Medicine supply chain data security method based on multi-mode sensor design
Environmental monitoring and data processing are carried out through multimodal sensors and edge computing technology, and combined with efficient encryption and distributed storage, the problems of incomplete monitoring of the pharmaceutical supply chain and insufficient data security are solved, and efficient monitoring of the pharmaceutical environment is achieved. Data security guarantee improves the overall efficiency and reliability of the supply chain.
Patent Information
- Application Number
- CN202411030163.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-29
AI Technical Summary
The existing pharmaceutical supply chain monitoring system is incomplete in the face of changes in various environmental parameters, has low processing efficiency, and insufficient data security, is vulnerable to cyber attacks, and cannot effectively protect the safety of drugs and the health of patients.
Using a multimodal sensor system combined with edge computing and efficient data encryption technology, the environment is monitored by integrating multiple sensors (such as temperature, humidity, vibration, light) and real-time processing and analysis at the data source, using the extended Kalman filtering algorithm Data fusion is carried out, deep learning analysis is used using Transformer model and autoencoder, combined with AES, ECC encryption and TLS protocols, distributed storage and blockchain technology ensure data security.
It has achieved comprehensive and accurate monitoring and rapid response to the pharmaceutical supply chain environment, improved data processing efficiency and system reliability, ensured data confidentiality and integrity, and improved drug safety and supply chain management efficiency.
Smart Images

Figure CN119047685A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of medical informationization and relates to a medical supply chain data security method based on multimodal sensor design. Background Art
[0002] With the rapid expansion of the global pharmaceutical market and the continuous improvement of drug quality requirements, pharmaceutical supply chain management is facing more and more challenges. From the production, transportation, storage to the final sale of drugs, environmental conditions (such as temperature, humidity, vibration and light) have a crucial impact on the quality of drugs. Any improper environmental changes may cause the drug to fail, thus affecting the treatment effect and health of patients. Therefore, real-time and accurate monitoring of environmental parameters in the pharmaceutical logistics process to ensure the safety and quality of drugs throughout the supply chain is a technical problem that needs to be solved urgently.
[0003] 1. Current status and challenges of pharmaceutical supply chain monitoring
[0004] Most existing pharmaceutical supply chain monitoring systems use a single or a small number of sensors, which can usually only monitor basic parameters such as temperature and humidity. However, with the increase in the number of drug types and the improvement in transportation conditions, a single monitoring parameter can no longer fully reflect the environmental changes of drugs during transportation and storage. The lack of comprehensive monitoring means leads to incomplete monitoring information, and it is impossible to timely warn and deal with environmental abnormalities, thus affecting the quality of drugs. At present, pharmaceutical supply chain monitoring systems mostly use a centralized data processing mode, and all sensor data needs to be transmitted to a central server for processing and analysis. This mode has low processing efficiency when facing a large amount of real-time data, and is easily affected by network bandwidth and latency, and cannot achieve real-time monitoring and rapid response to environmental parameters. In addition, there are problems such as packet loss and latency during data transmission, which further reduces the reliability of the monitoring system.
[0005] Data security is another key issue in pharmaceutical supply chain management. Pharmaceutical logistics data involves multiple links such as drug production, transportation, and storage, and the security of this data is crucial. Traditional monitoring systems mostly use simple encryption methods, which are at risk of being attacked by hackers, data tampering, and leakage, and cannot effectively guarantee the confidentiality and integrity of pharmaceutical logistics data. Once the data is tampered with or leaked, it will pose a serious threat to the safety of drugs and the health of patients. Therefore, improving the security of pharmaceutical supply chain data has become an urgent need in the industry.
[0006] 2. Development and application of multimodal sensor technology
[0007] Multimodal sensor technology refers to the integration of multiple sensors into a system that can monitor multiple environmental parameters (such as temperature, humidity, vibration, light, etc.) at the same time. Compared with single sensors, multimodal sensors have the advantages of strong comprehensiveness, strong data fusion and analysis capabilities, real-time monitoring and fast response. The development of multimodal sensor technology provides a new solution for environmental monitoring in the pharmaceutical supply chain.
[0008] In the pharmaceutical supply chain monitoring, the use of multimodal sensors can collect multiple environmental data at the same time. Through data fusion and analysis, it can more accurately identify the impact of environmental changes on drugs. This comprehensive monitoring method not only improves the accuracy and reliability of data, but also enables timely warning and processing of abnormal situations. In addition, the multimodal sensor system can perform preliminary processing and analysis at the data collection end through edge computing technology, reducing dependence on central servers and improving the overall processing efficiency and response speed of the system.
[0009] 3. Application of edge computing technology in pharmaceutical supply chain monitoring
[0010] Edge computing is a computing model that processes and analyzes data close to the data source. Compared with the traditional centralized computing model, edge computing can significantly reduce the latency and bandwidth requirements of data transmission, and improve the real-time performance and reliability of the system. In pharmaceutical supply chain monitoring, edge computing technology can realize real-time processing and analysis of data in multimodal sensor systems, reduce the amount of data transmission through localized processing, and improve data processing efficiency and system response speed.
[0011] Through edge computing technology, the pharmaceutical supply chain monitoring system can realize real-time processing and analysis of collected data, including data cleaning, feature extraction and preliminary analysis. This localized data processing method not only reduces delays and packet loss during data transmission, but also can identify and handle abnormal situations in a timely manner, providing faster response and early warning capabilities.
[0012] 4. The importance of efficient data encryption technology in pharmaceutical supply chain monitoring
[0013] Data encryption technology is the key to ensuring the security of pharmaceutical supply chain data. Traditional encryption methods are powerless in the face of increasingly complex network attacks and cannot provide adequate security. Modern data encryption technologies, such as quantum encryption and encryption algorithms based on artificial intelligence, can provide higher security and anti-attack capabilities, ensuring the confidentiality and integrity of pharmaceutical logistics data. In the pharmaceutical supply chain monitoring system, the use of efficient data encryption technology can effectively prevent data leakage and tampering during transmission and storage. By encrypting data end-to-end, it ensures that only authorized users can access and decrypt data, providing comprehensive data security protection.
[0014] In summary, with the increasing demand for pharmaceutical supply chain management, the application of multimodal sensor technology, edge computing technology and efficient data encryption technology in pharmaceutical supply chain monitoring and data security is particularly important. The integration and application of these cutting-edge technologies can not only provide comprehensive environmental monitoring and data security, but also improve the overall efficiency and reliability of the pharmaceutical supply chain, providing strong technical support for the development of the pharmaceutical industry. Summary of the invention
[0015] In view of this, an object of the present invention is to provide a pharmaceutical supply chain data security method based on multimodal sensor design.
[0016] In order to achieve the above object, the present invention provides the following technical solutions:
[0017] A pharmaceutical supply chain data security method based on multimodal sensor design includes the following steps:
[0018] 1. Multimodal sensor system design
[0019] The multimodal sensor system is the core component of the present invention. By integrating multiple types of sensors, comprehensive monitoring of the pharmaceutical logistics environment is achieved. Specific sensors include Bosch BME680 for temperature, humidity, air pressure and air quality monitoring, STMicroelectronics IIS3DWB for vibration monitoring, and Hamamatsu S1337-1010BQ for light monitoring. Each sensor converts analog signals into high-precision digital signals through a high-speed ADC (Analog Devices AD7606C-18, 18-bit ADC, sampling rate up to 1MSPS). The measurement range of the temperature and humidity sensor BME680 is -40℃ to 85℃, with an accuracy of ±0.5℃; the humidity measurement range is 0% to 100%RH, with an accuracy of ±3%RH; the air pressure measurement range is 300hPa to 1100 hPa, with an accuracy of ±1 hPa; the VOC (volatile organic compound) concentration measurement range in air quality detection is 0 ppb to 10000 ppb. The vibration sensor IIS3DWB has a measurement range of ±2g to ±16g, a bandwidth of 0.5 Hz to 6kHz, and an accuracy of ±0.1g. The light sensor S1337-1010BQ has a measurement range of 0.01 lx to 100000 lx and an accuracy of ±2%.
[0020] In order to achieve efficient collection and transmission of sensor data, the present invention adopts the NVIDIA Jetson AGX Xavier module, which provides 32 TOPS of AI computing performance based on its 512 CUDA cores and 64 Tensor cores, and can process sensor data in real time and perform preliminary analysis. Sensor data is transmitted to the Jetson AGX Xavier module through interfaces such as I2C, SPI, and UART. In the system design, multimodal sensors perform local processing and analysis of data through edge computing technology, and use the extended Kalman filter (EKF) algorithm to achieve multi-sensor data fusion. The EKF algorithm recursively estimates the system state through two steps of state estimation and observation update, and fuses data from different sensors. The state estimation formula is:
[0021]
[0022] The observation update formula is:
[0023]
[0024] P k|k =(IK k H k ) k|k-1
[0025] in, is the predicted state estimate, P k|k-1 is the predicted covariance matrix, Fk is the state transfer matrix, Bk is the control input matrix, uk is the control input, and Qk is the process noise covariance matrix. In the observation update, Kk is the Kalman gain, zk is the observation value, Hk is the observation matrix, Rk is the observation noise covariance matrix, and I is the identity matrix.
[0026] Through the EKF algorithm, the data of multimodal sensors can be fused into a comprehensive state estimate, thereby more accurately reflecting the real-time state of the pharmaceutical logistics environment. The application of edge computing technology has significantly improved data processing efficiency and system response speed, ensuring real-time monitoring of environmental parameters and rapid response to abnormal situations.
[0027] 2. Edge computing and real-time data processing
[0028] Edge computing technology plays an important role in the present invention. By processing and analyzing data close to the data source, the real-time performance and response speed of the system are greatly improved. The NVIDIA JetsonAGX Xavier module is used as the edge computing unit. Its 512 CUDA cores and 64 Tensor cores provide 32TOPS of AI computing performance, which can process and analyze a large amount of sensor data in real time, avoiding the delay and bandwidth bottleneck problems that may occur during the data transmission to the central server. In the data processing process, the JetsonAGX Xavier module first pre-processes the data collected by the multimodal sensor, including data cleaning, normalization and feature extraction. In the data cleaning step, outliers are identified and removed by the IQR (interquartile range) method to ensure the reliability and accuracy of the data. The normalization step converts data of different dimensions to the same scale to improve the effectiveness of subsequent data fusion and analysis. In the feature extraction step, key environmental change information is extracted by calculating the time domain and frequency domain features of the data, such as mean, variance, kurtosis, skewness and spectral energy. In order to achieve efficient processing and analysis of real-time data, the present invention adopts a deep learning model based on Transformer. The Transformer model can capture the long-term dependencies in sensor data through the self-attention mechanism and improve the prediction accuracy of environmental changes. The input of the Transformer model is the sensor data sequence, and the output is the predicted value of future environmental parameters. Its core formula includes:
[0029] Self-Attention Mechanism:
[0030]
[0031] Where T represents the transpose operation, Q, K, and V represent the query matrix, key matrix, and value matrix, respectively. k is the dimension of the key vector. Attention is a mechanism that converts these scores into weights by calculating the similarity scores between the query and the key and applying the soft maximization function to perform weighted summation on the values to capture the correlation between each position in the input sequence. Softmax represents a normalization function that is used to convert the input vector into a probability distribution and calculate the weights so that the sum of the weights is 1. Through the self-attention mechanism, the model can calculate the correlation between the data at each moment and the data at other moments and capture the long-term dependencies in the time series.
[0032] Environmental parameter prediction formula based on Transformer model:
[0033]
[0034] in, is the predicted future environmental parameter, Xt is the current and past sensor data sequence, θ is the model parameter. By optimizing the objective function (such as mean square error):
[0035]
[0036] The optimal model parameters can be trained to minimize the prediction error.
[0037] In addition, in order to improve the anomaly detection capability of the system, the present invention also integrates an anomaly detection algorithm based on an autoencoder. An autoencoder is an unsupervised learning model that learns the normal pattern of data by compressing the input data into a low-dimensional representation and then reconstructing it back into a high-dimensional space. The core formula for anomaly detection includes:
[0038] The loss function of the autoencoder is:
[0039]
[0040] Where N is the number of samples, i is the sample index, |||| represents the norm, which is used to measure the difference between the original sample and the reconstructed sample, and X i is the input data, f and g represent the encoder and decoder respectively, is the reconstruction error.
[0041] For new data, if the reconstruction error exceeds a certain threshold, it is judged as abnormal data. Anomaly detection judgment:
[0042]
[0043] Among them, ∈ is the preset anomaly detection threshold.
[0044] Through the application of edge computing and the above-mentioned advanced AI algorithms, the present invention can achieve real-time monitoring and intelligent analysis of the pharmaceutical supply chain environment, provide high-precision environmental parameter prediction and anomaly detection, and effectively improve the overall performance and reliability of the system.
[0045] 3. Efficient data encryption and secure transmission
[0046] In the pharmaceutical supply chain management, data security is of vital importance. In order to ensure the confidentiality, integrity and non-repudiation of pharmaceutical logistics data, the present invention adopts data encryption and secure transmission technology to ensure the security of data during collection, transmission and storage.
[0047] First, the data encryption part adopts a hybrid encryption mechanism, which combines the advantages of symmetric encryption and asymmetric encryption to ensure efficient encryption and secure transmission of data. In the specific implementation process, the Advanced Encryption Standard (AES) is selected as the symmetric encryption algorithm, and the Elliptic Curve Cryptography (ECC) is selected as the asymmetric encryption algorithm. For each sensor data packet, a random session key is first generated, and the data packet is encrypted using the AES algorithm. The AES algorithm has an efficient encryption and decryption speed and is suitable for real-time transmission of large amounts of data. The core formula of AES encryption is as follows:
[0048]
[0049] Among them, C is the ciphertext, P is the plaintext, and Respectively represent the use of session key K s The encryption and decryption process performed.
[0050] Session key K s The receiver's public key is used for encryption to ensure security during transmission. The core formula of ECC encryption is as follows:
[0051]
[0052] Among them, K pub and K priv Represent the public key and private key of the recipient respectively, and Represent the public key encryption and private key decryption processes respectively.
[0053] During the transmission process, the data packet includes the encrypted sensor data and the encrypted session key. After receiving the data packet, the receiver uses its private key to decrypt the session key, and then uses the decrypted session key to decrypt the data to ensure the secure transmission and decryption of the data. The entire encrypted transmission process not only ensures the efficiency of data transmission, but also enhances the security of the data. In order to further improve the security and reliability of data transmission, the present invention also adopts the Transport Layer Security (TLS) protocol. The TLS protocol provides data integrity, confidentiality and identity authentication by adding a layer of security between the application layer and the transport layer. The handshake process of the TLS protocol ensures the identity authentication and key negotiation of the communicating parties to prevent man-in-the-middle attacks and replay attacks.
[0054] During data transmission, the handshake process of the TLS protocol is as follows:
[0055] The client sends the supported encryption algorithms and protocol versions to the server.
[0056] The server selects the encryption algorithm and protocol version and sends the server certificate to the client.
[0057] The client verifies the server certificate and generates a pre-master secret, which it encrypts with the server’s public key and sends to the server.
[0058] The server uses its private key to decrypt the pre-master secret and generate a session key.
[0059] The client and server use the session key to encrypt subsequent data transmission.
[0060] Through the TLS protocol, the present invention provides strong security protection at the data transmission level to ensure the confidentiality and integrity of data during transmission.
[0061] In the process of data storage, the present invention adopts distributed storage and blockchain technology to further improve the security and traceability of data. The distributed storage system ensures the high availability and disaster recovery capability of data through data redundancy and multi-node backup. Blockchain technology ensures the immutability and traceability of data through on-chain records and hash algorithms, and enhances the credibility of data. The present invention constructs an efficient and secure pharmaceutical supply chain data encryption and transmission system through the application of hybrid encryption, TLS protocol, distributed storage and blockchain technology, which effectively ensures the security and reliability of pharmaceutical logistics data.
[0062] 4. Intelligent data analysis and decision support
[0063] In the management of pharmaceutical supply chain, intelligent data analysis and decision support are the key to achieving efficient and accurate management. The present invention uses artificial intelligence technology and big data analysis methods to deeply mine and analyze the collected multimodal sensor data, provide intelligent decision support, and improve the management level and efficiency of the pharmaceutical supply chain.
[0064] First, the present invention uses a deep learning model to predict and analyze the collected environmental data. Based on the combination of the Transformer model and the autoencoder, the present invention can effectively process complex time series data and multimodal data, and achieve high-precision prediction and anomaly detection of environmental parameters. In the specific implementation process, the Transformer model captures long-term dependencies through the self-attention mechanism, while the autoencoder is used to learn the normal pattern of the data and detect abnormal data. The core structure of the Transformer model includes multiple encoders and decoders, each of which is composed of a multi-head self-attention mechanism and a feedforward neural network. The calculation formula of the encoder part is as follows:
[0065]
[0066] IultiHead(Q,K,V)=Concat(head1,head2,…,head h )W O
[0067] FFN(x)=max(0,xW1+b1)W2+b2
[0068] Where Q, K, and V represent query, key, and value matrices, respectively. k is the dimension of the key vector, head h represents the output of the h-th attention head, W O , W1, W2, b1, b2 and represent the learnable weight and bias parameters respectively.
[0069] The autoencoder part reconstructs the data through the encoder-decoder structure and calculates the reconstruction error to detect anomalies. The loss function of the autoencoder is as follows:
[0070]
[0071] Among them, is the input data, f and g represent the encoder and decoder respectively, is the reconstruction error. By setting an appropriate anomaly detection threshold ∈, abnormal data can be identified and alarmed:
[0072]
[0073] Through the combination of the above models, the present invention can achieve high-precision prediction of environmental parameters and real-time anomaly detection, providing intelligent data analysis support for pharmaceutical supply chain management.
[0074] The present invention also introduces an intelligent decision support system based on deep reinforcement learning. The deep reinforcement learning algorithm continuously adjusts the strategy to maximize long-term returns through interaction with the environment. In pharmaceutical supply chain management, the intelligent decision support system can automatically optimize transportation routes, adjust transportation plans, and warn of potential risks based on real-time monitoring data and prediction results. One of the core algorithms of deep reinforcement learning is the deep Q network (DQN). DQN approximates the Q value function through a neural network and selects the optimal action to maximize the return. The update formula of the Q value function is as follows:
[0075] Q(s,a)←Q(s,a)+α(r+γmax a' Q(s',a')-Q(s,a))
[0076] Among them, s and a represent the current state and action respectively, r is the immediate reward, γ is the discount factor, and α is the learning rate. Through the DQN algorithm, the intelligent decision support system can learn and optimize the management strategy of the pharmaceutical supply chain to improve overall efficiency and safety.
[0077] The beneficial effects of the present invention are: in practical applications, the intelligent decision support system of the present invention can analyze and process massive environmental data in real time, automatically generate optimal decision suggestions, and help managers efficiently cope with various challenges in the pharmaceutical logistics process. For example, the system can dynamically adjust the transportation route based on real-time monitoring data and prediction results to avoid congestion and delays; when abnormal situations occur, the system can automatically issue an early warning and provide emergency treatment plans to ensure the safety and quality of drugs.
[0078] In summary, the present invention realizes intelligent analysis and decision support of pharmaceutical supply chain environmental data through deep learning and reinforcement learning algorithms. The system can not only provide high-precision environmental parameter prediction and real-time anomaly detection, but also automatically optimize transportation strategies, improve the overall management level and efficiency of the pharmaceutical supply chain, and provide strong technical support for the development of the pharmaceutical industry.
[0079] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0080] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:
[0081] Figure 1 It is the overall flow chart of the present invention;
[0082] Figure 2 It is a flow chart of multi-modal sensor data acquisition and fusion;
[0083] Figure 3 It is a flowchart for intelligent data analysis and decision support. DETAILED DESCRIPTION
[0084] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.
[0085] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.
[0086] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.
[0087] Figure 1 The overall architecture flow chart is presented, which describes the whole process from data collection to user feedback in detail. When the system starts, it first collects environmental data through multimodal sensors, including temperature, humidity, vibration and light sensors. The collected data undergoes data preprocessing steps, including data cleaning, normalization and feature extraction, to ensure data quality. The preprocessed data is transmitted to the NVIDIA JetsonAGX Xavier edge computing unit, where real-time data processing and analysis are performed. The edge computing unit uses the Transformer model and autoencoder algorithm to perform deep learning and anomaly detection on the data. The data is encrypted using AES symmetric encryption and ECC asymmetric encryption technology to ensure the security of the data during transmission. Then, the data is securely transmitted through the TLS protocol to ensure the confidentiality and integrity of the data during transmission. After the data arrives at the server, the intelligent data analysis and decision support system uses deep learning and reinforcement learning algorithms to provide high-precision environmental parameter prediction and intelligent decision support. The system can warn and respond to abnormal situations in real time, and store data through distributed storage and blockchain technology to ensure data security and traceability. Finally, the results are displayed and fed back through the user interfaces on the mobile and PC terminals to complete the operation of the entire system.
[0088] Figure 2The multimodal sensor data acquisition and fusion flow chart is shown, and the specific process of multimodal sensor data acquisition and processing is described in detail. At the beginning of the system, multimodal sensors are initialized, including temperature, humidity, vibration and light sensors such as BME680, IIS3DWB and S1337-1010BQ. The sensors collect environmental data and transmit these data to the AD7606C-18 high-precision ADC module for analog signal to digital signal conversion. The converted digital signal enters the data preprocessing stage, including data cleaning, normalization and feature extraction to improve data quality and consistency. The preprocessed data is fused through the extended Kalman filter (EKF) algorithm, which can effectively integrate data from different sensors to provide a comprehensive environmental state estimation. After data fusion, it is transmitted to the NVIDIA JetsonAGX Xavier edge computing unit for further processing and analysis. The entire process ensures the efficient acquisition, preprocessing and fusion of multimodal sensor data, providing a reliable data foundation for subsequent real-time data processing and analysis.
[0089] Figure 3 The intelligent data analysis and decision support flow chart is presented, and the whole process of data reception, analysis and decision support is described in detail. The system receives processed data from the edge computing unit, and first performs deep learning analysis through the Transformer model to capture the long-term dependencies in the data and make high-precision predictions of environmental parameters. Then, the autoencoder algorithm is used for anomaly detection. The autoencoder identifies and alarms abnormal data by learning the normal pattern of the data. For the detected abnormal data, the system automatically triggers a real-time early warning and response mechanism. Based on these analysis results, the intelligent decision support system uses a deep reinforcement learning algorithm for decision support. The deep Q network (DQN) algorithm optimizes the transportation route, adjusts the transportation plan, and provides emergency response solutions through interaction with the environment. The generated optimal decision suggestions are presented to managers through the user interface of the mobile terminal and PC terminal, helping them to efficiently cope with various challenges in the pharmaceutical logistics process. Finally, the system adjusts the strategy according to user feedback, continuously optimizes the management plan, and improves the overall efficiency and safety of the pharmaceutical supply chain.
[0090] Example 1: Pharmaceutical cold chain transportation environment monitoring and data security management
[0091] In this embodiment, the method of the present invention is applied to the cold chain transportation of medicines, and environmental monitoring and data security management of medicines during the cold chain transportation are achieved through a multimodal sensor system, edge computing technology, intelligent data analysis and decision support.
[0092] First, a multimodal sensor system is arranged inside the pharmaceutical cold chain transport vehicle, including temperature sensor BME680, humidity sensor BME680, vibration sensor IIS3DWB and light sensor S1337-1010BQ. The sensors collect environmental data in real time during transportation and convert analog signals into high-precision digital signals through the high-speed ADC module AD7606C-18. These digital signals undergo data preprocessing, including data cleaning, normalization and feature extraction, to ensure data accuracy and consistency. The preprocessed data is transmitted to the NVIDIA JetsonAGX Xavier edge computing unit, where real-time data processing and analysis are performed. The edge computing unit uses the Transformer model and autoencoder algorithm to perform deep learning and anomaly detection on environmental data. The Transformer model captures long-term dependencies through the self-attention mechanism to achieve high-precision prediction of environmental parameters such as temperature, humidity, vibration and light. The autoencoder identifies and alarms abnormal data by learning the normal pattern of the data. For detected abnormal data, the system automatically triggers a real-time early warning and response mechanism to notify relevant personnel to take corresponding emergency measures. At the same time, in order to ensure the security of data during transmission, the system encrypts the sensor data. First, a random session key is generated and the data is symmetrically encrypted using the AES algorithm. The encrypted data is ECC encrypted using the receiver's public key and securely transmitted through the TLS protocol. The receiver uses the private key to decrypt the session key, and then uses the decrypted session key to decrypt the data to ensure the confidentiality and integrity of the data. After the data reaches the server, the intelligent data analysis and decision support system uses deep learning and reinforcement learning algorithms to perform high-precision predictions of environmental parameters and intelligent decision support. Based on real-time monitoring data and prediction results, the system can automatically generate the optimal transportation route and plan to avoid congestion and delays and improve transportation efficiency. When abnormal situations occur, the system can provide emergency response plans to ensure the safety and quality of drugs.
[0093] Data storage uses distributed storage and blockchain technology. The data redundancy and multi-node backup of the distributed storage system ensure high data availability and disaster recovery capabilities. Blockchain technology ensures the immutability and traceability of data through on-chain records and hash algorithms, enhancing the credibility and security of data. Finally, through the user interface of the mobile terminal and PC terminal, managers can view the environmental data during transportation and the decision-making suggestions generated by the system in real time. The early warning information and emergency response solutions provided by the system help managers efficiently cope with various challenges in the cold chain transportation of medicines and ensure the quality and safety of medicines.
[0094] This embodiment demonstrates the specific application of the present invention in pharmaceutical cold chain transportation. Through a multimodal sensor system, edge computing, intelligent data analysis and decision support, the management level and efficiency of the pharmaceutical supply chain are improved, and the safety and quality of drugs are effectively guaranteed.
[0095] Example 2: Pharmaceutical storage environment monitoring and data security management
[0096] In this embodiment, the method of the present invention is applied to the pharmaceutical warehousing process, and environmental monitoring and data security management of the drugs in the warehousing process are realized through a multimodal sensor system, edge computing technology, intelligent data analysis and decision support.
[0097] First, a multimodal sensor system was arranged inside the pharmaceutical warehouse, including temperature sensor BME680, humidity sensor BME680, vibration sensor IIS3DWB and light sensor S1337-1010BQ. These sensors collect environmental data inside the warehouse in real time and convert analog signals into high-precision digital signals through the high-speed ADC module AD7606C-18. The converted digital signals undergo data preprocessing, including data cleaning, normalization and feature extraction, to ensure the accuracy and consistency of the data. The preprocessed data is transmitted to the NVIDIA JetsonAGX Xavier edge computing unit, where real-time data processing and analysis are performed. The edge computing unit uses the Transformer model and autoencoder algorithm to perform deep learning and anomaly detection on environmental data. The Transformer model captures long-term dependencies through the self-attention mechanism to achieve high-precision prediction of environmental parameters such as temperature, humidity, vibration and light. The autoencoder identifies and alarms abnormal data by learning the normal pattern of the data. For detected abnormal data, the system automatically triggers a real-time early warning and response mechanism to notify warehouse managers to take corresponding emergency measures. In order to ensure the security of data during transmission, the system encrypts the sensor data. First, a random session key is generated and the data is symmetrically encrypted using the AES algorithm. The encrypted data is ECC encrypted using the receiver's public key and securely transmitted using the TLS protocol. The receiver uses the private key to decrypt the session key, and then uses the decrypted session key to decrypt the data to ensure the confidentiality and integrity of the data. After the data reaches the server, the intelligent data analysis and decision support system uses deep learning and reinforcement learning algorithms to perform high-precision predictions of environmental parameters and intelligent decision support. Based on real-time monitoring data and prediction results, the system can automatically generate the optimal warehouse management strategy, including temperature and humidity adjustment, inventory optimization, and environmental anomaly handling. When an abnormal situation occurs, the system can provide emergency response plans to ensure the safety and quality of drugs.
[0098] Data storage uses distributed storage and blockchain technology. The data redundancy and multi-node backup of the distributed storage system ensure high data availability and disaster recovery capabilities. Blockchain technology ensures the immutability and traceability of data through on-chain records and hash algorithms, enhancing the credibility and security of data. Finally, through the user interface of the mobile terminal and PC terminal, warehouse managers can view the environmental data and management strategies generated by the system in the storage process in real time. The early warning information and emergency response solutions provided by the system help managers efficiently cope with various challenges in the pharmaceutical storage process and ensure the quality and safety of drugs.
[0099] This embodiment demonstrates the specific application of the present invention in pharmaceutical warehousing. Through a multimodal sensor system, edge computing, intelligent data analysis and decision support, the management level and efficiency of the pharmaceutical supply chain are improved, and the safety and quality of drugs are effectively guaranteed.
[0100] Technical terms that need to be explained to help understand the present invention
[0101] (1) Multimodal sensor: refers to a sensor system that can collect multiple types of environmental parameters at the same time, including temperature, humidity, vibration and light sensors. By integrating the data of multiple sensors, more comprehensive and accurate environmental monitoring can be achieved.
[0102] (2) Edge computing: A computing model that processes and analyzes data at the source of the data (such as near the sensor) instead of transmitting the data to a remote data center for processing. This model can significantly reduce data transmission latency and bandwidth requirements, and improve the real-time performance and response speed of the system.
[0103] (3) NVIDIA Jetson AGX Xavier: A high-performance edge computing device with 512 CUDA cores and 64 Tensor cores, providing 32TOPS of AI computing performance. It is used to process and analyze multimodal sensor data in real time.
[0104] (4) Extended Kalman Filter (EKF): A recursive state estimation algorithm for nonlinear systems that estimates the state of a system by combining a system model with measurement data. In the present invention, EKF is used to fuse data from multimodal sensors.
[0105] (5) Transformer model: A deep learning model based on the self-attention mechanism that can capture long-term dependencies in sequence data. It is widely used in natural language processing and time series data analysis. In the present invention, the Transformer model is used for high-precision prediction of environmental parameters.
[0106] (6) Autoencoder: An unsupervised learning model that learns the normal pattern of data by compressing and reconstructing the input data. When used for anomaly detection, the autoencoder can identify data that deviates significantly from the normal pattern.
[0107] (7) AES (Advanced Encryption Standard): A symmetric encryption algorithm with high encryption and decryption speed, suitable for real-time encryption of large amounts of data. In the present invention, it is used to encrypt sensor data.
[0108] (8) ECC (Elliptic Curve Cryptography): An asymmetric encryption algorithm that has higher security and smaller key size than other asymmetric encryption algorithms (such as RSA). In the present invention, it is used to encrypt session keys to ensure the security of data transmission.
[0109] (9) TLS (Transport Layer Security) protocol: An encryption protocol used to provide secure communications between computer networks. Through the handshake process, it ensures the identity authentication and key negotiation of both communicating parties and prevents man-in-the-middle attacks and replay attacks.
[0110] (10) Deep Q-Network (DQN): A deep reinforcement learning algorithm that approximates the Q-value function through a neural network and selects the best action to maximize long-term rewards. In the present invention, DQN is used to optimize the management strategy of the pharmaceutical supply chain.
[0111] (11) Distributed storage: A data storage mode that ensures high availability and disaster recovery of data by storing data copies on multiple nodes. In the present invention, distributed storage is used to improve the reliability of data storage.
[0112] (12) Blockchain technology: A distributed ledger technology that ensures the immutability and traceability of data through on-chain records and hash algorithms. In the present invention, blockchain technology is used to enhance the credibility and security of data.
[0113] (13) Data cleaning: The process of processing raw data to remove noise and outliers and improve data quality. In the present invention, data cleaning is used to ensure the accuracy of sensor data.
[0114] (14) Normalization: The process of converting data of different dimensions to the same scale to facilitate subsequent data processing and analysis. In the present invention, normalization is used to improve the processing efficiency of multimodal sensor data.
[0115] (15) Feature extraction: Extracting meaningful features from raw data to facilitate model training and prediction. In the present invention, feature extraction is used to extract key information from environmental data.
[0116] (16) Reconstruction error: The difference between the original data and the reconstructed data when the autoencoder reconstructs the data. When used for anomaly detection, data with a reconstruction error exceeding a preset threshold is judged as abnormal data.
[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.
Claims
1. A pharmaceutical supply chain data security method based on multimodal sensor design, characterized by: The method comprises the following steps: Real-time collection of environmental data in the pharmaceutical supply chain process through a multimodal sensor system, which includes temperature sensors, humidity sensors, vibration sensors, and light sensors; The collected environmental data is transmitted to the analog-to-digital converter (ADC) module for conversion from analog signals to digital signals; Perform data preprocessing on the converted digital signals, including data cleaning, normalization and feature extraction; The pre-processed data is transmitted to the edge computing unit for real-time data processing and analysis. The edge computing unit uses the NVIDIA Jetson AGX Xavier module; In the edge computing unit, the Transformer model is used to make high-precision predictions of environmental data, and the autoencoder algorithm is used for anomaly detection; Encrypting the processed data, including generating a random session key, symmetric encryption of the data using the AES algorithm, and asymmetric encryption of the session key using the Elliptic Curve Cryptography Algorithm (ECC) algorithm; The data is transmitted securely through the Transport Layer Security (TLS) protocol. The receiver uses the private key to decrypt the session key, and then uses the session key to decrypt the data. After the data reaches the server, deep learning and reinforcement learning algorithms are used to conduct intelligent analysis and decision support on environmental data to generate the optimal transportation route and emergency response plan; Use distributed storage and blockchain technology for data storage to ensure high availability, immutability and traceability of data; Environmental data and decision-making recommendations are displayed through mobile and PC user interfaces, providing real-time warning and emergency response.
2. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The temperature sensor and humidity sensor are Bosch BME680 temperature and humidity sensors, the vibration sensor is STMicroelectronics IIS3DWB vibration sensor, and the light sensor is Hamamatsu S1337-1010BQ light sensor.
3. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The data preprocessing comprises the following steps: Data cleaning to remove noise and outliers; Normalization, converting data of different dimensions to the same scale; Feature extraction, extracting time domain and frequency domain features of environmental data.
4. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The Transformer model in the edge computing unit captures the long-term dependencies in the environmental data through the self-attention mechanism. The formula is as follows: Where T represents the transpose operation, Q, K, and V represent the query matrix, key matrix, and value matrix, respectively. k is the dimension of the key vector. Attention is a mechanism that converts these scores into weights by calculating the similarity scores between the query and the key and applying the soft maximization function to perform weighted summation on the values to capture the correlation between each position in the input sequence. Softmax represents a normalization function that is used to convert the input vector into a probability distribution and calculate the weights so that the sum of the weights is 1. Through the self-attention mechanism, the model can calculate the correlation between the data at each moment and the data at other moments and capture the long-term dependencies in the time series.
5. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The autoencoder algorithm performs anomaly detection by learning the normal pattern of the data, and its loss function is as follows: Where N is the number of samples, i is the sample index, |||| represents the norm, which is used to measure the difference between the original sample and the reconstructed sample, and X i is the input data, f and g represent the encoder and decoder respectively, is the reconstruction error.
6. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The AES algorithm is used to symmetrically encrypt sensor data, and the encryption and decryption process is as follows: Among them, C is the ciphertext, P is the plaintext, and Respectively represent the use of session key K s The encryption and decryption process performed.
7. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The ECC algorithm is used to perform asymmetric encryption on the session key, and the encryption and decryption process is as follows: Among them, K pub and K priv Represent the public key and private key of the recipient respectively, and Represent the public key encryption and private key decryption processes respectively.
8. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The deep learning algorithm includes a Transformer model, and the reinforcement learning algorithm includes a deep Q-network (DQN), which is used to generate optimal transportation routes and emergency response plans.
9. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The distributed storage system ensures high availability of data through data redundancy and multi-node backup, and the blockchain technology ensures the immutability and traceability of data through on-chain records and hash algorithms.
10. The pharmaceutical supply chain data security method based on multimodal sensor design according to claim 1 is characterized by: The user interface displays environmental data and decision suggestions generated by the system through mobile terminals and PC terminals, and provides real-time warning information and emergency response plans.
Citation Information
Cited By
Intelligent cold chain transportation information management method, system and device and storage medium
CN119273262A
A smart cold chain transportation information management method, system, device and storage medium
CN119273262B
Insulin storage environment monitoring method
CN120008684A
Supply chain data encryption storage method based on edge calculation
CN120474839A
Edge computing-based encrypted storage method for supply chain data
CN120474839B
