Intelligent vaccine supply chain collaborative optimization system based on multi-source data fusion

Abstract

The invention provides an intelligent vaccine supply chain collaborative optimization system based on multi-source data fusion, and relates to the technical field of vaccine supply chains, and the system comprises a data acquisition and processing module which is used for collecting and preprocessing multi-source heterogeneous data in real time; the intelligent prediction and optimization module is used for constructing a hybrid prediction model and predicting vaccine demands; the storage and logistics management module is used for constructing a virtual warehouse model; the collaborative decision module is used for generating a multi-dimensional supply chain scheduling scheme and evaluating the feasibility and efficiency of the scheme; and the dynamic priority marking module is used for carrying out priority ranking on vaccines. By constructing the hybrid prediction model, the future vaccine demand can be accurately predicted, the prediction accuracy and reliability are improved, vaccine shortage or excess can be avoided, and the response speed and flexibility of a supply chain are improved; through accurate priority ranking of vaccines, when emergency situations such as epidemic situations and the like are coped, resources can be quickly allocated, and the inoculation requirement is met to the maximum extent.

Description

Technical Field

[0001] This invention relates to the field of vaccine supply chain technology, and in particular to a collaborative optimization system for intelligent vaccine supply chain based on multi-source data fusion. Background Technology

[0002] The development and application of vaccines are based on immunological principles. The core of this approach is to introduce specific antigenic substances to stimulate the body's immune system to produce a targeted immune response, thereby forming resistance to the corresponding pathogen. Early vaccine development can be traced back to the practice of smallpox vaccination, a process that utilized attenuated or inactivated pathogenic microorganisms to stimulate the body to produce immune memory. With advancements in molecular biology, technologies such as subunit vaccines and recombinant vaccines have matured. By extracting or synthesizing key antigenic components of pathogens (such as proteins and polysaccharides), these vaccines reduce safety risks while preserving immunogenicity.

[0003] The efficient operation of the vaccine supply chain is a key support for ensuring the orderly progress of vaccination. Its technical system covers multiple aspects such as demand forecasting, inventory management, transportation scheduling, quality monitoring, and emergency response.

[0004] Typical intelligent vaccine supply chain coordination systems often use single forecasting models, leading to significant errors in predicting future vaccine demand and making it difficult to accurately cover demand fluctuations within 30 days. This can directly cause vaccine shortages or surpluses, and the supply chain's response speed to changes in demand is slow and lacks flexibility. Furthermore, in emergencies such as pandemics, it is difficult to quickly identify and prioritize urgent needs, resulting in weak emergency response capabilities.

[0005] To address the shortcomings of the existing technologies, this technical solution proposes a collaborative optimization system for the intelligent vaccine supply chain based on multi-source data fusion. Summary of the Invention

[0006] This invention provides a collaborative optimization system for the intelligent vaccine supply chain based on multi-source data fusion, in order to address the deficiencies in existing technologies.

[0007] On the one hand, this invention provides a collaborative optimization system for a vaccine intelligent supply chain based on multi-source data fusion, comprising: The data acquisition and processing module is used to collect real-time data on vaccine production, warehouse dynamics, logistics and transportation, and vaccination site demand, and output multi-source heterogeneous data; and to preprocess the multi-source heterogeneous data and output a standard real-time data stream. The intelligent prediction and optimization module is used to build a hybrid prediction model based on historical data and combine it with standard real-time data streams to predict vaccine demand and output the demand prediction results. The warehousing and logistics management module is used to build a virtual warehouse model based on demand forecasting results and digital twin technology, combined with dynamic warehouse data and logistics transportation data; The collaborative decision-making module is used to generate multi-dimensional supply chain scheduling schemes based on the virtual warehouse model, integrating vaccine production data and vaccination site demand data, and to evaluate the feasibility and efficiency of the schemes through simulation, and output the optimal collaborative decision-making instructions. The dynamic priority tagging module is used to prioritize different batches and types of vaccines based on the optimal collaborative decision-making instructions and the real-time demand changes at vaccination sites, generate dynamic priority tags, and synchronize the priority tags to the warehousing and logistics management module.

[0008] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the data acquisition and processing module includes a multi-source data access unit, a data cleaning unit, a data conversion unit, and a data integration unit. The multi-source data access unit is used to receive vaccine production data, warehouse dynamic data, logistics and transportation data, and vaccination site demand data through multiple data interfaces to obtain multi-source heterogeneous data. The data cleaning unit is used to remove noise, missing values, and outliers from the multi-source heterogeneous data to obtain cleaned multi-source heterogeneous data. The data conversion unit is used to convert the cleaned multi-source heterogeneous data in different formats into a unified data format to obtain standard multi-source heterogeneous data. The data integration unit is used to correlate and integrate the standard multi-source heterogeneous data to form a standard real-time data stream.

[0009] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the intelligent prediction and optimization module includes a feature extraction unit, a model building unit, a model training unit, and a result prediction unit. The feature extraction unit is used to extract vaccine features from historical data through relevant new analysis. The model building unit is used to build a hybrid prediction model by combining statistical models and machine learning algorithms. The model training unit is used to train the hybrid prediction model using vaccine features to obtain the final hybrid prediction model. The result prediction unit is used to input standard real-time data streams into the final hybrid prediction model for prediction to obtain demand prediction results.

[0010] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the step of the model building unit constructing a hybrid prediction model includes: The vaccine features were divided into training, validation, and test sets in chronological order. Use autocorrelation function and partial autocorrelation function plots to determine the parameters of the statistical model; Based on the statistical model parameters, fit the ARIMA model on the training set; based on the ARIMA model, evaluate the prediction error on the validation set, adjust the parameters to the optimal value, and obtain the final statistical model. Set the model parameters of the XGBoost regression model, train the XGBoost regression model using the training set based on the model parameters, and evaluate and adjust the parameters on the validation set to obtain the final learned prediction model. Calculate the root mean square error of the final statistical model and the final learned prediction model on the validation set, and assign weights based on the derivative of the root mean square error; Weighted fusion is performed on the test set to obtain a hybrid prediction model.

[0011] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the warehousing and logistics management module includes a digital twin unit and an inventory optimization unit; the digital twin unit is used to construct a virtual mapping of the physical warehouse and the transportation network based on warehouse dynamic data and logistics transportation data, so as to obtain a virtual warehouse model driven by IoT data in real time; the inventory optimization unit is used to dynamically adjust the inventory strategy based on the demand forecast results and the virtual warehouse model.

[0012] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the steps of obtaining a virtual warehouse model driven by real-time IoT data by the digital twin unit include: Based on warehouse dynamic data, a 3D virtual model is established; and the 3D virtual model is simplified by reducing the number of polygons and optimizing texture materials to obtain a simplified 3D virtual model. Based on the transportation network, construct a transportation network model; Establish a unified data transmission protocol to convert dynamic data from warehouses and transportation networks into a format recognizable by the virtual model, thus obtaining standard transmission data; Based on standard transmission data, a virtual warehouse model is obtained by binding the corresponding physical entity data source to the 3D virtual model and the transportation network model. Based on physical entity data sources, high-frequency dynamic data is processed through edge computing nodes, and updates to the virtual warehouse model are triggered according to priority.

[0013] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the collaborative decision-making module includes a scheme generation unit, a simulation unit, and a scheme optimization unit. The scheme generation unit is used to generate a preliminary multi-dimensional supply chain scheduling scheme based on a virtual warehouse model and vaccine production data. The simulation unit is used to build a simulation environment, input the preliminary multi-dimensional supply chain scheduling scheme into it for simulation operation, and collect evaluation indicators during the operation. The scheme optimization unit is used to adjust and optimize the preliminary multi-dimensional supply chain scheduling scheme according to the evaluation indicators, obtain multiple sets of alternative schemes, and select the optimal collaborative decision instruction through comparative analysis.

[0014] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the simulation environment construction steps of the simulation unit include: Obtain the physical parameters of each link in the supply chain; build a virtual simulation scenario that matches the scale of the actual supply chain scenario based on the physical parameters; Set the simulation time step and simulation termination conditions; Import simulation parameters from multi-source heterogeneous data and initialize the virtual simulation scene; Run a virtual simulation scenario and record the operational status data of each stage in real time.

[0015] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the dynamic priority marking module includes a data monitoring unit, a priority calculation unit, and a tag generation and synchronization unit. The data monitoring unit is used to monitor the demand change data of vaccination sites in real time. The priority calculation unit is used to calculate the priority values ​​of different batches and types of vaccines according to the scheduling priority rules in the optimal collaborative decision-making instruction and in combination with the demand change data, using a preset priority algorithm. The tag generation and synchronization unit is used to generate corresponding dynamic priority tags according to the priority values ​​and synchronize the dynamic priority tags to the warehousing and logistics management module through the data transmission interface.

[0016] According to the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by the present invention, the priority calculation unit performs priority numerical calculations in the following steps: The core influencing factors contained in the scheduling priority rules are extracted from the optimal collaborative decision-making instructions, and the basic weight range of the influencing factors in priority evaluation is determined. The demand change data is quantified and transformed to obtain quantified demand change data; Based on the impact factor and combined with the basic attribute data of each vaccine, the initial priority score is calculated according to the basic weight range. Based on the quantitative data on changes in demand, a dynamic adjustment coefficient is generated; The initial priority score is weighted and fused with the dynamic adjustment coefficient to obtain the final priority value; The final priority values ​​are verified. If the differences in values ​​for the same batch of vaccines exceed the preset fluctuation range, a second correction is made based on the inventory distribution data in the virtual warehouse model.

[0017] This invention provides a vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion. By constructing a hybrid prediction model that combines historical data and real-time data streams, it can accurately predict vaccine demand for the next 30 days, improving the accuracy and reliability of predictions. This helps avoid vaccine shortages or surpluses and enhances the responsiveness and flexibility of the supply chain. By using digital twin technology to construct a virtual warehouse model, it achieves a precise virtual mapping between physical warehouses and transportation networks. Combined with edge computing and other technologies, it ensures real-time model updates, allowing managers to intuitively grasp supply chain dynamics. Simultaneously, it employs reinforcement learning to optimize inventory strategies, effectively improving inventory turnover and reducing inventory costs, while ensuring controllable vaccine quality during storage and transportation. By accurately prioritizing vaccines based on real-time demand changes and decision-making instructions, and leveraging blockchain technology to ensure the reliability and traceability of labels, it ensures priority supply of vaccines in emergency situations, enhancing the supply chain's emergency response capabilities and ensuring the orderly conduct of vaccination programs. Especially in response to emergencies such as epidemics, it can quickly allocate resources to maximize the fulfillment of vaccination needs. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the structure of a vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided in an embodiment of the present invention; Figure 2 This is a flowchart of a vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention. Example 1

[0021] The following is combined with Figures 1-2 This invention describes a collaborative optimization system for a vaccine intelligent supply chain based on multi-source data fusion.

[0022] like Figures 1-2As shown in the figure, the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided in this embodiment of the invention includes: a data acquisition and processing module, an intelligent prediction and optimization module, a warehousing and logistics management module, a collaborative decision-making module, and a dynamic priority marking module.

[0023] The data acquisition and processing module is used to collect real-time data on vaccine production, warehouse dynamics, logistics and transportation, and vaccination site demand, outputting multi-source heterogeneous data. It also preprocesses this multi-source heterogeneous data and outputs a standard real-time data stream.

[0024] The data acquisition and processing module includes a multi-source data access unit, a data cleaning unit, a data conversion unit, and a data integration unit. The multi-source data access unit receives vaccine production data, warehouse dynamic data, logistics and transportation data, and vaccination site demand data through various data interfaces to obtain multi-source heterogeneous data. It receives vaccine production data (such as production batches, output, and quality testing data), warehouse dynamic data (such as inventory levels, storage conditions, and inbound / outbound records), logistics and transportation data (such as transport vehicle locations, temperature and humidity monitoring, and estimated arrival times), and vaccination site demand data (such as vaccination plans, vaccine consumption rates, and emergency needs) through various data interfaces, supporting message queues, file transfer, and other interface protocols to obtain multi-source heterogeneous data.

[0025] The data cleaning unit is used to remove noise, missing values, and outliers from multi-source heterogeneous data, resulting in cleaned multi-source heterogeneous data. Sliding window filtering, interpolation methods (such as linear interpolation and spline interpolation), and the isolated forest algorithm are used to remove noise, missing values, and outliers from the multi-source heterogeneous data. For time-series data, an anomaly detection strategy based on time series decomposition is designed to obtain the cleaned multi-source heterogeneous data.

[0026] The data transformation unit is used to convert cleaned multi-source heterogeneous data in different formats into a unified data format, resulting in standard multi-source heterogeneous data. ETL tools (such as Apache NiFi and Talend) are used to convert the cleaned multi-source heterogeneous data in different formats into a unified columnar storage format and standardize the data types.

[0027] The data integration unit is used to associate and integrate standard multi-source heterogeneous data to form a standard real-time data stream. Based on a graph database, standard multi-source heterogeneous data is associated and integrated to construct a vaccine supply chain knowledge graph, forming a standard real-time data stream that includes entity relationships (such as vaccine-manufacturer-warehouse-transportation route-vaccination point) and attribute associations.

[0028] The intelligent prediction and optimization module is used to build a hybrid prediction model based on historical data and combine it with standard real-time data streams to predict vaccine demand and output the demand prediction results.

[0029] The intelligent prediction and optimization module includes a feature extraction unit, a model building unit, a model training unit, and a result prediction unit. The feature extraction unit extracts vaccine features from historical data through correlation analysis. It extracts vaccine features from historical data through correlation analysis, applies time series decomposition to extract seasonal, trend, and periodic features, and combines SHAP value analysis to identify key influencing factors, such as geographical factors, seasonal factors, and epidemic outbreak events.

[0030] The model building unit is used to combine statistical models and machine learning algorithms to construct a hybrid prediction model. The steps for constructing a hybrid prediction model include: Vaccine features were divided into training set (70%), validation set (15%) and test set (15%) in chronological order, and cross-validation was performed using a time sliding window.

[0031] The parameters of the ARIMA model are determined using autocorrelation and partial autocorrelation function plots, and the model is selected in conjunction with the information criterion.

[0032] Based on the statistical model parameters, an ARIMA model is fitted on the training set, and grid search is applied to optimize the differencing order and seasonality parameters. Based on the ARIMA model, the prediction error is evaluated on the validation set, and the parameters are adjusted to their optimal values ​​to obtain the final statistical model.

[0033] Configure the model parameters of the XGBoost regression model, including learning rate, tree depth, and regularization coefficient. Based on the model parameters, train the XGBoost regression model using the training set, evaluate it on the validation set, and perform hyperparameter tuning using Bayesian optimization to obtain the final learned prediction model.

[0034] Calculate the root mean square error of the final statistical model and the final learned prediction model on the validation set, and assign weights based on the derivative of the root mean square error.

[0035] We perform weighted fusion on the test set and use a stacking fusion strategy to build a hybrid prediction model, which outputs point predictions and probability distribution predictions.

[0036] The model training unit is used to train the hybrid prediction model using vaccine features. It introduces an early stop mechanism and regularization to prevent overfitting. The model training is pipelined through TensorFlow Extended to obtain the final hybrid prediction model.

[0037] The result prediction unit is used to input the standard real-time data stream into the final hybrid prediction model for prediction, generating vaccine demand prediction results for the next 30 days, including the daily demand, confidence interval, and abnormal event trigger threshold for each vaccination site, thus obtaining the demand prediction results.

[0038] The warehousing and logistics management module is used to build a virtual warehouse model based on demand forecasting results and digital twin technology, combined with dynamic warehouse data and logistics transportation data.

[0039] The warehousing and logistics management module includes a digital twin unit and an inventory optimization unit. The digital twin unit is used to construct a virtual mapping of the physical warehouse and transportation network based on dynamic warehouse data and logistics transportation data, resulting in a virtual warehouse model driven by real-time IoT data. The steps to obtain the virtual warehouse model include: Based on dynamic warehouse data, namely warehouse CAD drawings and point cloud data, a 3D virtual model was built using the Unity 3D engine to accurately model the shelving layout, handling equipment, and storage areas. LOD (Level of Detail) technology was used to reduce the number of polygons, and texture compression and instantiation techniques were applied to optimize texture materials, improving model rendering performance by over 40% to obtain a simplified 3D virtual model.

[0040] Based on the transportation network, a transportation network model is constructed, including elements such as road network, traffic nodes, and transportation vehicles, supporting real-time traffic updates and route optimization.

[0041] Establish a unified data transmission protocol to convert dynamic data from warehouses (such as temperature and humidity sensor data and AGV location data) and transportation networks (such as GPS positioning and vehicle status) into a format recognizable by the virtual model. Then, use an edge computing gateway to perform protocol conversion and obtain standard transmission data.

[0042] Based on standard transmission data, to simplify the binding of the 3D virtual model and the transportation network model to the corresponding physical entity data source, a data twin mapping relationship is established to support bidirectional data interaction, resulting in a virtual warehouse model.

[0043] Based on physical entity data sources, high-frequency dynamic data is processed through edge computing nodes, and updates to the virtual warehouse model are triggered according to priority, with key indicator update latency controlled within 500ms.

[0044] The inventory optimization unit is used to dynamically adjust inventory strategies based on demand forecasting results and a virtual warehouse model. It applies reinforcement learning to optimize inventory replenishment points and safety stock levels, thereby improving inventory turnover.

[0045] The collaborative decision-making module is used to generate multi-dimensional supply chain scheduling schemes based on the virtual warehouse model, integrating vaccine production data and vaccination site demand data. It also evaluates the feasibility and efficiency of the schemes through simulation and outputs the optimal collaborative decision-making instructions.

[0046] The collaborative decision-making module includes a solution generation unit, a simulation unit, and a solution optimization unit. The solution generation unit is used to generate preliminary multi-dimensional supply chain scheduling solutions based on a virtual warehouse model and vaccine production data, applying constraint satisfaction problems, including production planning, warehousing allocation, and transportation route planning.

[0047] The simulation unit is used to build a simulation environment, input the preliminary multi-dimensional supply chain scheduling plan into it for simulation operation, and collect evaluation indicators during the operation process.

[0048] The simulation environment construction steps include: The system acquires physical parameters for each link in the supply chain, such as warehouse storage capacity, vehicle load capacity, and production equipment capacity, and collects and updates the parameter database in real time via IoT devices. Based on these physical parameters, a virtual simulation scenario is built that closely matches the actual supply chain scenario, including physical entities such as production plants, warehouses, transportation networks, and inoculation points.

[0049] Set the simulation time step and simulation termination conditions. Import simulation parameters from multi-source heterogeneous data and initialize the virtual simulation scene.

[0050] Run virtual simulation scenarios to record operational status data of each link in real time, and apply discrete event simulation and system dynamics methods to simulate the dynamic behavior of the supply chain.

[0051] The scheme optimization unit is used to adjust and optimize the preliminary multi-dimensional supply chain scheduling scheme based on the evaluation indicators. It applies the NSGA-II multi-objective optimization algorithm to obtain multiple sets of alternative schemes and selects the optimal collaborative decision instruction through comparative analysis.

[0052] The dynamic priority tagging module is used to prioritize different batches and types of vaccines based on the optimal collaborative decision-making instructions and the real-time demand changes at vaccination sites, generate dynamic priority tags, and synchronize the priority tags to the warehousing and logistics management module.

[0053] The dynamic priority labeling module includes a data monitoring unit, a priority calculation unit, and a tag generation and synchronization unit. The data monitoring unit is used to monitor changes in demand at vaccination sites in real time, analyze epidemic-related information from social media and news reports using natural language processing technology, and construct a demand early warning indicator system by combining it with medical system data.

[0054] The priority calculation unit is used to calculate the priority values ​​for different batches and types of vaccines based on the scheduling priority rules in the optimal collaborative decision-making instruction and combined with demand change data, using a preset priority algorithm. Specific steps include: The core influencing factors included in the scheduling priority rules were extracted from the optimal collaborative decision-making instructions, including vaccine expiration date, urgency of vaccination site needs, and epidemic risk level. The basic weight range of these influencing factors in priority assessment was determined using the Analytic Hierarchy Process (AHP).

[0055] The demand change data is quantified and transformed, and fuzzy mathematics theory is applied to handle uncertain demand to obtain quantified demand change data.

[0056] Based on the impact factor and the basic attribute data of each vaccine, an initial priority score is calculated according to the basic weight range. The specific calculation method is as follows: .

[0057] Among them, P initial The initial priority score is a weighted linear combination of multiple impact factors. m represents the total number of impact factors, and w... i It is the weight of the i-th influence factor, satisfying f i (x) is the normalized scoring function of the i-th influencing factor, where x is the basic attribute data vector of the vaccine. The influencing factors include the remaining percentage of vaccine shelf life, the urgency of demand at vaccination sites, the vaccine type matching degree, batch quality score, and transportation distance cost.

[0058] Construct an n×n judgment matrix, and then calculate the eigenvectors based on the judgment matrix, as follows: .

[0059] Where w is the judgment matrix and w' is the weight vector of the influencing factors. λ max λ is the largest eigenvalue of the judgment matrix A. When the judgment matrices are completely identical, λ max =n, and the corresponding feature vector w is the weight vector. In practice, the judgment may be contradictory, in which case λ max >n, can be obtained through λ max Calculate the consistency index Then, the consistency ratio is obtained by combining the average random consistency index RI. If CR < 0.1, the judgment matrix is ​​acceptable; otherwise, the judgment needs to be revised.

[0060] Based on quantitative demand change data, a time series forecasting model is applied to generate dynamic adjustment coefficients, supporting differentiated processing of short-term fluctuations and long-term trends.

[0061] The initial priority score is weighted and fused with the dynamic adjustment coefficient to obtain the final priority value. A non-linear weighting function is used to enhance the responsiveness to urgent needs.

[0062] The final priority values ​​are verified. If the differences in values ​​for the same batch of vaccines exceed the preset fluctuation range, a second correction is made based on the inventory distribution data in the virtual warehouse model to ensure the consistency and fairness of the priority ranking.

[0063] The tag generation and synchronization unit generates corresponding dynamic priority tags based on priority values, and uses blockchain technology to ensure the immutability and traceability of the tags. It also uses a message queue to synchronize the dynamic priority tags to the warehousing and logistics management module, achieving a tag update response within 100ms.

[0064] In summary, the vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion provided by this invention, by constructing a hybrid prediction model and combining historical data and real-time data streams, can accurately predict vaccine demand for the next 30 days, improving the accuracy and reliability of predictions. This helps avoid vaccine shortages or surpluses and enhances the responsiveness and flexibility of the supply chain. By using digital twin technology to construct a virtual warehouse model, a precise virtual mapping between physical warehouses and transportation networks is achieved. Combined with edge computing and other technologies, real-time model updates are ensured, allowing managers to intuitively grasp supply chain dynamics. Simultaneously, reinforcement learning is used to optimize inventory strategies, effectively improving inventory turnover and reducing inventory costs, while ensuring controllable vaccine quality during storage and transportation. By accurately prioritizing vaccines based on real-time demand changes and decision-making instructions, and leveraging blockchain technology to ensure the reliability and traceability of labels, priority supply of vaccines is ensured in emergency situations, enhancing the supply chain's emergency response capabilities and guaranteeing the orderly conduct of vaccination programs. Especially in response to emergencies such as epidemics, resources can be quickly allocated to maximize the fulfillment of vaccination needs.

[0065] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion, characterized in that, include: The data acquisition and processing module is used to collect vaccine production data, warehouse dynamic data, logistics and transportation data, and vaccination site demand data in real time, and output multi-source heterogeneous data. The multi-source heterogeneous data is preprocessed to output a standard real-time data stream; The intelligent prediction and optimization module is used to build a hybrid prediction model based on historical data and combine it with the standard real-time data stream to predict vaccine demand and output the demand prediction results. The warehousing and logistics management module is used to construct a virtual warehouse model based on the demand forecast results and digital twin technology, combined with the warehouse dynamic data and the logistics transportation data; The collaborative decision-making module is used to generate a multi-dimensional supply chain scheduling scheme based on the virtual warehouse model, by integrating the vaccine production data and the vaccination site demand data, and to evaluate the feasibility and efficiency of the scheme through simulation and output the optimal collaborative decision-making instruction. The dynamic priority tagging module is used to prioritize different batches and types of vaccines according to the optimal collaborative decision-making instructions and the real-time demand changes of vaccination sites, generate dynamic priority tags, and synchronize the priority tags to the warehousing and logistics management module.

2. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 1, characterized in that, The data acquisition and processing module includes a multi-source data access unit, a data cleaning unit, a data conversion unit, and a data integration unit. The multi-source data access unit receives vaccine production data, warehouse dynamic data, logistics and transportation data, and vaccination site demand data through various data interfaces to obtain the multi-source heterogeneous data. The data cleaning unit removes noise, missing values, and outliers from the multi-source heterogeneous data to obtain cleaned multi-source heterogeneous data. The data conversion unit converts the cleaned multi-source heterogeneous data in different formats into a unified data format to obtain standard multi-source heterogeneous data. The data integration unit integrates the standard multi-source heterogeneous data to form the standard real-time data stream.

3. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 1, characterized in that, The intelligent prediction and optimization module includes a feature extraction unit, a model building unit, a model training unit, and a result prediction unit. The feature extraction unit is used to extract vaccine features from the historical data through relevant new analysis. The model building unit is used to construct the hybrid prediction model by combining statistical models and machine learning algorithms. The model training unit is used to train the hybrid prediction model using the vaccine features to obtain the final hybrid prediction model. The result prediction unit is used to input the standard real-time data stream into the final hybrid prediction model for prediction to obtain the demand prediction result.

4. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 3, characterized in that, The steps for the model building unit to build the hybrid prediction model include: The vaccine features were divided into training set, validation set and test set in chronological order. Use autocorrelation function and partial autocorrelation function plots to determine the parameters of the statistical model; Based on the statistical model parameters, an ARIMA model is fitted on the training set; based on the ARIMA model, the prediction error is evaluated on the validation set, and the parameters are adjusted to the optimal value to obtain the final statistical model. Set the model parameters of the XGBoost regression model, train the XGBoost regression model using the training set based on the model parameters, and evaluate and adjust the parameters on the validation set to obtain the final learned prediction model; Calculate the root mean square error of the final statistical model and the final learning prediction model on the validation set, and assign weights based on the derivative of the root mean square error; The weighted fusion is performed on the test set to obtain the hybrid prediction model.

5. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 1, characterized in that, The warehousing and logistics management module includes a digital twin unit and an inventory optimization unit; The digital twin unit is used to construct a virtual mapping between the physical warehouse and the transportation network based on the warehouse dynamic data and the logistics transportation data, thereby obtaining a virtual warehouse model driven by real-time IoT data. The inventory optimization unit is used to dynamically adjust the inventory strategy based on the demand forecast results and the virtual warehouse model.

6. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 5, characterized in that, The steps for the digital twin unit to obtain a virtual warehouse model driven by real-time IoT data include: Based on the warehouse dynamic data, a three-dimensional virtual model is established; and the three-dimensional virtual model is simplified by reducing the number of polygons and optimizing the texture materials to obtain a simplified three-dimensional virtual model. Based on the aforementioned transportation network, a transportation network model is constructed; A unified data transmission protocol is established to convert the dynamic data of the warehouse and the dynamic data of the transportation network into a format that the virtual model can recognize, thereby obtaining standard transmission data; Based on the standard transmission data, the simplified 3D virtual model and the transportation network model are bound to the corresponding physical entity data source to obtain the virtual warehouse model; Based on the physical entity data source, high-frequency dynamic data is processed through edge computing nodes, and the update of the virtual warehouse model is triggered according to priority.

7. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 1, characterized in that, The collaborative decision-making module includes a scheme generation unit, a simulation unit, and a scheme optimization unit; the scheme generation unit is used to generate a preliminary multi-dimensional supply chain scheduling scheme based on the virtual warehouse model and the vaccine production data. The simulation unit is used to build a simulation environment, input the preliminary multi-dimensional supply chain scheduling scheme into it for simulation operation, and collect evaluation indicators during the operation process; the scheme optimization unit is used to adjust and optimize the preliminary multi-dimensional supply chain scheduling scheme according to the evaluation indicators, obtain multiple sets of alternative schemes, and select the optimal collaborative decision instruction through comparative analysis.

8. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 7, characterized in that, The simulation environment construction steps of the simulation unit include: Obtain the physical parameters of each link in the supply chain; build a virtual simulation scenario with the same scale as the actual supply chain scenario based on the physical parameters; Set the simulation time step and simulation termination conditions; Import the simulation parameters from the multi-source heterogeneous data to initialize the virtual simulation scene; Run the virtual simulation scenario and record the operational status data of each stage in real time.

9. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 1, characterized in that, The dynamic priority marking module includes a data monitoring unit, a priority calculation unit, and a tag generation and synchronization unit; the data monitoring unit is used to monitor the real-time demand changes at vaccination sites. The priority calculation unit is used to calculate the priority values ​​of different batches and types of vaccines based on the scheduling priority rules in the optimal collaborative decision instruction and the demand change data, using a preset priority algorithm; the tag generation and synchronization unit is used to generate corresponding dynamic priority tags based on the priority values ​​and synchronize the dynamic priority tags to the warehousing and logistics management module through the data transmission interface.

10. The vaccine intelligent supply chain collaborative optimization system based on multi-source data fusion according to claim 9, characterized in that, The priority calculation unit performs priority value calculations in the following steps: The core influencing factors contained in the scheduling priority rules are extracted from the optimal collaborative decision-making instructions, and the basic weight range of the influencing factors in priority evaluation is determined. The aforementioned demand change data is then quantified to obtain quantified demand change data. Based on the aforementioned influencing factors and combined with the basic attribute data of each vaccine, an initial priority score is calculated according to the aforementioned basic weight range. Based on the aforementioned quantitative demand change data, a dynamic adjustment coefficient is generated; The initial priority score is weighted and fused with the dynamic adjustment coefficient to obtain the final priority value; The final priority value is verified. If the difference in the value of the same batch of vaccines exceeds the preset fluctuation range, a second correction is made based on the inventory distribution data in the virtual warehouse model.

Images