and diffusion rate of the quality deterioration of goods.

[0073] In this embodiment, the quality degradation feature tensor is recorded as a three-dimensional tensor F∈R H×W×d , where H and W represent the height and width of the spatial dimension respectively, and d represents the dimension of the quality degradation feature. The microenvironment temperature and humidity gradient matrix is represented as a two-dimensional matrix G∈R H×W , each element in the matrix represents the comprehensive gradient amplitude of temperature and humidity at the corresponding position in the cargo box. The amplitude calculation formula is: Where G(i,j) represents the comprehensive amplitude of temperature and humidity gradient at position (i,j), T represents the temperature field, and H represents the humidity field. are the directional gradient values of the temperature and humidity fields at the spatial position (i, j).

[0074] In some embodiments, to determine the diffusion path of quality degradation characteristics along the direction of the maximum temperature and humidity gradient, this embodiment introduces a graph network model to perform topological modeling on the above spatial data. In this topological model, each spatial location (i, j) is considered a node in the graph network. The topological connection between nodes is determined by the directionality of the temperature and humidity gradient. The connection weight between two adjacent nodes (i, j) and (p, q) is determined by the difference in temperature and humidity gradients. The weight calculation formula is expressed as: The weight W(i,j),(p,q) It represents the diffusion connection strength between node (i, j) and node (p, q). The greater the difference in temperature and humidity gradients between nodes, the greater the connection weight, and vice versa.

[0075] In some embodiments, the degradation diffusion rate field is modeled based on the above-mentioned graph network structure, and the kinetic equation of degradation diffusion is further defined. In this embodiment, the diffusion process is mathematically modeled using a discrete diffusion rate equation, specifically expressed as: Where V(i,j) represents the degradation diffusion rate at the spatial position (i,j), N(i,j) represents the set of neighboring nodes of node (i,j), and W (i,j),(p,q) is the connection weight between nodes, ||F(i,j,:)-F(p,q,:)||2 is the Euclidean distance of the quality degradation feature tensor between node (i,j) and node (p,q), which reflects the degree of difference in quality degradation features between spatial locations.

[0076] In this embodiment, after the degradation diffusion rate of each spatial position is calculated by the above model, a complete degradation diffusion rate field matrix V∈R is formed. H×W This rate field clearly characterizes the diffusion path and propagation rate of quality degradation along the direction of maximum temperature and humidity gradient, serving as the basic input information for mechanical path coupling and frequency domain stability analysis in subsequent steps.

[0077] S302, performing mechanical path coupling processing on the degradation diffusion rate field and the stack pressure spatial distribution thermodynamic map in the environmental layer data to obtain a pressure path correlation tensor, wherein the mechanical path coupling processing calculates the local degradation acceleration effect through geometric superposition of the pressure distribution and the diffusion path;

[0078] In some embodiments, the mechanical path coupling process is to calculate the local degradation acceleration effect by geometric superposition of pressure distribution and diffusion path, including local stress coefficient extraction process and pressure diffusion gain fusion process.

[0079] In this embodiment, the mechanical path coupling processing is based on the mutual influence between the local pressure distribution generated by fresh goods under stacking conditions and the quality deterioration diffusion path. By geometrically superposing and numerically calculating the two, the accelerating effect of local pressure on the deterioration diffusion rate is quantified, and then a three-dimensional tensor representation that comprehensively reflects the coupling state of the two is constructed.

[0080] In this embodiment, the degradation diffusion rate field is obtained from the previous step S301 and is recorded as the matrix V∈R H×W , where V(i,j) represents the quality degradation diffusion rate at position (i,j). The stack pressure spatial distribution heat map is recorded as matrix P∈R H×W, where P(i,j) represents the local pressure at position (i,j). Mechanical path coupling superimposes the distribution characteristics of the pressure field and the diffusion field to identify potential accelerated degradation trends in high-pressure regions and form a three-dimensional tensor reflecting the coupling relationship between the two.

[0081] In some embodiments, the local stress coefficient extraction process performs a normalization operation on the stack pressure matrix P, and maps the original pressure value into a dimensionless local stress coefficient, which is recorded as the matrix P * ∈R H×W The normalized pressure coefficient matrix P * (i, j) is distributed in the interval [0,1], reflecting the relative strength of different positions relative to the overall pressure distribution.

[0082] In some embodiments, the normalized pressure matrix P is transformed into * It is superimposed with the degradation diffusion rate field V according to the corresponding position of the element to obtain the mechanical path coupling gain field M∈R H×W The gain field is used to quantify the accelerating effect of pressure on degradation diffusion, and the expression is: M(i,j)=V(i,j)×(1+α·α·P * (i, j)), where α is the coupling gain coefficient, which is used to control the degree to which local pressure amplifies the diffusion rate, and M(i, j) represents the degradation diffusion rate after pressure superposition at position ((i, j). The coefficient α is selected based on the actual characteristics of fresh produce and compression deformation, and is determined through experimental data regression or parameter identification algorithm.

[0083] In some embodiments, in order to represent the coupling state of diffusion path and pressure distribution in three-dimensional space, this embodiment converts the degraded diffusion rate field V, the normalized pressure matrix P * and the fusion gain field M are constructed together to form the pressure path correlation tensor R∈R H×W×3 , its three-dimensional vector is expressed as follows: R(i,j,:)=[V(i,j),P * (i,j),M(i,j)], where the first dimension of R(i,j,:) is the original diffusion rate, the second dimension is the local pressure coefficient, and the third dimension is the mechanical coupling gain rate. The combined representation of the three reflects the degree of interaction between the mechanical pressure and quality degradation diffusion at each position.

[0084] In this embodiment, the obtained pressure path correlation tensor will be used as the input of the frequency domain stability analysis in the subsequent steps to further evaluate the disturbance effect of the transportation path vibration on the degradation diffusion in the high-pressure area, and provide more accurate multi-dimensional data support for the overall quality degradation environment coupling model.

[0085] S303, performing frequency domain stability analysis on the pressure path correlation tensor and the transport path vibration spectrum in the environmental layer data to obtain a quality degradation environment coupling parameter set. The frequency domain stability analysis quantifies the disturbance intensity of the vibration on the diffusion path through the frequency band energy distribution.

[0086] In some embodiments, the frequency domain stability analysis process is to quantify the disturbance intensity of vibration on the diffusion path through frequency band energy distribution, including vibration energy spectrum extraction process and frequency band coupling factor calculation process.

[0087] In this embodiment, the frequency domain stability analysis and processing is based on the vibration signal of the transportation path collected by the high-sensitivity vibration sensor. By quantitatively calculating the energy distribution characteristics of the vibration signal in different frequency bands, the degree of disturbance effect of the vibration on the quality degradation diffusion path is evaluated, thereby determining the frequency band energy coupling relationship related to the degradation process.

[0088] In this embodiment, the transport path vibration signal is recorded as x(t), and the sampling frequency is defined as f s , the recording time is T. Perform discrete Fourier transform (DFT) on x(t) to obtain the expression of the vibration signal in the frequency domain X(ω k ), where ω k Represents the kth discrete angular frequency component, k = 0, 1, ..., K-1, K represents the number of discrete points. The mathematical expression of DFT is: Where Δt=1 / f s , N is the total number of sampling points, By X(ω k ) is used to perform modular square operation to obtain the power spectrum distribution of the vibration signal: S(ω k )=|X(ω k )| 2 .

[0089] In some embodiments, the vibration energy spectrum extraction process includes S(ω k ) Perform segmented integration or regional summation within multiple frequency bands to calculate the energy value of the corresponding frequency band and form a multi-dimensional frequency band energy vector. Define the frequency band division interval as {Ω1,Ω2,...,Ω B}, where B represents the number of frequency bands, and each frequency band Corresponding to a certain angular frequency range, the energy value of the vibration signal in this frequency band is expressed as

[0090] In this embodiment, the pressure path correlation tensor is generated by step S302 and is recorded as R(i,j,:)=[V(i,j),P * (i,j),M(i,j)], where V(i,j) is the degradation diffusion rate, P *(i, j) is the local pressure coefficient, M(i, j) is the mechanical coupling gain rate. In order to quantify the disturbance effect of vibration on the local degradation path, the frequency band coupling factor calculation process is introduced, and the segmented energy vector {E1, E2, ..., E B} is coupled with the pressure path tensor. Define a set of weight coefficients {β1,β2,...,β B} represents the degree of disturbance influence in each frequency band, and the vibration coupling disturbance factor at position (i, j) is recorded as where β b The value of is based on the actual vibration characteristics and the sensitivity of the goods to vibrations in different frequency bands. M(i,j) is the mechanical coupling gain rate, which represents the local degradation gain value formed at position (i,j) by the combined action of stacking pressure and diffusion rate.

[0091] In some embodiments, the vibration coupling perturbation factor D(i, j) is calculated point by point for the entire spatial position ((i, j)), and the vibration perturbation distribution matrix D∈R is generated on the two-dimensional plane. H×W The disturbance distribution matrix is then integrated with the existing pressure path correlation tensor. Specifically, the vibration disturbance distribution matrix is expanded to make its dimension consistent with the pressure path correlation tensor, forming a new three-dimensional matrix D′∈R H×W×1 , and then through the tensor connection operation, the expanded matrix D′ is spliced with the pressure path association tensor in the third direction to form the quality degradation environment coupling parameter set tensor E∈R H×W×4 The specific mathematical expression of the above integration operation is expressed as: E(i,j,:) = Concat[R(i,j,:),D′(i,j)], i = 1,...,H, j = 1,...,W, where the tensor concatenation result E∈R H×W×4 , clearly expressing the interaction between the four-dimensional characteristics of degradation diffusion rate, pressure coefficient, mechanical coupling gain rate, and vibration disturbance intensity at the spatial position (i, j). In this embodiment, the perturbation effect of multi-band vibration energy on the diffusion path determined by the above-mentioned frequency domain stability analysis provides quantifiable frequency domain perturbation parameter support for the subsequent multi-dimensional dynamic constraint field modeling of cold chain operation layer data, thereby making the global coupling model of quality degradation more complete and accurate at the spatial, mechanical, and vibration levels.

[0092] S4, performing multi-dimensional dynamic constraint field modeling processing on the quality degradation environment coupling parameter set and the operation layer data, thereby obtaining a degradation constraint distribution matrix;

[0093] For details, please refer to the attached Figure 4 As shown, this step includes the following sub-steps:

[0094] S401, performing pulse sequence trajectory tensor processing on the spatiotemporal stamp records of loading and unloading events in the operation layer data to obtain a loading and unloading pulse trajectory tensor, wherein the trajectory tensor processing is to reconstruct the loading and unloading energy impact distribution through the spatiotemporal trajectory of the pulse event;

[0095] In some embodiments, the trajectory tensorization process is to reconstruct the loading and unloading energy impact distribution through the spatiotemporal trajectory of the pulse event, including spatiotemporal gridding mapping process and pulse intensity normalization process.

[0096] In this embodiment, the spatiotemporal gridding mapping process generates a three-dimensional voxel grid by discretizing the spatiotemporal domain of loading and unloading events. The original data form of the spatiotemporal stamp record of loading and unloading events is a discrete event set where t k is the time when the kth loading and unloading event occurs, (x k ,y k ) is a two-dimensional space coordinate. In this embodiment, the spatiotemporal gridding mapping is defined as dividing the continuous time axis into N t The space plane is divided into N equally spaced intervals. x ×N y equal-sized grid cells to form a three-dimensional space-time voxel grid Each voxel ((i,j,m) corresponds to the time interval [t i ,t i+1 ) and spatial grid unit (x j ,x j+1 )×(y m ,y m+1 ), which is used to count the frequency and intensity of loading and unloading events within this time and space unit.

[0097] In this embodiment, the pulse intensity normalization process realizes the energy quantification of the loading and unloading event through the time decay function and the spatial weight factor. The time decay function φ(t) is defined as an exponential decay form: Where λ is the attenuation coefficient, t min With t max are the minimum and maximum timestamps of the loading and unloading events, respectively. The spatial weight factor ψ(x,y) is calculated based on the kinematic model of the loading and unloading manipulator, and is expressed as: Where (x c ,y c ) is the coordinate of the operation center of the loading and unloading robot arm, and σ is the spatial diffusion coefficient, which represents the propagation range of the loading and unloading energy in space.

[0098] In some embodiments, the spatiotemporal trajectory of the impulse event is reconstructed by mapping the discrete events to a spatiotemporal voxel grid using a convolution operation. k ,x k ,y k), its energy contribution in the space-time grid is E k (i, j, m) is calculated as: E k (i,j,m)=φ(t k -t i )·ψ(x k -x j ,y k -y m ), where t i is the starting time of the ith interval of the time grid, (x j ,y m ) is the center coordinate of the spatial grid cell. The initial energy distribution matrix is generated by traversing all loading and unloading events and accumulating their energy contributions. Its elements

[0099] In this embodiment, the generation of the loading and unloading pulse trajectory tensor further includes time smoothing and spatial interpolation. Time smoothing is achieved by a sliding average filter with a filter window length of W. t , the smoothed energy distribution matrix ε * Calculated as: Spatial interpolation uses a bicubic interpolation algorithm to expand the discrete grid energy distribution into a continuous space field. The tensor after interpolation Where H and W are the spatial resolutions after interpolation.

[0100] In some embodiments, the final tensorization of the loading and unloading energy impact distribution is completed by multi-scale feature fusion. where K s is a Gaussian kernel of size s×s, which is used to extract the energy concentration characteristics at different spatial scales. The fused loading and unloading pulse trajectory tensor Calculated as: Where s is the scale parameter, is the Gaussian kernel function, σ s Proportional to the scale s.

[0101] In this embodiment, through the above-mentioned spatiotemporal grid mapping processing and pulse intensity normalization processing, the loading and unloading pulse trajectory tensor completely characterizes the energy impact distribution of the loading and unloading events in the spatiotemporal dimension. Its four-dimensional structure (time, spatial height, spatial width, and scale) provides high-resolution input data for subsequent multi-dimensional dynamic constraint field modeling, realizing the quantitative expression of the disturbance effect of loading and unloading operations on the quality deterioration of fresh goods.

[0102] S402: Performing temperature and humidity disturbance evolution modeling on the cold storage door opening and closing frequency time series data in the operation layer data to obtain a dynamic distribution of cold chain stability. The evolution modeling process is to construct a cold chain disturbance dynamic response model by coupling the time series disturbance curve with the humidity gradient;

[0103] In some embodiments, the evolutionary modeling process is to construct a cold chain disturbance dynamic response model by coupling the time series disturbance curve with the humidity gradient, including time series disturbance curve decomposition processing and humidity gradient coupling processing.

[0104] In this embodiment, the preprocessing of the cold storage door opening and closing frequency time series data includes event frequency accumulation and time series interpolation processing. The raw data of the cold storage door opening and closing events are collected by magnetic induction or photoelectric sensors and recorded in the form of discrete time stamp sequences. where t n Indicates the time when the nth cold storage door opening and closing action occurs. In this embodiment, the event frequency is accumulated by counting the number of opening and closing times per unit time through a sliding time window. The time window length is defined as ΔT, and the time window sliding step is δT. The cumulative frequency sequence F(t) is calculated as: Where Ι(·) is the characteristic function, when t n The value is 1 when it is within the time window [t-ΔT, t), otherwise it is 0. The time series interpolation process converts the discrete frequency sequence F(t) into a continuous time function F through the cubic spline interpolation algorithm. * (t)∈C 2 (R), ensuring the smoothness and differentiability of the timing perturbation curve.

[0105] In this embodiment, the time series disturbance curve decomposition process extracts the frequency domain energy distribution characteristics through Fourier transform. The cold storage door opening and closing disturbance signal is defined as the interpolated frequency function F * (t), its continuous Fourier transform is: Where ω is the angular frequency, The power spectral density S(ω) is calculated as: S(ω)=|F(ω)| 2 , by setting the frequency band division threshold {ω1,ω2,...ω B The power spectrum is divided into B frequency bands, and the energy integral value of each frequency band is E b for: Form the frequency domain energy vector E=[E1,E2,...,E B ] T , characterizing the energy distribution characteristics of the cold storage door opening and closing disturbance in different frequency bands.

[0106] In some embodiments, the humidity gradient coupling process spatially and temporally associates the frequency domain energy with the humidity gradient field through a dynamic response function. The humidity gradient matrix H(x, y, t) in the environmental layer data is collected by the temperature and humidity sensor array, and its spatial gradient field Indicates the spatial change rate of humidity in the cargo box. In this embodiment, the dynamic response function R(x, y, t) is defined as the convolution operation of the humidity gradient field and the time series disturbance energy: where βb is the coupling coefficient of the b-th frequency band, g b (t) is the impulse response function of the corresponding frequency band, and its expression is: σ b The center frequency of the band Inversely proportional to the delay parameter, satisfying κ is the proportional constant. The convolution operation* represents the weighted integral of the humidity gradient field and the impulse response function in the time dimension. The mathematical form is: In this embodiment, the generation of the dynamic distribution of cold chain stability is achieved by superimposing the normalized dynamic response field and the humidity gradient perturbation. The stability distribution matrix S(x, y, t) is defined as: Where γ is the attenuation factor, which is used to adjust the impact of disturbance on stability. is the Euclidean norm of the humidity gradient field. Through the above calculation, the value range of S(x, y, t) is [0, 1]. The closer the value is to 1, the higher the cold chain stability at that location at time t is. Otherwise, it indicates that the location is significantly affected by the disturbance.

[0107] In some embodiments, the spatiotemporal discretization process maps the continuous stability distribution S(x, y, t) to a discrete grid. Define the time resolution as Δt and the spatial resolution as Δx×Δy. The discretized stability dynamic distribution matrix S∈R T ×X×Y The calculation is S(i,j,m)=S(s j ,y m ,t i ), where t i =iΔt,x j =jΔx,y m =mΔy,i=1,2,...,T,j=1,2,...,X,m=1,2,...,Y. In this embodiment, through the above-mentioned time-series perturbation curve decomposition and humidity gradient coupling processing, the cold chain stability dynamic distribution matrix S fully characterizes the perturbation effect of the cold storage door opening and closing operation on the humidity field inside the cargo box and its spatiotemporal evolution law. It provides a quantitative stability indicator for subsequent multidimensional dynamic constraint field modeling and supports the global coupling analysis of the fresh food quality deterioration prediction model.

[0108] S403, performing multi-dimensional dynamic constraint mapping processing on the loading and unloading pulse trajectory tensor, the dynamic distribution of cold chain stability and the quality degradation environment coupling parameter set, so as to obtain the degradation constraint distribution matrix. The mapping processing forms a constraint balance field through space-time-mechanical multivariate weight calculation.

[0109] In some embodiments, the mapping process forms a constrained balance field through space-time-mechanical multivariate weight calculation, including space-time dimension alignment processing and multivariate weight fusion processing.

[0110] In this embodiment, the space-time dimension alignment process unifies the space-time resolution of the input data through interpolation operation. H×W×4 The spatial dimension of is H×W, and the time dimension is implicit in the dynamic evolution of the environmental parameters; the loading and unloading pulse trajectory tensor T * ∈R Nt×H×W×S The time dimension is N t , the spatial dimension is H×W; the cold chain stability dynamic distribution matrix S∈R T×X×Y The time dimension of is T, and the space dimension is X×Y. In this embodiment, the spatial resolution of S is adjusted from X×Y to H×W by bilinear interpolation algorithm, and the time dimension is adjusted from N to N by linear interpolation. t Align with T to the common time axis length L to generate the aligned stability distribution tensor S * ∈R L×H×W .

[0111] In this embodiment, the multivariate weight fusion process realizes the construction of the constraint balance field by defining the spatial weight, time weight and mechanical weight. space ∈RH×W is calculated based on the local pressure coefficient and vibration disturbance intensity in the environmental coupling parameter set E, and the expression is: Where E(i,j,2) is the local pressure coefficient at position (i,j), E(i,j,4) is the vibration disturbance intensity, ⊙ represents element-by-element multiplication, and the denominator is the global maximum normalization factor. Time weight W time ∈RL based on the loading and unloading pulse trajectory tensor T * The time energy accumulation value is calculated as follows: Mechanical weight W mech ∈R H×W Based on the calculation of the degradation diffusion rate and mechanical coupling gain rate in the environmental coupling parameter set E, the expression is: Where α is the mechanical gain coefficient, E(i,j,1) is the degradation diffusion rate, and E(i,j,3) is the mechanical coupling gain rate. In some embodiments, the generation of the constraint balance field is achieved by a tensor product operation of space-time-mechanical weights. Define the constraint balance field C∈R L×H×W = C(l,i,j) = W space (i,j)·W time (l)·W mech (i,j)·S * (l,i,j), where S * (l,i,j) represents the aligned cold chain stability distribution. By element-by-element multiplication, the constrained equilibrium field comprehensively characterizes the multidimensional coupling effects of spatial pressure-vibration disturbance, temporal loading and unloading energy impact, mechanical diffusion gain, and cold chain stability.

[0112] In this embodiment, the generation of the degradation constraint distribution matrix is completed by the time-space integration and normalization of the constraint balance field. H×W for: Where η is the attenuation coefficient, which is used to suppress the oversaturation effect in areas with high cumulative constraints. Normalization ensures that the matrix D ranges from [0 to 1]. Values closer to 1 indicate a higher degree of suppression of quality degradation at that spatial location due to the multidimensional constraint effect, while values lower than 1 indicate a weaker constraint effect.

[0113] In some embodiments, post-processing optimization smoothes local mutations in the constraint distribution matrix by anisotropic diffusion filtering. The filtering equation uses the Perona-Malik model: Where τ is the pseudo-time parameter, is the edge preservation diffusion function, K is the gradient threshold. The finite difference method is used to iteratively solve until convergence, and the smoothed degradation constraint distribution matrix D is obtained. * ∈R H×W .

[0114] In this embodiment, through the above-mentioned space-time dimension alignment processing and multivariate weight fusion processing, the degradation constraint distribution matrix D * The synergistic constraints of loading and unloading operations, cold chain stability, and environmental coupling parameters on the quality degradation of fresh goods are fully quantified, providing spatially resolved constraint strength input for the final adaptive degradation trajectory evolution processing and supporting the precise division of four-level quality labels.

[0115] S5, adaptive degradation trajectory evolution processing is performed on the acoustic resonance frequency response curve and degradation constraint distribution matrix in the sensory layer data to obtain a four-level quality label;

[0116] For details, please refer to the attached Figure 5 As shown, this step includes the following sub-steps:

[0117] S501, performing a resonance modal coherence analysis on the acoustic wave resonance frequency response curve in the sensory layer data to obtain a spectrum coherence drift factor, wherein the coherence analysis is to calculate the degree of acoustic wave mode shift through the temporal coherence of the resonance frequency band distribution;

[0118] In some embodiments, the coherence analysis process is to calculate the degree of acoustic wave mode shift through the temporal coherence of the resonance frequency band distribution, which includes a frequency band segmentation process and a temporal coherence calculation process.

[0119] In this embodiment, the frequency band segmentation process extracts the multi-scale frequency band features of the acoustic resonance frequency response curve through continuous wavelet transform. The acoustic resonance frequency response curve is collected by the acoustic excitation-response sensing system, and the original data is in the form of time series. Where f(t) represents the resonant frequency amplitude at time point t. In this embodiment, the continuous wavelet transform is defined as: Where a is the scale parameter, b is the translation parameter, and ψ is the Morlet wavelet basis function, and the expression is: ω0 is the center frequency of the wavelet. By setting the scale parameter set Extract the resonant frequency bands related to the internal structure of fresh produce Time-varying energy distribution E in each frequency band m (t) is calculated as: E m (t)=|W(a m ,t)| 2 In this embodiment, the temporal coherence calculation process quantifies the phase consistency between frequency bands through sliding window cross-correlation analysis. The time window length is defined as W, the sliding step is δ, and the coherence coefficient ρ between the mth frequency band and the nth frequency band in the window is defined as mn (k) is: in is the energy mean of the mth frequency band in the window, and k is the starting time point of the window. By traversing all time windows and frequency band pairs, the full-band coherence tensor P∈R is generated. K×M×M , where K is the total number of windows P(k,m,n)=ρ mn (k). In some embodiments, the quantification of the degree of acoustic mode deviation is achieved by principal component analysis dimensionality reduction and drift factor calculation. The coherence matrix P(k,:,:) of each time window is subjected to eigenvalue decomposition, and the first P principal component vectors are extracted. Construct the time-varying principal component trajectory matrix V∈R K×P×M The spectral coherence drift factor Δ(k) is defined as the Euclidean distance between the principal component trajectories of adjacent windows: In this embodiment, the drift factor normalization process is performed to achieve global comparability through exponential smoothing and range scaling. * (k) is calculated as: Where ξ is the attenuation coefficient, which is used to suppress the accumulated drift noise at the end of the time series.

[0120] In some embodiments, post-processing optimization smoothes the temporal mutation of the drift factor by Kalman filtering. The state equation and observation equation are defined as: x(k) = x(k-1) + ω(k), z(k) = x(k) + v(k), where x(k) is the hidden state and z(k) = Δ *(k) is the observed value, w(k)~N(0,Q) and v(k)~N(0,R) are process noise and observation noise. Through recursive prediction and update steps, the smoothed spectrum coherence drift factor sequence Δ is obtained. ** (k).

[0121] In this embodiment, through the above-mentioned frequency band division process and time series coherence calculation process, the spectrum coherence drift factor Δ ** (k) The degree of deviation of the acoustic resonance mode in the time dimension is accurately quantified. The larger the value, the more significant the acoustic mode instability effect caused by the deterioration of the internal structure of fresh goods. This provides frequency domain dynamic feature input for subsequent nonlinear degradation trajectory mapping processing.

[0122] S502, nonlinear degradation trajectory mapping processing is performed on the spectral coherence drift factor and the degradation constraint distribution matrix to obtain four-level quality labels. The trajectory mapping processing is to calculate the quality evolution critical point through multi-dimensional degradation space projection. The four-level quality labels include compliance label, warning label, abnormal label, and irreversible label.

[0123] In some embodiments, the trajectory mapping process is to calculate the critical point of quality evolution through multi-dimensional degradation space projection, including a multi-dimensional feature space construction process and a critical point determination process.

[0124] In this embodiment, the multi-dimensional feature space construction process fuses the spectrum coherence drift factor and the degradation constraint distribution matrix through tensor splicing and normalization operations. ** (k)∈R K Indicates the degree of acoustic mode shift in the time dimension, the degradation constraint distribution matrix D * ∈R H×W In this embodiment, the spatiotemporal alignment process interpolates Δ ** (k) Expand to the spatial dimension and generate the drift factor distribution matrix Δ map ∈R H×W , whose elements where δ ij (k) is the spatiotemporal correlation factor, reflecting the degradation constraint weight of position (i, j) at time k. The fused multidimensional feature tensor F∈R H×W×2 Defined as: F(i,j,1)=D * (i, j), F(i, j, 2) = Δ map (i, j), the normalization process ensures F(i, j, d)∈[0,1], d=1,2.

[0125] In this embodiment, the critical point determination process identifies the quality evolution boundary through hyperplane segmentation and density clustering. Define a two-dimensional degradation space D = [0, 1]2 , the horizontal axis is the constraint strength D * , the vertical axis is the drift intensity Δ map . Hyperplane segmentation function h(D * ,Δ map )=θ1D * +θ2Δ map -θ0, divides the space into four types of regions, where θ0, θ1, and θ2 are segmentation parameters determined by fitting historical degradation data. Density clustering further optimizes the boundaries based on the kernel density estimation function: Identify the center of high-density areas as critical points In some embodiments, the logic for generating four-level quality labels is as follows:

[0126] Compliance Label: D * ≥τ1 and Δ map ≤γ1, indicating that the constraint strength is high and the acoustic mode is stable, and the quality of the goods meets the storage standards;

[0127] Warning label: D * ∈[τ2,τ1) and Δ map ∈(γ1,γ2], indicating that the constraint strength decreases or the acoustic mode shifts slightly, triggering a quality degradation warning;

[0128] Abnormal label: D * ∈[τ3,τ2) and Δ map ∈(γ2,γ3], indicating that the constraint effect is significantly weakened or the acoustic mode is moderately unstable, and the product quality enters an abnormal state;

[0129] Irreversible label: D * <τ3 and Δ map >γ3, indicating that the constraint effect fails and the acoustic mode is severely shifted, and the quality of the goods deteriorates irreversibly.

[0130] In this embodiment, the threshold parameters {τ1, τ2, τ3} and {γ1, γ2, γ3} are determined by historical data statistics and ROC curve analysis to ensure that the classification results are consistent with the actual degradation stage. The label mapping function L(i, j) is defined as: In some embodiments, post-processing validation evaluates label classification accuracy by using confusion matrix and Kappa coefficient. and the predicted label set Calculate the precision, recall and overall Kappa consistency coefficient for each category to ensure that the model output is consistent with the actual quality test results.

[0131] In this embodiment, through the above-mentioned multidimensional feature space construction and critical point determination processing, the four-level quality label L(i,j) accurately quantifies the quality status of fresh goods at different spatial locations, providing a hierarchical decision-making basis for cold chain management: compliance labels guide normal storage, early warning labels trigger inspection intervention, abnormal labels initiate quality re-inspection, and irreversible labels execute product rejection, thereby achieving full-cycle control of deterioration risks.

[0132] Based on the description of the embodiment of the method for predicting fresh produce quality deterioration based on multimodal dynamic coupling, the embodiment of the present application also discloses a system for predicting fresh produce quality deterioration based on multimodal dynamic coupling. The system for predicting fresh produce quality deterioration based on multimodal dynamic coupling can be a computer program (including program code) that runs the method for predicting fresh produce quality deterioration based on multimodal dynamic coupling. Figure 6 As shown in the figure, the fresh produce quality deterioration prediction system based on multimodal dynamic coupling can run the following units:

[0133] Acquisition unit 110 is used to acquire data to be processed, wherein the data to be processed includes sensory layer data, environmental layer data, and operational layer data. The sensory layer data includes visible light images of fresh produce surfaces, near-infrared reflectance spectra, and acoustic resonance frequency response curves. The environmental layer data includes a temperature and humidity gradient matrix of the microenvironment within the cargo box, a heat map of the spatial distribution of stacking pressure, and a vibration spectrum of the transportation path. The operational layer data includes time and space stamp records of loading and unloading events and time series data on the opening and closing frequency of cold storage doors.

[0134] a feature fusion unit 120 for extracting and fusing quality degradation features on the sensory layer data to obtain a quality degradation feature tensor;

[0135] A coupling modeling unit 130 is configured to perform dynamic coupling modeling processing on the quality degradation feature tensor and the environmental layer data, thereby obtaining a quality degradation environment coupling parameter set;

[0136] The constraint modeling unit 140 is configured to perform multi-dimensional dynamic constraint field modeling on the quality degradation environment coupling parameter set and the operation layer data, thereby obtaining a degradation constraint distribution matrix;

[0137] The hierarchical quality label unit 150 is configured to perform adaptive degradation trajectory evolution processing on the acoustic resonance frequency response curve in the sensory layer data and the degradation constraint distribution matrix, thereby obtaining four-level quality labels.

[0138] The foregoing description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein and should not be construed as excluding other embodiments. Rather, the present invention can be used in various other combinations, modifications, and environments and can be modified within the scope of the concept described herein through the above teachings or techniques or knowledge in the relevant field. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention are intended to be protected by the appended claims.

Claims

1. A method for predicting fresh produce quality deterioration based on multimodal dynamic coupling, characterized in that: The method comprises the following steps: S1. Obtain data to be processed, including sensory layer data, environmental layer data, and operational layer data. The sensory layer data includes visible light images of fresh produce surfaces, near-infrared reflectance spectra, and acoustic resonance frequency response curves. The environmental layer data includes a temperature and humidity gradient matrix of the microenvironment within the cargo box, a heat map of the spatial distribution of stacking pressure, and a vibration spectrum of the transportation path. The operational layer data includes time and space stamp records of loading and unloading events and time series data on the opening and closing frequency of cold storage doors. S2, extracting and fusing quality degradation features of the sensory layer data to obtain a quality degradation feature tensor; S3, performing dynamic coupling modeling processing on the quality degradation feature tensor and the environmental layer data, thereby obtaining a quality degradation environment coupling parameter set; S4, performing multi-dimensional dynamic constraint field modeling processing on the quality degradation environment coupling parameter set and the operation layer data, thereby obtaining a degradation constraint distribution matrix; S5, performing adaptive degradation trajectory evolution processing on the acoustic wave resonance frequency response curve in the sensory layer data and the degradation constraint distribution matrix, thereby obtaining a four-level quality label.

2. The method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to claim 1, characterized in that: The S2 step includes the following sub-steps: S201, performing color difference quantization processing on the surface visible light image in the sensory layer data to obtain a surface color difference heat map, wherein the color difference quantization processing is to extract surface color abnormal areas through color space transformation; S202, performing chemical bond feature extraction processing on the near-infrared reflectance spectrum in the sensory layer data to obtain a water loss feature vector, wherein the chemical bond feature extraction processing is performed by screening characteristic bands related to quality degradation through a PLS regression model; S203, performing spatial chemical fusion processing on the surface color difference heat map and the water loss feature vector to obtain a quality degradation feature tensor, wherein the spatial chemical fusion processing maps chemical features to color difference abnormality areas through a spatial weighted matrix.

3. The method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to claim 1, characterized in that: The S3 step includes the following sub-steps: S301, performing diffusion path topology modeling processing on the quality degradation feature tensor and the microenvironment temperature and humidity gradient matrix in the environmental layer data to obtain a degradation diffusion rate field, wherein the diffusion path topology modeling processing simulates the propagation path of degradation along the direction of the maximum temperature and humidity gradient through a graph network model; S302, performing mechanical path coupling processing on the degradation diffusion rate field and the stack pressure spatial distribution thermodynamic map in the environmental layer data to obtain a pressure path correlation tensor, wherein the mechanical path coupling processing calculates the local degradation acceleration effect through geometric superposition of the pressure distribution and the diffusion path; S303, performing frequency domain stability analysis on the pressure path correlation tensor and the transport path vibration spectrum in the environmental layer data to obtain a quality degradation environment coupling parameter set. The frequency domain stability analysis quantifies the disturbance intensity of the vibration on the diffusion path through the frequency band energy distribution.

4. A method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to any one of claims 1 to 3, characterized in that: The S4 step includes the following sub-steps: S401, performing pulse sequence trajectory tensor processing on the spatiotemporal stamp records of loading and unloading events in the operation layer data to obtain a loading and unloading pulse trajectory tensor, wherein the trajectory tensor processing is to reconstruct the loading and unloading energy impact distribution through the spatiotemporal trajectory of the pulse event; S402: Performing temperature and humidity disturbance evolution modeling on the cold storage door opening and closing frequency time series data in the operation layer data to obtain a dynamic distribution of cold chain stability. The evolution modeling process is to construct a cold chain disturbance dynamic response model by coupling the time series disturbance curve with the humidity gradient; S403, performing multi-dimensional dynamic constraint mapping processing on the loading and unloading pulse trajectory tensor, the dynamic distribution of cold chain stability and the quality degradation environment coupling parameter set, so as to obtain the degradation constraint distribution matrix. The mapping processing forms a constraint balance field through space-time-mechanical multivariate weight calculation.

5. The method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to claim 4, characterized in that: The S5 step includes the following sub-steps: S501, performing a resonance modal coherence analysis on the acoustic wave resonance frequency response curve in the sensory layer data to obtain a spectrum coherence drift factor, wherein the coherence analysis is to calculate the degree of acoustic wave mode shift through the temporal coherence of the resonance frequency band distribution; S502, nonlinear degradation trajectory mapping processing is performed on the spectral coherence drift factor and the degradation constraint distribution matrix to obtain four-level quality labels. The trajectory mapping processing is to calculate the quality evolution critical point through multi-dimensional degradation space projection. The four-level quality labels include compliance label, warning label, abnormal label, and irreversible label.

6. The method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to claim 3, characterized in that: The mechanical path coupling processing in step S302 is to calculate the local degradation acceleration effect through the geometric superposition of pressure distribution and diffusion path, including local stress coefficient extraction processing and pressure diffusion gain fusion processing; the frequency domain stability analysis processing in step S303 is to quantify the disturbance intensity of vibration on the diffusion path through frequency band energy distribution, including vibration energy spectrum extraction processing and frequency band coupling factor calculation processing.

7. The method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to claim 4, characterized in that: The trajectory tensorization processing in step S401 is to reconstruct the loading and unloading energy impact distribution through the spatiotemporal trajectory of the pulse event, including spatiotemporal grid mapping processing and pulse intensity normalization processing.

8. The method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to claim 4, characterized in that: The evolutionary modeling process in the step S402 is to construct a cold chain disturbance dynamic response model by coupling the time series disturbance curve with the humidity gradient, including the time series disturbance curve decomposition process and the humidity gradient coupling process; the mapping process in the step 403 is to form a constrained equilibrium field by space-time-mechanical multivariate weight calculation, including space-time dimension alignment process and multivariate weight fusion process.

9. The method for predicting fresh produce quality deterioration based on multimodal dynamic coupling according to claim 5, characterized in that: The coherence analysis processing in step S501 is to calculate the degree of acoustic wave mode deviation through the temporal coherence of the resonant frequency band distribution, including frequency band segmentation processing and temporal coherence calculation processing; the trajectory mapping processing in step S502 is to calculate the critical point of quality evolution through multi-dimensional degradation space projection, including multi-dimensional feature space construction processing and critical point determination processing.

10. A fresh produce quality deterioration prediction system based on multimodal dynamic coupling, characterized in that: The system comprises: An acquisition unit is configured to acquire data to be processed, the data to be processed including sensory layer data, environmental layer data, and operational layer data. The sensory layer data includes visible light images of fresh produce surfaces, near-infrared reflectance spectra, and acoustic resonance frequency response curves. The environmental layer data includes a temperature and humidity gradient matrix of the microenvironment within the cargo box, a thermal map of the spatial distribution of stacking pressure, and a vibration spectrum of the transport path. The operational layer data includes time and space stamp records of loading and unloading events and time series data on the opening and closing frequency of cold storage doors. a feature fusion unit, configured to extract and fuse quality degradation features of the sensory layer data to obtain a quality degradation feature tensor; A coupling modeling unit, configured to perform dynamic coupling modeling processing on the quality degradation feature tensor and the environmental layer data, thereby obtaining a quality degradation environment coupling parameter set; a constraint modeling unit, configured to perform multi-dimensional dynamic constraint field modeling processing on the quality degradation environment coupling parameter set and the operation layer data, thereby obtaining a degradation constraint distribution matrix; The hierarchical quality label unit is used to perform adaptive degradation trajectory evolution processing on the acoustic wave resonance frequency response curve in the sensory layer data and the degradation constraint distribution matrix, so as to obtain four-level quality labels.

Images