A method and system for logistics detection

Abstract

The application provides a logistics detection method and system, and relates to the technical field of logistics detection. The method comprises the following steps: generating a medical biological gel matrix packaging unit property change risk distribution map; determining a plurality of property deterioration suspected points based on the medical biological gel matrix packaging unit property change risk distribution map; determining a property deterioration influence boundary based on image photoelectric detection information of the plurality of property deterioration suspected points; determining a plurality of matrix component qualitative review points corresponding to the property deterioration influence boundary based on image photoelectric detection information of the plurality of property deterioration suspected points and medium electromagnetic property detection data of the property deterioration influence boundary; performing matrix deterioration qualitative identification based on the plurality of matrix component qualitative review points corresponding to the property deterioration influence boundary; and completing cold chain logistics detection of the medical biological gel matrix. The method can efficiently and accurately determine the component review points of the medical biological gel matrix that need to be qualitatively identified in the cold chain logistics.

Description

Technical Field

[0001] This invention relates to the field of logistics inspection technology, specifically to a logistics inspection method and system. Background Technology

[0002] Cold chain logistics for pharmaceutical and biological products is crucial for ensuring the bioactivity of these products and the safety of patient medication. Transparent tracking of their entire process directly impacts product efficacy. Currently, cold chain tracking primarily relies on temperature and humidity sensors to record environmental fluctuations and assess transportation risks by comparing them to preset thresholds. However, this method, based on external environmental monitoring, has significant limitations. Firstly, environmental monitoring data lacks correlation with changes in the internal state of the product. Traditional monitoring only acquires macroscopic parameters within the vehicle, failing to detect how external temperature shocks or mechanical vibrations affect the encapsulation unit, and thus struggling to characterize the non-uniformity and gradual degradation process of the biogel matrix under continuous load. Secondly, the granularity of damage identification is insufficient. Due to a lack of understanding of the internal evolution of the gel matrix, traditional methods cannot identify the initiation location, diffusion path, and boundary of degradation effects, leading to a severe lag in early warning of latent deterioration. This situation, characterized by abundant external data but blind spots in internal state data, renders cold chain tracking largely a passive, post-event verification process, unable to provide proactive risk assessment support. This results in highly sensitive pharmaceutical products facing extremely high risks of spoilage during distribution, causing significant resource waste.

[0003] Therefore, how to efficiently and accurately determine the key components of pharmaceutical biogel matrices that require qualitative identification in cold chain logistics is an urgent problem to be solved. Summary of the Invention

[0004] The main technical problem addressed by this invention is how to efficiently and accurately determine the key components of a pharmaceutical biogel matrix that require qualitative identification in cold chain logistics.

[0005] According to a first aspect, the present invention provides a logistics inspection method, comprising: acquiring environmental damage load data at preset intervals during the transportation of a pharmaceutical biogel matrix packaging unit; generating a risk distribution map of morphological changes in the pharmaceutical biogel matrix packaging unit based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit; identifying multiple suspected points of morphological deterioration based on the risk distribution map of morphological changes in the pharmaceutical biogel matrix packaging unit; acquiring photoelectric detection information of the multiple suspected points of morphological deterioration; determining the boundary of morphological degradation impact based on the photoelectric detection information of the multiple suspected points of morphological deterioration; acquiring electromagnetic property detection data of the medium at the boundary of morphological degradation impact; determining multiple qualitative verification points of matrix components corresponding to the boundary of morphological degradation impact based on the photoelectric detection information of the multiple suspected points of morphological deterioration and the electromagnetic property detection data of the medium at the boundary of morphological degradation impact; performing qualitative identification of matrix deterioration based on the multiple qualitative verification points of matrix components corresponding to the boundary of morphological degradation impact, thereby completing the cold chain logistics inspection of the pharmaceutical biogel matrix.

[0006] In one possible implementation, the environmental damage load data includes spatial temperature distribution data and load vibration sequence data.

[0007] In one possible implementation, determining the boundary of trait degradation impact based on the image photoelectric detection information of the multiple suspected trait degradation points includes: clustering the image photoelectric detection information of the multiple suspected trait degradation points to obtain multiple trait abnormality feature clusters; generating an image feature anomaly distribution map of the gel matrix based on the multiple trait abnormality feature clusters; determining multiple suspected trait degradation areas and degradation information of each suspected trait degradation area based on the image feature anomaly distribution map of the gel matrix and the trait change risk distribution map of the pharmaceutical biological gel matrix encapsulation unit; constructing a trait evolution map, the trait evolution map including multiple suspected trait degradation area nodes and edges between multiple nodes, the node feature of each suspected trait degradation area node being the degradation information of the suspected trait degradation area; and processing the trait evolution map based on a graph neural network to determine the boundary of trait degradation impact.

[0008] In one possible implementation, determining multiple matrix component qualitative verification points corresponding to the morphological degradation influence boundary based on the image photoelectric detection information of the multiple suspected morphological degradation points and the medium electromagnetic property detection data of the morphological degradation influence boundary includes: determining severely deteriorated detection points and diffusion information of the morphological degradation influence boundary based on the image photoelectric detection information of the multiple suspected morphological degradation points and the medium electromagnetic property detection data of the morphological degradation influence boundary; determining potential damage points of the morphological degradation influence boundary based on the diffusion information of the morphological degradation influence boundary; and determining multiple matrix component qualitative verification points corresponding to the morphological degradation influence boundary based on the severely deteriorated detection points and the potential damage points of the morphological degradation influence boundary.

[0009] According to a second aspect, the present invention provides a logistics inspection system, comprising: The data acquisition module is used to acquire environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit; The risk distribution map generation module is used to generate a risk distribution map of the changes in the properties of the pharmaceutical biogel matrix packaging unit based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit. The suspected point identification module is used to identify multiple suspected points of phenotypic deterioration based on the phenotypic change risk distribution map of the pharmaceutical biogel matrix encapsulation unit. The detection information acquisition module is used to acquire photoelectric detection information of multiple suspected points of phenotypic deterioration; The degradation boundary determination module is used to determine the degradation influence boundary based on the image photoelectric detection information of the multiple suspected degradation points. The detection data acquisition module is used to acquire detection data of the electromagnetic properties of the medium at the boundary affected by the degradation of its properties; The verification point determination module is used to determine multiple matrix component qualitative verification points corresponding to the boundary of morphological degradation based on the image photoelectric detection information of the multiple suspected morphological degradation points and the medium electromagnetic property detection data of the boundary of morphological degradation influence. The identification and tracking module is used to perform qualitative identification of matrix deterioration based on multiple matrix component qualitative verification points corresponding to the boundary of the property deterioration effect, and to complete the cold chain logistics detection of the pharmaceutical biogel matrix.

[0010] In one possible implementation, the environmental damage load data includes spatial temperature distribution data and load vibration sequence data.

[0011] In one possible implementation, the degradation boundary determination module is further configured to: cluster multiple trait aberration feature clusters based on the image photoelectric detection information of the multiple suspected trait degradation points; generate an image feature aberration distribution map of the gel matrix based on the multiple trait aberration feature clusters; determine multiple suspected trait degradation areas and degradation information of each suspected trait degradation area based on the image feature aberration distribution map of the gel matrix and the trait change risk distribution map of the pharmaceutical biological gel matrix encapsulation unit; construct a trait evolution map, the trait evolution map including multiple suspected trait degradation area nodes and edges between multiple nodes, the node feature of each suspected trait degradation area node being the degradation information of the suspected trait degradation area; and process the trait evolution map based on a graph neural network to determine the trait degradation influence boundary.

[0012] In one possible implementation, the verification point determination module is further configured to: determine the severe degradation detection point of the morphological degradation influence boundary and the diffusion information of the morphological degradation influence boundary based on the image photoelectric detection information of the plurality of suspected morphological degradation points and the medium electromagnetic property detection data of the morphological degradation influence boundary; determine the potential damage point of the morphological degradation influence boundary based on the diffusion information of the morphological degradation influence boundary; and determine the plurality of matrix component qualitative verification points corresponding to the morphological degradation influence boundary based on the severe degradation detection point of the morphological degradation influence boundary and the potential damage point of the morphological degradation influence boundary.

[0013] According to a third aspect, embodiments of the present invention provide an electronic device, including: a processor; a memory; and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method described above, the method including: acquiring environmental damage load data at preset intervals during the transportation of a pharmaceutical biogel matrix encapsulation unit; generating a risk distribution map of morphological changes in the pharmaceutical biogel matrix encapsulation unit based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix encapsulation unit; and generating a risk distribution map of morphological changes in the pharmaceutical biogel matrix encapsulation unit based on the risk distribution map of morphological changes in the pharmaceutical biogel matrix encapsulation unit. The process involves: identifying multiple suspected points of morphological degradation through a layout map; acquiring photoelectric detection information of these suspected points; determining the morphological degradation impact boundary based on this information; acquiring electromagnetic property detection data of the medium at the morphological degradation impact boundary; determining multiple qualitative verification points of matrix components corresponding to the morphological degradation impact boundary based on the photoelectric detection information of these suspected points and the electromagnetic property detection data of the medium at the morphological degradation impact boundary; and performing qualitative identification of matrix degradation based on these multiple qualitative verification points of matrix components corresponding to the morphological degradation impact boundary to complete the cold chain logistics detection of the pharmaceutical biogel matrix.

[0014] According to the fourth aspect, this embodiment provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the program implements the aforementioned logistics detection method. The method includes: acquiring environmental damage load data at preset intervals during the transportation of a pharmaceutical biogel matrix packaging unit; generating a risk distribution map of morphological changes in the pharmaceutical biogel matrix packaging unit based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit; identifying multiple suspected points of morphological deterioration based on the risk distribution map of morphological changes in the pharmaceutical biogel matrix packaging unit; acquiring photoelectric detection information of the multiple suspected points of morphological deterioration; determining the boundary of morphological degradation impact based on the photoelectric detection information of the multiple suspected points of morphological deterioration; acquiring electromagnetic property detection data of the medium at the boundary of morphological degradation impact; determining multiple qualitative verification points of matrix components corresponding to the boundary of morphological degradation impact based on the photoelectric detection information of the multiple suspected points of morphological deterioration and the electromagnetic property detection data of the medium at the boundary of morphological degradation impact; performing qualitative identification of matrix deterioration based on the multiple qualitative verification points of matrix components corresponding to the boundary of morphological degradation impact, thereby completing the cold chain logistics detection of the pharmaceutical biogel matrix.

[0015] This invention provides a logistics inspection method and system. The method includes acquiring environmental damage load data at preset intervals during the transportation of pharmaceutical biogel matrix packaging units; generating a risk distribution map of morphological changes in the pharmaceutical biogel matrix packaging units based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging units; identifying multiple suspected points of morphological deterioration based on the risk distribution map of morphological changes in the pharmaceutical biogel matrix packaging units; acquiring photoelectric detection information of the multiple suspected points of morphological deterioration; determining the morphological degradation impact boundary based on the photoelectric detection information of the multiple suspected points of morphological deterioration; acquiring electromagnetic property detection data of the medium at the morphological degradation impact boundary; determining multiple qualitative verification points of matrix components corresponding to the morphological degradation impact boundary based on the photoelectric detection information of the multiple suspected points of morphological deterioration and the electromagnetic property detection data of the medium at the morphological degradation impact boundary; and performing qualitative identification of matrix deterioration based on the multiple qualitative verification points of matrix components corresponding to the morphological degradation impact boundary, thereby completing the cold chain logistics inspection of pharmaceutical biogel matrix. This method can efficiently and accurately determine the component verification points of pharmaceutical biogel matrix that most need qualitative identification in cold chain logistics. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a logistics inspection method provided in an embodiment of the present invention; Figure 2 A schematic diagram of a pharmaceutical biogel matrix encapsulation unit provided in an embodiment of the present invention; Figure 3A flowchart illustrating the determination of the boundary of morphological degradation effects, provided in an embodiment of the present invention; Figure 4 This is a flowchart illustrating a process for determining multiple matrix component qualitative verification points corresponding to the boundary of morphological degradation impact, as provided in an embodiment of the present invention. Figure 5 This is a schematic diagram of a logistics inspection system provided in an embodiment of the present invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the invention. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to the present invention are not shown or described in the specification. This is to avoid obscuring the core parts of the invention with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0018] In this embodiment of the invention, the following are provided: Figure 1 The illustrated logistics inspection method includes steps S1 to S8: Step S1: Obtain environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix encapsulation unit.

[0019] Pharmaceutical biogel matrices are biocompatible gel-like functional materials used in the pharmaceutical field. They can be used as drug carriers, tissue repair substrates, or medical barrier materials, such as hyaluronic acid gel, collagen gel, and carbomer medical gel. The physicochemical properties of pharmaceutical biogel matrices are easily affected by environmental factors such as temperature and vibration, and they are important components of pharmaceutical cold chain storage and transportation.

[0020] The pharmaceutical biogel matrix encapsulation unit is a cold chain sealing and protection device used for storing and transporting pharmaceutical biogel matrices. Figure 2 This is a schematic diagram of a pharmaceutical biogel matrix encapsulation unit provided in an embodiment of the present invention.

[0021] The environmental damage load data at preset intervals records the physical impact of the environment on the packaging unit at each mileage node of the cold chain logistics vehicle. The environmental damage load data includes spatial temperature distribution data and load vibration sequence data.

[0022] Spatial temperature distribution data describes the temperature conditions at different spatial locations within the transport vehicle. This data includes temperature values ​​at different three-dimensional coordinate points within the vehicle, temperature gradient indices between different areas, and the highest and lowest temperature values ​​within a preset mileage interval.

[0023] Load vibration sequence data records the changes in mechanical vibration experienced by the packaging unit over time during transportation. The load vibration sequence data includes the amplitude sequence of the transport vehicle in three axes, real-time vibration frequency values, peak impact acceleration, and vibration duration.

[0024] Step S2: Generate a risk distribution map of the changes in the properties of the pharmaceutical biogel matrix packaging unit based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit.

[0025] In some embodiments, a risk analysis model can be used to generate a risk distribution map of morphological changes in the pharmaceutical biogel matrix encapsulation unit. The risk analysis model is a Transformer model. The input to the risk analysis model is environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix encapsulation unit, and the output of the risk analysis model is the risk distribution map of morphological changes in the pharmaceutical biogel matrix encapsulation unit.

[0026] The Transformer model consists of two core modules: an encoder and a decoder. The encoder contains a multi-layered self-attention mechanism and a feedforward neural network. It performs deep feature extraction and representation learning on the input sequence data and captures long-range dependencies between data points. The decoder also incorporates a self-attention mechanism and a feedforward neural network, and adds a cross-attention mechanism to the encoder output. The decoder can generate target sequences or spatial distribution classes that meet the task requirements based on the features extracted by the encoder.

[0027] The risk distribution map of morphological changes in pharmaceutical biogel matrix encapsulation units is a two-dimensional image that displays the probability of property changes in the gel matrix at different spatial coordinate locations within the encapsulation unit. Different color gradients and numerical labels in the risk distribution map of morphological changes in pharmaceutical biogel matrix encapsulation units represent the probability of morphological degradation in different parts of the encapsulation unit.

[0028] Environmental damage load data recorded at preset intervals during the transportation of the pharmaceutical biogel matrix encapsulation unit revealed details of temperature field fluctuations and continuous mechanical oscillations accumulated over distance. The spatial temperature distribution data within the environmental damage load data revealed thermal conditions that could lead to gel matrix melting, while the load vibration sequence data reflected mechanical stress conditions that could damage the matrix's network structure.

[0029] The Transformer model receives and compares load vibration sequence data and spatial temperature distribution data at different mileage nodes through a self-attention module in the encoder. The self-attention mechanism can capture the superimposed effect of severe vibration and specific high-temperature stages during transportation, and assign a higher attention weight to this combination of hazards. The decoder of the Transformer model can map the deeply extracted cross-mileage damage correlation features onto the spatial coordinates of the encapsulation unit, and then calculate the deterioration probability value of each coordinate point after experiencing cumulative damage effects, thereby generating a risk distribution map of the morphological changes of the pharmaceutical biogel matrix encapsulation unit.

[0030] Step S3: Based on the risk distribution map of the morphological changes of the pharmaceutical biogel matrix encapsulation unit, identify multiple suspected points of morphological deterioration.

[0031] In some embodiments, a suspected qualitative change point determination model can be used to identify multiple suspected trait deterioration points. The suspected qualitative change point determination model is a convolutional neural network model. The input to the suspected qualitative change point determination model is the risk distribution map of trait changes in the pharmaceutical biogel matrix encapsulation unit, and the output of the suspected qualitative change point determination model is multiple suspected trait deterioration points.

[0032] Convolutional Neural Network (CNN) models are a type of deep feedforward artificial neural network specifically designed to process data with a grid-like structure. The core components of a CNN include convolutional layers, pooling layers, and fully connected layers. Convolutional layers perform sliding operations on the input image or matrix using kernels of a preset size, extracting low-level visual and spatial features such as local edges and textures. Pooling layers downsample the feature maps output by the convolutions, compressing the data dimensionality while preserving salient features, thus enhancing the model's tolerance to translation and deformation. Fully connected layers flatten the extracted two-dimensional features into a one-dimensional representation, which is then mapped to the final classification or regression result through a non-linear activation function.

[0033] Multiple suspected deterioration points were identified using a suspected deterioration point determination model, representing the specific three-dimensional spatial coordinates of areas with a high probability of degradation within the pharmaceutical biogel matrix encapsulation unit. Each suspected deterioration point represents a sampling center where anomalies may occur in the gel matrix regarding structural continuity, color intensity, or light transmittance.

[0034] The risk distribution map of the biomaterial properties of pharmaceutical biogel matrix encapsulation units can clearly show the numerical gradient of risk probability in different regions within a space. The distribution map includes fluctuations in risk probability and the geometric outline of high-risk areas.

[0035] Convolutional neural networks (CNNs) can take the risk distribution map of phenotypic changes in pharmaceutical biogel matrix encapsulation units as an input matrix, and then scan the probability gradient changes on the distribution map using convolutional kernels at multiple scales in the convolutional layers. CNNs can identify local extreme regions where probability values ​​rise sharply in the distribution map, as well as the edge contour features of risk propagation. Fully connected layers can calculate the precise spatial location based on these extracted high-risk spatial features, thereby identifying multiple suspected points of phenotypic deterioration.

[0036] Step S4: Obtain photoelectric detection information of multiple suspected phenotypic deterioration points.

[0037] The photoelectric detection information of suspected morphological alteration points records the feedback status of matrix visual and light intensity characteristics at the suspected morphological alteration points. The photoelectric detection information of multiple suspected morphological alteration points is obtained by close-range imaging and detection of the spatial locations corresponding to multiple suspected morphological alteration points using endoscopic imaging equipment and spectral detection instruments.

[0038] The photoelectric detection information of suspected morphological deterioration points includes matrix discoloration information at the suspected morphological deterioration point, matrix transmittance change information, and matrix bubble distribution information within a preset radius centered on the suspected morphological deterioration point.

[0039] Matrix color change information reflects the degree to which the color of the gel matrix deviates from its normal color. Matrix color change information includes the center color coordinates of the color-changed area, the color deviation index, and the blurring of the color-changed edge.

[0040] The matrix transmittance change information records how the gel matrix changes its ability to block light. The matrix transmittance change information includes the light transmittance value of a specific wavelength band, the light scattering intensity level, and the matrix turbidity index.

[0041] Matrix bubble distribution information describes the state of gas voids generated inside the gel matrix. Matrix bubble distribution information includes the diameter sequence of bubbles, the number density of bubbles per unit volume, and the spatial distribution orientation of bubble groups.

[0042] Step S5: Determine the boundary of trait degradation influence based on the image photoelectric detection information of the multiple suspected trait deterioration points.

[0043] In some embodiments, Figure 3 This is a flowchart illustrating the determination of the morphological degradation influence boundary according to an embodiment of the present invention. The determination of the morphological degradation influence boundary includes steps S51 to S55: Step S51: Cluster the image photoelectric detection information of the multiple suspected trait deterioration points to obtain multiple trait abnormality feature clusters.

[0044] In some embodiments, the clustering is the K-means clustering algorithm.

[0045] K-means clustering is an unsupervised iterative distance-based partitioning algorithm. It classifies data by dividing a dataset into a predetermined number of non-overlapping subsets. The value of K in K-means clustering can be pre-defined manually.

[0046] Multiple morphological abnormality feature clusters are obtained by using the K-means clustering algorithm to cluster the photoelectric detection information of multiple suspected morphological deterioration points according to the feature similarity of the gel matrix detection feedback state. The photoelectric detection information in each cluster has similar matrix image feature abnormalities, and different clusters correspond to different types of matrix image feature abnormalities.

[0047] The photoelectric detection information of multiple suspected deterioration points includes image features and abnormal features such as matrix discoloration, transmittance changes, and bubble distribution at each suspected point. The features of each suspected point have similarities and differences. These differences and similarities provide a classification basis for the K-means clustering algorithm. The algorithm can classify abnormal features by calculating these multi-dimensional features and discover common deterioration feature types.

[0048] The process of clustering image photoelectric detection information of multiple suspected morphological deterioration points using the K-means clustering algorithm is as follows: First, K data samples are randomly selected as initial cluster centers, and the Euclidean distance from all other samples to each initial cluster center is calculated. Then, the samples are assigned to the cluster to the nearest cluster center. Subsequently, the feature mean of all samples within each cluster is recalculated and used as the new cluster center. The distance from the samples to the new cluster centers is calculated again, and the clusters are reassigned. This process is iterated repeatedly until the position of the cluster centers no longer changes significantly, and the sum of squared distance errors of samples within a cluster reaches its minimum value. Finally, image photoelectric detection information with similar features is grouped into the same cluster, resulting in multiple morphological anomalous feature clusters.

[0049] Clustering the photoelectric detection information of multiple suspected morphological deterioration points into multiple morphological anomaly feature clusters allows suspected points with similar color change patterns, transmittance changes, and bubble distribution characteristics to be grouped together, thereby achieving structured integration of discrete detection data. This classification method can effectively distinguish different morphological degradation patterns and avoid blindly processing a large number of discrete points.

[0050] Step S52: Generate an image feature anomaly distribution map of the gel matrix based on the multiple abnormal feature clusters.

[0051] In some embodiments, a second risk analysis model can be used to generate an image feature anomaly distribution map of the gel matrix. The second risk analysis model is a Transformer model. The input to the second risk analysis model is the plurality of trait anomaly feature clusters, and the output of the second risk analysis model is the image feature anomaly distribution map of the gel matrix.

[0052] The image feature anomaly distribution map of the gel matrix is ​​a spatially continuous probability image representing the severity of image anomalies within the pharmaceutical and biological gel matrix encapsulation unit.

[0053] In the distribution map of abnormal features in the image of the gel matrix, different hues are used to represent the types of abnormal features, and color saturation and brightness are used to indicate the numerical intensity of color change and turbidity at that location.

[0054] By using multiple clusters of anomalous features, discrete detection point information can be transformed into structurally correlated feature combinations, thereby reflecting the qualitative changes in the gel matrix in different regions. These feature clusters not only encompass the types of anomalous features but also their distribution density in the feature space.

[0055] The Transformer model utilizes a self-attention mechanism to receive and process multiple clusters of anomalous features, capturing the topological relationships and evolutionary logic between different feature clusters across spatial spans. The encoder performs deep representation of the image features of each cluster. The decoder, through a cross-attention mechanism, maps these features back to a three-dimensional spatial coordinate system, and then calculates the anomalous probability of non-sampling points through interpolation and feature fusion, thereby generating an image feature anomaly distribution map covering the entire gel matrix.

[0056] Step S53: Based on the abnormal distribution map of the image features of the gel matrix and the risk distribution map of the morphological changes of the pharmaceutical biological gel matrix encapsulation unit, determine multiple suspected morphological deterioration areas and the deterioration information of each suspected morphological deterioration area.

[0057] In some embodiments, a suspected qualitative change point determination model can be used to determine multiple suspected areas of trait deterioration and the deterioration information of each suspected trait deterioration area. The suspected qualitative change point determination model is a convolutional neural network model. The input to the suspected qualitative change point determination model is an image feature anomaly distribution map of the gel matrix and a trait change risk distribution map of the pharmaceutical biogel matrix encapsulation unit. The output of the suspected qualitative change point determination model is multiple suspected trait deterioration areas and the deterioration information of each suspected trait deterioration area.

[0058] Multiple suspected trait degradation zones are identified within the encapsulation unit by a trait degradation suspected point determination model. These zones are spatial regions formed by the aggregation of multiple trait degradation suspected points and exhibit a continuous probability of trait degradation. Each suspected trait degradation zone is a continuous three-dimensional spatial region within the encapsulation unit with a certain range.

[0059] The degradation information for each suspected trait degradation area refers to the information on the deterioration state of the gel matrix within that suspected area. The degradation information for each suspected trait degradation area includes the spatial coordinates of the suspected area, the dominant degradation type, and the average degradation grade index.

[0060] The dominant metamorphic types include matrix network collapse, localized thermal melting, and phase separation induced by mechanical stress.

[0061] The image feature anomaly distribution map of the gel matrix can represent the spatial numerical distribution of the matrix in terms of color shift, transmittance, and bubble density. The risk distribution map of the property changes of the pharmaceutical and biological gel matrix encapsulation unit provides the probability of deterioration risk at each spatial location after being subjected to temperature and vibration loads during transportation. By comparing the values ​​of the two in the same spatial coordinates, the model can identify areas where physical property anomalies and external damage loads highly match.

[0062] A convolutional neural network (CNN) can use the distribution map of abnormal image features of the gel matrix and the risk distribution map of morphological changes in the encapsulation unit of the pharmaceutical biogel matrix as dual input matrices. Then, a shared convolutional layer performs a simultaneous sliding scan of the two distribution maps. Multi-scale convolutional kernels extract local spatial features from the two distribution maps, including clustered regions of abnormal image features, high-probability regions, and spatial overlap features between the two types of features. By fusing the extracted dual-map features, the model can identify continuous spatial regions with both obvious image feature anomalies and high degradation probabilities; these continuous regions are the core range of suspected morphological degradation areas. Pooling layers downsample the fused feature maps to preserve the core spatial contour features of the suspected areas. Fully connected layers quantize the features to determine the precise spatial coordinates of each suspected morphological degradation area. Simultaneously, the fully connected layers combine the abnormal image feature distribution map to extract the degradation type of the suspected areas and the risk distribution map to extract the average degradation level index, while simultaneously extracting the geometric contour features of the degradation areas. Finally, multiple suspected morphological degradation areas and their degradation information are integrated.

[0063] Step S54: Construct a trait evolution map, which includes multiple suspected trait transformation zone nodes and multiple edges between nodes. The node features of each suspected trait transformation zone node are the transformation information of the suspected trait transformation zone.

[0064] The trait evolution map is a constructed graph structure data used to represent the correlation between multiple suspected trait degradation intervals within a pharmaceutical biogel matrix encapsulation unit. The trait evolution map consists of multiple suspected trait degradation interval nodes and edges connecting the nodes. The node feature of each suspected trait degradation interval node is the degradation information of the corresponding suspected trait degradation interval, and the edges between nodes represent the similarity of degradation information between different suspected trait degradation intervals.

[0065] In some embodiments, deep neural networks can be used to determine the similarity of metamorphic information among suspected metamorphic regions of different traits.

[0066] Step S55: Process the trait evolution map based on graph neural network to determine the boundary of trait degradation influence.

[0067] Graph Neural Networks (GNNs) are deep learning models used to process graph-structured data. By aggregating the features of a node itself and the features of its neighboring nodes, GNNs can achieve deep feature extraction and representation learning of graph-structured data. GNNs can accurately capture the relationships and topological features between nodes in a graph, thereby completing the analysis and prediction of nodes, edges, or the entire graph. GNNs are suitable for regional analysis tasks with spatial associations and propagation relationships. The input of the GNN is the trait evolution graph, and the output of the GNN is the boundary of trait degradation influence.

[0068] The property degradation impact boundary is the three-dimensional spatial contour boundary between the area of ​​the gel matrix where property degradation has occurred and the area where property degradation has not occurred within the pharmaceutical biogel matrix encapsulation unit. The property degradation impact boundary precisely defines the overall spatial range of the degraded matrix within the encapsulation unit.

[0069] By constructing a trait evolution map, discrete potential trait degradation zones within the encapsulation unit can be connected, revealing their spatial relationships and degradation transmission paths. Since trait degradation in the gel matrix spreads continuously along the network structure, degradation in one region leads to decreased stability in adjacent locations. Therefore, identifying this spatial connectivity between regions is crucial for defining the boundaries of trait degradation impact. Using degradation information from potential trait degradation zones as node features allows the model to directly utilize the specific trait data of each potential zone. This helps the model understand how different regions interpenetrate and change, thereby improving the accuracy of boundary determination. Processing this map data using graph neural networks can simulate the transmission logic of degradation information between potential zones, thus more scientifically defining the scope of trait degradation impact.

[0070] Graph neural networks (Graph Neural Networks) can aggregate features of each suspected trait degradation zone node in the trait evolution map through graph convolutional layers. These layers not only extract the degradation information features of the node itself but also aggregate the degradation information features of its neighboring nodes based on edge weights, capturing the correlation between the degradation features of each node and its neighbors. Through deep learning with multiple graph convolutional layers, Graph Neural Networks can uncover the degradation propagation patterns of all nodes in the entire trait evolution map and analyze the diffusion trend and range of degradation from high-severity suspected zones to low-severity suspected zones and from nearby suspected zones to distant suspected zones. Subsequently, the Graph Neural Network can quantify the aggregated node features and the topological features of the trait evolution map to identify the maximum spatial range that degradation features can cover. By combining the actual spatial coordinates of the suspected trait degradation zones corresponding to each node within the encapsulation unit, the Graph Neural Network can connect and fit the boundary points of these maximum spatial ranges, forming a continuous three-dimensional spatial contour, and ultimately determining the boundary of trait degradation influence within the pharmaceutical biogel matrix encapsulation unit.

[0071] Step S6: Obtain electromagnetic property detection data of the medium affecting the boundary of property degradation.

[0072] Electromagnetic property detection data of the medium at the boundary of property degradation is data reflecting the physical properties of the material obtained by scanning the determined degradation boundary with an electromagnetic induction device.

[0073] The electromagnetic property detection data of the medium affecting the boundary of the property degradation is obtained by detecting the signal feedback passing through the medium through non-contact electromagnetic sensors arranged around the periphery of the packaging unit.

[0074] The electromagnetic property detection data of the dielectric at the boundary affected by property degradation include the real part of the complex permittivity, the dielectric loss tangent, the conductivity distribution sequence, and the electromagnetic wave attenuation rate.

[0075] Step S7: Based on the image photoelectric detection information of the multiple suspected morphological deterioration points and the medium electromagnetic property detection data of the morphological degradation influence boundary, determine multiple matrix component qualitative verification points corresponding to the morphological degradation influence boundary.

[0076] In some embodiments, Figure 4 This is a flowchart illustrating a method for determining multiple matrix component qualitative verification points corresponding to a boundary of morphological degradation impact, as provided in an embodiment of the present invention. The multiple matrix component qualitative verification points corresponding to the boundary of morphological degradation impact include steps S71 to S73: Step S71: Based on the image photoelectric detection information of the multiple suspected deterioration points and the medium electromagnetic property detection data of the deterioration influence boundary, determine the severe deterioration detection point of the deterioration influence boundary and the diffusion information of the deterioration influence boundary.

[0077] In some embodiments, a degradation diffusion analysis model can be used to determine the severely degraded detection points and diffusion information of the degradation influence boundary. The degradation diffusion analysis model is a Transformer model. The inputs to the degradation diffusion analysis model are the image photoelectric detection information of the plurality of suspected degradation points and the medium electromagnetic property detection data of the degradation influence boundary. The outputs of the degradation diffusion analysis model are the severely degraded detection points and diffusion information of the degradation influence boundary.

[0078] The severe degradation detection point at the boundary of property degradation is a spatially located point near the boundary where the medium structure has been physically damaged. The severe degradation detection point at the boundary of property degradation includes precise three-dimensional coordinates, the structural damage index at that point, and the corresponding electromagnetic reflection intensity value.

[0079] The diffusion information at the boundary of morphological degradation refers to the predictive parameters of the trend of qualitative change penetrating into the healthy region. This diffusion information includes the diffusion direction vector, the diffusion displacement per unit time, and the concentration gradient distribution at the degradation front.

[0080] Photoelectric detection data from multiple suspected sites of morphological degradation revealed discoloration and bubbling in the matrix from a microscopic perspective, while electromagnetic property detection data of the medium affecting the boundary of morphological degradation reflected changes in molecular polarization and conductivity at a macroscopic level. By combining the surface features of the images with the internal physical properties of the electromagnetic field, the model can more accurately determine the true penetration depth of degradation at the boundary.

[0081] The Transformer model, through its self-attention mechanism, can process in parallel the photoelectric detection information of suspected deterioration points and the electromagnetic property detection data of the medium at the boundary of deterioration influence. It can also capture the numerical correspondence between the bubble density sequence within the matrix and the electromagnetic wave attenuation rate. The encoder of the Transformer model can identify spatial regions with drastic fluctuations in dielectric loss tangent and conductivity, and correlate these electromagnetic data features with the color deviation index in the matrix discoloration information on spatial coordinates. Subsequently, the decoder uses this correlation feature to screen points near the boundary of deterioration influence that exhibit the largest deviation of the real part of the complex permittivity and the most significant decrease in transmittance, serving as severe deterioration detection points for the boundary of deterioration influence. Simultaneously, by analyzing the rate of change of the electromagnetic field distribution sequence outside the boundary, the decoder can calculate the attenuation characteristics of the medium properties with spatial distance, thereby determining the diffusion information of the boundary of deterioration influence.

[0082] Step S72: Determine the potential damage points of the morphological degradation influence boundary based on the diffusion information of the morphological degradation influence boundary.

[0083] In some embodiments, a potential damage point determination model can be used to determine potential damage points on the boundary of trait degradation influence. The potential damage point determination model is a deep neural network. The input to the potential damage point determination model is the diffusion information of the trait degradation influence boundary, and the output of the potential damage point determination model is the potential damage points on the trait degradation influence boundary.

[0084] Deep neural networks (DNNs) are multilayer perceptron structures consisting of an input layer, multiple hidden layers, and an output layer. DNNs can be optimized for parameters using backpropagation algorithms. Each layer of the model contains a large number of neurons and nonlinear activation functions, enabling hierarchical feature mapping and high-level abstraction of complex input data. Deep neural networks excel in tasks such as regression prediction and nonlinear pattern recognition, learning deep functional mapping relationships from historical data.

[0085] Potentially damaged points at the boundary of morphological degradation are spatial coordinates that predict potential qualitative changes or early-stage micro-deterioration within a short period. These potential damaged points include their coordinate locations, the estimated time window for qualitative change, and the probability of initial degradation.

[0086] The diffusion information of the property degradation effect boundary records the direction vector of the outward extension of the deteriorated region and the displacement value per unit time, which can reflect the spatial penetration law of the matrix deterioration state within the encapsulation unit.

[0087] Deep neural networks utilize multiple hidden layers to nonlinearly transform the direction vectors and gradient data in the diffusion information of the morphological degradation boundary. Through hierarchical feature mapping, deep neural networks can analyze the stability evolution trend of the matrix network structure under degradation osmotic pressure, thereby identifying spatial points along the diffusion path that currently have normal image features and electromagnetic properties but are under high degradation risk probability coverage. By quantifying and filtering the distance weights and risk probability values ​​of these points along the diffusion path, deep neural networks can accurately locate high-risk coordinates within a preset range outside the morphological degradation boundary, thus determining the potential damaged points of the morphological degradation boundary.

[0088] In some embodiments, determining the potential damage points of the trait degradation influence boundary based on the diffusion information of the trait degradation influence boundary includes steps S721 to S723: Step S721: Based on the diffusion information of the property degradation affecting the boundary, determine the expected arrival time series of degradation at each coordinate point outside the boundary and the displacement span range of the boundary outward expansion.

[0089] In some embodiments, a deep neural network can be used to determine the expected time series of degradation at each coordinate point outside the boundary and the displacement span range of the boundary outward expansion.

[0090] The predicted arrival time series of degradation at each coordinate point outside the boundary is a set of specific time points at which the degradation effect of traits arrives at different coordinate locations outside the boundary, predicted by a deep neural network.

[0091] The estimated arrival time series of degradation at each coordinate point outside the boundary records the mapping relationship between each three-dimensional spatial coordinate and the corresponding penetration timestamp.

[0092] The displacement span of the boundary expansion is a numerical range representing the physical distance range by which the boundary affected by the degradation of the property extends outward within a predetermined time period in the future.

[0093] The displacement span of the boundary expansion is defined by the minimum and maximum expansion displacements to limit the spatial depth that the degradation may affect.

[0094] Deep neural networks can model the diffusion information of the boundary effects of morphological degradation by utilizing the nonlinear mapping capabilities of multi-layered neurons. They can analyze the boundary movement vectors and expansion rate characteristics recorded in the diffusion information and simulate the transmission path of the degradation effect within the gel matrix encapsulation unit. By calculating the diffusion momentum in different directions, deep neural networks can predict the estimated time it takes for the degradation edge to reach specific spatial coordinates outside the boundary, thus generating a time series of predicted degradation arrival at each coordinate point outside the boundary. Simultaneously, by evaluating the attenuation trend of diffusion kinetic energy with spatial distance, deep neural networks can determine the minimum and maximum displacement boundaries that the degradation boundary may reach within a target time point, thereby determining the displacement span range of the boundary's outward expansion.

[0095] Step S722: Based on the predicted arrival time series of degradation at each coordinate point outside the boundary and the displacement span range of the boundary expansion, determine multiple predicted diffusion spatial regions, the matrix structure damage probability value of each predicted spatial region, and the matrix risk level distribution score of each predicted spatial region.

[0096] In some embodiments, a deep neural network can be used to determine multiple diffusion prediction spatial regions, the matrix structure damage probability value of each prediction spatial region, and the matrix risk level distribution score of each prediction spatial region.

[0097] Multiple diffusion prediction spatial regions are multiple local spatial ranges with the potential for component state evolution, defined by deep neural networks outside the boundary of morphological degradation influence based on degradation infiltration trajectories.

[0098] The probability of matrix structure damage in each expected spatial region is a percentage of the likelihood that the biomedical gel matrix in a specific spatial region will experience network structure breakage or performance degradation.

[0099] The matrix risk level distribution score for each projected spatial region is a quantitative score that identifies the severity of the matrix degradation effect within a specific spatial region.

[0100] The matrix risk level distribution score for each projected spatial area reflects the density of damage risk at different locations within the area, based on the score.

[0101] Deep neural networks can perform feature correlation analysis on the predicted degradation arrival time series of coordinate points outside the boundary and the displacement span of the outward expansion of the boundary through hidden layers. Deep neural networks can identify key sets of degradation stress concentrations in the spatiotemporal coordinates and delineate these spatiotemporally continuous coordinate points into multiple predicted diffusion spatial regions. Subsequently, by analyzing the energy distribution within the displacement span, deep neural networks can quantify the severity of perturbation of the matrix molecular chains in each region, thereby determining the probability of matrix structural damage in each predicted spatial region. Furthermore, deep neural networks can assess the threat weight of each region to overall stability, ultimately calculating the matrix damage risk level distribution score for each predicted spatial region.

[0102] Step S723: Based on the multiple diffusion prediction spatial regions, the matrix structure damage probability value of each prediction spatial region, and the matrix risk level distribution score of each prediction spatial region, determine the potential damage points of the trait degradation impact boundary.

[0103] In some embodiments, deep neural networks can be used to determine potential damage points at the boundary of trait degradation effects.

[0104] Deep neural networks can logically assess the geometric features of multiple predicted diffusion spatial regions, the probability of matrix structural damage in each region, and the risk level distribution of matrix involvement in each region. Based on the probability values ​​and risk scores, the model identifies the local center with the most significant stress response within each predicted diffusion spatial region. Then, a global optimization algorithm eliminates overly dense redundant regions and extracts the precise spatial coordinates of high-risk core locations. Through this precise mapping and feature filtering from region to point, deep neural networks can pinpoint a set of the most representative high-risk coordinate points within the potential spillover range of morphological degradation, thereby identifying potential damage points at the boundary of morphological degradation impact.

[0105] Step S73: Based on the severe degradation detection points of the property degradation influence boundary and the potential damage points of the property degradation influence boundary, determine multiple matrix component qualitative verification points corresponding to the property degradation influence boundary.

[0106] In some embodiments, a verification point determination model can be used to determine multiple qualitative verification points of matrix components corresponding to the boundary of trait degradation. The verification point determination model is a deep neural network. The inputs of the verification point determination model are the severely deteriorated detection points of the boundary of trait degradation and the potential damaged points of the boundary of trait degradation; the output of the verification point determination model is the multiple qualitative verification points of matrix components corresponding to the boundary of trait degradation.

[0107] The multiple matrix component qualitative verification points corresponding to the boundary of property degradation are the three-dimensional spatial coordinate points within the pharmaceutical and biological gel matrix encapsulation unit where professional qualitative testing of gel matrix components is required at the boundary of property degradation.

[0108] Multiple matrix component qualitative verification points can comprehensively cover severely degraded areas, potentially damaged areas, and key areas of deterioration diffusion at the boundary.

[0109] The qualitative verification point of matrix components is the core point for matrix component detection.

[0110] The severe degradation detection points at the boundary of the degradation effect record the spatial location of the severe shift in the physical properties of the medium, while the potential damage points at the boundary of the degradation effect identify high-risk coordinates on the outward extension path of the degradation. These two types of point data cover multiple spatial samples with different degrees of matrix damage within the encapsulation unit. By extracting the coordinates of these points and their corresponding degradation probabilities, the deep neural network can compare and analyze the matrix distribution characteristics at different degradation stages, thereby screening out multiple qualitative verification points for matrix components in the boundary and its outer regions for component re-examination.

[0111] Deep neural networks can utilize the hierarchical structure of neurons to perform logical operations on the spatial coordinates and attribute data of severely degraded detection points and potentially damaged points at the boundary of trait degradation, in order to analyze the spatial representativeness and distribution density of different points within the encapsulation unit. Deep neural networks evaluate the response strength of each point to changes in degradation traits through nonlinear mapping, automatically identifying and eliminating redundant coordinates that are spatially close and have overlapping features, and optimizing the sampling distribution structure. Based on the calculated weight values, the model can filter out feature points that reflect changes in matrix composition in the damaged area and the diffusion front region, ultimately determining multiple qualitative verification points of matrix composition corresponding to the boundary of trait degradation.

[0112] Step S8: Based on the qualitative verification points of multiple matrix components corresponding to the boundary of the property deterioration, perform qualitative identification of matrix deterioration and complete the cold chain logistics detection of the pharmaceutical biogel matrix.

[0113] Once multiple qualitative verification points of matrix components corresponding to the boundary of property degradation in the target pharmaceutical biogel matrix encapsulation unit are identified, the deterioration qualitative identification of matrix components is performed based on these multiple qualitative verification points of matrix components corresponding to the boundary of property degradation.

[0114] In some embodiments, the chemical structural characteristics of each core point can be obtained by a molecular energy level detection device, thereby determining whether a chemical deterioration has occurred at the core point, and ultimately confirming the quality damage of the pharmaceutical biogel matrix during cold chain transportation.

[0115] Based on the same inventive concept Figure 5 This is a schematic diagram of a logistics inspection system provided in an embodiment of the present invention. The logistics inspection system includes: Data acquisition module 91 is used to acquire environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit; The risk distribution map generation module 92 is used to generate a risk distribution map of the changes in the properties of the pharmaceutical biogel matrix packaging unit based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit. The suspected point determination module 93 is used to determine multiple suspected points of phenotypic deterioration based on the phenotypic change risk distribution map of the pharmaceutical biogel matrix encapsulation unit. The detection information acquisition module 94 is used to acquire photoelectric detection information of multiple suspected points of phenotypic deterioration. The degradation boundary determination module 95 is used to determine the degradation influence boundary based on the image photoelectric detection information of the multiple suspected degradation points. The detection data acquisition module 96 is used to acquire detection data of the electromagnetic properties of the medium at the boundary affected by the property degradation. The verification point determination module 97 is used to determine multiple matrix component qualitative verification points corresponding to the morphological degradation influence boundary based on the image photoelectric detection information of the multiple suspected morphological degradation points and the medium electromagnetic property detection data of the morphological degradation influence boundary. The identification and tracking module 98 is used to perform qualitative identification of matrix deterioration based on multiple matrix component qualitative verification points corresponding to the boundary of the property deterioration effect, and to complete the cold chain logistics detection of the pharmaceutical biogel matrix.

[0116] It should be noted that, in order to simplify the descriptions disclosed herein and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of embodiments of this specification may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this method of disclosure does not imply that the subject matter of this specification requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of a single embodiment disclosed above.

[0117] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A logistics inspection method, characterized in that, include: Acquire environmental damage load data at preset intervals during the transportation of pharmaceutical biogel matrix encapsulation units; Based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix encapsulation unit, a risk distribution map of the property changes of the pharmaceutical biogel matrix encapsulation unit is generated. Based on the risk distribution map of phenotypic changes in the aforementioned pharmaceutical biogel matrix encapsulation unit, several suspected points of phenotypic deterioration were identified. Acquire photoelectric detection information of multiple suspected sites of phenotypic deterioration; Based on the image photoelectric detection information of the multiple suspected trait deterioration points, the boundary of trait degradation influence is determined; Acquire electromagnetic property detection data of the medium at the boundary affected by morphological degradation; Based on the image photoelectric detection information of the multiple suspected morphological deterioration points and the medium electromagnetic property detection data of the morphological degradation influence boundary, multiple matrix component qualitative verification points corresponding to the morphological degradation influence boundary are determined. Qualitative identification of matrix deterioration is performed based on multiple matrix component qualitative verification points corresponding to the boundary of the aforementioned property degradation, thus completing the cold chain logistics detection of pharmaceutical biogel matrix.

2. The logistics inspection method as described in claim 1, characterized in that, The environmental damage load data includes spatial temperature distribution data and load vibration sequence data.

3. The logistics inspection method as described in claim 1, characterized in that, The determination of the trait degradation impact boundary based on the image photoelectric detection information of the multiple suspected trait degradation points includes: Multiple trait aberration feature clusters are obtained by clustering the image photoelectric detection information of the multiple suspected trait deterioration points; An image feature anomaly distribution map of the gel matrix is ​​generated based on the multiple trait anomaly feature clusters; Based on the abnormal distribution map of the image features of the gel matrix and the risk distribution map of the morphological changes of the pharmaceutical and biological gel matrix encapsulation unit, multiple suspected morphological deterioration areas and deterioration information of each suspected morphological deterioration area were identified. Construct a trait evolution map, which includes multiple nodes of suspected trait transformation areas and edges between multiple nodes. The node features of each suspected trait transformation area node are the transformation information of the suspected trait transformation area. The trait evolution map is processed using a graph neural network to determine the boundary of trait degradation influence.

4. The logistics inspection method as described in claim 1, characterized in that, The determination of multiple matrix component qualitative verification points corresponding to the morphological degradation influence boundary based on the image photoelectric detection information of the multiple suspected morphological degradation points and the medium electromagnetic property detection data of the morphological degradation influence boundary includes: Based on the photoelectric detection information of the multiple suspected points of property deterioration and the electromagnetic property detection data of the medium at the boundary of property degradation, the severe degradation detection points at the boundary of property degradation and the diffusion information of the boundary of property degradation are determined. Based on the diffusion information of the morphological degradation impact boundary, potential damage points of the morphological degradation impact boundary are determined; Based on the severe degradation detection points of the property degradation influence boundary and the potential damage points of the property degradation influence boundary, multiple matrix component qualitative verification points corresponding to the property degradation influence boundary are determined.

5. A logistics inspection system, characterized in that, include: The data acquisition module is used to acquire environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit; The risk distribution map generation module is used to generate a risk distribution map of the changes in the properties of the pharmaceutical biogel matrix packaging unit based on the environmental damage load data at preset intervals during the transportation of the pharmaceutical biogel matrix packaging unit. The suspected point identification module is used to identify multiple suspected points of phenotypic deterioration based on the phenotypic change risk distribution map of the pharmaceutical biogel matrix encapsulation unit. The detection information acquisition module is used to acquire photoelectric detection information of multiple suspected points of phenotypic deterioration; The degradation boundary determination module is used to determine the degradation influence boundary based on the image photoelectric detection information of the multiple suspected degradation points. The detection data acquisition module is used to acquire detection data of the electromagnetic properties of the medium at the boundary affected by the degradation of its properties; The verification point determination module is used to determine multiple matrix component qualitative verification points corresponding to the boundary of morphological degradation based on the image photoelectric detection information of the multiple suspected morphological degradation points and the medium electromagnetic property detection data of the boundary of morphological degradation influence. The identification and tracking module is used to perform qualitative identification of matrix deterioration based on multiple matrix component qualitative verification points corresponding to the boundary of the property deterioration effect, and to complete the cold chain logistics detection of the pharmaceutical biogel matrix.

6. The logistics detection system as described in claim 5, characterized in that, The environmental damage load data includes spatial temperature distribution data and load vibration sequence data.

7. The logistics detection system as described in claim 5, characterized in that, The degradation boundary determination module is also used for: Multiple trait aberration feature clusters are obtained by clustering the image photoelectric detection information of the multiple suspected trait deterioration points; An image feature anomaly distribution map of the gel matrix is ​​generated based on the multiple trait anomaly feature clusters; Based on the abnormal distribution map of the image features of the gel matrix and the risk distribution map of the morphological changes of the pharmaceutical and biological gel matrix encapsulation unit, multiple suspected morphological deterioration areas and deterioration information of each suspected morphological deterioration area were identified. Construct a trait evolution map, which includes multiple nodes of suspected trait transformation areas and edges between multiple nodes. The node features of each suspected trait transformation area node are the transformation information of the suspected trait transformation area. The trait evolution map is processed using a graph neural network to determine the boundary of trait degradation influence.

8. The logistics detection system as described in claim 5, characterized in that, The verification point determination module is also used for: Based on the photoelectric detection information of the multiple suspected points of property deterioration and the electromagnetic property detection data of the medium at the boundary of property degradation, the severe degradation detection points at the boundary of property degradation and the diffusion information of the boundary of property degradation are determined. Based on the diffusion information of the morphological degradation impact boundary, potential damage points of the morphological degradation impact boundary are determined; Based on the severe degradation detection points of the property degradation influence boundary and the potential damage points of the property degradation influence boundary, multiple matrix component qualitative verification points corresponding to the property degradation influence boundary are determined.

9. An electronic device, characterized in that, include: processor; Memory; And a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the logistics detection method as claimed in any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the logistics detection method as described in any one of claims 1 to 4.

Images