Grain logistics process data-based security assessment method and system

Abstract

The invention discloses a safety assessment method and system based on grain logistics process data, and the method comprises the steps: 1, collecting the cross-modal state data of the current batch of grain logistics process, and the grain logistics process comprises a grain purchase and collection stage, a transportation and transfer stage, and a storage stage; 2, constructing a grain logistics process data lake, and storing the cross-modal state data of the current batch of grain logistics process into the grain logistics process data lake in original and unstructured forms; 3, inputting the cross-modal state data, stored in the grain logistics process data lake, of the current batch of grain logistics process into the cross-modal safety evaluation model, performing feature extraction and cross-modal fusion by the cross-modal safety evaluation model, generating a risk coefficient, and calculating the risk coefficient of the current batch of grain logistics process; according to the invention, the cross-modal safety assessment model is utilized to carry out safety risk assessment on the grain logistics process, and the assessment result has integrity and comprehensiveness.

Description

Technical Field

[0001] The present invention relates to the field of grain quality control, and in particular to a safety assessment method and system based on grain logistics process data. Background Art

[0002] The grain industry's logistics process involves multiple implementation phases, and numerous uncontrollable factors can contribute to food quality and safety issues, leading to food safety concerns. Furthermore, the complexity of the logistics process makes it difficult to connect information, making safety assessments of the logistics process more challenging.

[0003] Most existing methods and systems use streaming data for security risk assessment, lacking a structure for data storage and processing. Furthermore, they all rely on a single source modality for security risk assessment, resulting in a lack of holistic and comprehensive assessment results. Summary of the Invention

[0004] The purpose of the present invention is to provide a safety assessment method and system based on food logistics process data, aiming to solve the problem that most of the existing disclosed related methods or systems use streaming data for safety risk assessment, lack a structure for data storage and processing, and use a single modality of source data for safety risk assessment, resulting in a lack of integrity and comprehensiveness in the assessment results.

[0005] In view of the above problems, this application provides a safety assessment method and system based on food logistics process data.

[0006] The first aspect disclosed in the present application provides a safety assessment method based on food logistics process data, the method comprising: Step 1: Collect cross-modal state data of the current batch of grain logistics process, where the grain logistics process includes the grain procurement and collection stage, the transportation and transit stage, and the warehousing stage; Step 2: Build a grain logistics process data lake and store the cross-modal status data of the current batch of grain logistics process in the grain logistics process data lake in raw and unstructured form; Step 3: Input the cross-modal status data of the current batch of grain logistics process stored in the grain logistics process data lake into the cross-modal safety assessment model. The cross-modal safety assessment model performs feature extraction and cross-modal fusion to generate a risk coefficient.

[0007] Preferably, the step 1 specifically includes: Step 1.1: Establish a phased data collection domain based on the grain logistics process, where the data collection domain includes the acquisition and integration phase domain, the transportation and transit phase domain, and the storage phase domain; Step 1.2: The acquisition and merging stage domain, the transportation and transit stage domain, and the warehousing stage domain respectively collect cross-modal state data of the acquisition and merging stage, the transportation and transit stage, and the warehousing stage of the current batch of grain logistics process.

[0008] Preferably, the step 2 specifically includes: Step 2.1: Build a grain logistics process data lake, including a main data area and a data registration area. The main data area divides the storage space into multiple storage units of different levels, which are used to store the cross-modal state data of each stage of the current batch of grain logistics process according to the size. The data registration area is used to store the cross-modal state data of each stage of the current batch of grain logistics process in the form of two tuples and pointers to the corresponding storage units in the main data area; Step 2.2: Use the snowflake algorithm to assign a unique character identifier to the cross-modal state data at each stage of the current batch of grain logistics process, and establish a mapping relationship between them; Step 2.3: Store the cross-modal status data of each stage in the current batch of grain logistics process into the main data area of ​​the grain logistics process data lake, and form a tuple of the corresponding character identifier and the pointer to the storage unit stored in the main data area into the data registration area.

[0009] Preferably, the step 3 specifically includes: Step 3.1: Based on the binary information stored in the data registration area of ​​the grain logistics process data lake, obtain the cross-modal state data of each stage of the current batch of grain logistics process stored in the main data area of ​​the grain logistics process data lake; Step 3.2: Preprocess the cross-modal state data of each stage in the current batch of grain logistics process, and then convert all data into tensor format to generate cross-modal structured data of each stage in the current batch of grain logistics process, wherein the preprocessing is a missing value filling operation; Step 3.3: Input the cross-modal structured data from each stage of the current batch of grain logistics into the cross-modal security assessment model. The cross-modal security assessment model extracts image modal data through a convolutional neural network model and text modal data through a bidirectional long short-term memory recurrent neural network. Cross-modal fusion is then performed based on a graph neural network and a self-attention mechanism neural network to generate a semantic feature vector for security assessment. Step 3.4: Input the security assessment semantic feature vector into the cascaded fully connected neural network and finally output the risk coefficient.

[0010] Preferably, the step 3.3 specifically includes: Step 3.3.1: Input the cross-modal structured data of each stage of the current batch of grain logistics process into the cross-modal safety assessment model; Step 3.3.2: For image modality data, the cross-modal security assessment model uses the N-layer filter preset in the convolutional neural network to scan and extract the spatial features of the image modality data. It then uses a pooling layer between the N-layer filters to aggregate local region features and generate a feature vector containing the image spatial features. Step 3.3.3: For text modal data, the cross-modal security assessment model uses the forward and reverse long short-term memory units of the bidirectional long short-term memory recurrent neural network to extract the front-end and back-end dependencies of the text modal data and generate a feature vector containing contextual semantic features; Step 3.3.4: Combine the feature vectors generated from the cross-modal structured data at each stage of the current batch of grain logistics into a first feature vector group. The cross-modal safety assessment model uses a graph neural network to process the first feature vector group. The first feature vector group is used as the node input of the graph neural network. The edges between the nodes represent the logical connections between the various stages. Through the graph convolution operation, the features of each stage are aggregated to generate a second feature vector group. Step 3.3.5: The cross-modal security assessment model uses the self-attention mechanism to calculate the association weights of all feature vectors in the second feature vector group, and performs a weighted sum based on the association weights to generate a third feature vector group containing cross-modal semantics; Step 3.3.6: Concatenate and integrate the feature vectors of the third feature vector group to generate a security assessment semantic feature vector.

[0011] Preferably, the step 3.4 specifically includes: Step 3.4.1: Input the security assessment semantic feature vector into a cascaded fully connected neural network, where the fully connected neural network sequentially includes M preset hidden layers and one output layer; Step 3.4.2: Each hidden layer of the fully connected neural network extracts information layer by layer through a linear transformation in the form of matrix multiplication and a ReLU nonlinear activation function; Step 3.4.3: The output layer of the fully connected neural network uses a linear transformation in the form of dot product and a sigmoid nonlinear activation function to map the vector output by the M hidden layers into a scalar value as the risk coefficient.

[0012] A second aspect disclosed in the present application provides a safety assessment system based on food logistics process data, the system being used in the above-mentioned safety assessment method based on food logistics process data, the system comprising: A collection module, the collection module is used to collect cross-modal status data of the current batch of grain logistics process, wherein the grain logistics process includes the grain procurement and collection stage, the transportation and transit stage, and the warehousing stage; A data lake module is used to build a grain logistics process data lake, storing cross-modal status data of the current batch of grain logistics process in the grain logistics process data lake in raw and unstructured form; A safety assessment module is used to input the cross-modal status data of the current batch of grain logistics processes stored in the grain logistics process data lake into a cross-modal safety assessment model. The cross-modal safety assessment model performs feature extraction and cross-modal fusion, and generates a risk coefficient through a safety assessment neural network.

[0013] The third aspect disclosed in the present application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned safety assessment method based on food logistics process data when executing the computer program.

[0014] The fourth aspect disclosed in the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned safety assessment method based on food logistics process data.

[0015] The fifth aspect disclosed in the present application provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the steps of the above-mentioned safety assessment method based on food logistics process data.

[0016] The beneficial effects of the present invention are: (1) Construct a data lake storage architecture to store the status data of the grain logistics process and conduct subsequent safety assessments based on big data processing methods. This architecture not only has the characteristics of real-time acquisition and assessment of streaming data, but also has the characteristics of easy storage and management, which is conducive to subsequent process traceability. (2) Use cross-modal multi-source data and a cross-modal security assessment model to conduct security risk assessment, so that the assessment results are holistic and comprehensive. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 This is an overall flow chart of a safety assessment method based on food logistics process data.

[0019] Figure 2 This is the overall structure diagram of a safety assessment system based on food logistics process data. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions 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 the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] Example 1: like Figure 1 As shown, the embodiment of the present application provides a safety assessment method based on food logistics process data, including: Step 1: Collect cross-modal status data of the current batch of grain logistics process, where the grain logistics process includes the grain procurement and collection stage, transportation and transit stage, and warehousing stage.

[0022] Step 1 specifically includes the following steps: Step 1.1: Establish a phased data collection domain based on the food logistics process, where the data collection domain includes the acquisition and integration phase domain, the transportation and transit phase domain, and the warehousing phase domain.

[0023] Step 1.2: The acquisition and merging stage domain, the transportation and transit stage domain, and the warehousing stage domain respectively collect cross-modal status data of the acquisition and merging stage, transportation and transit stage, and warehousing stage of the current batch of grain logistics process. Among them, the acquisition and merging stage domain collects grain quality inspection reports in PDF or text format, weighing sensor data, and spectral data of rapid pest detectors. The transportation and transit stage domain collects vehicle-mounted temperature and humidity sensor data, text-based GPS or Beidou trajectory data, and transfer station loading and unloading records. The warehousing stage domain collects grain pile temperature and humidity gradient matrices, gas concentration monitoring sequences, and manual inspection logs in unstructured text format.

[0024] Step 2: Build a grain logistics process data lake and store the cross-modal status data of the current batch of grain logistics process in the grain logistics process data lake in raw and unstructured form.

[0025] Step 2 specifically includes the following steps: Step 2.1: Build a grain logistics process data lake, including a main data area and a data registration area. The main data area divides the storage space into multiple storage units of different levels, which are used to store the cross-modal state data of each stage of the current batch of grain logistics process according to the size. The data registration area is used to store the cross-modal state data of each stage of the current batch of grain logistics process in the form of two tuples and pointers to the corresponding storage units in the main data area; Step 2.2: Use the snowflake algorithm to assign a unique character identifier to the cross-modal state data at each stage of the current batch of grain logistics process, and establish a mapping relationship between them; Step 2.3: Store the cross-modal status data of each stage in the current batch of grain logistics process into the main data area of ​​the grain logistics process data lake, and form a tuple of the corresponding character identifier and the pointer to the storage unit stored in the main data area into the data registration area.

[0026] Step 3: Input the cross-modal status data of the current batch of grain logistics process stored in the grain logistics process data lake into the cross-modal safety assessment model. The cross-modal safety assessment model performs feature extraction and cross-modal fusion to generate a risk coefficient.

[0027] Step 3 specifically includes the following steps: Step 3.1: Based on the binary information stored in the data registration area of ​​the grain logistics process data lake, obtain the cross-modal state data of each stage of the current batch of grain logistics process stored in the main data area of ​​the grain logistics process data lake; Step 3.2: Preprocess the cross-modal state data of each stage in the current batch of grain logistics process, and then convert all data into tensor format to generate cross-modal structured data of each stage in the current batch of grain logistics process. The preprocessing is a missing value filling operation; Step 3.3: Input the cross-modal structured data from each stage of the current batch of grain logistics into the cross-modal safety assessment model. The cross-modal safety assessment model extracts image modal data through a convolutional neural network model and text modal data through a bidirectional long short-term memory recurrent neural network. Cross-modal fusion is then performed based on a graph neural network and a self-attention mechanism neural network to generate a safety assessment semantic feature vector. Step 3.4: Input the security assessment semantic feature vector into the cascaded fully connected neural network and finally output the risk coefficient.

[0028] Step 3.3 specifically includes the following steps: Step 3.3.1: Input the cross-modal structured data of each stage of the current batch of grain logistics process into the cross-modal safety assessment model; Step 3.3.2: For image modality data, the cross-modal security assessment model uses the N-layer filter preset in the convolutional neural network to scan and extract the spatial features of the image modality data. It then uses a pooling layer between the N-layer filters to aggregate local region features and generate a feature vector containing the image spatial features. Step 3.3.3: For text modal data, the cross-modal security assessment model uses the forward and reverse long short-term memory units of the bidirectional long short-term memory recurrent neural network to extract the front-end and back-end dependencies of the text modal data and generate a feature vector containing contextual semantic features; Step 3.3.4: Combine the feature vectors generated from the cross-modal structured data at each stage of the current batch of grain logistics into a first feature vector group. The cross-modal safety assessment model uses a graph neural network to process the first feature vector group. The first feature vector group is used as the node input of the graph neural network. The edges between the nodes represent the logical connections between the various stages. Through the graph convolution operation, the features of each stage are aggregated to generate a second feature vector group. Step 3.3.5: The cross-modal security assessment model uses the self-attention mechanism to calculate the association weights of all feature vectors in the second feature vector group, and performs a weighted sum based on the association weights to generate a third feature vector group containing cross-modal semantics; Step 3.3.6: Concatenate and integrate the feature vectors of the third feature vector group to generate a security assessment semantic feature vector.

[0029] Step 3.4 specifically includes the following steps: Step 3.4.1: Input the security assessment semantic feature vector into a cascaded fully connected neural network, where the fully connected neural network sequentially includes M preset hidden layers and one output layer; Step 3.4.2: Each hidden layer of the fully connected neural network extracts information layer by layer through a linear transformation in the form of matrix multiplication and a ReLU nonlinear activation function; Step 3.4.3: The output layer of the fully connected neural network uses a linear transformation in the form of a dot product and a sigmoid nonlinear activation function to map the vector output by the M hidden layers into a scalar value, which serves as the risk coefficient. This coefficient ranges from 0 to 1, with higher values ​​indicating a higher safety risk in the food logistics process, such as 0.8 representing high risk and 0.3 representing low risk. Through this cascaded fully connected neural network, the model is able to gradually strip away irrelevant information from the raw cross-modal data, ultimately focusing on the core risk factors affecting food security and quantifying them.

[0030] In summary, the safety assessment method based on food logistics process data provided by the embodiments of the present application has the following technical effects: (1) Construct a data lake storage architecture to store the status data of the grain logistics process and conduct subsequent safety assessments based on big data processing methods. This architecture not only has the characteristics of real-time acquisition and assessment of streaming data, but also has the characteristics of easy storage and management, which is conducive to subsequent process traceability. (2) Use cross-modal multi-source data and a cross-modal security assessment model to conduct security risk assessment, so that the assessment results are holistic and comprehensive.

[0031] Example 2: Based on the same inventive concept as the method for safety assessment based on food logistics process data in Example 1, Figure 2 As shown, the present application provides a safety assessment system based on food logistics process data, the system comprising: A collection module, the collection module is used to collect cross-modal status data of the current batch of grain logistics process, wherein the grain logistics process includes the grain procurement and collection stage, the transportation and transit stage, and the warehousing stage; A data lake module is used to build a grain logistics process data lake, storing cross-modal status data of the current batch of grain logistics process in the grain logistics process data lake in raw and unstructured form; A safety assessment module is used to input the cross-modal status data of the current batch of grain logistics processes stored in the grain logistics process data lake into a cross-modal safety assessment model. The cross-modal safety assessment model performs feature extraction and cross-modal fusion, and generates a risk coefficient through a safety assessment neural network.

[0032] Through the above detailed description of a safety assessment method based on food logistics process data in this specification, those skilled in the art can clearly understand a safety assessment system based on food logistics process data in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For relevant matters, please refer to the method section.

[0033] Example 3: In the third embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the above-mentioned safety assessment method based on food logistics process data when executing the computer program.

[0034] Example 4: In a fourth embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned safety assessment method based on food logistics process data are implemented.

[0035] Embodiment 5: In the fifth embodiment, a computer program product is provided, including a computer program or instructions, which implement the steps of the above-mentioned safety assessment method based on food logistics process data when executed by a processor.

[0036] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0037] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A safety assessment method based on food logistics process data, characterized in that: The method comprises: Step 1: Collect cross-modal state data of the current batch of grain logistics process, where the grain logistics process includes the grain procurement and collection stage, the transportation and transit stage, and the warehousing stage; Step 2: Build a grain logistics process data lake and store the cross-modal status data of the current batch of grain logistics process in the grain logistics process data lake in raw and unstructured form; Step 3: Input the cross-modal status data of the current batch of grain logistics process stored in the grain logistics process data lake into the cross-modal safety assessment model. The cross-modal safety assessment model performs feature extraction and cross-modal fusion to generate a risk coefficient.

2. A food logistics process data-based safety assessment method according to claim 1, characterized in that: The step 1 specifically includes: Step 1.1: Establish a phased data collection domain based on the grain logistics process, where the data collection domain includes the acquisition and integration phase domain, the transportation and transit phase domain, and the storage phase domain; Step 1.2: The acquisition and merging stage domain, the transportation and transit stage domain, and the warehousing stage domain respectively collect cross-modal state data of the acquisition and merging stage, the transportation and transit stage, and the warehousing stage of the current batch of grain logistics process.

3. A food logistics process data-based safety assessment method according to claim 2, characterized in that: The step 2 specifically includes: Step 2.1: Build a grain logistics process data lake, including a main data area and a data registration area. The main data area divides the storage space into multiple storage units of different levels, which are used to store the cross-modal state data of each stage of the current batch of grain logistics process according to the size. The data registration area is used to store the cross-modal state data of each stage of the current batch of grain logistics process in the form of two tuples and pointers to the corresponding storage units in the main data area; Step 2.2: Use the snowflake algorithm to assign a unique character identifier to the cross-modal state data at each stage of the current batch of grain logistics process, and establish a mapping relationship between them; Step 2.3: Store the cross-modal status data of each stage in the current batch of grain logistics process into the main data area of ​​the grain logistics process data lake, and form a tuple of the corresponding character identifier and the pointer to the storage unit stored in the main data area into the data registration area.

4. A food logistics process data-based safety assessment method according to claim 3, characterized in that: The step 3 specifically includes: Step 3.1: Based on the binary information stored in the data registration area of ​​the grain logistics process data lake, obtain the cross-modal state data of each stage of the current batch of grain logistics process stored in the main data area of ​​the grain logistics process data lake; Step 3.2: Preprocess the cross-modal state data of each stage in the current batch of grain logistics process, and then convert all data into tensor format to generate cross-modal structured data of each stage in the current batch of grain logistics process, wherein the preprocessing is a missing value filling operation; Step 3.3: Input the cross-modal structured data from each stage of the current batch of grain logistics into the cross-modal safety assessment model. The cross-modal safety assessment model extracts image modal data through a convolutional neural network model and text modal data through a bidirectional long short-term memory recurrent neural network. Cross-modal fusion is then performed based on a graph neural network and a self-attention mechanism neural network to generate a safety assessment semantic feature vector. Step 3.4: Input the security assessment semantic feature vector into the cascaded fully connected neural network and finally output the risk coefficient.

5. A food logistics process data-based safety assessment method according to claim 4, characterized in that: The step 3.3 specifically includes: Step 3.3.1: Input the cross-modal structured data of each stage of the current batch of grain logistics process into the cross-modal safety assessment model; Step 3.3.2: For image modality data, the cross-modal security assessment model uses the N-layer filter preset in the convolutional neural network to scan and extract the spatial features of the image modality data. It then uses a pooling layer between the N-layer filters to aggregate local region features and generate a feature vector containing the image spatial features. Step 3.3.3: For text modal data, the cross-modal security assessment model uses the forward and reverse long short-term memory units of the bidirectional long short-term memory recurrent neural network to extract the front-end and back-end dependencies of the text modal data and generate a feature vector containing contextual semantic features; Step 3.3.4: Combine the feature vectors generated from the cross-modal structured data at each stage of the current batch of grain logistics into a first feature vector group. The cross-modal safety assessment model uses a graph neural network to process the first feature vector group. The first feature vector group is used as the node input of the graph neural network. The edges between the nodes represent the logical connections between the various stages. Through the graph convolution operation, the features of each stage are aggregated to generate a second feature vector group. Step 3.3.5: The cross-modal security assessment model uses the self-attention mechanism to calculate the association weights of all feature vectors in the second feature vector group, and performs a weighted sum based on the association weights to generate a third feature vector group containing cross-modal semantics; Step 3.3.6: Concatenate and integrate the feature vectors of the third feature vector group to generate a security assessment semantic feature vector.

6. A food logistics process data-based safety assessment method according to claim 5, characterized in that: The step 3.4 specifically includes: Step 3.4.1: Input the security assessment semantic feature vector into a cascaded fully connected neural network, where the fully connected neural network sequentially includes M preset hidden layers and one output layer; Step 3.4.2: Each hidden layer of the fully connected neural network extracts information layer by layer through a linear transformation in the form of matrix multiplication and a ReLU nonlinear activation function; Step 3.4.3: The output layer of the fully connected neural network uses a linear transformation in the form of dot product and a sigmoid nonlinear activation function to map the vector output by the M hidden layers into a scalar value as the risk coefficient.

7. A safety assessment system based on food logistics process data, characterized in that: The system comprises: A collection module, the collection module is used to collect cross-modal status data of the current batch of grain logistics process, wherein the grain logistics process includes the grain procurement and collection stage, the transportation and transit stage, and the warehousing stage; A data lake module is used to build a grain logistics process data lake, storing cross-modal status data of the current batch of grain logistics process in the grain logistics process data lake in raw and unstructured form; A safety assessment module is used to input the cross-modal status data of the current batch of grain logistics processes stored in the grain logistics process data lake into a cross-modal safety assessment model. The cross-modal safety assessment model performs feature extraction and cross-modal fusion, and generates a risk coefficient through a safety assessment neural network.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the processor implements the steps of a safety assessment method based on food logistics process data as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a safety assessment method based on food logistics process data according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of a safety assessment method based on food logistics process data described in any one of claims 1 to 6 are implemented.

Images