A multi-level collaborative intelligent order scheduling and tracking method and system for a medical supply chain

Abstract

The application provides a medical supply chain multi-level collaborative intelligent order scheduling and tracking method and system, acquires supply chain node information, inventory batch information and to-be-scheduled order information; constructs a material scheduling tendency potential energy matrix between nodes, calculates demand potential energy by using an S-shaped gain function according to the quantity of medical institution order demand, calculates node state disturbance values in combination with historical operation interruption frequency and storage environment compliance data of the nodes, calculates the time value of each batch by an exponential decay function according to the shelf life of the inventory batch, and generates a potential energy matrix by comprehensively integrating the three; generates an order candidate scheduling path based on the gradient of the potential energy matrix, accumulates the path comprehensive disturbance value of the path node state disturbance value, and selects the minimum value as the scheduling path; performs scheduling according to the path, tracks logistics information in real time, updates the historical operation interruption frequency of the related nodes when the logistics information deviates, adjusts the node state disturbance value, and optimizes subsequent order scheduling.

Description

Technical Field

[0001] This application belongs to the field of order scheduling, and in particular relates to a multi-level collaborative intelligent order scheduling and tracking method and system for the medical supply chain. Background Technology

[0002] The healthcare supply chain is responsible for the warehousing, transportation, and distribution of pharmaceuticals, biological products, diagnostic reagents, and specialized medical devices. Scheduling methods for supply chain orders typically employ optimization models based on minimum cost or shortest path, such as classic algorithms for the Traveling Salesman Problem or vehicle routing problems. However, these models struggle to meet the demands of modern healthcare services for responsiveness, reliability, and intelligence in the face of uncertain healthcare supply chain scheduling scenarios. Scheduling models neglect several core influencing factors in the healthcare supply chain. For example, they rarely integrate and unify information such as the urgency of needs from various medical institutions, the potential risks arising from historical operational disruptions or non-compliant environments at warehousing and transit nodes, and the varying time-sensitive value of different inventory batches nearing their expiration dates. This can lead to scheduling schemes that, while achieving optimal distance or cost, allocate urgent supplies to high-risk routes, resulting in safety hazards and resource waste. Existing scheduling systems lack a closed-loop feedback and self-optimization mechanism. When unexpected situations occur during actual transportation, adjustments cannot be made based on real-time information, nor can these deviations be accumulated as empirical data to correct and optimize future scheduling strategies. This results in weak resilience and learning / evolutionary capabilities of the entire supply chain system in the face of emergencies. Summary of the Invention

[0003] First, this invention proposes a multi-level collaborative intelligent order scheduling and tracking method for the medical supply chain, which addresses the problems of existing scheduling systems lacking a closed-loop feedback and self-optimization mechanism, being unable to adjust based on real-time information, and being unable to accumulate deviation events as empirical data. This includes: Obtain information on each node in the medical supply chain, inventory batch information, and order information awaiting dispatch; A potential energy matrix representing the material scheduling tendency among nodes is constructed. The construction steps include: calculating the demand potential energy using an S-shaped gain function based on the order demand quantity of medical institutions; calculating the node state disturbance value based on the historical operation interruption frequency and warehousing environment compliance data of each node; calculating the time-effectiveness value of each inventory batch using an exponential decay function based on the material validity period in the inventory batch information; generating the potential energy matrix by combining the demand potential energy, the time-effectiveness value, and the node state disturbance value; and optimizing the potential energy matrix. Based on the gradient of the optimized potential energy matrix, a set of candidate scheduling paths are generated for the orders to be scheduled; the node state disturbance values ​​of the nodes on each candidate scheduling path are accumulated to calculate the comprehensive path disturbance value, and the path with the smallest comprehensive path disturbance value is selected as the scheduling path. Order scheduling is performed according to the scheduling path, and logistics information during the scheduling process is tracked in real time. When the tracked logistics information deviates, the historical operation interruption frequency data of the relevant nodes is updated according to the deviation, and the node status disturbance value is adjusted based on the updated data to optimize the scheduling of subsequent orders.

[0004] Secondly, this invention also relates to a multi-level collaborative intelligent order scheduling and tracking system for the medical supply chain, comprising the following modules: The acquisition module is used to acquire information on each node in the medical supply chain, inventory batch information, and order information to be scheduled. A construction module is used to construct a potential energy matrix representing the material scheduling tendency among nodes. The construction steps include: calculating the demand potential energy using an S-shaped gain function based on the order demand quantity of medical institutions; calculating the node state disturbance value based on the historical operation interruption frequency and warehousing environment compliance data of each node; calculating the time-effectiveness value of each inventory batch using an exponential decay function based on the material validity period in the inventory batch information; generating the potential energy matrix by combining the demand potential energy, the time-effectiveness value, and the node state disturbance value; and optimizing the potential energy matrix. The selection module is used to generate a set of candidate scheduling paths for the order to be scheduled based on the gradient of the optimized potential energy matrix; calculate the comprehensive path disturbance value by accumulating the node state disturbance values ​​of the nodes on each candidate scheduling path, and select the path with the smallest comprehensive path disturbance value as the scheduling path. The optimization module is used to execute order scheduling according to the scheduling path and track logistics information in real time during the scheduling process. When the tracked logistics information deviates, the module updates the historical operation interruption frequency data of the relevant nodes according to the deviation and adjusts the node status disturbance value based on the updated data to optimize the scheduling of subsequent orders.

[0005] Finally, the present invention also proposes a computer program that, when executed by a processor, implements the steps of the above method.

[0006] This invention constructs a potential energy matrix that integrates demand potential energy, timeliness value, and node state disturbances. It incorporates the urgency of medical institution needs, the near-expiration status of inventory materials, and potential risk factors at warehousing and transportation nodes into a unified decision-making model. Further optimization of the potential energy matrix ensures that the generated scheduling scheme balances efficiency, safety, and cost, proactively avoiding high-risk nodes with historically frequent interruptions or non-compliant environments. By calculating and selecting the path with the minimum comprehensive disturbance value, the overall reliability of the execution scheme is ensured. When actual scheduling deviates, updating historical data for relevant nodes ensures that the risk assessments used to base scheduling decisions are closer to reality, thereby continuously improving the stability of the entire supply chain scheduling scheme and reducing the risk of material waste and transportation delays. Attached Figure Description

[0007] Figure 1 A flowchart of the first embodiment; Figure 2 This is a schematic diagram of the potential energy matrix for material scheduling between nodes; Figure 3 A schematic diagram illustrating the generation of multi-level collaborative intelligent order scheduling paths; Figure 4 This is a schematic diagram of node state perturbation update based on feedback. Detailed Implementation

[0008] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0009] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0010] In the first embodiment, the present invention proposes a multi-level collaborative intelligent order scheduling and tracking method for the medical supply chain, such as... Figure 1 ,include: S1, obtain information on each node in the medical supply chain, inventory batch information, and order information to be scheduled; The system integrates with enterprise resource planning (ERP) systems, warehouse management systems, and transportation management systems via application programming interfaces (APIs) to collect data in real time. Node information includes the geographical location, storage capacity, operating hours, and available transportation capacity of warehouses at all levels, transit centers, and medical institutions. Inventory batch information includes the unique identifier, name, quantity, production date, expiration date, and storage conditions such as cold chain requirements for each batch of materials. Order information awaiting dispatch includes the medical institution initiating the order, the name and quantity of the required materials, and the expected delivery time.

[0011] S2, Construct a potential energy matrix representing the material scheduling tendency among nodes. The construction steps include: calculating the demand potential energy using an S-shaped gain function based on the order demand quantity of medical institutions; calculating the node state disturbance value based on the historical operation interruption frequency and warehousing environment compliance data of each node; calculating the time-sensitive value of each inventory batch using an exponential decay function based on the material validity period in the inventory batch information; generating the potential energy matrix by combining the demand potential energy, the time-sensitive value, and the node state disturbance value; and optimizing the potential energy matrix. The Logistic function is used as the sigmoid gain function to map the quantity of order demand to a demand potential value. For an order with quantity X, the order demand potential P is expressed by the formula... The calculation is performed, where L is the maximum potential energy value, k controls the steepness of the curve, and X0 is the inflection point of the demand quantity. This allows a small amount of regular demand to generate low potential energy, while when the demand reaches a certain scale, the potential energy increases rapidly, reflecting urgency. After exceeding a certain threshold, the growth of potential energy slows down, which is used to distinguish between emergency demand and strategic reserves.

[0012] The historical operational interruption frequency of each node, such as the average number of monthly transportation delays or sorting equipment failures, is normalized to obtain an interruption risk score. Simultaneously, an environmental compliance score is given based on compliance data of the storage environment, such as the duration or frequency of exceeding temperature and humidity limits in monitoring records. A weighted summation is then performed, adding the interruption risk score to the result minus the environmental compliance score, to obtain a comprehensive node state disturbance value. For environmentally sensitive cold chain pharmaceuticals, environmental compliance carries a higher weight.

[0013] For each inventory batch, based on the number of days t remaining until the expiration date of the batch, use the formula... Calculate the timeliness value V. Here, V0 is the initial value, preferably 1, and λ is a normal decay factor. This function ensures that the further away from the expiration date a batch is, the higher its timeliness value and the slower its decay; while as a batch approaches its expiration date, its value decreases more rapidly, thus assigning higher scheduling priority to batches nearing their expiration date in scheduling decisions.

[0014] The potential energy matrix is ​​a two-dimensional table. Rows represent all available inventory batches and their corresponding nodes, while lists represent all healthcare institution nodes that issued orders. The element value at row i and column j in the matrix indicates the tendency to schedule batch i to institution j. This element value is calculated using a comprehensive formula, such as multiplying the demand potential energy of target institution j by the timeliness value of batch i, and then subtracting a multiple of the node state perturbation value of the node containing batch i. The scheduling combination of high demand, high timeliness value, and originating from low-risk nodes will yield the highest potential energy value, such as... Figure 2 In one embodiment, the potential energy matrix is ​​further optimized; however, those skilled in the art should understand that further optimization of the potential energy matrix is ​​not necessary. Optionally, the transportation distance between each node in the supply chain is calculated to construct a distance matrix; the initial potential energy matrix is ​​subtracted from the normalized distance matrix, or the initial potential energy matrix is ​​multiplied by a distance-based attenuation factor, so that under the same demand potential energy and timeliness value, nodes that are closer have a higher relative potential energy difference, while links with optimized potential energy values ​​lower than a preset threshold are set to zero, thereby prioritizing the matching of nearby nodes and eliminating long-distance, high-cost, invalid scheduling paths while ensuring the effectiveness of materials.

[0015] In some embodiments, calculating the demand potential energy using an S-shaped gain function based on the order quantity required by the medical institution includes: The Logistic function is used, which is based on the order demand quantity and parameterized by preset maximum demand potential, growth rate and inflection point value of demand quantity.

[0016] By using a function to represent the urgency of the receiving node's demand for resources, the limitation of linear models in failing to reflect demand saturation is avoided. The expression for the Logistic function is: ,in is the potential energy demanded by receiving node j when the demand quantity is x, L is the preset maximum potential energy demand, representing the upper limit of the node's processing capacity, and k is the growth rate, which determines the steepness of the curve. This is the inflection point value, representing a quantitative point sensitive to demand response. The three parameters are pre-set based on the node's historical data and business attributes. For example, the maximum demand potential L of a regional distribution center j is set to 100, the growth rate k is set to 0.05, and the inflection point value... Let's assume the number of orders is 1000. When the number of pending orders at the center is 800, the demand potential is close to 0, indicating that the demand has not yet reached an urgent level. When the number of orders increases to 1100, the demand potential is approximately 99.3, indicating that the demand is close to saturation and the scheduling priority is very high. The parameterized model structure can flexibly adapt to nodes of different sizes and types, with the input being the real-time updated number of orders and the output being a standardized demand potential value.

[0017] In an optional embodiment, calculating the node state disturbance value based on the historical operational interruption frequency and warehousing environment compliance data of each node includes: The node state disturbance value is obtained by weighted summation of the node's historical operation interruption frequency and the degree of non-compliance of the warehousing environment. The historical operation interruption frequency is determined based on the average daily interruption frequency within a preset time period, and the degree of non-compliance of the warehousing environment is determined based on a compliance score.

[0018] For historical operation interruption frequency, the number of times a node i experienced operation interruptions due to equipment failure or process errors within the past 90 days is statistically analyzed. For example, if there were 18 interruptions in 90 days, the average daily interruption frequency is 0.2 times per day. To facilitate calculation, this value can be normalized, for example, by mapping the value to the interval between 0 and 1 to obtain the interruption frequency index. For the degree of non-compliance in the storage environment, a compliance score is given based on sensor data or periodic inspection results, with a maximum score of 100. If the temperature and humidity control compliance score of node i is 70, then the degree of non-compliance of that node can be calculated as 0.3, yielding the non-compliance index. The node state perturbation value is obtained by weighted summation of the two indicators. The calculation formula is: ,in and It is a preset weight, and For example, if operational disruption is considered a greater risk than environmental non-compliance, settings can be configured... Continuing with the previous example, if the normalized interruption frequency index of node i... The value is 0.4, which is a non-compliant indicator. If the value is 0.3, then the node state disturbance value is... The model structure comprises three modules: data acquisition, indicator calculation, and weighted fusion. It integrates multi-source risk information into a unified perturbation value to evaluate the stability of a node as a link in the path.

[0019] To assess the degree to which the economic or use value of inventory batch k changes over time, in an optional embodiment, the calculation of the time-sensitive value of each inventory batch using an exponential decay function based on the expiration dates of the materials in the inventory batch information includes: The time-sensitive value is calculated using an exponential decay function, which is determined based on the initial value of the material, the storage time, and the decay coefficient corresponding to the type of material.

[0020] The expression for the exponential decay function is: ,in It represents the time-sensitivity value of batch k stored at node i after a storage time of t. This is the initial value of the batch of goods when it was received into the warehouse. This is the attenuation coefficient related to this type of material, where t is the time elapsed since the material entered the warehouse. Attenuation coefficient Based on the preset attributes of the goods, such as fresh food The value is relatively high, while that of electronic components is... The value is low. For example, a batch of initial values... For a drug k worth 20,000 yuan, the value decay coefficient is... The value is set to 0.01 per day. When the batch of medicine has been stored in warehouse i for 30 days, i.e., t=30, the real-time value is calculated as follows: Yuan. In contrast, another batch of drugs with an initial value of 5000 yuan had a decay coefficient... At a rate of only 0.0001 per day, after 30 days of storage, the value would be approximately 4985 yuan, demonstrating a very slow value decay. The model structure transforms material information in the inventory management system, including category, entry time, and initial value, into a time-sensitive value indicator, enabling scheduling decisions to prioritize materials with rapid value loss.

[0021] In an optional embodiment, generating the potential energy matrix by integrating the demand potential energy, the timeliness value, and the node state perturbation value includes: The elements in the potential energy matrix represent the potential energy allocated from inventory batch k at sending node i to receiving node j. The calculation method is as follows: The node state perturbation value of node i will be emitted. The node state perturbation value of receiving node j The product of the two, divided by the required potential energy of receiving node j. The time-sensitive value of inventory batch k The product of , the formula is: .

[0022] The numerator of the formula This represents the combined risk at the start and end points of the scheduling path. If both the sending node i and the receiving node j are unstable, i.e. and If both are relatively high, their product will increase, thereby raising the scheduling potential and making the system more inclined to avoid the higher-risk scheduling. The denominator of the formula... This represents the driver of scheduling. If the receiving node j requires potential energy... The timeliness value of the inventory batch k to be scheduled is very high. It's also very high, representing a high-value and high-demand emergency scheduling operation. Increasing the denominator reduces the scheduling potential energy, making it a preferred choice. For a single operation scheduling batch k from node i to node j, let the state perturbation value of node i be... The state perturbation value of node j Node j has strong demand, and its demand potential is high. Batch k represents high-time-sensitivity goods, and the time-sensitivity value of this batch... Therefore, the potential energy of this scheduling is... A low potential energy value indicates that the scheduling is a low-risk, high-reward preferred option. Conversely, if the issuing node... Receive node Demand potential Timeliness value Then the potential energy is The potential energy value is much higher than the former, indicating that it is a high-risk, low-return scheduling, which should be avoided as much as possible. Multi-dimensional information should be integrated into a single potential energy scalar.

[0023] Due to the substitutability and mutual exclusion of drugs, in some embodiments, the potential energy matrix is ​​further optimized. Specifically, the potential energy matrix is ​​optimized based on a pharmacological equivalence substitution relationship knowledge graph and a mixed storage mutual exclusion tensor of materials. The pharmacological equivalence substitution relationship knowledge graph is traversed to extract a set of substitute nodes that have an equivalent relationship with the materials in the order to be scheduled. The efficacy utility decay coefficient is calculated based on the clinical substitution recommendation level. For example, the association edge attributes between the materials in the order to be scheduled and the candidate substitute materials in the knowledge graph are extracted to obtain the clinical substitution recommendation evidence level L and the treatment index risk factor k of the target material; the efficacy utility index decays to... For specially controlled drugs with narrow therapeutic windows, a higher risk factor is set. The potential energy of the stockpiled materials at the substitute inventory nodes is weighted by the efficacy decay coefficient and projected onto the potential energy matrix plane of the current order materials through coordinate transformation. This forms a virtual low-potential-energy attraction domain (or potential well) in the potential energy matrix, allowing the algorithm to automatically guide the path to inventory nodes with equivalent substitutes based on the principle of minimum potential energy when the original drug or preferred material is scarce. Using the material mixed storage mutual exclusion tensor, a vector product operation is performed between the material to be scheduled and the currently stored materials of each candidate path node to calculate the chemical / biological property repulsion index. For nodes with a non-zero repulsion index, a positive Gaussian barrier is superimposed at the corresponding coordinate position in the potential energy matrix. By significantly increasing the potential energy value at that position, a high potential energy impedance region is created, forcing the gradient-based path search algorithm to generate a topological deflection due to incompatibility when generating the path, thereby automatically avoiding transit nodes with hazardous chemical conflicts or cross-contamination risks.

[0024] S3, Based on the gradient of the optimized potential energy matrix, generate a set of candidate scheduling paths for the order to be scheduled; accumulate the node state disturbance values ​​of the nodes on each candidate scheduling path to calculate the comprehensive path disturbance value, and select the path with the smallest comprehensive path disturbance value as the scheduling path. The potential energy matrix is ​​treated as a potential energy field. For an order to be scheduled, starting from the node where the available inventory batch is located, the direction of the fastest potential energy growth is sought to determine the next node. This process is iterated until the target medical institution is reached, thus generating the first candidate path. In one embodiment, the gradient of the potential energy matrix is ​​the potential energy difference between adjacent nodes.

[0025] By adjusting the initial selection or search strategy, this process is repeated to generate, for example, three to five different high-potential candidate paths. For each candidate path, the node state perturbation values ​​of the originating node, all transit nodes, and the target node included in the path are summed to obtain the comprehensive perturbation value of the path. Among all candidate paths, the one with the lowest comprehensive perturbation value is selected as the scheduling path.

[0026] In an optional embodiment, generating a set of candidate scheduling paths for the order to be scheduled based on the gradient of the optimized potential energy matrix includes: Starting from the node where the inventory is located, a greedy algorithm is used to iteratively select the neighboring node with the lowest potential energy until the target node is reached, generating the first candidate path; By disabling some nodes in the generated path and repeating the greedy algorithm, a preset number of subsequent candidate paths are generated.

[0027] Suppose a batch of medicine needs to be transported from origin A to destination E. Possible intermediate nodes in the logistics network are B, C, and D. Starting from A, evaluate all directly connected neighboring nodes, namely B and C. It calculates the scheduling potential energy from A to B. and the scheduling potential energy from A to C Assuming , According to the greedy strategy, the path with the lowest potential energy is chosen, so the first step is to move to B. Starting from B, the potential energy of the neighboring nodes C and D is evaluated. If... Below If the path is not found, move to D. Move from D to E to complete path generation. The first candidate path is A to B to D to E.

[0028] Other candidate paths are generated by temporarily disabling a node in the first path. For example, node B is disabled, and the greedy algorithm is re-executed from A to E. Since B is unavailable, the first step only allows moving to C. Iterating from C until reaching E may generate a new path, such as A to C to D to E. B is restored, node D is disabled, and the algorithm is run again, potentially resulting in a third path, A to B to C to E. This process of disabling and replanning is repeated until a predetermined number of candidate paths, such as 5, are generated. The model structure is an iterative path finder that combines the efficiency of the greedy algorithm with the diversity of node disabling strategies, enabling the rapid generation of a set of high-quality and differentiated scheduling schemes, such as... Figure 3 .

[0029] S4. Execute order scheduling according to the scheduling path and track logistics information in real time during the scheduling process; when the tracked logistics information deviates, update the historical operation interruption frequency data of the relevant nodes according to the deviation, and adjust the node status disturbance value based on the updated data to optimize the scheduling of subsequent orders.

[0030] The determined scheduling routes are translated into specific operational instructions, including outbound instructions, picking instructions, and transportation tasks, and then issued to the corresponding warehouse management system and transportation management system to initiate physical logistics operations. Simultaneously, logistics information is continuously collected through GPS devices deployed on transport vehicles and IoT devices such as temperature and humidity sensors inside the cargo compartment. This information includes the vehicle's real-time geographical location, speed, estimated arrival time, and environmental parameters inside the cargo compartment, and is transmitted back to the central monitoring platform in real time.

[0031] When the monitoring platform detects a deviation in logistics information, such as a vehicle experiencing a delay exceeding a preset threshold near a certain road segment or node, or abnormal fluctuations in the temperature of a cold chain cargo container, it records the deviation event and associates it with the geographically nearest node or road segment. It also updates the historical operational interruption frequency database for that node. Figure 4 For example, the monthly delay event count for that node can be incremented by one. Based on the updated frequency data, the node state disturbance value is recalculated, increasing the disturbance value. In subsequent order scheduling, the overall disturbance value of paths passing through that node will increase, leading to a preference for selecting other more reliable paths, thereby achieving continuous improvement of the scheduling strategy.

[0032] In an optional embodiment, updating the historical operational interruption frequency data of the relevant nodes based on the deviation includes: When logistics delays exceed a preset threshold, it is determined as a deviation event; Identify the responsible node for the deviation event and update the historical operation interruption frequency of the node using an exponential moving average method with a preset smoothing coefficient.

[0033] Real-time monitoring of transportation tasks compares actual arrival times with planned arrival times. For example, a transportation task planned for 24 hours may actually take 32 hours, a delay of 8 hours. If the preset delay threshold is 6 hours, this transportation is considered a deviation event. By analyzing the transportation logs and node operation records for each segment, the main cause of the delay is determined. If it is found that the medicines were delayed for 10 hours at a transfer center B en route, far exceeding the planned 2 hours, and the cause was equipment failure, then node B is identified as the responsible node for this deviation event.

[0034] The historical outage frequency of the responsible node is updated using the exponential moving average method, and the formula is as follows: ,in This is the updated frequency. This is the original historical outage frequency of this node. This is the quantified value of this event; typically, it is 1 if an interruption occurs on the same day, and 0 if no interruption occurs. This is a smoothing coefficient, ranging from 0 to 1, which determines the weight of the impact of new events on historical data. For example, the original historical interruption frequency of node B. The value is 0.15. This node is determined to be the responsible node, therefore... If the smoothing coefficient If we set it to 0.1, then the updated frequency will be... The updated value will be used to calculate the perturbation value of this node in the future, so that the risk assessment is closer to the actual performance in the near future.

[0035] In the second embodiment, the present invention also proposes a multi-level collaborative intelligent order scheduling and tracking system for the medical supply chain, comprising the following modules: The acquisition module is used to acquire information on each node in the medical supply chain, inventory batch information, and order information to be scheduled. A construction module is used to construct a potential energy matrix representing the material scheduling tendency among nodes. The construction steps include: calculating the demand potential energy using an S-shaped gain function based on the order demand quantity of medical institutions; calculating the node state disturbance value based on the historical operation interruption frequency and warehousing environment compliance data of each node; calculating the time-effectiveness value of each inventory batch using an exponential decay function based on the material validity period in the inventory batch information; generating the potential energy matrix by combining the demand potential energy, the time-effectiveness value, and the node state disturbance value; and optimizing the potential energy matrix. The selection module is used to generate a set of candidate scheduling paths for the order to be scheduled based on the gradient of the optimized potential energy matrix; calculate the comprehensive path disturbance value by accumulating the node state disturbance values ​​of the nodes on each candidate scheduling path, and select the path with the smallest comprehensive path disturbance value as the scheduling path. The optimization module is used to execute order scheduling according to the scheduling path and track logistics information in real time during the scheduling process. When the tracked logistics information deviates, the module updates the historical operation interruption frequency data of the relevant nodes according to the deviation and adjusts the node status disturbance value based on the updated data to optimize the scheduling of subsequent orders.

[0036] In an optional embodiment, the step of calculating the demand potential energy using an S-shaped gain function based on the order quantity required by the medical institution includes: The Logistic function is used, which is based on the order demand quantity and parameterized by preset maximum demand potential, growth rate and inflection point value of demand quantity.

[0037] In an optional embodiment, calculating the node state disturbance value based on the historical operational interruption frequency and warehousing environment compliance data of each node includes: The node state disturbance value is obtained by weighted summation of the node's historical operation interruption frequency and the degree of non-compliance of the warehousing environment. The historical operation interruption frequency is determined based on the average daily interruption frequency within a preset time period, and the degree of non-compliance of the warehousing environment is determined based on a compliance score.

[0038] In an optional embodiment, the step of calculating the time-sensitive value of each inventory batch based on the expiration dates of the materials in the inventory batch information using an exponential decay function includes: The time-sensitive value is calculated using an exponential decay function, which is determined based on the initial value of the material, the storage time, and the decay coefficient corresponding to the type of material.

[0039] In an optional embodiment, generating the potential energy matrix by integrating the demand potential energy, the timeliness value, and the node state perturbation value includes: The elements in the potential energy matrix represent the potential energy allocated from inventory batch k at sending node i to receiving node j. The calculation method is as follows: The node state perturbation value of node i will be emitted. The node state perturbation value of receiving node j The product of the two, divided by the required potential energy of receiving node j. The time-sensitive value of inventory batch k The product of.

[0040] In an optional embodiment, generating a set of candidate scheduling paths for the order to be scheduled based on the gradient of the optimized potential energy matrix includes: Starting from the node where the inventory is located, a greedy algorithm is used to iteratively select the neighboring node with the lowest potential energy until the target node is reached, generating the first candidate path; By disabling some nodes in the generated path and repeating the greedy algorithm, a preset number of subsequent candidate paths are generated.

[0041] In an optional embodiment, updating the historical operational interruption frequency data of the relevant nodes based on the deviation includes: When logistics delays exceed a preset threshold, it is determined as a deviation event; Identify the responsible node for the deviation event and update the historical operation interruption frequency of the node using an exponential moving average method with a preset smoothing coefficient.

[0042] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0043] The functional modules shown in the above-described block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0044] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0045] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0046] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A multi-level collaborative intelligent order scheduling and tracking method for the medical supply chain, characterized in that, include: Obtain information on each node in the medical supply chain, inventory batch information, and order information awaiting dispatch; Constructing a potential energy matrix representing the material scheduling tendency among nodes includes: calculating the demand potential energy using an S-shaped gain function based on the order demand quantity of medical institutions; calculating the node state disturbance value based on the historical operation interruption frequency and warehousing environment compliance data of each node; calculating the time-sensitive value of each inventory batch using an exponential decay function based on the material expiration date in the inventory batch information; generating a potential energy matrix by combining the demand potential energy, time-sensitive value, and node state disturbance value; and optimizing the potential energy matrix. Based on the gradient of the optimized potential energy matrix, a set of candidate scheduling paths are generated for the orders to be scheduled; the node state disturbance values ​​of each node on each candidate scheduling path are accumulated to calculate the comprehensive disturbance value of the path, and the path with the smallest comprehensive disturbance value is selected as the scheduling path. Order scheduling is executed according to the scheduling path, and logistics information during the scheduling process is tracked in real time. When the tracked logistics information deviates, the historical operation interruption frequency data of the relevant nodes is updated according to the deviation, and the node status disturbance value is adjusted based on the updated data to optimize the scheduling of subsequent orders. Based on the historical operation interruption frequency and warehousing environment compliance data of each node, the node status disturbance value is calculated, including: the node status disturbance value is obtained by weighted summation of the node's historical operation interruption frequency and the degree of non-compliance of the warehousing environment; the historical operation interruption frequency is determined based on the daily average interruption frequency within a preset time period, and the degree of non-compliance of the warehousing environment is determined based on the compliance score; Based on the expiration dates of materials in the inventory batch information, the time-limited value of each inventory batch is calculated using an exponential decay function, including: the calculation of the time-limited value uses an exponential decay function, which is determined based on the initial value of the materials, the storage time, and the decay coefficient corresponding to the material category. A potential energy matrix is ​​generated by combining demand potential energy, timeliness value, and node status perturbation values. The potential energy matrix is ​​a two-dimensional table. The rows represent all available inventory batches and their respective nodes, and the columns represent all medical institution nodes that issued orders. It includes the elements in the potential energy matrix, namely the potential energy of scheduling inventory batch k from issuing node i to receiving node j. The calculation method is as follows: the node state perturbation value of the sending node i is... The node state perturbation value of receiving node j The product of the two, divided by the required potential energy of receiving node j. The time-sensitive value of inventory batch k The product; Based on the order demand from medical institutions, the demand potential energy is calculated using the S-shaped gain function, including: using the Logistic function; the expression for the Logistic function is... ,in is the potential energy demanded by receiving node j when the demand quantity is x, L is the preset maximum potential energy demand, representing the upper limit of the node's processing capacity, and k is the growth rate, which determines the steepness of the curve. It is an inflection point value, representing a quantity point that is sensitive to demand response.

2. The method according to claim 1, characterized in that, Based on the gradient of the optimized potential energy matrix, a set of candidate scheduling paths is generated for the orders to be scheduled, including: starting from the node where the inventory is located, using a greedy algorithm to iteratively select the neighboring node with the lowest potential energy until the target node is reached, generating the first candidate path; by disabling some nodes in the generated path and repeating the greedy algorithm, a preset number of subsequent candidate paths are generated.

3. The method according to claim 2, characterized in that, The historical operation interruption frequency data of relevant nodes is updated based on the deviation, including: when the logistics delay exceeds a preset threshold, it is determined as a deviation event; the responsible node of the deviation event is identified, and the historical operation interruption frequency of the node is updated using the exponential moving average method with a preset smoothing coefficient.

4. A multi-level collaborative intelligent order scheduling and tracking system for the medical supply chain, applied to the method described in claim 3, characterized in that, include: The acquisition module is used to acquire information on each node in the medical supply chain, inventory batch information, and order information to be scheduled. The module is used to construct a potential energy matrix representing the material scheduling tendency among nodes. The construction steps include: calculating the demand potential energy using an S-shaped gain function based on the order demand quantity of medical institutions; calculating the node state disturbance value based on the historical operation interruption frequency and warehousing environment compliance data of each node; calculating the time-sensitive value of each inventory batch using an exponential decay function based on the material validity period in the inventory batch information; generating the potential energy matrix by combining the demand potential energy, time-sensitive value, and node state disturbance value; and optimizing the potential energy matrix. The selection module is used to generate a set of candidate scheduling paths for the orders to be scheduled based on the gradient of the optimized potential energy matrix; the node state disturbance values ​​of each candidate scheduling path are accumulated to calculate the comprehensive disturbance value of the path, and the path with the smallest comprehensive disturbance value is selected as the scheduling path. The optimization module is used to execute order scheduling according to the scheduling path and track logistics information in real time during the scheduling process. When the tracked logistics information deviates, the module updates the historical operation interruption frequency data of the relevant nodes according to the deviation, and adjusts the node status disturbance value based on the updated data to optimize the scheduling of subsequent orders.

5. The system according to claim 4, characterized in that, Based on the order demand quantity of medical institutions, the demand potential energy is calculated using an S-shaped gain function, including: using a Logistic function, which is based on the order demand quantity and parameterized by preset maximum demand potential energy, growth rate, and inflection point value of demand quantity.

6. The system according to claim 5, characterized in that, Based on the historical operation interruption frequency and warehousing environment compliance data of each node, the node status disturbance value is calculated, including: the node status disturbance value is obtained by weighted summation of the node's historical operation interruption frequency and the degree of non-compliance of the warehousing environment; the historical operation interruption frequency is determined based on the daily average interruption frequency within a preset time period, and the degree of non-compliance of the warehousing environment is determined based on the compliance score.

Images