Supply chain management system and method based on operation data
Through quantum Bayesian deduction and multi-task learning model combined with distributed agent system, the problems of inaccurate prediction and uneven resource scheduling in the existing technology are solved, dynamic resource allocation and closed-loop optimization of the supply chain are realized, and the overall Efficiency and market response capabilities.
Patent Information
- Application Number
- CN202411283453.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-13
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing supply chain management technology is unable to capture market dynamic changes in a timely manner, resulting in inaccurate prediction results and lack of real-time response and dynamic optimization capabilities, resulting in uneven resource allocation and inefficient efficiency.
By collecting multi-dimensional operational data, using quantum Bayesian deduction and multi-task learning models to generate prediction data in the market scenario, and combining distributed agent systems and resource pooling technology, dynamic allocation and adaptive scheduling of supply chain resources are realized.
It improves the overall efficiency of the supply chain and market response capabilities, solves the problems of inaccurate predictions and uneven resource scheduling, and realizes cross-regional dynamic allocation and closed-loop optimization of supply chain resources.
Smart Images

Figure CN118917628A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of supply chain management, and in particular to a supply chain management system and method based on operation data. Background Art
[0002] Each link in the supply chain, such as raw material procurement, production, inventory management and logistics distribution, relies on a large amount of data. By analyzing these data in real time, the response speed and efficiency of the supply chain can be improved, and resource waste and inventory backlog can be reduced. However, the existing technology mainly relies on manual prediction and simple models based on historical data. For example, in the existing solution (Chinese invention patent, publication number: CN118333531B, name: A smart supply chain management method based on artificial intelligence), a commodity demand prediction model based on artificial intelligence is adopted. This method predicts procurement and inventory demand by analyzing historical sales data, location data and other auxiliary information. However, this technology has the following defects: since it cannot capture the dynamic changes of the market in time, the accuracy of the prediction results depends on historical data and cannot cope with the drastic fluctuations in market demand; in addition, the traditional method is mainly static decision-making, lacks the ability of real-time response and dynamic optimization, and easily leads to uneven resource allocation and inefficiency. Summary of the invention
[0003] In response to the many problems existing in the above-mentioned prior arts, the present invention provides a supply chain management system and method based on operational data. The present invention collects multi-dimensional operational data in the supply chain, uses quantum Bayesian deduction to generate forecast data under various market scenarios, and combines multi-task learning models to optimize the resources of supply chain nodes. Through a distributed intelligent system and resource pooling technology, dynamic allocation and adaptive scheduling of supply chain resources are achieved, effectively improving the overall efficiency and market response capabilities of the supply chain.
[0004] A supply chain management system based on operational data, comprising: The data collection module is used to collect raw operation data from the dimensions of market demand, consumer behavior, production, logistics and inventory through IoT devices and edge computing nodes, and clean, format and standardize the raw operation data to generate standardized operation data; A data deduction and prediction module is used to extract features based on standardized operational data and generate demand forecast data and production scheduling data through a multi-task learning model and quantum Bayesian deduction, wherein the quantum Bayesian deduction is based on market demand feature data, production efficiency feature data, and logistics status feature data for processing; Distributed collaborative decision-making module, which is used to make distributed autonomous decisions on procurement, production and logistics nodes based on demand forecast data and production scheduling data, using multi-agent systems and self-organizing criticality algorithms to generate procurement data, production scheduling data and logistics adjustment data; The resource allocation and management module is used to dynamically allocate production equipment, storage space and logistics resources in the supply chain based on procurement data, production scheduling data and logistics adjustment data, using resource pooling and virtualization, and adaptive dynamic scheduling algorithms to generate final resource scheduling data; The execution and feedback module is used to execute supply chain resource allocation based on the final resource scheduling data, monitor the execution status of each link in the supply chain through IoT sensors, generate execution feedback data, perform self-correction based on the execution feedback data, and generate corrected resource scheduling data to ensure closed-loop optimization of the supply chain.
[0005] Preferably, the multi-task learning model in the data deduction and prediction module simultaneously performs multi-task feature extraction on the market demand characteristic data, the production efficiency characteristic data and the logistics status characteristic data to generate shared feature representations of different tasks, so that the optimization processes of multiple tasks are related to each other.
[0006] Preferably, the quantum Bayesian deduction is deduced based on the Bayesian theorem, wherein the market demand characteristic data, the production efficiency characteristic data and the logistics state characteristic data are processed by quantum state superposition and quantum parallel computing to generate the demand forecast data under various market scenarios, and the calculation process of the quantum Bayesian deduction includes the following expression: in, is the posterior distribution, indicating that given data In the case of The probability distribution of is the likelihood function, which means that given the parameters When, data Probability of occurrence; is the prior distribution, which means that when the data is observed Before, the parameter The initial probability distribution of is a standardization constant, indicating that the data The total probability of is used to ensure the normalization of the posterior distribution.
[0007] Preferably, the prior distribution of the quantum Bayesian deduction is constructed based on historical supply chain data and external market intelligence, and multiple market scenarios are simultaneously deduced through the superposition of quantum states during the quantum computing process. The quantum Bayesian deduction results are used to provide multiple possible paths for market demand and production scheduling, and quantify the credibility of each path.
[0008] Preferably, the multi-agent system in the distributed collaborative decision-making module includes multiple independent supply chain nodes, each node makes autonomous decisions based on the local market demand characteristic data and the production status data, and the nodes communicate in real time through a distributed network and share decision parameters to achieve globally optimal supply chain scheduling and resource allocation.
[0009] Preferably, the self-organizing criticality algorithm monitors the resource consumption status of each node, and when the resource status of the node approaches a critical value, adjusts the priority and resource allocation strategy of the node.
[0010] Preferably, the distributed collaborative decision-making module performs global optimization on resource allocation among nodes based on the following calculation expression: in, For the The resource consumption function of nodes represents the node Use Resources the costs incurred; For the and The resource complementarity function between nodes represents the node and nodes In their respective resources and The benefits generated by resource synergy under the configuration; It represents the total number of nodes in the supply chain. By optimizing the resource allocation of each node, the resource consumption of the entire supply chain is minimized, while the resource complementarity effect between the nodes is maximized to generate the procurement data, the production scheduling data and the logistics adjustment data.
[0011] Preferably, the resource allocation and management module virtualizes all physical resources in the supply chain, including production equipment, storage space and logistics resources, through resource pooling technology, and uniformly allocates them through the virtual resource pool to achieve cross-regional scheduling of supply chain resources.
[0012] Preferably, the adaptive dynamic scheduling algorithm dynamically adjusts the resource allocation strategy according to the market demand data, the production efficiency data and the logistics status data acquired in real time, performs local resource optimization at each node in the supply chain through edge computing nodes, and generates final resource scheduling data.
[0013] A method for implementing the supply chain management system based on operational data comprises the following steps: Collect raw operational data from the dimensions of market demand, consumer behavior, production, logistics, and inventory through IoT devices and edge computing nodes, and clean, format, and standardize the raw operational data to generate standardized operational data; Extract features based on the standardized operation data, and generate demand forecast data and production scheduling data through a multi-task learning model and quantum Bayesian deduction, wherein the quantum Bayesian deduction is processed based on market demand feature data, production efficiency feature data, and logistics status feature data; Based on the demand forecast data and the production scheduling data, a multi-agent system and a self-organizing criticality algorithm are used to make distributed autonomous decisions on procurement, production and logistics nodes to generate procurement data, production scheduling data and logistics adjustment data; Based on the procurement data, the production scheduling data and the logistics adjustment data, the production equipment, storage space and logistics resources in the supply chain are dynamically allocated using resource pooling and virtualization and adaptive dynamic scheduling algorithms to generate final resource scheduling data; Supply chain resource allocation is performed based on the final resource scheduling data, the execution status of each link in the supply chain is monitored through the Internet of Things sensors, execution feedback data is generated, self-correction is performed based on the execution feedback data, and corrected resource scheduling data is generated to ensure closed-loop optimization of the supply chain.
[0014] Compared with the prior art, the advantages and beneficial effects of the present invention are: The present invention uses quantum Bayesian deduction technology to perform parallel deductions in multiple market scenarios and generate multiple demand forecast data, thus solving the problem of inaccurate forecasts caused by market fluctuations in the prior art. The present invention adopts a multi-task learning model to extract features and generate shared feature representations of different tasks, thus achieving deep correlation optimization of demand forecasting and production scheduling, and solving the problem of disconnection between forecasting and execution in traditional methods. The present invention uses a multi-agent system and a self-organizing criticality algorithm to make distributed autonomous decisions, ensuring optimal resource allocation for each node and solving the problem of uneven resource scheduling in existing solutions. The resource pooling and virtualization technical means of the present invention realizes the cross-regional dynamic allocation of supply chain resources, greatly improving the elasticity and response speed of the supply chain. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 It is a structural block diagram of the system of the present invention; Figure 2 It is a block diagram of the decision-making process in the distributed collaborative decision-making module of the present invention; Figure 3 It is a closed-loop optimization flow chart of the execution and feedback module in the present invention; Figure 4 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION
[0016] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0017] like Figure 1 As shown, a supply chain management system based on operational data includes: The data collection module is used to collect raw operation data from the dimensions of market demand, consumer behavior, production, logistics and inventory through IoT devices and edge computing nodes, and clean, format and standardize the raw operation data to generate standardized operation data; The data acquisition module realizes real-time data collection of each supply chain link through IoT devices and edge computing nodes. These IoT devices can include smart sensors, RFID tags, industrial cameras, etc., which are widely distributed in various nodes in the supply chain, such as production workshops, logistics centers, storage facilities and market ends. Through these devices, various data related to supply chain operations can be captured in real time. For example, market demand data can be collected through customer orders, sales terminal equipment, etc.; production data comes from the operating status and production capacity of production equipment; logistics data is provided by transport vehicles, cargo tracking equipment, etc. These raw data have not been processed in any way, and there are problems such as inconsistent data formats and noise interference. They must be pre-processed before entering the next step of analysis.
[0018] Edge computing nodes are an important part of the data collection module, and are particularly suitable for real-time requirements in supply chain management. Its core principle is to delegate the initial processing of data from the central server to the edge node close to the data source to reduce latency and bandwidth usage during data transmission. Edge computing nodes can clean, format and standardize the raw data nearby. The cleaning process is mainly used to filter out invalid data and noise to ensure data quality; the formatting process converts data from different sources into a unified format for subsequent processing and analysis; standardization is used to normalize the data so that data of different dimensions can be compared and processed on the same scale. For example, logistics data in different regions must be uniformly standardized due to different unit systems and expressions before they can be applied in the same supply chain decision-making process.
[0019] In this process, the data collection module combines the wide coverage of IoT devices and the real-time processing capabilities of edge computing to ensure that the massive data obtained from each supply chain link can be quickly and accurately processed into a standardized format, thereby supporting the subsequent data analysis and decision optimization of the supply chain management system. This architectural design not only reduces the load on the central server, but also improves the response speed of supply chain data processing.
[0020] Through the effective application of the data acquisition module, the present invention can greatly improve the accuracy and real-time performance of data in each link of the supply chain, and reduce decision-making errors caused by data delays or inconsistencies. For example, inventory data collected and processed in real time can reflect accurate information about the current inventory status, avoiding overproduction or undersupply. In addition, the standardized processing of data ensures that information from different data sources can be seamlessly integrated and used for subsequent decision-making analysis, improving the overall efficiency of supply chain management.
[0021] A specific implementation example can be a supply chain management system for a large retail enterprise. Through the IoT devices deployed in various storage centers, the system can collect the inventory status, inbound and outbound logistics data of each warehouse in real time. The edge computing node cleans and standardizes this data, eliminates the differences in data expression in different warehouse systems, and generates unified standardized inventory data. Based on this data, the system can quickly adjust the replenishment plan and optimize the allocation of logistics resources, reduce inventory backlogs and logistics costs, and achieve efficient management of the supply chain.
[0022] A data deduction and prediction module is used to extract features based on standardized operational data and generate demand forecast data and production scheduling data through a multi-task learning model and quantum Bayesian deduction, wherein the quantum Bayesian deduction is based on market demand feature data, production efficiency feature data, and logistics status feature data for processing; The data deduction and prediction module performs feature extraction based on standardized operational data. The so-called feature extraction is actually a deep processing of the original data, which simplifies the subsequent analysis process by extracting the core information with predictive value from the data. Standardized operational data usually contains multi-dimensional information such as market demand, consumer behavior, production efficiency, and logistics status. The purpose of feature extraction is to extract representative and operational features from this complex information to reduce data redundancy and improve the computational efficiency and accuracy of the model. For example, the features in consumer behavior data may include historical purchase frequency, single product preference, etc.; the features in logistics status data may be transportation time, warehousing pressure, etc.
[0023] After feature extraction is completed, the system uses a multi-task learning model to further deduce the data. The application of the multi-task learning model can effectively handle the complexity and correlation of multi-dimensional supply chain data. The basic principle of this model is to simultaneously process multiple related tasks in a shared learning framework, and multiple tasks share some features or parameters, thereby improving the learning effect of each task. For example, market demand forecasting and production scheduling are actually two closely related tasks. Changes in market demand directly affect the adjustment of production scheduling. Therefore, multi-task learning allows the model to optimize production plans while predicting demand. Specifically, the model uses a shared layer to unify the feature representation of market demand feature data, production efficiency feature data, and logistics status feature data, and realizes collaborative learning between different tasks. This approach not only improves the prediction accuracy, but also reduces the risk of overfitting of single-task models.
[0024] On this basis, quantum Bayesian deduction further enhances the accuracy and deduction efficiency of data prediction. The core of quantum Bayesian deduction is to introduce the state superposition and parallel computing characteristics in quantum computing, and to perform probabilistic deduction of data through Bayesian theorem. Bayesian theorem allows the system to infer future market changes based on existing prior knowledge (such as historical data or experience). The superposition of quantum states enables the system to process multiple possible market scenarios in parallel, greatly improving the computational efficiency of prediction. Specifically, quantum Bayesian deduction will construct different market change scenarios based on market demand characteristic data, production efficiency characteristic data, and logistics status characteristic data, and deduce and quantify the demand and production scheduling under these scenarios to generate demand forecast data and production scheduling data.
[0025] The data deduction and prediction module combines the advantages of multi-task learning models and quantum Bayesian deduction to achieve precise optimization of demand forecasting and production scheduling in the supply chain. Its effect is not only reflected in the prediction of market demand, but also in the effective scheduling of production resources to ensure the efficient operation of the supply chain. For example, when market demand fluctuates, the system can adjust the production plan in real time to avoid overproduction or insufficient supply, and improve the response speed and flexibility of the supply chain.
[0026] Example: For example, a global electronic product manufacturer can use this module to combine sales data from different markets around the world to extract the market demand characteristics of each region. At the same time, through multi-task learning models and quantum Bayesian deduction, the company can accurately predict market demand fluctuations in the next few months and adjust the production scheduling plans of each production plant based on these predicted data. Quantum Bayesian deduction can also effectively handle uncertainties between different markets, ensuring that production plans can respond in a timely manner when demand fluctuates drastically, reducing inventory backlogs or supply shortages.
[0027] Preferably, the multi-task learning model in the data deduction and prediction module simultaneously performs multi-task feature extraction on the market demand characteristic data, the production efficiency characteristic data and the logistics status characteristic data to generate shared feature representations of different tasks, so that the optimization processes of multiple tasks are related to each other.
[0028] The multi-task learning model is a machine learning framework that can utilize the connections between tasks when processing multiple related tasks, and improve the overall performance of the model by sharing some features or model parameters. In the present invention, the tasks in supply chain management include market demand forecasting, production scheduling optimization, and logistics status monitoring, and there is a natural correlation between these tasks. Fluctuations in market demand will directly affect production plans, and changes in production efficiency will affect logistics arrangements. Therefore, these tasks cannot be processed in isolation, but need to be collaboratively optimized under a unified framework.
[0029] The multi-task learning model generates shared feature representations by jointly extracting features from the data of these tasks. The so-called shared feature representation refers to high-level features shared between multiple tasks, that is, some features play an important role in different tasks. For example, trend features in market demand not only affect demand forecasts, but also have important significance for capacity scheduling in production plans. Similarly, delay features in logistics status data can not only affect production scheduling decisions, but also reflect bottleneck problems in logistics links. By putting the data of these related tasks into the same model for feature extraction, the model can learn the implicit associations between multiple tasks and optimize the representation of these shared features during the training process.
[0030] In the specific implementation, the market demand feature data, production efficiency feature data, and logistics status feature data are first preliminarily encoded through the input layer of the model to extract task-specific features. Then, in the shared layer, the multi-task learning model jointly extracts and shares the features of different tasks. This process allows the model to share some feature representations between different tasks, thereby improving the performance of another task while optimizing one task. For example, the optimization of market demand forecasting not only improves the accuracy of production scheduling, but also optimizes logistics status forecasting, making the allocation of logistics resources more accurate.
[0031] By sharing feature representations, multi-task learning models can achieve collaborative optimization between multiple tasks. The principle of this collaborative optimization is that in the process of optimizing a certain task, the model can use the data of other tasks to learn the relationship between tasks and improve the performance of all tasks as a whole through shared feature representations. Specifically, when the model is processing market demand forecasts, it not only learns the laws of demand changes, but also grasps the changing trends of logistics status and production efficiency through the shared layer, which provides a more accurate decision-making basis for production scheduling and logistics arrangements.
[0032] The application of multi-task learning models significantly improves the coordination and overall efficiency of the supply chain management system. By sharing features between tasks, the model can better capture the mutual influence of different links in the supply chain, ensuring that when one link is optimized, the performance of other links can be improved simultaneously. For example, when market demand changes suddenly, the system can quickly adjust production plans and logistics scheduling to avoid overproduction or logistics delays.
[0033] Let's take an example to illustrate: A global manufacturing company uses this multi-task learning model to optimize its supply chain. In the actual application of this company, fluctuations in market demand directly affect the production arrangements and logistics transportation plans of production bases around the world. Through the multi-task learning model, the system can optimize production scheduling and logistics arrangements while forecasting demand, ensuring that production capacity and logistics resources in different regions of the world can be adjusted synchronously. When market demand surges in a certain area, the system will automatically adjust the global production plan, concentrate production capacity in high-demand areas, and ensure that goods can be delivered to the area in a timely manner through optimized logistics solutions. This collaborative optimization not only enables companies to improve production and logistics efficiency, but also significantly reduce inventory costs and transportation delays.
[0034] Preferably, the quantum Bayesian deduction is deduced based on the Bayesian theorem, wherein the market demand characteristic data, the production efficiency characteristic data and the logistics state characteristic data are processed by quantum state superposition and quantum parallel computing to generate the demand forecast data under various market scenarios, and the calculation process of the quantum Bayesian deduction includes the following expression: in, is the posterior distribution, indicating that given data In the case of The probability distribution of is the likelihood function, which means that given the parameters When, data Probability of occurrence; is the prior distribution, which means that when the data is observed Before, the parameter The initial probability distribution of is a standardization constant, indicating that the data The total probability of is used to ensure the normalization of the posterior distribution.
[0035] Quantum Bayesian deduction is based on Bayesian theorem. Its core is to make probabilistic inferences about unknown parameters by calculating prior distribution, likelihood function and posterior distribution. For demand forecasting in supply chain management, parameters Usually represents some unknown parameters of the demand model, such as market demand growth rate or production capacity, data It is the market demand characteristic data, production efficiency characteristic data and logistics status characteristic data collected from each supply chain node.
[0036] In the present invention, quantum Bayesian deduction accelerates and expands this deduction process through quantum state superposition and quantum parallel computing. Quantum state superposition allows the system to process multiple possible market scenarios at the same time, while quantum parallel computing can perform parallel calculations on multiple possible parameter combinations without increasing computational complexity. This enables the system to generate demand forecast data for multiple market scenarios in a short period of time when faced with a large amount of complex supply chain data, thereby improving the flexibility and responsiveness of the supply chain system.
[0037] In practical applications, quantum Bayesian deduction can generate multiple possible future demand scenarios by inputting market demand characteristic data and combining the changing trends of production efficiency and logistics status. The superposition characteristics of quantum states allow different scenarios to be evaluated in the same time frame. In each scenario, the system simulates the demand changes and calculates its posterior probability, thereby providing a basis for subsequent supply chain scheduling decisions.
[0038] For example, in a global supply chain, quantum Bayesian deduction can generate multiple market demand scenarios for the next few months based on past sales data, current market trends, and supply chain status. For each scenario, the system predicts possible demand changes and calculates the probability of their occurrence, helping companies make optimal production and logistics scheduling decisions when facing demand fluctuations.
[0039] Through quantum Bayesian deduction, supply chain management systems can more effectively handle complex uncertainties and dynamic changes. This method not only improves the accuracy of market demand forecasts, but also greatly improves the computational efficiency of forecasts. In traditional Bayesian deduction, the computational cost of processing large-scale, multi-dimensional data is extremely high, but through the parallel processing capabilities of quantum computing, the present invention significantly reduces the computational time of real-time demand forecasting in supply chain management.
[0040] Quantum Bayesian deduction enables the supply chain management system to flexibly respond to different scenarios of market demand changes and provide a clear probabilistic basis for decision-making. For example, the system can adjust production scheduling and inventory strategies in real time based on current production efficiency and logistics status. If it is predicted that a certain market demand has a high probability of increasing significantly, the system can adjust the production plan in advance and optimize logistics arrangements to avoid a situation where supply exceeds demand. For low-probability market scenarios, the system can also reserve corresponding emergency resources to ensure the resilience of the supply chain.
[0041] Example: In a high-tech manufacturing enterprise, quantum Bayesian deduction is used to handle the task of predicting global market demand. The enterprise obtains real-time market demand data through a global sales network, combined with production efficiency data of production plants and status data of global logistics networks. Based on these data, quantum Bayesian deduction generates multiple future market demand scenarios in a short period of time. For example, for new products to be released, the system can predict changes in demand in different global markets in the next few months, and adjust the capacity allocation of each production base based on the predicted posterior probability. In this way, enterprises can not only respond to market demand in a timely manner, but also optimize logistics arrangements to ensure that products can be delivered on time.
[0042] Preferably, the prior distribution of the quantum Bayesian deduction is constructed based on historical supply chain data and external market intelligence, and multiple market scenarios are simultaneously deduced through the superposition of quantum states during the quantum computing process. The quantum Bayesian deduction results are used to provide multiple possible paths for market demand and production scheduling, and quantify the credibility of each path.
[0043] In supply chain management, prior distributions are used to represent the system's initial estimates of future market demand and production changes. These estimates usually come from two main sources: historical supply chain data and external market intelligence. Historical supply chain data includes past sales records, production efficiency, logistics status, etc., which can provide a solid foundation for forecasting. For example, historical demand trends can provide the system with preliminary expectations about demand fluctuations, while production efficiency data can help estimate the reasonable range of production scheduling. External market intelligence comes from market research, competitor information, industry dynamics, etc., which helps to make corresponding judgments on sudden changes in the market. For example, when external market intelligence shows that a region's economy is recovering or new competitors are emerging, the system can adapt to these changes by adjusting the prior distribution.
[0044] Quantum Bayesian deduction uses the quantum state superposition characteristics of quantum computing to process multiple different market scenarios simultaneously during the deduction process. Unlike traditional computing, the quantum state superposition in quantum computing allows multiple states to be calculated in parallel, that is, the system can perform probabilistic deductions on multiple possible market scenarios within the same time frame. Specifically, each market scenario represents the evolution path of market demand and production capacity under different scenarios. Through quantum state superposition, the system can quickly deduce future demand and production scheduling data under multiple market scenarios.
[0045] After the deduction is completed, the results of quantum Bayesian deduction are presented in the form of multiple possible paths. Each path represents a scenario that may occur in the supply chain in the future, including changes in market demand, production scheduling adjustments, inventory management optimization, etc. These paths not only describe the different market demand fluctuations that the supply chain may face, but also include how the system adjusts production and logistics scheduling in different scenarios to adapt to these changes. In addition, quantum Bayesian deduction quantifies each path and calculates its probability or credibility. The calculation of credibility is based on historical data and market intelligence, which provides a clear basis for risk assessment and decision-making for the supply chain management system.
[0046] The prior distribution and quantitative path calculation of quantum Bayesian deduction have extremely high practical value in supply chain management. First, the prior distribution constructed based on historical supply chain data and external market intelligence can help the system grasp future trends more accurately. In the supply chain, demand fluctuations and uncertainty in production capacity often lead to low supply chain efficiency and even cause supply disruptions or excess inventory. By combining historical data and external information, the system can provide reasonable estimates for each possible market scenario, which provides a stable foundation for supply chain decision-making.
[0047] Secondly, quantum state superposition and parallel computing significantly improve the efficiency and accuracy of deduction. Compared with traditional Bayesian deduction, quantum Bayesian deduction can not only deduce multiple scenarios at the same time, but also complete the processing of large amounts of data in a short time. This means that the system can still maintain efficient responsiveness when facing complex supply chain networks and multi-dimensional data. Especially when the market changes rapidly, the system can quickly deduce a variety of possible future scenarios and provide companies with timely decision-making references.
[0048] To give a specific example: a large retail company is planning its global supply chain layout for the next few months. Through quantum Bayesian deduction, the company can construct multiple possible paths for market demand based on past sales data and external economic reports. These paths include market demand growth forecasts in different regions of the world, possible changes in production capacity, and fluctuations in logistics status. Each path is assigned a credibility value, indicating the possibility of the path. Based on these forecast data, the company can prioritize production scheduling arrangements for high-credibility paths and reserve emergency resources for low-credibility but high-risk paths. Through the deduction of this quantitative path, companies can not only effectively avoid inventory backlogs or supply disruptions, but also quickly adjust their supply chain strategies to adapt to rapid changes in the market.
[0049] Distributed collaborative decision-making module, which is used to make distributed autonomous decisions on procurement, production and logistics nodes based on demand forecast data and production scheduling data, using multi-agent systems and self-organizing criticality algorithms to generate procurement data, production scheduling data and logistics adjustment data; The distributed collaborative decision-making module uses a multi-agent system to manage different nodes in the supply chain. A multi-agent system (MAS) is a distributed computing framework that consists of multiple independent but collaborative agents, each of which represents an independent decision-making subject. In the context of supply chain management, each agent can be regarded as a node in the supply chain, such as a procurement node, a production node, or a logistics node. Each node has its own decision-making power and can make decisions autonomously based on local data. This autonomy gives the system a high degree of flexibility in dealing with the complexity and uncertainty in the supply chain. For example, a production node can autonomously schedule production based on demand forecast data and its own production capacity without waiting for instructions from the central system.
[0050] These agents communicate in real time through a distributed network and share necessary decision-making data and information. Since each agent has independent computing and decision-making capabilities, the entire system no longer relies on a central decision point, but achieves global optimization through the collaboration of various nodes. This distributed feature makes the system highly robust. Even if a node fails, other nodes can still continue to perform their tasks, thereby maintaining the overall operation of the supply chain.
[0051] On this basis, the self-organizing criticality algorithm was introduced to cope with the dynamics and complexity of resource allocation and load changes in the supply chain. The basic principle of the self-organizing criticality algorithm is to monitor the resource status of each node in the system. When the resource status of the node approaches the critical point (such as overload or resource exhaustion), the system will automatically adjust the priority and resource allocation strategy of the node. This mechanism can effectively prevent excessive consumption or waste of resources in the supply chain. For example, when the production capacity of a production node approaches its limit, the system can automatically reduce its task priority and reallocate more tasks to other nodes. This resource allocation mechanism based on the principle of self-organizing criticality ensures the continuous operation of each node in the supply chain and avoids system paralysis caused by excessive load on individual nodes.
[0052] During the implementation process, the distributed collaborative decision-making module will automatically generate adjustment strategies for procurement, production, and logistics based on demand forecast data and production scheduling data. For example, in the case of a surge in demand, the system will prioritize production tasks to nodes with sufficient production capacity based on the capacity status of each production node, and adjust procurement and logistics plans accordingly to ensure that goods can be supplied in a timely manner. In the case of a decrease in demand, the system will reduce production scheduling, adjust inventory, and optimize the utilization of logistics resources.
[0053] The distributed collaborative decision-making module greatly improves the automation level and response speed of supply chain management through the combination of multi-agent system and self-organizing criticality algorithm. First, through the distributed decision-making of the multi-agent system, each node in the supply chain can optimize its decision autonomously and no longer rely on the instructions of the central system. This distributed decision-making architecture provides a high degree of flexibility and autonomy for the supply chain. For example, when a logistics node causes a delay in cargo transportation due to traffic congestion, the system will automatically adjust the scheduling plans of other logistics nodes to ensure that the overall logistics efficiency is not affected. This autonomous scheduling capability improves the overall flexibility of the supply chain, especially in a complex and changing market environment, and can quickly adapt to changes in demand and supply fluctuations.
[0054] Secondly, the self-organizing criticality algorithm ensures that each node in the supply chain can maintain balanced operation when the load changes. By monitoring the resource usage of the nodes, the system can identify the critical point of resource usage in advance and automatically adjust the resource allocation strategy. This dynamic resource allocation method effectively prevents certain nodes in the supply chain system from collapsing due to overload, thereby ensuring the continuous and stable operation of the supply chain. Especially in the face of supply chain emergencies, such as a sudden increase in market demand or logistics interruptions, the system can respond quickly and make corresponding adjustments to avoid systemic collapse.
[0055] Example: In a global consumer goods manufacturer, a distributed collaborative decision-making module is used to manage its global supply chain network. The company has production bases, storage centers, and logistics networks in many countries around the world. Through the distributed collaborative decision-making module, each production base acts as an intelligent entity and can make autonomous decisions based on local production capacity and market demand. Assuming that the market demand in a certain country suddenly increases, and the production base in the region has limited production capacity, the system will automatically assign part of the production tasks to other production bases with sufficient production capacity through a self-organizing criticality algorithm. At the same time, the system will adjust the procurement and logistics strategies accordingly to ensure that the raw materials required for production can be delivered in time. In this way, companies can respond quickly to changes in demand and avoid supply shortages or logistics bottlenecks.
[0056] Preferably, Figure 2 As shown, the multi-agent system in the distributed collaborative decision-making module includes multiple independent supply chain nodes. Each node makes autonomous decisions based on the local market demand characteristic data and the production status data. The nodes communicate in real time through a distributed network and share decision parameters to achieve globally optimal supply chain scheduling and resource allocation.
[0057] A multi-agent system (MAS) is a distributed system architecture in which each "agent" represents an independent entity or node with independent perception, decision-making and execution capabilities. The autonomy of the agent is a key feature of the multi-agent system, that is, each node can make decisions independently based on its own goals and the environment in which it is located. In the present invention, each supply chain node (such as a production base, logistics center, storage facility, etc.) can be regarded as an agent. Through real-time analysis of local market demand feature data and production status data, the node can independently determine the production, procurement and logistics strategies without waiting for instructions from the central decision-making system.
[0058] The processing of local data is the basis for each node to make autonomous decisions. For example, a production node can make adjustments based on current inventory levels, changes in market demand, and production capacity. When production demand increases, the node may increase production, or when production capacity reaches its upper limit, prioritize tasks to other nodes with larger production capacity. This autonomous scheduling method based on local data enables each node to respond quickly to local changes, improving the overall response speed and decision-making efficiency of the system.
[0059] However, the nodes in the supply chain do not exist independently, and their decisions will affect each other. To this end, the nodes in the multi-agent system communicate in real time through a distributed network and share key decision parameters. This distributed communication mechanism ensures that each node can obtain the status information of other nodes in a timely manner (such as current production capacity, logistics delays, changes in market demand, etc.), so that when making local decisions, it can take into account the global demand and resource allocation. Through this information sharing, the system can avoid suboptimal decisions made by individual nodes due to information asymmetry.
[0060] In order to achieve the global optimal supply chain scheduling, the multi-agent system forms a collaborative optimization decision-making mechanism through shared decision parameters, such as current market demand, inventory status, logistics bottlenecks, etc. This means that each node takes into account the global resource allocation of the entire supply chain while optimizing its own goals. For example, if a production node predicts that a logistics node may be delayed, it will autonomously reduce production tasks to avoid inventory backlogs or waste of resources. Similarly, the logistics node will adjust the transportation plan according to the production capacity of the production node to ensure maximum transportation efficiency. This collaborative optimization mechanism achieves global coordination and efficient operation of the supply chain through a distributed network.
[0061] Through the multi-agent system, the supply chain management system can achieve a high degree of autonomy and collaborative optimization among various nodes. This autonomy enables each node to make real-time decisions based on local demand and production status, reducing dependence on the central system. At the same time, real-time communication in the distributed network ensures that each node can synchronously obtain global information, giving the system global scheduling capabilities.
[0062] The specific effects are reflected in two aspects: First, through local decision-making capabilities, the system improves the response speed of the supply chain. In a market environment where demand changes rapidly, supply chain nodes can quickly make adjustments based on real-time data, avoiding the command transmission and response delays of traditional centralized systems. Second, through global information sharing and collaborative optimization, the system can avoid the global suboptimal phenomenon caused by local decision-making. For example, when a production node increases its production capacity, the system will simultaneously adjust the transportation capacity of the logistics node to ensure efficient allocation of resources.
[0063] Let's take an example of actual implementation: a multinational automobile manufacturer manages its global supply chain network through a multi-agent system. Each production plant, logistics center, and warehousing node acts as an agent, independently analyzing the market demand and production capacity of the area it is responsible for. Assuming that the demand for automobiles in the European market surges in a short period of time, and the production base in the region cannot meet the demand due to limited production capacity, the system will notify production bases in other regions (such as Asia or North America) in real time through a distributed network. Other production bases decide whether to increase production to meet the needs of the European market based on their own production status and inventory levels. At the same time, the logistics nodes will also adjust the transportation plan according to the scheduling of the production base to ensure that the products can be delivered to the European market in a timely manner. The entire process does not need to rely on the instructions of the central system. Each node makes independent decisions, which improves the responsiveness and flexibility of the supply chain.
[0064] In this way, enterprises can achieve global production optimization and resource allocation, ensuring that the supply chain has the ability to respond quickly to changes in the global market. The multi-agent system greatly improves the overall efficiency and robustness of the supply chain by combining local decision-making with global optimization.
[0065] Preferably, the self-organizing criticality algorithm monitors the resource consumption status of each node, and when the resource status of the node approaches a critical value, adjusts the priority and resource allocation strategy of the node.
[0066] The Self-Organized Criticality (SOC) algorithm is an optimization algorithm for dynamically balancing system resources. The basic principle of this algorithm is to continuously monitor the resource consumption of each node in the system, and automatically make adjustments when the resource status of the node approaches the critical value (such as overload or resource exhaustion), to ensure that the system can still run smoothly when it approaches the critical state. Specifically in supply chain management, the node resource status includes production capacity, logistics capacity, inventory level, etc., and the critical value is the maximum load or minimum safety level of these resources.
[0067] In daily operations, each node in the supply chain is constantly consuming or replenishing resources. For example, production nodes consume raw materials, logistics nodes allocate transportation vehicles, and storage nodes manage inventory. When the resource consumption of a node reaches a certain critical point, such as the production capacity of the production base is close to the limit, the number of vehicles at the logistics node is insufficient, or the inventory of the storage center is close to the minimum safe stock, the system will identify this critical state through the self-organizing criticality algorithm. The system analyzes the current load of the node and its future resource requirements to determine whether the node is close to the critical state, and adjusts the resource allocation strategy when necessary, or prioritizes reducing the task load of the node.
[0068] An important feature of the self-organizing criticality algorithm is its adaptability. The system is able to autonomously adjust the resource allocation strategy without external instructions or human intervention. The core of this adjustment mechanism lies in the dynamic management of priorities. Specifically, when the resources of a node are about to reach a critical value, the system will lower the priority of the node, reduce its task load in the supply chain, and allocate more resources and tasks to other nodes that have not reached the critical state. For example, when the production capacity of a production node is saturated, the system will reduce the production tasks of the node and allocate new production tasks to other nodes with more sufficient production capacity. Correspondingly, the resources of the logistics nodes will also be dynamically adjusted to ensure the global optimal allocation of resources.
[0069] Through such dynamic allocation, the system can achieve global balance of supply chain resources and avoid the situation where some nodes are overloaded with resources and cause system performance to decline. At the same time, the self-organizing criticality algorithm can also predict the changing trend of nodes when they are about to reach a critical state, and take countermeasures in advance to avoid sudden resource shortages or overload problems.
[0070] The self-organizing criticality algorithm provides the supply chain management system with real-time dynamic resource allocation capabilities, ensuring that the system maintains stable operation in a high-load or rapidly changing environment. The algorithm can significantly reduce the problems caused by resource imbalances at supply chain nodes through dynamic adjustments to priorities and resource allocation. For example, in the event of a sudden increase in demand, the system can identify production nodes that are about to reach their capacity limits and reduce the task load of the node in advance to avoid capacity bottlenecks caused by overproduction. Similarly, when the inventory of a storage node is about to reach a critical value, the algorithm will be dynamically adjusted to replenish inventory or reallocate storage resources in a timely manner.
[0071] The effect of this algorithm not only improves the overall efficiency of the supply chain, but also increases the robustness of the supply chain system. Even in the face of emergencies, the system can quickly make adjustments. For example, when market demand increases sharply and a production node is operating at a high intensity, the system will automatically transfer part of the production tasks to other nodes with unsaturated production capacity, thereby effectively avoiding supply chain disruptions. At the same time, the system can also ensure that preventive measures are taken when resources are close to exhaustion to maintain the sustainable operation of the overall supply chain.
[0072] Example: A global electronics manufacturer uses the self-organizing criticality algorithm to manage resources in its global production and logistics network. In a certain market area, sudden seasonal demand caused the production base in the area to quickly approach its capacity limit, and the inventory at the storage center was about to run out. The self-organizing criticality algorithm identified this situation through real-time monitoring, and the system automatically reduced the task load of the production base and transferred some production tasks to production bases in other regions. At the same time, the inventory allocation strategy of the storage node was also adjusted accordingly, and products were allocated from other regions to the market, ensuring that production and logistics resources were reasonably allocated while meeting market demand.
[0073] In this embodiment, the self-organizing criticality algorithm not only avoids the capacity bottleneck caused by the overload of resources at a single node, but also ensures the global optimal allocation of supply chain resources and improves the flexibility of the system in the face of demand fluctuations.
[0074] Preferably, the distributed collaborative decision-making module performs global optimization on resource allocation among nodes based on the following calculation expression: in, For the The resource consumption function of nodes represents the node Use Resources the costs incurred; For the and The resource complementarity function between nodes represents the node and nodes In their respective resources and The benefits generated by resource synergy under the configuration; It represents the total number of nodes in the supply chain. By optimizing the resource allocation of each node, the resource consumption of the entire supply chain is minimized, while the resource complementarity effect between the nodes is maximized to generate the procurement data, the production scheduling data and the logistics adjustment data.
[0075] The first part of the expression Indicates the total resource consumption of each node. The resource consumption of express, Is a node The amount of resources used. For production nodes in the supply chain, the resource consumption function can be expressed as the cost of using raw materials, energy, time, or other production resources required in the production process. For example, the resource consumption function of a production plant can include the energy and raw material costs required to produce a product. The optimization goal of this part is to reduce the resource consumption of a single node as much as possible while meeting demand, thereby reducing the operating cost of the overall supply chain.
[0076] Part 2 Represents the resource complementarity effect between nodes. Resource complementarity function Used to quantify the benefits of two nodes when they share resources or work together. For example, a production node and logistics nodes The collaboration between nodes can optimize resource usage by coordinating production and transportation time, reduce logistics waiting time, or reduce inventory pressure by sharing inventory information. The role of this function is to encourage different nodes to form synergy in the resource allocation process, thereby further reducing the cost of the overall supply chain or improving the operating efficiency of the system.
[0077] Through the comprehensive optimization of these two functions, the supply chain management system can achieve global resource allocation for procurement, production and logistics. The goal of optimization is to ensure that each node minimizes costs when using resources, while at the same time, through information sharing and collaboration between nodes, more efficient resource utilization and task allocation are formed. The final result is to generate specific procurement data, production scheduling data and logistics adjustment data to guide the actual operation of the supply chain.
[0078] The main effect of this global optimization method is to improve the overall efficiency of supply chain management and reduce operating costs. Through accurate resource consumption calculation and resource coordination between nodes, the system can effectively deal with complex supply chain scheduling problems. In a distributed supply chain network, the operating conditions of each node (such as production progress, logistics status, inventory level) are not completely independent. The resource consumption or status change of any node will affect the operation of other nodes. Through the global optimization model, the system can maximize the complementary effect of resources, so that the supply chain system can quickly adjust when facing changes to ensure the stable operation of the overall system.
[0079] Let's take a practical example: In the supply chain of a global electronic product manufacturer, the enterprise coordinates resource allocation among multiple production bases, logistics centers, and warehousing nodes around the world. The enterprise optimizes production and logistics tasks through this distributed collaborative decision-making module. For example, if the production efficiency of a certain production base decreases, resulting in inventory backlogs at the local logistics center, the system can identify possible resource redundancies in production nodes in other regions through the optimization model and transfer tasks to these production nodes in a timely manner. At the same time, the resource complementarity function of the logistics center can calculate how logistics nodes in other regions can work together effectively to reduce inventory pressure and transportation costs by adjusting inventory allocation or transportation routes. Through this dual optimization of resource consumption and synergy, the entire supply chain ensures efficient use of resources worldwide and reduces enterprise operating costs.
[0080] Through this global optimization algorithm, the system can automatically calculate the resource consumption of each node, monitor the collaboration efficiency between nodes, and make corresponding adjustments based on real-time changes. The result is significant cost savings and improved responsiveness and flexibility of the entire supply chain.
[0081] The resource allocation and management module is used to dynamically allocate production equipment, storage space and logistics resources in the supply chain based on procurement data, production scheduling data and logistics adjustment data, using resource pooling and virtualization, and adaptive dynamic scheduling algorithms to generate final resource scheduling data; The operating principle of the resource allocation and management module is based on two core concepts: resource pooling and virtualization, and adaptive dynamic scheduling algorithm.
[0082] Resource pooling and virtualization: Resource pooling refers to integrating various scattered resources in the supply chain (such as production equipment, storage space and logistics resources) into a unified resource pool for centralized management. In traditional supply chains, production equipment, storage space and logistics resources are often distributed in different regions, departments or nodes, and it is difficult to coordinate resources between them. Resource pooling virtualizes these physical resources to break the limitations of geographical location and organizational structure and achieve unified resource scheduling. Virtualization uses software technology to map physical resources into logical resources, so that each supply chain node can flexibly call global resources when demand changes. For example, when the equipment utilization rate of a production base is low, the system can virtualize its idle resources and allocate them to nodes that need more equipment. This virtualization allows all resources in the supply chain to be dynamically allocated between different nodes according to actual needs, maximizing resource utilization efficiency.
[0083] Adaptive dynamic scheduling algorithm: The role of the adaptive dynamic scheduling algorithm is to intelligently schedule various resources in the resource pool based on real-time procurement data, production scheduling data, and logistics adjustment data. In the supply chain, the frequency of use of production equipment, the occupancy of storage space, and the scheduling requirements of logistics resources are dynamically changing. The adaptive dynamic scheduling algorithm can monitor these changes in real time and automatically adjust resource allocation. The algorithm dynamically allocates resources based on the current status of each node, demand forecast, and logistics conditions. For example, when the production equipment of a certain node faces high load, the algorithm will automatically allocate additional production equipment resources to the node, or adjust the equipment utilization of other nodes to ensure that there is no bottleneck in resources. In addition, adaptive scheduling can also reallocate transportation vehicles and storage space according to delays or changes in logistics to ensure the continuous operation of the logistics chain.
[0084] The adaptability of this algorithm lies in its ability to iteratively update based on real-time data and self-correct based on feedback within the system. This means that the system can not only allocate resources based on the current situation, but also make corresponding adjustments in advance when predicting possible load changes in certain links in the future. For example, when predicting that a certain logistics node may be delayed, the system can adjust the production plan or warehousing resources in advance to reduce the negative impact on the overall supply chain.
[0085] The resource allocation and management module achieves efficient allocation and dynamic optimization of supply chain resources through resource pooling, virtualization and adaptive scheduling. Its effects are reflected in the following aspects: Maximizing resource utilization: By pooling and managing dispersed resources, each node in the supply chain can flexibly call on global resources to avoid resource waste or local resource overload. Virtualization technology breaks the physical limitations of resources, allowing production equipment, storage space, and logistics resources to be deployed across nodes, thereby maximizing resource utilization. For example, when a warehouse is fully loaded, the system can virtualize the free space of other storage nodes and allocate it to the fully loaded node to avoid inventory backlogs or logistics stagnation.
[0086] Improved flexibility and responsiveness of the supply chain: Adaptive dynamic scheduling algorithms can respond to changes in demand at each link in the supply chain in real time, quickly adjust resources, and ensure the smooth operation of the supply chain. When demand fluctuates or emergencies occur in the supply chain (such as logistics delays or production equipment failures), the system can quickly reallocate resources through algorithms to avoid supply chain disruptions or reduced efficiency. For example, when a production line stops working due to equipment failure, the system can immediately allocate other backup equipment or reallocate production tasks to other nodes with unsaturated production capacity.
[0087] Global optimization and cost control: Through global optimization analysis of procurement data, production scheduling data, and logistics adjustment data, the system can effectively control the operating costs of the supply chain. For example, it can reduce the idle time and transportation costs of logistics resources through intelligent scheduling, or reduce storage costs by reasonably allocating storage space. The adaptive dynamic scheduling algorithm can continuously adjust the scheduling strategy according to the actual operation data to ensure that the supply chain can save resource costs while maximizing efficiency.
[0088] Example: Take a global consumer electronics company as an example. The company has production bases, storage centers and logistics networks in many regions around the world. Through the resource allocation and management module, the company has achieved unified management and dynamic allocation of production equipment, storage space and logistics resources. In a certain quarter, the demand in the North American market suddenly surged. The system quickly allocated more production equipment and logistics resources to the North American production base through an adaptive dynamic scheduling algorithm to meet order demand. At the same time, due to insufficient storage space, the system allocated part of the idle storage space in Europe to North America through virtualization technology, avoiding supply chain disruptions caused by insufficient storage.
[0089] This dynamic allocation method not only improves the overall efficiency of the enterprise supply chain, but also reduces the risk of idle and overloaded resources at each node. When demand decreases, the system will automatically recycle excess resources to avoid waste of resources. Through global scheduling and optimization, enterprises can not only effectively respond to fluctuations in global market demand, but also reduce operating costs.
[0090] Preferably, the resource allocation and management module virtualizes all physical resources in the supply chain, including production equipment, storage space and logistics resources, through resource pooling technology, and uniformly allocates them through the virtual resource pool to achieve cross-regional scheduling of supply chain resources.
[0091] Resource pooling technology is a process of abstracting physical resources in the supply chain into virtual resources. Its core idea is to integrate the originally physically dispersed production equipment, storage space and logistics resources into a virtual resource pool, and manage and allocate them through the virtualization layer. In this way, the physical location and actual ownership of resources become unimportant, and each node in the supply chain can flexibly call these virtualized resources driven by system requirements without being restricted by physical conditions.
[0092] Production equipment virtualization: Production equipment virtualization is to abstract the equipment of all production nodes in the supply chain so that the equipment resources of each node can be uniformly scheduled across geographical boundaries. After virtualization, the equipment resources of each production base can be flexibly allocated according to demand. For example, when the equipment of a production base is overloaded, the system can virtualize the underutilized production equipment in other regions and deploy it to the node to alleviate the equipment shortage problem. In this way, the utilization rate of production equipment is maximized, while reducing the production capacity bottleneck caused by uneven equipment distribution.
[0093] Storage space virtualization: Storage space virtualization integrates the storage resources of each storage node to achieve cross-regional inventory allocation and optimization. When the inventory of a warehouse is close to saturation, the system can integrate the free space of other storage nodes through virtualization technology, so that the inventory space in the entire supply chain can be dynamically allocated according to demand. For example, when the storage center in a region is close to saturation, but there is free storage space in other regions, the system can automatically reallocate the inventory to the idle storage nodes, thereby avoiding inventory overflow or waste of warehouse resources.
[0094] Logistics resource virtualization: Logistics resource virtualization is the abstract management of logistics resources such as transportation vehicles, transportation routes, and transportation time windows. This virtualization enables the system to dynamically adjust the allocation of transportation resources according to real-time logistics needs. For example, when a transportation route is delayed, the system can dispatch spare vehicles or alternative routes from the virtual resource pool to ensure the continuity of the logistics chain. In this way, the use of logistics resources becomes more flexible, reducing the risk of logistics bottlenecks and transportation delays.
[0095] An important feature of resource pooling technology is that it enables cross-regional scheduling in the supply chain. Unlike traditional supply chains, where resource allocation is usually limited to nodes within a certain region, resource pooling technology integrates all physical resources into a cross-regional resource pool through virtualization. In this way, when production, warehousing or logistics resources are in short supply in a certain region, the system can mobilize resources from other regions to form a broader global optimization. For example, when logistics nodes in Asia face high demand, the system can mobilize logistics resources from Europe or America to ensure the stable operation of the supply chain.
[0096] The application of resource pooling technology in supply chain management has significant effects, especially when dealing with resource imbalance, local overload or demand fluctuations in the supply chain: Improve resource utilization: Resource pooling allows each resource in the supply chain to be reallocated according to actual needs, avoiding idle or overloaded resources. Whether it is production equipment, storage space or logistics resources, global scheduling can be achieved to maximize resource utilization efficiency. For example, when the equipment utilization rate of a production base is low, the virtualized resources of the equipment can be allocated to other production bases that need to expand production capacity, thereby improving overall equipment utilization.
[0097] Improve the flexibility and responsiveness of the supply chain: Through the dynamic scheduling of the virtual resource pool, various resources in the supply chain can quickly adapt to changes in market demand. Whether it is a sudden increase in orders or a bottleneck in logistics nodes, the system can quickly respond through resource allocation to ensure the efficient operation of the supply chain. For example, when the inventory of a warehouse is close to the limit, the system can quickly call other storage resources for cross-regional allocation, thereby avoiding the problem of excess or insufficient inventory.
[0098] Global optimization and cost control: By uniformly allocating various resources, the system can achieve the global optimal resource allocation and reduce operating costs. Especially in the global supply chain network, resource pooling technology can effectively solve the problem of regional resource imbalance and avoid additional costs caused by uneven resource distribution. For example, cross-regional scheduling of logistics resources can avoid empty driving costs and reduce the cost increase caused by excessive concentration or shortage of logistics resources.
[0099] Example: A global automobile manufacturer applies the resource allocation and management module to manage its global production, warehousing and logistics resources through resource pooling technology. When the demand for automobiles in the European market surges, the system calls the production equipment in North America and Asia through the resource pool, and adjusts the allocation of warehousing and logistics resources to cope with changes in market demand. Production equipment is virtually allocated from the Americas and Asia to production nodes in Europe, meeting production capacity needs in a short period of time; at the same time, inventory pressure is dispersed through cross-regional warehousing allocation, and logistics resources are also mobilized from low-demand areas to high-demand areas. In this way, the company avoids production bottlenecks and logistics delays caused by local resource shortages and maintains the stability of the global supply chain.
[0100] Preferably, the adaptive dynamic scheduling algorithm dynamically adjusts the resource allocation strategy according to the market demand data, the production efficiency data and the logistics status data acquired in real time, performs local resource optimization at each node in the supply chain through edge computing nodes, and generates final resource scheduling data.
[0101] The basic principle of the adaptive dynamic scheduling algorithm is to make resource optimization decisions based on real-time data. Each node in the supply chain, including production, warehousing and logistics nodes, obtains and processes their respective operational data in real time through IoT sensors and edge computing devices. Real-time data includes market demand data (such as order volume and sales trends), production efficiency data (such as capacity utilization, equipment operating status) and logistics status data (such as delays in transportation routes and inventory status). These data can accurately reflect the current status and future needs of each node in the supply chain, thereby providing a data basis for scheduling decisions.
[0102] Edge computing is a key technology in this invention, which can realize localized data processing and optimization decisions at each node of the supply chain. Edge computing avoids the delay problem of transmitting large amounts of data to the central server by performing calculations at edge nodes close to the data source, and can respond quickly based on local conditions. For example, when a production node obtains real-time market demand data, the edge computing node will immediately make production scheduling optimization decisions, adjust production plans and resource allocations based on current production efficiency and inventory status. This local optimization method significantly improves the responsiveness and adaptability of the supply chain.
[0103] The adaptive dynamic scheduling algorithm can dynamically adjust the resources in the supply chain by analyzing the data of each node in real time. For example, when the system detects an increase in market demand and the production efficiency of a production node reaches a certain load, the algorithm will optimize the resource allocation strategy of the node in real time through the edge computing node, adjust the production plan or reallocate production equipment. Similarly, the status data of logistics nodes (such as delays in transportation routes or vehicle utilization) will also be updated in real time through edge computing to ensure a quick response in the event of resource shortages or delays.
[0104] This real-time resource scheduling method means that the system can quickly adjust to changes in the supply chain without waiting for unified instructions from the central system, greatly improving the supply chain's adaptability. At the same time, due to the distributed nature of edge computing, the scheduling algorithm can locally optimize each node, so that each node in the supply chain can achieve the best in its resource allocation and task scheduling.
[0105] Although the adaptive dynamic scheduling algorithm can perform local optimization at each node, its ultimate goal is to achieve global resource optimization of the supply chain. The system will form a global resource scheduling diagram through the local scheduling data generated by each node, so as to make global optimization decisions. For example, when the production capacity of some nodes is saturated, the system can reallocate production tasks from other nodes, or transfer production equipment or inventory resources from other areas to the node through logistics adjustments. In this way, the resource allocation of the entire supply chain can not only adapt to the changes of local nodes, but also ensure global coordination and optimization.
[0106] The application of adaptive dynamic scheduling algorithms has produced significant results in supply chain management, especially through real-time data-driven scheduling decisions, which improves resource utilization efficiency and supply chain response speed.
[0107] Through localized optimization of edge computing nodes, the system can achieve accurate resource allocation and task allocation at each node. This not only avoids over-utilization or idleness of resources, but also ensures that the production, warehousing and logistics resources of each node can be optimally configured according to actual needs. For example, when market demand fluctuates, the system can quickly adjust production and logistics plans to ensure that the resources of production nodes and logistics nodes are fully utilized to avoid overproduction or logistics bottlenecks.
[0108] The adaptive dynamic scheduling algorithm can significantly shorten the response time of the supply chain to market changes by driving decisions through real-time data. Traditional supply chains usually rely on the decision transmission of the central system, but the present invention realizes decentralized rapid response capabilities through edge computing. For example, when the transportation route of a logistics node is delayed, the edge computing node will immediately make a decision to reallocate the transportation vehicle or adjust the route to avoid supply chain disruptions caused by delays. This rapid response capability is crucial to responding to emergencies and dynamic changes in the supply chain.
[0109] Although the scheduling of each node is optimized locally, the system can form a global resource scheduling strategy and achieve global optimization through the local resource scheduling data generated by edge computing. When resources are scarce, the system can ensure the stable operation of the supply chain through cross-node resource allocation, reducing the increase in overall supply chain costs caused by local resource shortages or improper allocation. For example, the system can reduce transportation costs, inventory management costs, and equipment usage costs by optimizing production and logistics scheduling.
[0110] Example: A global electronics manufacturing company applies an adaptive dynamic scheduling algorithm to manage its global supply chain network. The company monitors the real-time operational data of various production bases, storage centers, and logistics nodes around the world through edge computing nodes. The system obtains and analyzes market demand and production efficiency data in real time. When the production capacity of a production base approaches the upper limit, the edge computing node will adjust the production plan based on real-time data and assign part of the production tasks to other nodes with sufficient production capacity. At the same time, the edge computing nodes of the logistics nodes monitor the delays of the transportation routes and reallocate transportation resources when necessary to ensure efficient coordination of production and logistics.
[0111] In this embodiment, the adaptive dynamic scheduling algorithm not only improves the company's flexibility in responding to changes in global market demand, but also reduces production and logistics costs by optimizing resource allocation, ensuring that the supply chain can remain efficient and stable even when operating at high loads.
[0112] The execution and feedback module is used to execute supply chain resource allocation based on the final resource scheduling data, monitor the execution status of each link in the supply chain through IoT sensors, generate execution feedback data, perform self-correction based on the execution feedback data, and generate corrected resource scheduling data to ensure closed-loop optimization of the supply chain.
[0113] like Figure 3 As shown in Figure 1, the principle of the execution and feedback module is based on the idea of a closed-loop control system, which uses sensors to monitor the execution of each supply chain node in real time, generate feedback data, and continuously adjust and optimize resource scheduling strategies based on these data. This process is divided into three main stages: execution, monitoring and feedback, and self-correction.
[0114] Execution phase: In the execution phase, the system first allocates various resources in the supply chain based on the generated final resource scheduling data. These resources include production equipment, storage space, and logistics resources. The final resource scheduling data is generated by the adaptive dynamic scheduling algorithm in the previous phase and represents the optimal resource allocation plan for the current supply chain. The system performs specific operations according to this plan, such as allocating production tasks, adjusting logistics routes, and configuring storage resources.
[0115] Monitoring and feedback stage: During the supply chain resource scheduling execution process, IoT sensors monitor the use and status of resources in real time at each node (including production, warehousing, logistics, etc.). IoT sensors can collect key data such as the operating status of production equipment, real-time changes in inventory, and delays in logistics routes. For example, sensors can detect problems such as reduced operating efficiency of a certain production equipment, delays in logistics vehicles, and warehouses approaching storage limits. All of this data is fed back to the system in real time through sensors and generates execution feedback data.
[0116] Self-correction stage: Based on the execution data fed back by the sensors, the system will review the current resource scheduling strategy. If the feedback data shows that there is a problem with the current resource scheduling plan or it needs to be optimized (such as production equipment failure, excess inventory, or logistics delays), the system will automatically adjust the scheduling strategy based on these feedback data and generate corrected resource scheduling data. For example, when a logistics vehicle is delayed, the system may need to reallocate other vehicles or change the transportation route to ensure the continuity of the supply chain; when the warehouse is close to full, the system will readjust the warehousing strategy based on the inventory status and transfer the inventory to other storage nodes.
[0117] The entire process forms a closed-loop control system, that is, the system continuously obtains execution status data through sensors and makes dynamic adjustments based on this data to ensure continuous optimization of resource scheduling and efficient operation of the supply chain.
[0118] The effects of the execution and feedback module in supply chain management are mainly reflected in two aspects: real-time and closed-loop optimization.
[0119] Real-time: Through real-time monitoring by IoT sensors, the system can quickly obtain the execution status of each node in the supply chain and respond immediately. This real-time performance ensures the efficient operation of the supply chain, especially when faced with emergencies or unplanned changes, the system can quickly identify problems and respond through feedback data. For example, if a logistics node is delayed due to weather or traffic problems, the system can immediately capture this information through sensors and automatically adjust the logistics scheduling plan to avoid interruptions to the entire supply chain.
[0120] Closed-loop optimization: Closed-loop optimization is the core advantage of this module. The system does not simply perform a one-time scheduling, but continuously provides feedback and adjustments during the execution process. Through this closed-loop control, the supply chain system can continuously self-correct and optimize during operation. For example, when a production equipment fails, the system will automatically reduce the resource allocation to the equipment and assign the task to other equipment; when inventory pressure increases, the system will automatically allocate other storage resources. This closed-loop feedback and optimization mechanism enables the system to adapt to the dynamically changing market demand and supply chain environment, ensuring that the supply chain can still maintain efficient and stable operation under complex circumstances.
[0121] Example: A global consumer goods manufacturer applied the execution and feedback module to ensure the dynamic optimization of its global supply chain. In the production process, IoT sensors monitor the equipment operation of each production line in real time, including the equipment's working efficiency, maintenance requirements, etc. The system self-corrects production tasks based on the equipment status data fed back by the sensors. When the equipment of a production line fails or the production capacity is reduced, the system automatically allocates resources from other production lines and generates revised production scheduling data to ensure that the overall production task is not affected.
[0122] In the logistics process, the system uses sensors to monitor the location and operating status of transport vehicles in real time. When the logistics route is delayed or the inventory of a storage node is close to full, the system quickly adjusts logistics resources and reallocates transportation tasks through feedback data. In this way, the system realizes dynamic optimization of logistics scheduling, ensuring that products in various regions around the world can be delivered on time, while reducing inventory pressure and transportation costs.
[0123] like Figure 4 As shown, a method for executing the supply chain management system based on operational data comprises the following steps: The original operation data is collected from the dimensions of market demand, consumer behavior, production, logistics and inventory through IoT devices and edge computing nodes, and the original operation data is cleaned, formatted and standardized to generate standardized operation data; the present invention is based on the supply chain management of operation data, and collects multi-dimensional data of market demand, consumer behavior, production, logistics and inventory in real time through IoT devices and edge computing nodes to form a dynamic supply chain data set. These original data are cleaned, formatted and standardized to generate unified standardized operation data, which provides a basis for subsequent predictions and decisions.
[0124] Based on the standardized operation data, feature extraction is performed, and demand forecast data and production scheduling data are generated through multi-task learning models and quantum Bayesian deduction. The quantum Bayesian deduction is based on market demand feature data, production efficiency feature data and logistics status feature data for processing; in the data processing stage, the system extracts features from the standardized data through multi-task learning models and quantum Bayesian deduction to generate demand forecast data and production scheduling data. Quantum Bayesian deduction is based on multi-dimensional feature data (such as market demand, production efficiency, and logistics status), and can handle prediction problems in multiple market scenarios at the same time, thereby improving the accuracy and flexibility of predictions.
[0125] Based on the demand forecast data and the production scheduling data, the multi-agent system and the self-organizing criticality algorithm are used to make distributed autonomous decisions on the procurement, production and logistics nodes, and generate procurement data, production scheduling data and logistics adjustment data; based on the forecast results, the system uses the multi-agent system and the self-organizing criticality algorithm to make distributed autonomous decisions on the procurement, production and logistics nodes, and generate specific procurement, production scheduling and logistics adjustment data. This distributed decision-making mechanism can optimize the decision-making of each node based on real-time data, ensuring the overall coordination and optimization of the supply chain.
[0126] Based on the procurement data, the production scheduling data and the logistics adjustment data, the production equipment, storage space and logistics resources in the supply chain are dynamically allocated using resource pooling and virtualization, and adaptive dynamic scheduling algorithms to generate final resource scheduling data; in the resource allocation stage, the system dynamically allocates production equipment, storage space and logistics resources through resource pooling, virtualization technology, and adaptive dynamic scheduling algorithms to generate the final resource scheduling plan. This process ensures that resources in the supply chain can be flexibly allocated between different nodes to achieve optimal resource allocation.
[0127] The system performs supply chain resource allocation based on the final resource scheduling data, monitors the execution status of each link of the supply chain through the IoT sensors, generates execution feedback data, performs self-correction based on the execution feedback data, generates corrected resource scheduling data, and ensures closed-loop optimization of the supply chain. The system monitors the execution status of each node in real time through IoT sensors, generates execution feedback data, and performs self-correction based on the feedback to generate new resource scheduling data to achieve closed-loop optimization. Through this closed-loop feedback mechanism, the supply chain can continuously adapt to market changes, optimize resource utilization, and improve overall operational efficiency.
[0128] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims
1. A supply chain management system based on operational data, characterized in that: include: The data collection module is used to collect raw operation data from the dimensions of market demand, consumer behavior, production, logistics and inventory through IoT devices and edge computing nodes, and clean, format and standardize the raw operation data to generate standardized operation data; A data deduction and prediction module is used to extract features based on standardized operational data and generate demand forecast data and production scheduling data through a multi-task learning model and quantum Bayesian deduction, wherein the quantum Bayesian deduction is based on market demand feature data, production efficiency feature data, and logistics status feature data for processing; Distributed collaborative decision-making module, which is used to make distributed autonomous decisions on procurement, production and logistics nodes based on demand forecast data and production scheduling data, using multi-agent systems and self-organizing criticality algorithms to generate procurement data, production scheduling data and logistics adjustment data; The resource allocation and management module is used to dynamically allocate production equipment, storage space and logistics resources in the supply chain based on procurement data, production scheduling data and logistics adjustment data, using resource pooling and virtualization, and adaptive dynamic scheduling algorithms to generate final resource scheduling data; The execution and feedback module is used to execute supply chain resource allocation based on the final resource scheduling data, monitor the execution status of each link in the supply chain through IoT sensors, generate execution feedback data, perform self-correction based on the execution feedback data, and generate corrected resource scheduling data to ensure closed-loop optimization of the supply chain.
2. The supply chain management system based on operational data according to claim 1, characterized in that: The multi-task learning model in the data deduction and prediction module simultaneously performs multi-task feature extraction on the market demand feature data, the production efficiency feature data and the logistics status feature data to generate shared feature representations of different tasks, so that the optimization processes of multiple tasks are related to each other.
3. The supply chain management system based on operational data according to claim 2, characterized in that: The quantum Bayesian deduction is deduced based on the Bayesian theorem, wherein the market demand characteristic data, the production efficiency characteristic data and the logistics state characteristic data are processed by quantum state superposition and quantum parallel calculation to generate the demand forecast data under various market scenarios. The calculation process of the quantum Bayesian deduction includes the following expression: in, is the posterior distribution, indicating that given data In the case of The probability distribution of is the likelihood function, which means that given the parameters When, data Probability of occurrence; is the prior distribution, which means that when the data is observed Before, the parameter The initial probability distribution of is a standardization constant, indicating that the data The total probability of is used to ensure the normalization of the posterior distribution.
4. The supply chain management system based on operational data according to claim 3, characterized in that: The prior distribution of the quantum Bayesian deduction is constructed based on historical supply chain data and external market intelligence, and multiple market scenarios are simultaneously deduced through the superposition of quantum states during the quantum computing process. The quantum Bayesian deduction results are used to provide multiple possible paths for market demand and production scheduling, and quantify the credibility of each path.
5. The supply chain management system based on operational data according to claim 1, characterized in that: The multi-agent system in the distributed collaborative decision-making module includes multiple independent supply chain nodes. Each node makes autonomous decisions based on the local market demand characteristic data and the production status data. The nodes communicate in real time through a distributed network and share decision parameters to achieve globally optimal supply chain scheduling and resource allocation.
6. The supply chain management system based on operational data according to claim 5, characterized in that: The self-organizing criticality algorithm monitors the resource consumption state of each node, and when the resource state of the node approaches a critical value, adjusts the priority and resource allocation strategy of the node.
7. The supply chain management system based on operational data according to claim 6, characterized in that: The distributed collaborative decision-making module performs global optimization on resource allocation among nodes based on the following calculation expression: in, For the The resource consumption function of nodes represents the node Use Resources the costs incurred; For the and The resource complementarity function between nodes represents the node and nodes In their respective resources and The benefits generated by resource synergy under the configuration; It represents the total number of nodes in the supply chain. By optimizing the resource allocation of each node, the resource consumption of the entire supply chain is minimized, while the resource complementarity effect between the nodes is maximized to generate the procurement data, the production scheduling data and the logistics adjustment data.
8. The supply chain management system based on operational data according to claim 1, characterized in that: The resource allocation and management module virtualizes all physical resources in the supply chain, including production equipment, storage space and logistics resources, through resource pooling technology, and uniformly allocates them through virtual resource pools to achieve cross-regional scheduling of supply chain resources.
9. The supply chain management system based on operational data according to claim 8, characterized in that: The adaptive dynamic scheduling algorithm dynamically adjusts the resource allocation strategy according to the market demand data, the production efficiency data and the logistics status data acquired in real time, optimizes local resources at each node of the supply chain through edge computing nodes, and generates final resource scheduling data.
10. A method for executing the supply chain management system based on operational data according to any one of claims 1 to 9, characterized in that: The following steps are involved: Collect raw operational data from the dimensions of market demand, consumer behavior, production, logistics, and inventory through IoT devices and edge computing nodes, and clean, format, and standardize the raw operational data to generate standardized operational data; Extract features based on the standardized operation data, and generate demand forecast data and production scheduling data through a multi-task learning model and quantum Bayesian deduction, wherein the quantum Bayesian deduction is processed based on market demand feature data, production efficiency feature data, and logistics status feature data; Based on the demand forecast data and the production scheduling data, a multi-agent system and a self-organizing criticality algorithm are used to make distributed autonomous decisions on procurement, production and logistics nodes to generate procurement data, production scheduling data and logistics adjustment data; Based on the procurement data, the production scheduling data and the logistics adjustment data, the production equipment, storage space and logistics resources in the supply chain are dynamically allocated using resource pooling and virtualization and adaptive dynamic scheduling algorithms to generate final resource scheduling data; Supply chain resource allocation is performed based on the final resource scheduling data, the execution status of each link in the supply chain is monitored through the Internet of Things sensors, execution feedback data is generated, self-correction is performed based on the execution feedback data, and corrected resource scheduling data is generated to ensure closed-loop optimization of the supply chain.
Citation Information
Patent Citations
Logistics and supply chain management system based on RFID
CN118333501A
Industrial internet-based supply chain artificial intelligence processing method and system
CN118394536A
Cited By
Real-time commodity sales prediction system and method based on edge calculation
CN119444305A
Cooperative processing method based on digital service supply chain
CN119599598A
A digital service supply chain collaborative processing method
CN119599598B
Fuel optimization scheduling method based on coal supply chain prediction
CN119904026A
Order dispatching optimization method and device based on automobile service supply chain, and storage medium
CN119940880A
