A blockchain-based drug supply chain traceability system

Abstract

The present invention discloses a drug supply chain traceability system based on blockchain, including a data management module, a drug identity authentication module, an anti-counterfeiting traceability module, a real-time monitoring module, a blockchain smart contract management module and a feedback interaction module. The present invention belongs to the field of drug supply chain management technology, specifically a drug supply chain traceability system based on blockchain. The present invention adopts an optimized design combining drug identity authentication, anti-counterfeiting traceability and real-time monitoring, and combines blockchain technology to improve the overall automation, intelligence and scalability of the drug supply chain traceability system design. By applying optimized search algorithms and optimized encryption algorithms in blockchain smart contract management, the security of blockchain technology as a technical foundation is also improved; drug identity authentication is performed using a feature attention improvement model combined with a pre-trained network; and real-time monitoring is performed using a temporal convolution long-term parallel network with local polynomial filtering.

Description

Technical Field

[0001] The present invention belongs to the technical field of drug supply chain management, and specifically is a drug supply chain traceability system based on blockchain. Background Art

[0002] The blockchain-based drug supply chain traceability system is a solution that uses blockchain technology to ensure the transparency, traceability and security of the entire process from drug production to consumer. The system records key drug data (such as production, transportation, storage and other information) on the blockchain in an unalterable form, allowing all parties involved (such as manufacturers, distributors, retailers and consumers) to access drug information in real time to ensure the authenticity and safety of drugs. Its main functions include preventing counterfeit and substandard drugs from entering the market, improving the efficiency and transparency of drug supervision, and enhancing consumers' trust in drug safety. By combining deep learning technology for anomaly detection and early warning, the system can also achieve real-time monitoring and prevention of potential risks in the drug circulation process.

[0003] However, in the existing blockchain-based drug supply chain traceability system, there is a technical problem that the traditional methods for optimizing the traceability of the drug supply chain simply focus on the model structure design of the blockchain, but ignore the multi-stage assistance and support of intelligent methods for drug traceability, resulting in the drug supply chain traceability still relying on manual inspection in essence, and the traditional methods are difficult to provide an objective and automatic drug supply chain traceability solution; in the existing drug identity verification methods, there is a technical problem that the core problem of intelligent methods in applying blockchain drug supply chain traceability lies in computational efficiency, and the existing technology is difficult to efficiently extract drug identity verification information, and the accuracy is also insufficient; in the existing anti-counterfeiting traceability methods, there is a technical problem that drug anti-counterfeiting cannot only consider the detection of drug labels, and the abnormalities and disguises of drug packaging must also be taken into consideration; in the existing real-time monitoring methods, there is a technical problem that the drug supply chain must not only be inspected at the supplier level, but errors may also occur in the transportation of drugs and other production links that can be monitored, which leads to reduced availability of intelligent detection results. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a drug supply chain traceability system based on blockchain. In view of the existing drug supply chain traceability system based on blockchain, there is a problem that the traditional method for optimizing the traceability of the drug supply chain simply focuses on the model structure design of the blockchain, but ignores the multi-stage assistance and support of the intelligent method for drug traceability, resulting in the drug supply chain traceability still relying on manual detection in essence, and the traditional method is difficult to provide an objective and automatic drug supply chain traceability solution. The technical problem, this solution creatively adopts the optimization design combining the three intelligent functions of drug identity authentication, anti-counterfeiting traceability and real-time monitoring, and combines blockchain technology to design the drug supply chain traceability system to improve the overall automation, intelligence and scalability of the solution. At the same time, the three intelligent functions adopt a separate design, which can be maintained and replaced separately, which improves the versatility and practicality of the system itself. By applying the optimized search algorithm and the optimized encryption algorithm in the blockchain smart contract management, the security of the blockchain technology as the technical basis is also improved. In view of the existing drug identity authentication method, the core problem of the intelligent method in the application of blockchain drug supply chain traceability is the computational efficiency. The existing technology is difficult to efficiently extract drug identity. In order to solve the technical problem that the drug identity verification information is insufficient and the accuracy is insufficient, this solution creatively adopts a feature attention improvement model combined with a pre-trained network to perform drug identity verification. By extracting the key tags and information of drug identity verification and using a pre-trained model combined with an encoding and decoding improvement model, it can not only improve the computational efficiency of drug identity verification and reduce the consumption of computing resources, but also improve the overall verification efficiency. In view of the technical problem that in the existing anti-counterfeiting and traceability methods, the anti-counterfeiting of drugs cannot only consider the detection of drug labels, but also the abnormalities and disguises of drug packaging need to be taken into consideration, this solution creatively adopts a feature attention improvement model Anti-counterfeiting traceability of drug packaging is carried out, and comprehensive anti-counterfeiting detection is carried out by combining drug identity authentication information, which improves the detectability of anti-counterfeiting traceability at the supplier level; in the existing real-time monitoring methods, there is a technical problem that the drug supply chain must not only be detected from the supplier level, but errors may also occur in the transportation of drugs and other production links that can be monitored, which in turn leads to reduced availability of intelligent detection results. This solution creatively uses a local polynomial filtered temporal convolution long-short-term parallel network for real-time monitoring. By also monitoring the transportation process in real time, it provides strong support and auxiliary reference for the overall drug supply chain traceability.

[0005] The technical solution adopted by the present invention is as follows: A drug supply chain traceability system based on blockchain provided by the present invention includes a data management module, a drug identity authentication module, an anti-counterfeiting traceability module, a real-time monitoring module, a blockchain smart contract management module and a feedback interaction module;

[0006] The data management module is used to collect, process and store drug supply chain process data and perform data transmission and interaction between modules. Through data management, a drug supply chain traceability data set is obtained, the drug supply chain traceability data set is uploaded to the blockchain, and sent to the drug identity authentication module, the anti-counterfeiting traceability module and the real-time monitoring module;

[0007] The drug identity verification module is used to verify the link information of the drug through the label information, obtain the authenticity verification data of the drug information through the drug identity verification, upload the authenticity verification data of the drug information to the blockchain, and send it to the anti-counterfeiting traceability module, the blockchain smart contract management module and the feedback interaction module;

[0008] The anti-counterfeiting and traceability module is used to analyze and detect the counterfeiting of drugs through packaging images, obtain the anti-counterfeiting evaluation data of drug packaging through anti-counterfeiting and traceability, upload the anti-counterfeiting evaluation data of drug packaging to the blockchain, and send it to the real-time monitoring module, the blockchain smart contract management module and the feedback interaction module;

[0009] The real-time monitoring module is used to track the transportation and storage status of drugs in real time and enhance the traceability of the supply chain. Through real-time monitoring, the evaluation data of the drug transportation process is obtained, and the evaluation data of the drug transportation process is uploaded to the blockchain and sent to the blockchain smart contract management module and the feedback interaction module;

[0010] The blockchain smart contract management module is used to manage transaction data in the drug supply chain through the blockchain index, obtain data verification signature information through blockchain smart contract management, and execute the drug supply chain smart contract by verifying the aforementioned information based on the data;

[0011] The feedback interaction module is used to provide a user-interactive feedback function, and through feedback interaction, users can query, use and manage the traceability of the drug supply chain.

[0012] Furthermore, the data management is specifically to obtain the original data set of the drug supply chain through data collection from the drug production link, and to obtain the drug supply chain traceability data set by data classification and preprocessing;

[0013] The drug supply chain original data set specifically includes identity authentication original data, anti-counterfeiting and traceability original data, and real-time monitoring original data;

[0014] The data classification and preprocessing specifically refers to obtaining a drug supply chain traceability data set by sequentially performing data classification operations and data preprocessing operations;

[0015] The data classification operation specifically refers to dividing the drug supply chain original data set into identity authentication original data, anti-counterfeiting and traceability original data, and real-time monitoring original data according to the data type;

[0016] The data preprocessing operation is specifically to preprocess the identity authentication original data, anti-counterfeiting and traceability original data and real-time monitoring original data in sequence to obtain a drug supply chain traceability data set, including the following steps: identity authentication data preprocessing, anti-counterfeiting and traceability data preprocessing, real-time monitoring data preprocessing and comprehensive preprocessing;

[0017] The identity verification data preprocessing specifically refers to performing denoising and image enhancement operations on the drug label image, and performing missing value processing operations on the drug label information to obtain identity verification optimization data;

[0018] The anti-counterfeiting and traceability data preprocessing specifically refers to performing denoising, graying, contrast adjustment and forged drug image sample generation operations on the drug appearance image, performing missing value processing and standardization operations on the drug label information, and obtaining anti-counterfeiting and traceability optimization data;

[0019] The real-time monitoring data preprocessing specifically refers to performing missing value filling, standardization and denoising operations on the transport process sensor data and the transport process positioning data to obtain real-time monitoring optimization data;

[0020] The comprehensive preprocessing specifically refers to performing a comprehensive preprocessing operation through the identity verification data preprocessing, the anti-counterfeiting traceability data preprocessing and the real-time monitoring data preprocessing to obtain a drug supply chain traceability data set;

[0021] The pharmaceutical supply chain traceability data set specifically includes identity authentication optimization data, anti-counterfeiting traceability optimization data and real-time monitoring optimization data.

[0022] Furthermore, the drug identity verification is specifically based on the identity verification optimization data in the drug supply chain traceability data set, using a feature attention improvement model combined with a pre-trained network to perform drug identity verification to obtain drug information authenticity verification data, and specifically includes the following steps: constructing a pre-trained network layer, constructing a feature attention improved U-type model, drug identity verification model training and drug identity verification;

[0023] The construction of the pre-trained network layer specifically involves constructing a pre-trained target detection model as an initial feature extraction layer to obtain initial feature data of drug labels;

[0024] The pre-trained target detection model specifically adopts the Yolov8 model;

[0025] The construction of the feature attention improved U-type model specifically includes constructing a feature encoding improvement layer, constructing a three-part feature parallel fusion layer, constructing a joint attention improvement layer and a feature decoding layer;

[0026] The construction of the feature coding improvement layer is used to optimize the extraction capability of local and global features of the image, and specifically includes an improved transformer layer and an efficient subnet layer. The construction steps include: constructing an improved transformer layer and constructing an efficient subnet layer;

[0027] The improved transformer layer is constructed to optimize computational efficiency and extract local features, specifically by introducing a self-attention mechanism, replacing the standard multi-head self-attention head with a double-layer alternating self-attention head structure including a windowed self-attention layer and a window translation self-attention layer, and respectively constructing a residual-connected multi-layer perceptron activated by a nonlinear activation function in the double-layer alternating self-attention head structure, performing window partition transformer improvement, and obtaining improved transformer feature data;

[0028] The efficient subnet layer is constructed to optimize computing resource consumption and extract low-level image features, specifically, a standard efficient network is connected after the improved transformer layer as an efficient subnet layer, and the improved transformer feature data is sequentially subjected to efficient convolution extraction, feature fusion and upsampling pooling operations to obtain improved feature encoding data;

[0029] The three-dimensional feature parallel fusion layer is constructed to improve the feature processing capability of different image perspectives and scales, specifically to construct a three-dimensional feature parallel fusion module for global feature fusion, the three-dimensional feature parallel fusion module specifically includes three parallel convolution branches, each convolution branch specifically includes a 3×3 convolution block, a squeeze excitation block and a 1×1 convolution block, and through the rate deviation setting, the rate of the 3×3 convolution block of the first convolution branch is set to 1, the rate of the 3×3 convolution block of the second convolution branch is set to 3, and the rate of the 3×3 convolution block of the third convolution branch is set to 5, to obtain a branch feature data set, and by constructing a softmax classifier, a pixel attention atlas is calculated, and feature map fusion is performed using element-by-element multiplication to obtain fused feature map data;

[0030] The joint attention improvement layer is constructed to optimize the recognition ability of irregular shapes, specifically by extracting dual-channel attention features according to the fused feature map data through a joint channel attention module and a spatial attention module, and aggregating channel attention features through a parallel connection method to obtain joint attention feature encoding data;

[0031] The feature decoding layer is used to perform deep feature decoding, specifically, performing a block scaling operation on the joint attention feature encoding data to obtain a deep scaling feature, and performing a feature upsampling operation to obtain a high-resolution feature map, and then constructing two consecutive improved transformer layers, and by using a skip connection, the shallow features and the deep features are feature-connected to perform feature decoding, and by constructing a standard convolution layer and an upsampling pooling layer, feature decoding is performed to obtain a decoded feature map, and a classifier layer is constructed to output the model result according to the decoded feature map;

[0032] The drug identity verification model training is specifically to train the drug identity verification model through the pre-trained network layer and the feature attention improved U-type model to obtain the drug identity verification model Model ID ;

[0033] The drug identity verification is specifically to use the drug identity verification model Model ID , performing drug identity authentication based on the identity authentication optimization data in the drug supply chain traceability data set to obtain drug information authenticity verification data;

[0034] The drug information authenticity verification data specifically refers to the drug information authenticity verification result obtained by comparing the drug information parsed from the drug label image with the drug label information stored in the blockchain. The drug information authenticity verification result specifically includes real information verification data class and forged information verification data class.

[0035] Furthermore, the anti-counterfeiting traceability is specifically based on the anti-counterfeiting traceability optimization data in the drug supply chain traceability data set and the drug information authenticity verification data, using the feature attention improvement model to perform anti-counterfeiting traceability to obtain drug packaging anti-counterfeiting evaluation data, specifically including the following steps: data reconstruction, construction of a feature attention improved U-type model, anti-counterfeiting traceability model training and anti-counterfeiting traceability;

[0036] The data reconstruction is specifically to fuse the anti-counterfeiting traceability optimization data and the drug information authenticity verification data, and obtain an anti-counterfeiting traceability training data set through manual labeling, wherein the anti-counterfeiting traceability training data set is used for anti-counterfeiting traceability model training;

[0037] The construction of the feature attention improved U-type model specifically includes constructing a feature encoding improvement layer, constructing a three-part feature parallel fusion layer, constructing a joint attention improvement layer, and a feature decoding layer;

[0038] The anti-counterfeiting traceability model training is specifically to improve the U-type model through the data reconstruction and the feature attention, and perform model training to obtain the anti-counterfeiting traceability model Model DR ;

[0039] The anti-counterfeiting traceability is specifically to use the anti-counterfeiting traceability model Model DR , based on the anti-counterfeiting traceability optimization data and the drug information authenticity verification data, anti-counterfeiting traceability is performed to obtain drug packaging anti-counterfeiting evaluation data;

[0040] The drug packaging anti-counterfeiting assessment data specifically includes real drug packaging data and counterfeit drug security data.

[0041] Furthermore, the real-time monitoring is specifically based on the real-time monitoring optimization data in the drug supply chain traceability data set and the drug packaging anti-counterfeiting evaluation data, using a local polynomial filtered temporal convolution long-term parallel network to perform real-time monitoring to obtain drug transportation process evaluation data, specifically including the following steps:

[0042] Local polynomial filtering, building time series convolutional long-term parallel networks, abnormal monitoring mechanism design, real-time monitoring model training and real-time monitoring;

[0043] The local polynomial filtering is specifically to use a Savitzky-Gorey filter to perform smoothing filtering and denoising on the sensor data in the real-time monitoring optimization data to obtain smoothed optimization signal data;

[0044] The construction of the time series convolution long-short term parallel network is used to optimize short-term fluctuation data through the time series convolution network and process long-term trend fluctuation data through the long short-term memory neural network. Specifically, a standard time series convolution network and a standard long short-term memory neural network are sequentially constructed, and the standard time series convolution network and the standard long short-term memory neural network are processed in parallel by constructing a parallel architecture, and the outputs of the standard time series convolution network and the standard long short-term memory neural network of the parallel architecture are connected by weighted summation to obtain comprehensive network output data;

[0045] The abnormal monitoring mechanism is designed to specifically set an abnormal threshold value. When the output of the integrated network output data is greater than the abnormal threshold value, abnormal feedback is performed. The abnormal feedback specifically includes abnormal time, abnormal location, actual value of abnormal temperature and humidity value, and abnormal temperature and humidity predicted value;

[0046] The real-time monitoring model training is specifically to perform model training through the local polynomial filtering, the time series convolution long-term parallel network and the abnormal monitoring mechanism design to obtain the real-time monitoring model Model TL ;

[0047] The real-time monitoring is specifically to use the real-time monitoring model Model TL, based on the real-time monitoring optimization data in the drug supply chain traceability data set and the drug packaging anti-counterfeiting evaluation data, real-time monitoring is performed to obtain drug transportation process evaluation data.

[0048] Furthermore, the blockchain smart contract management specifically includes building a drug supply chain traceability blockchain security model, performing blockchain smart contract management, obtaining data verification signature information, and executing the drug supply chain smart contract by verifying the aforementioned information based on the data;

[0049] The drug supply chain traceability blockchain security model specifically includes a sending entity, a receiving entity, a transaction entity, an optimized search algorithm, and an optimized encryption algorithm;

[0050] The sending entity specifically refers to the sender of the drug smart contract, who creates and starts the smart contract after passing the drug identity authentication and anti-counterfeiting traceability, and provides drug information and drug entity;

[0051] The receiving entity specifically refers to the recipient of the drug smart contract who performs drug traceability management through real-time monitoring, and confirms drug counterfeiting by combining drug identity authentication and anti-counterfeiting traceability, completes smart contract management, and obtains data verification signature information;

[0052] The optimization search algorithm specifically refers to a parallel search algorithm based on a directed acyclic graph maximum matching algorithm, which is specifically constructed in the blockchain network topology to solve the maximum matching problem and serve as an optimization search algorithm;

[0053] The optimized encryption algorithm specifically refers to the hash chain of Chebyshev polynomials.

[0054] Furthermore, the feedback interaction is specifically to realize the user's query, use and management of drug supply chain traceability based on blockchain by sequentially executing data management, drug identity verification, anti-counterfeiting traceability, real-time monitoring and blockchain smart contract management.

[0055] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0056] (1) In the existing blockchain-based drug supply chain traceability system, there is a problem that traditional methods for optimizing drug supply chain traceability simply focus on the design of blockchain model structure, but ignore the multi-stage assistance and support of intelligent methods for drug traceability, resulting in the drug supply chain traceability still relying on manual inspection. Traditional methods are difficult to provide an objective and automatic drug supply chain traceability solution. This solution creatively adopts the optimization design of combining three intelligent functions: drug identity authentication, anti-counterfeiting traceability and real-time monitoring, and combines blockchain technology to design a drug supply chain traceability system to improve the overall automation, intelligence and scalability of the solution. At the same time, the three intelligent functions adopt a separate design and can be maintained and replaced separately, which improves the versatility and practicality of the system itself. By applying optimized search algorithms and optimized encryption algorithms in blockchain smart contract management, the security of blockchain technology as the technical foundation is also improved.

[0057] (2) In view of the fact that the core problem of applying intelligent methods to blockchain drug supply chain traceability is computational efficiency, the existing technology is difficult to efficiently extract drug authentication information, and the accuracy is insufficient, this solution creatively uses a feature attention improved model combined with a pre-trained network to perform drug authentication. By extracting key tags and information of drug authentication, and using a pre-trained model combined with an encoding and decoding improved model, it can not only improve the computational efficiency of drug authentication and reduce computing resource consumption, but also improve the overall verification efficiency.

[0058] (3) In view of the technical problem that the existing anti-counterfeiting and traceability methods cannot only consider the detection of drug labels, but also need to take into account the abnormalities and disguises of drug packaging, this solution creatively adopts the feature attention improvement model to carry out the anti-counterfeiting and traceability of drug packaging, and combines the drug identity verification information for comprehensive anti-counterfeiting detection, thereby improving the detectability of anti-counterfeiting and traceability at the supplier level;

[0059] (4) In view of the technical problem that the existing real-time monitoring methods have the problem that the drug supply chain must not only be monitored at the supplier level, but errors may also occur in the transportation of drugs and other production links that can be monitored, which in turn leads to reduced availability of intelligent detection results, this solution creatively adopts a local polynomial filtered temporal convolutional long-term parallel network for real-time monitoring. By also monitoring the transportation process in real time, it provides strong support and auxiliary reference for the overall drug supply chain traceability. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic diagram of the structure of a drug supply chain traceability system based on blockchain provided by the present invention;

[0061] Figure 2 A flowchart of the steps performed by the data preprocessing operation in the data management module;

[0062] Figure 3 A flowchart of the steps performed by the drug identity verification module;

[0063] Figure 4 Flowchart of the steps performed to build the feature attention improved U-model;

[0064] Figure 5 This is a schematic diagram of the result visualization subgraph cluster of the anti-counterfeiting traceability module;

[0065] Figure 6 Schematic diagram of the subgraph cluster for the result visualization of the real-time monitoring module.

[0066] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0067] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0068] Example 1, see Figure 1 , the technical solution adopted by the present invention is as follows: a drug supply chain traceability system based on blockchain provided by the present invention includes a data management module, a drug identity authentication module, an anti-counterfeiting traceability module, a real-time monitoring module, a blockchain smart contract management module and a feedback interaction module;

[0069] The data management module is used to collect, process and store drug supply chain process data and perform data transmission and interaction between modules. Through data management, a drug supply chain traceability data set is obtained, the drug supply chain traceability data set is uploaded to the blockchain, and sent to the drug identity authentication module, the anti-counterfeiting traceability module and the real-time monitoring module;

[0070] The drug identity verification module is used to verify the link information of the drug through the label information, obtain the authenticity verification data of the drug information through the drug identity verification, upload the authenticity verification data of the drug information to the blockchain, and send it to the anti-counterfeiting traceability module, the blockchain smart contract management module and the feedback interaction module;

[0071] The anti-counterfeiting and traceability module is used to analyze and detect the counterfeiting of drugs through packaging images, obtain the anti-counterfeiting evaluation data of drug packaging through anti-counterfeiting and traceability, upload the anti-counterfeiting evaluation data of drug packaging to the blockchain, and send it to the real-time monitoring module, the blockchain smart contract management module and the feedback interaction module;

[0072] The real-time monitoring module is used to track the transportation and storage status of drugs in real time and enhance the traceability of the supply chain. Through real-time monitoring, the evaluation data of the drug transportation process is obtained, and the evaluation data of the drug transportation process is uploaded to the blockchain and sent to the blockchain smart contract management module and the feedback interaction module;

[0073] The blockchain smart contract management module is used to manage transaction data in the drug supply chain through the blockchain index, obtain data verification signature information through blockchain smart contract management, and execute the drug supply chain smart contract by verifying the aforementioned information based on the data;

[0074] The feedback interaction module is used to provide a user-interactive feedback function, and through feedback interaction, users can query, use and manage the traceability of the drug supply chain.

[0075] Embodiment 2, this embodiment is based on the above embodiment, see Figure 1 and Figure 2 The data management specifically includes obtaining the original data set of the drug supply chain through data collection from the drug production link, and performing data classification and preprocessing to obtain the drug supply chain traceability data set;

[0076] The drug production process specifically includes direct drug production, drug transportation and drug distribution;

[0077] The drug supply chain original data set specifically includes identity authentication original data, anti-counterfeiting and traceability original data, and real-time monitoring original data;

[0078] The identity verification original data specifically refers to the drug label image and drug label information;

[0079] The drug label information specifically includes the drug QR code identifier, drug batch number, manufacturer, production date and expiration date information;

[0080] The anti-counterfeiting and traceability original data specifically refers to the drug appearance image and drug label information;

[0081] The drug appearance images specifically include real drug appearance images and forged drug appearance images;

[0082] The real-time monitoring raw data specifically refers to the transport process sensor data and the transport process positioning data;

[0083] The transport process sensing data specifically includes temperature data and humidity data of the drug transport process;

[0084] The transportation process positioning data specifically refers to the transportation process geographic location calibration data synchronized with the timestamp;

[0085] The data classification and preprocessing specifically refers to obtaining a drug supply chain traceability data set by sequentially performing data classification operations and data preprocessing operations;

[0086] The data classification operation specifically refers to dividing the drug supply chain original data set into identity authentication original data, anti-counterfeiting and traceability original data, and real-time monitoring original data according to the data type;

[0087] The data preprocessing operation is specifically to preprocess the identity authentication original data, anti-counterfeiting and traceability original data and real-time monitoring original data in sequence to obtain a drug supply chain traceability data set, including the following steps: identity authentication data preprocessing, anti-counterfeiting and traceability data preprocessing, real-time monitoring data preprocessing and comprehensive preprocessing;

[0088] The identity verification data preprocessing specifically refers to performing denoising and image enhancement operations on the drug label image, and performing missing value processing operations on the drug label information to obtain identity verification optimization data;

[0089] The anti-counterfeiting and traceability data preprocessing specifically refers to performing denoising, graying, contrast adjustment and forged drug image sample generation operations on the drug appearance image, performing missing value processing and standardization operations on the drug label information, and obtaining anti-counterfeiting and traceability optimization data;

[0090] The real-time monitoring data preprocessing specifically refers to performing missing value filling, standardization and denoising operations on the transport process sensor data and the transport process positioning data to obtain real-time monitoring optimization data;

[0091] The comprehensive preprocessing specifically refers to performing a comprehensive preprocessing operation through the identity verification data preprocessing, the anti-counterfeiting traceability data preprocessing and the real-time monitoring data preprocessing to obtain a drug supply chain traceability data set;

[0092] The pharmaceutical supply chain traceability data set specifically includes identity authentication optimization data, anti-counterfeiting traceability optimization data and real-time monitoring optimization data.

[0093] Embodiment 3, this embodiment is based on the above embodiment, see Figure 3 and Figure 4, the drug identity verification, specifically, is to perform drug identity verification based on the identity verification optimization data in the drug supply chain traceability data set, using a feature attention improvement model combined with a pre-trained network, to obtain drug information authenticity verification data, specifically including the following steps: constructing a pre-trained network layer, constructing a feature attention improved U-type model, drug identity verification model training and drug identity verification;

[0094] The construction of the pre-trained network layer specifically involves constructing a pre-trained target detection model as an initial feature extraction layer to obtain initial feature data of drug labels;

[0095] The pre-trained target detection model specifically adopts the Yolov8 model;

[0096] The construction of the feature attention improved U-type model specifically includes constructing a feature encoding improvement layer, constructing a three-part feature parallel fusion layer, constructing a joint attention improvement layer and a feature decoding layer;

[0097] The construction of the feature coding improvement layer is used to optimize the extraction capability of local and global features of the image, and specifically includes an improved transformer layer and an efficient subnet layer. The construction steps include: constructing an improved transformer layer and constructing an efficient subnet layer;

[0098] The improved transformer layer is constructed to optimize the computational efficiency and extract local features. Specifically, the self-attention mechanism is introduced, and the standard multi-head self-attention head is replaced by a double-layer alternating self-attention head structure including a windowed self-attention layer and a window translation self-attention layer. In the double-layer alternating self-attention head structure, a residual connected multi-layer perceptron activated by a nonlinear activation function is constructed, and the window partition transformer is improved to obtain the improved transformer feature data. The calculation formula is:

[0099] ;

[0100] In the formula, is the improved transformer feature data output of the lth windowed shifted self-attention layer, l is the level index of the double-layer alternating self-attention head structure, WMSA(·) is the windowed self-attention layer operation function, LN(·) is the layer normalization function, is the improved transformer feature data output of the l-1th windowed self-attention layer, is the improved transformer feature data output of the lth windowed self-attention layer, is the improved transformer feature data output of the l+1th window-shifted self-attention layer, MLP(·) is the residual connected multi-layer perceptron operation function activated by the nonlinear activation function, SWMSA(·) is the window-shifted self-attention layer operation function, It is the improved transformer feature data output of the l+1th windowed self-attention layer;

[0101] The efficient subnet layer is constructed to optimize computing resource consumption and extract low-level image features, specifically, a standard efficient network is connected after the improved transformer layer as an efficient subnet layer, and the improved transformer feature data is sequentially subjected to efficient convolution extraction, feature fusion and upsampling pooling operations to obtain improved feature encoding data;

[0102] The three-way feature parallel fusion layer is constructed to improve the feature processing capability of different image perspectives and scales. Specifically, a three-way feature parallel fusion module is constructed to perform global feature fusion. The three-way feature parallel fusion module specifically includes three parallel convolution branches, each of which specifically includes a 3×3 convolution block, a squeeze excitation block and a 1×1 convolution block. By setting the rate deviation, the rate of the 3×3 convolution block of the first convolution branch is set to 1, the rate of the 3×3 convolution block of the second convolution branch is set to 3, and the rate of the 3×3 convolution block of the third convolution branch is set to 5 to obtain a branch feature data set, and by constructing a softmax classifier, a pixel attention atlas is calculated, and feature map fusion is performed using element-by-element multiplication to obtain fused feature map data. The calculation formula is:

[0103] ;

[0104] In the formula, F fusion is the fused feature map data, A is the pixel attention map of the first convolution branch, and F A is the branch feature data of the first convolution branch, B is the pixel attention map of the second convolution branch, and F B is the branch feature data of the second convolution branch, C is the pixel attention map of the third convolution branch, and F C is the branch feature data of the third convolution branch, is the element-wise multiplication operator;

[0105] The joint attention improvement layer is constructed to optimize the recognition ability of irregular shapes, specifically by extracting dual-channel attention features according to the fused feature map data through a joint channel attention module and a spatial attention module, and aggregating channel attention features through a parallel connection method to obtain joint attention feature encoding data;

[0106] The feature decoding layer is used to perform deep feature decoding, specifically, performing a block scaling operation on the joint attention feature encoding data to obtain a deep scaling feature, and performing a feature upsampling operation to obtain a high-resolution feature map, and then constructing two consecutive improved transformer layers, and by using a skip connection, the shallow features and the deep features are feature-connected to perform feature decoding, and by constructing a standard convolution layer and an upsampling pooling layer, feature decoding is performed to obtain a decoded feature map, and a classifier layer is constructed to output the model result according to the decoded feature map;

[0107] The drug identity verification model training is specifically to train the drug identity verification model through the pre-trained network layer and the feature attention improved U-type model to obtain the drug identity verification model Model ID ;

[0108] The drug identity verification is specifically to use the drug identity verification model Model ID , performing drug identity authentication based on the identity authentication optimization data in the drug supply chain traceability data set to obtain drug information authenticity verification data;

[0109] The drug information authenticity verification data specifically refers to the drug information authenticity verification result obtained by comparing the drug information parsed from the drug label image with the drug label information stored in the blockchain. The drug information authenticity verification result specifically includes real information verification data class and forged information verification data class.

[0110] By performing the above operations, in view of the fact that in the existing drug identity authentication methods, the core problem of applying intelligent methods to blockchain drug supply chain traceability is computational efficiency, and the existing technology is difficult to efficiently extract drug identity authentication information, and the accuracy is also insufficient, this solution creatively adopts a feature attention improved model combined with a pre-trained network to perform drug identity authentication. By extracting the key tags and information of drug identity authentication, and using a pre-trained model combined with an encoding and decoding improved model, it can not only improve the computational efficiency of drug identity authentication and reduce the consumption of computing resources, but also improve the overall verification efficiency.

[0111] Embodiment 4, this embodiment is based on the above embodiment, see Figure 5 The anti-counterfeiting traceability is specifically to use the feature attention improvement model to perform anti-counterfeiting traceability based on the anti-counterfeiting traceability optimization data in the drug supply chain traceability data set and the drug information authenticity verification data to obtain drug packaging anti-counterfeiting evaluation data, which specifically includes the following steps: data reconstruction, construction of a feature attention improved U-type model, anti-counterfeiting traceability model training and anti-counterfeiting traceability;

[0112] The data reconstruction is specifically to fuse the anti-counterfeiting traceability optimization data and the drug information authenticity verification data, and obtain an anti-counterfeiting traceability training data set through manual labeling, wherein the anti-counterfeiting traceability training data set is used for anti-counterfeiting traceability model training;

[0113] The construction of the feature attention improved U-type model specifically includes constructing a feature encoding improvement layer, constructing a three-part feature parallel fusion layer, constructing a joint attention improvement layer, and a feature decoding layer;

[0114] The anti-counterfeiting traceability model training is specifically to improve the U-type model through the data reconstruction and the feature attention, and perform model training to obtain the anti-counterfeiting traceability model Model DR ;

[0115] The anti-counterfeiting traceability is specifically to use the anti-counterfeiting traceability model Model DR , based on the anti-counterfeiting traceability optimization data and the drug information authenticity verification data, anti-counterfeiting traceability is performed to obtain drug packaging anti-counterfeiting evaluation data;

[0116] The drug packaging anti-counterfeiting assessment data specifically includes real drug packaging data and counterfeit drug security data.

[0117] By executing the above operations, in order to address the technical problem that in the existing anti-counterfeiting and traceability methods, the anti-counterfeiting of drugs cannot only consider the detection of drug labels, but also the abnormalities and disguises of drug packaging need to be taken into consideration, this solution creatively adopts the feature attention improvement model to perform anti-counterfeiting and traceability of drug packaging, and combines drug identity authentication information for comprehensive anti-counterfeiting detection, thereby improving the detectability of anti-counterfeiting and traceability at the supplier level.

[0118] Embodiment 5, this embodiment is based on the above embodiment, see Figure 6 The real-time monitoring is specifically based on the real-time monitoring optimization data in the drug supply chain traceability data set and the drug packaging anti-counterfeiting evaluation data, using a local polynomial filtered temporal convolution long-term parallel network to perform real-time monitoring to obtain drug transportation process evaluation data, specifically including the following steps:

[0119] Local polynomial filtering, building time series convolutional long-term parallel networks, abnormal monitoring mechanism design, real-time monitoring model training and real-time monitoring;

[0120] The local polynomial filtering is specifically to use the Savitzky-Gorey filter to smooth and filter the sensor data in the real-time monitoring optimization data and remove noise to obtain smooth optimization signal data. The calculation formula is:

[0121] ;

[0122] In the formula, y t is the smoothing optimization signal data, t is the time index, M is the smoothing coefficient, k is the filter coefficient index, k=-k is the smoothing range before and after the filter window, a k is the filter coefficient, which is obtained by the least squares method, x t+k It is the original data, specifically the sensor data in the real-time monitoring optimization data;

[0123] The construction of the time series convolution long-short term parallel network is used to optimize short-term fluctuation data through the time series convolution network and process long-term trend fluctuation data through the long short-term memory neural network. Specifically, a standard time series convolution network and a standard long short-term memory neural network are sequentially constructed, and the standard time series convolution network and the standard long short-term memory neural network are processed in parallel by constructing a parallel architecture, and the outputs of the standard time series convolution network and the standard long short-term memory neural network of the parallel architecture are connected by weighted summation to obtain comprehensive network output data;

[0124] The abnormal monitoring mechanism is designed to specifically set an abnormal threshold value. When the output of the integrated network output data is greater than the abnormal threshold value, abnormal feedback is performed. The abnormal feedback specifically includes abnormal time, abnormal location, actual value of abnormal temperature and humidity value, and abnormal temperature and humidity predicted value;

[0125] The real-time monitoring model training is specifically to perform model training through the local polynomial filtering, the time series convolution long-term parallel network and the abnormal monitoring mechanism design to obtain the real-time monitoring model Model TL ;

[0126] The real-time monitoring is specifically to use the real-time monitoring model Model TL , based on the real-time monitoring optimization data in the drug supply chain traceability data set and the drug packaging anti-counterfeiting evaluation data, real-time monitoring is performed to obtain drug transportation process evaluation data.

[0127] By executing the above operations, in order to address the technical problem that in existing real-time monitoring methods, the drug supply chain must not only be inspected at the supplier level, but errors may also occur in the transportation of drugs and other production links that can be monitored, thereby reducing the availability of intelligent detection results, this solution creatively adopts a local polynomial filtered temporal convolution long-short-term parallel network for real-time monitoring. By also monitoring the transportation process in real time, it provides strong support and auxiliary reference for the overall drug supply chain traceability.

[0128] Embodiment 6, this embodiment is based on the above embodiment, see Figure 1, the blockchain smart contract management specifically includes building a drug supply chain traceability blockchain security model, performing blockchain smart contract management, obtaining data verification signature information, and executing the drug supply chain smart contract by verifying the aforementioned information based on the data;

[0129] The drug supply chain traceability blockchain security model specifically includes a sending entity, a receiving entity, a transaction entity, an optimized search algorithm, and an optimized encryption algorithm;

[0130] The sending entity specifically refers to the sender of the drug smart contract, who creates and starts the smart contract after passing the drug identity authentication and anti-counterfeiting traceability, and provides drug information and drug entity;

[0131] The receiving entity specifically refers to the recipient of the drug smart contract who performs drug traceability management through real-time monitoring, and confirms drug counterfeiting by combining drug identity authentication and anti-counterfeiting traceability, completes smart contract management, and obtains data verification signature information;

[0132] The optimization search algorithm specifically refers to a parallel search algorithm based on a directed acyclic graph maximum matching algorithm, which is specifically constructed in the blockchain network topology to solve the maximum matching problem and serve as an optimization search algorithm;

[0133] The optimized encryption algorithm specifically refers to the hash chain of Chebyshev polynomials;

[0134] The calculation formula of the hash chain of the Chebyshev polynomial is:

[0135] ;

[0136] In the formula, H n is the hash value corresponding to the nth hash chain index, Hash(·) is the standard hash function SHA-256, T j (H n-1 ) is the mapping result of applying Chebyshev polynomial to the hash value corresponding to the n-1th hash chain index, j is the order of Chebyshev polynomial, x is the input data mapped to the interval [-1,1], used to represent the data to be encrypted, cos(·) is the cosine function, and arccos(·) is the inverse cosine function;

[0137] The feedback interaction specifically implements data management, drug identity authentication, anti-counterfeiting traceability, real-time monitoring and blockchain smart contract management in sequence, so as to enable users to query, use and manage the drug supply chain traceability based on blockchain.

[0138] By executing the above operations, in the existing blockchain-based drug supply chain traceability system, there is a problem that traditional methods for optimizing drug supply chain traceability simply focus on the model structure design of the blockchain, but ignore the multi-stage assistance and support of intelligent methods for drug traceability, resulting in the drug supply chain traceability still relying on manual inspection in essence, and the traditional method is difficult to provide an objective and automatic drug supply chain traceability solution. Technical problems, this solution creatively adopts an optimized design that combines the three intelligent functions of drug identity authentication, anti-counterfeiting traceability and real-time monitoring, and combines blockchain technology to design a drug supply chain traceability system to improve the overall automation, intelligence and scalability of the solution. At the same time, the three intelligent functions adopt a separate design and can be maintained and replaced separately, which improves the versatility and practicality of the system itself. By applying optimized search algorithms and optimized encryption algorithms in blockchain smart contract management, the security of blockchain technology as the technical foundation is also improved.

[0139] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process and method including a series of elements include not only those elements, but also other elements not explicitly listed, or also include elements inherent to such process and method.

[0140] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0141] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A blockchain-based drug supply chain traceability system, characterized by: It includes data management module, drug identity authentication module, anti-counterfeiting traceability module, real-time monitoring module, blockchain smart contract management module and feedback interaction module; The data management module is used to collect, process and store drug supply chain process data and perform data transmission and interaction between modules. Through data management, a drug supply chain traceability data set is obtained, the drug supply chain traceability data set is uploaded to the blockchain, and sent to the drug identity authentication module, the anti-counterfeiting traceability module and the real-time monitoring module; The drug identity verification module is used to verify the link information of the drug through the label information, and adopts the feature attention improved model combined with the pre-trained network to perform drug identity verification, obtain drug information authenticity verification data, upload the drug information authenticity verification data to the blockchain, and send it to the anti-counterfeiting traceability module, the blockchain smart contract management module and the feedback interaction module; The drug identity verification is specifically to perform drug identity verification based on the identity verification optimization data in the drug supply chain traceability data set, using a feature attention improvement model combined with a pre-trained network to obtain drug information authenticity verification data, and specifically includes the following steps: constructing a pre-trained network layer, constructing a feature attention improved U-type model, drug identity verification model training and drug identity verification; The anti-counterfeiting and tracing module is used to analyze and detect the counterfeiting of drugs through packaging images, use the feature attention improved model to perform anti-counterfeiting and tracing, obtain the anti-counterfeiting evaluation data of drug packaging, upload the anti-counterfeiting evaluation data of drug packaging to the blockchain, and send it to the real-time monitoring module, the blockchain smart contract management module and the feedback interaction module; The anti-counterfeiting traceability is specifically to use the feature attention improvement model to perform anti-counterfeiting traceability based on the anti-counterfeiting traceability optimization data in the drug supply chain traceability data set and the drug information authenticity verification data to obtain drug packaging anti-counterfeiting evaluation data, including the following steps: data reconstruction, construction of a feature attention improved U-type model, anti-counterfeiting traceability model training and anti-counterfeiting traceability; The real-time monitoring module is used to track the transportation and storage status of drugs in real time and enhance the traceability of the supply chain. It uses a local polynomial filtered temporal convolution long-term parallel network to perform real-time monitoring to obtain drug transportation process evaluation data, upload the drug transportation process evaluation data to the blockchain, and send it to the blockchain smart contract management module and feedback interaction module; The real-time monitoring is specifically to perform real-time monitoring based on the real-time monitoring optimization data in the drug supply chain traceability data set and the drug packaging anti-counterfeiting evaluation data, using a local polynomial filtered time series convolution long-term parallel network to obtain drug transportation process evaluation data, specifically including the following steps: local polynomial filtering, constructing a time series convolution long-term parallel network, designing an abnormal monitoring mechanism, training a real-time monitoring model, and real-time monitoring; The blockchain smart contract management module is used to manage transaction data in the drug supply chain through the blockchain index, obtain data verification signature information through blockchain smart contract management, and execute the drug supply chain smart contract by verifying the aforementioned information based on the data; The feedback interaction module is used to provide a user-interactive feedback function, and through feedback interaction, users can query, use and manage the traceability of the drug supply chain.

2. A blockchain-based drug supply chain traceability system according to claim 1, characterized in that: The data management specifically involves obtaining the original data set of the drug supply chain through data collection from the drug production process, and performing data classification and preprocessing to obtain the drug supply chain traceability data set; The drug production process specifically includes direct drug production, drug transportation and drug distribution; The drug supply chain original data set specifically includes identity authentication original data, anti-counterfeiting and traceability original data, and real-time monitoring original data; The data classification operation specifically refers to dividing the drug supply chain original data set into identity authentication original data, anti-counterfeiting and traceability original data, and real-time monitoring original data according to the data type; The data preprocessing operation is specifically to preprocess the identity authentication original data, anti-counterfeiting and traceability original data and real-time monitoring original data in sequence to obtain a drug supply chain traceability data set, including the following steps: identity authentication data preprocessing, anti-counterfeiting and traceability data preprocessing, real-time monitoring data preprocessing and comprehensive preprocessing; The pharmaceutical supply chain traceability data set specifically includes identity authentication optimization data, anti-counterfeiting traceability optimization data and real-time monitoring optimization data.

3. A blockchain-based drug supply chain traceability system according to claim 2, characterized in that: The drug identity verification process includes the following steps: constructing a pre-trained network layer, constructing a feature attention improved U-shaped model, drug identity verification model training, and drug identity verification; The construction of the pre-trained network layer specifically involves constructing a pre-trained target detection model as an initial feature extraction layer to obtain initial feature data of drug labels; The pre-trained target detection model specifically adopts the Yolov8 model; The construction of the feature attention improved U-type model specifically includes constructing a feature encoding improvement layer, constructing a three-part feature parallel fusion layer, constructing a joint attention improvement layer and a feature decoding layer; The construction of the feature coding improvement layer is used to optimize the extraction capability of local and global features of the image, and specifically includes an improved transformer layer and an efficient subnet layer. The construction steps include: constructing an improved transformer layer and constructing an efficient subnet layer; The improved transformer layer is constructed to optimize computational efficiency and extract local features, specifically by introducing a self-attention mechanism, replacing the standard multi-head self-attention head with a double-layer alternating self-attention head structure including a windowed self-attention layer and a window translation self-attention layer, and respectively constructing a residual-connected multi-layer perceptron activated by a nonlinear activation function in the double-layer alternating self-attention head structure, performing window partition transformer improvement, and obtaining improved transformer feature data; The efficient subnet layer is constructed to optimize computing resource consumption and extract low-level image features, specifically, a standard efficient network is connected after the improved transformer layer as an efficient subnet layer, and the improved transformer feature data is sequentially subjected to efficient convolution extraction, feature fusion and upsampling pooling operations to obtain improved feature encoding data; The three-dimensional feature parallel fusion layer is constructed to improve the feature processing capability of different image perspectives and scales. Specifically, a three-dimensional feature parallel fusion module is constructed to perform global feature fusion. The three-dimensional feature parallel fusion module specifically includes three parallel convolution branches to obtain a branch feature data set, and a softmax classifier is constructed to calculate a pixel attention atlas, and feature map fusion is performed using element-by-element multiplication to obtain fused feature map data; The joint attention improvement layer is constructed to optimize the recognition ability of irregular shapes, specifically by extracting dual-channel attention features according to the fused feature map data through a joint channel attention module and a spatial attention module, and aggregating channel attention features through a parallel connection method to obtain joint attention feature encoding data; The feature decoding layer is used to perform deep feature decoding, specifically, performing a block scaling operation on the joint attention feature encoding data to obtain a deep scaling feature, and performing a feature upsampling operation to obtain a high-resolution feature map, and then constructing two consecutive improved transformer layers, and by using a skip connection, the shallow features and the deep features are feature-connected to perform feature decoding, and by constructing a standard convolution layer and an upsampling pooling layer, feature decoding is performed to obtain a decoded feature map, and a classifier layer is constructed to output the model result according to the decoded feature map; The drug identity verification model training is specifically to train the drug identity verification model through the pre-trained network layer and the feature attention improved U-type model to obtain the drug identity verification model Model ID ; The drug identity verification is specifically to use the drug identity verification model Model ID , performing drug identity authentication based on the identity authentication optimization data in the drug supply chain traceability data set to obtain drug information authenticity verification data; The drug information authenticity verification data specifically refers to the drug information authenticity verification result obtained by comparing the drug information parsed from the drug label image with the drug label information stored in the blockchain. The drug information authenticity verification result specifically includes real information verification data class and forged information verification data class.

4. A drug supply chain traceability system based on blockchain according to claim 3, characterized in that: The anti-counterfeiting tracing method includes the following steps: data reconstruction, construction of a feature attention improved U-shaped model, anti-counterfeiting tracing model training and anti-counterfeiting tracing; The data reconstruction is specifically to fuse the anti-counterfeiting traceability optimization data and the drug information authenticity verification data, and obtain an anti-counterfeiting traceability training data set through manual labeling, wherein the anti-counterfeiting traceability training data set is used for anti-counterfeiting traceability model training; The construction of the feature attention improved U-type model specifically includes constructing a feature encoding improvement layer, constructing a three-part feature parallel fusion layer, constructing a joint attention improvement layer, and a feature decoding layer; The anti-counterfeiting traceability model training is specifically to perform model training through the data reconstruction and the feature attention improvement U-type model to obtain the anti-counterfeiting traceability model ModelDR; The anti-counterfeiting traceability specifically uses the anti-counterfeiting traceability model ModelDR to perform anti-counterfeiting traceability based on the anti-counterfeiting traceability optimization data and the drug information authenticity verification data to obtain drug packaging anti-counterfeiting evaluation data; The drug packaging anti-counterfeiting assessment data specifically includes real drug packaging data and counterfeit drug security data.

5. A blockchain-based drug supply chain traceability system according to claim 4, characterized in that: The real-time monitoring comprises the following steps: Local polynomial filtering, building time series convolutional long-term parallel networks, abnormal monitoring mechanism design, real-time monitoring model training and real-time monitoring; The local polynomial filtering is specifically to use a Savitzky-Gorey filter to perform smoothing filtering and denoising on the sensor data in the real-time monitoring optimization data to obtain smoothed optimization signal data; The construction of the time series convolution long-short term parallel network is used to optimize short-term fluctuation data through the time series convolution network and process long-term trend fluctuation data through the long short-term memory neural network. Specifically, a standard time series convolution network and a standard long short-term memory neural network are sequentially constructed, and the standard time series convolution network and the standard long short-term memory neural network are processed in parallel by constructing a parallel architecture, and the outputs of the standard time series convolution network and the standard long short-term memory neural network of the parallel architecture are connected by weighted summation to obtain comprehensive network output data; The abnormal monitoring mechanism is designed to specifically set an abnormal threshold value. When the output of the integrated network output data is greater than the abnormal threshold value, abnormal feedback is performed. The abnormal feedback specifically includes abnormal time, abnormal location, actual value of abnormal temperature and humidity value, and abnormal temperature and humidity predicted value; The real-time monitoring model training is specifically to perform model training through the local polynomial filtering, the time series convolution long-term parallel network and the abnormal monitoring mechanism design to obtain the real-time monitoring model Model TL ; The real-time monitoring is specifically to use the real-time monitoring model Model TL , based on the real-time monitoring optimization data in the drug supply chain traceability data set and the drug packaging anti-counterfeiting evaluation data, real-time monitoring is performed to obtain drug transportation process evaluation data.

6. A blockchain-based drug supply chain traceability system according to claim 5, characterized in that: The blockchain smart contract management specifically involves building a drug supply chain traceability blockchain security model, performing blockchain smart contract management, obtaining data verification signature information, and executing the drug supply chain smart contract by verifying the aforementioned information based on the data.

7. A blockchain-based drug supply chain traceability system according to claim 6, characterized in that: The drug supply chain traceability blockchain security model specifically includes a sending entity, a receiving entity, a transaction entity, an optimized search algorithm, and an optimized encryption algorithm; The sending entity specifically refers to the sender of the drug smart contract who creates and starts the smart contract after passing the drug identity authentication and anti-counterfeiting traceability, and provides drug information and drug entity; The receiving entity specifically refers to the recipient of the drug smart contract who performs drug traceability management through real-time monitoring, and confirms drug counterfeiting by combining drug identity authentication and anti-counterfeiting traceability, completes smart contract management, and obtains data verification signature information; The optimization search algorithm specifically refers to a parallel search algorithm based on a directed acyclic graph maximum matching algorithm, which is specifically constructed in the blockchain network topology to solve the maximum matching problem and serve as an optimization search algorithm; The optimized encryption algorithm specifically refers to the hash chain of Chebyshev polynomials.

8. A drug supply chain traceability system based on blockchain according to claim 7, characterized in that: The feedback interaction specifically implements data management, drug identity authentication, anti-counterfeiting traceability, real-time monitoring and blockchain smart contract management in sequence, so as to enable users to query, use and manage the drug supply chain traceability based on blockchain.

Images