Internet of Things platform supply chain data processing method and system based on artificial intelligence

Abstract

The invention relates to the technical field of supply chain management, and discloses an Internet of Things platform supply chain data processing method and system based on artificial intelligence, and the method comprises the steps: constructing an Internet of Things information sensing network of a supply chain through the deployment of a sensor unit and an intelligent terminal, and carrying out the digital feedback simulation; a supply chain model is constructed, model parameters are adjusted through real-time information collection, supply chain process link actual conditions and article operation track data are obtained and substituted into the model, supply chain project features are deeply analyzed by using a pre-trained artificial intelligence algorithm, and analysis features of all analysis projects are extracted and transmitted to an application service layer; and adjusting execution parameters to output a supply chain optimization decision, realizing intelligent decision through real-time data perception, analysis and optimization of a supply chain process and deep analysis based on an AI model, reducing human errors and delay, optimizing each link of the supply chain, and solving the problem of difficulty in effective processing and analysis of supply chain data of the Internet of Things in the prior art. And the intelligent decision-making problem is supported.

Description

Technical Field

[0001] The present invention relates to the technical field of supply chain management, and particularly to a method and system for processing supply chain data of an Internet of Things platform based on artificial intelligence. Background Art

[0002] With the rapid development of globalization and informatization, supply chain management has become a core link in modern enterprise operations. Traditional supply chain management methods rely on manual operations and basic information systems, and often face problems such as poor real-time performance, slow response, inaccurate information, and insufficient collaboration.

[0003] Internet of Things (IoT) technology can sense and collect data in real time at all links in the supply chain (such as production, transportation, warehousing, distribution, etc.) through hardware devices such as sensors, RFID tags, and GPS. These sensed data can not only reflect various physical states in the supply chain (such as temperature, humidity, inventory, location, etc.), but also provide important information for enterprises in aspects such as production progress, transportation status, and inventory management.

[0004] However, simply collecting data through IoT technology is not sufficient to achieve intelligent supply chain management. The data collected by IoT systems is often massive, complex, and multi-dimensional, making it difficult to effectively process and analyze this data to support intelligent decision-making. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for processing supply chain data of an Internet of Things platform based on artificial intelligence, aiming to solve the problem in the prior art that it is difficult to effectively process and analyze IoT supply chain data to support intelligent decision-making.

[0006] The present invention is implemented as follows. In the first aspect, the present invention provides a method for processing supply chain data of an Internet of Things platform based on artificial intelligence, including: Construct an information perception layer of the Internet of Things platform through a pre-deployed set of sensor units, and perform perception supplementation on the information perception layer through an intelligent terminal linked to the information perception layer to obtain a supply chain perception network; Construct a supply chain perception network model corresponding to the supply chain perception network, and perform real-time information collection on the supply chain perception network to synchronously adjust the model parameters of the supply chain perception network model; Based on the supply chain perception network, perform information perception on the supply chain in terms of the actual situation dimension of the supply chain process link and the operation trajectory dimension of the supply chain item to obtain the actual situation perception information of the supply chain process link and the operation trajectory perception information of the supply chain item, and substitute them into the supply chain perception network model; Performing feature extraction on the supply chain for a specified analysis item based on the supply chain perception network model to construct a supply chain item feature distribution corresponding to each specified analysis item; Performing in-depth analysis on each of the supply chain item feature distributions through an artificially intelligent algorithm model pre-trained in the data processing layer of the Internet of Things platform to obtain analysis features for each item; Substituting each of the item analysis features into the application service layer of the Internet of Things platform, and enabling the supply chain operation execution algorithm model in the application service layer to output specific execution decisions for the supply chain according to each of the item analysis features.

[0007] In a second aspect, the present invention provides an Internet of Things platform supply chain data processing system based on artificial intelligence for implementing the method for processing supply chain data of an Internet of Things platform based on artificial intelligence according to any one of the first aspect.

[0008] The present invention provides a method for processing supply chain data of an Internet of Things platform based on artificial intelligence, having the following beneficial effects: By deploying sensor units and intelligent terminals, the present invention constructs an Internet of Things information perception network for the supply chain and conducts digital feedback simulation, constructs a supply chain model and adjusts model parameters through real-time information collection, obtains the actual situation of the supply chain process links and the data of the item operation trajectory and substitutes them into the model, uses a pre-trained artificial intelligence algorithm to conduct in-depth analysis on the supply chain item features, extracts the analysis features of each analysis item and transmits them to the application service layer, adjusts the execution parameters and outputs supply chain optimization decisions, optimizes the supply chain process through real-time data perception and analysis, improves the overall efficiency, realizes intelligent decision-making based on in-depth analysis of the AI model, reduces human errors and delays, optimizes each link of the supply chain, and solves the problem in the prior art that it is difficult to effectively process and analyze Internet of Things supply chain data to support intelligent decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 is a schematic diagram of the steps of a method for processing supply chain data of an Internet of Things platform based on artificial intelligence provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0010] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0011] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0012] Refer to Figure 1 as shown, which is a preferred embodiment provided by the present invention.

[0013] In a first aspect, the present invention provides an Internet of Things platform supply chain data processing method based on artificial intelligence, including: S1: Construct an information perception layer of the Internet of Things platform through a pre-deployed set of sensor units, and perform perception supplementation on the information perception layer through an intelligent terminal linked to the information perception layer to obtain a supply chain perception network; S2: Perform digital feedback simulation on the supply chain perception network to obtain a supply chain perception network model, and perform real-time information collection on the supply chain perception network to synchronously adjust the model parameters of the supply chain perception network model; S3: Based on the supply chain perception network, perform information perception on the supply chain in terms of the actual situation dimension of the supply chain process link and the item operation trajectory dimension of the supply chain to obtain the actual situation perception information and item operation trajectory perception information of the supply chain process link, and substitute them into the supply chain perception network model; S4: Based on the supply chain perception network model, extract features of specified analysis items for the supply chain to construct a supply chain item feature distribution corresponding to each specified analysis item; S5: Through an artificial intelligence algorithm model pre-trained in the data processing layer of the Internet of Things platform, perform in-depth analysis of each supply chain item feature distribution for the corresponding specified analysis item to obtain the item analysis features of each specified analysis item of the Internet of Things platform; S6: Substitute each item analysis feature into the application service layer of the Internet of Things platform, and enable the supply chain operation execution algorithm model in the application service layer to synchronously adjust the execution parameters according to each item analysis feature to output specific execution decisions of the supply chain.

[0014] Specifically, in step S1 of the embodiment provided by the present invention, according to the requirements of the supply chain, different types of sensors are selected, such as temperature and humidity sensors, position tracking sensors (such as GPS, RFID), environmental sensors (gas, noise), item status monitoring sensors (pressure, vibration, etc.), etc. These sensor units are pre-deployed in each link of the supply chain (such as warehousing, transportation, production, etc.). These sensors will monitor the environment, item status and their changes in real time and transmit the information to the Internet of Things platform. The sensors can cover all links of the supply chain, collect the status data of the environment and items in real time, ensure that comprehensive information can be obtained, and through the data transmission function of the sensors, the status changes of each link in the supply chain can be monitored in real time, and potential risks or anomalies can be captured in time.

[0015] More specifically, the data collected by sensors is transmitted to the information perception layer of the Internet of Things platform through wired or wireless networks. At this level, the data undergoes preliminary processing, filtering, and sorting to provide basic data for subsequent analysis and decision-making. The information perception layer is not just a data collection channel but also needs to have the functions of data integration and management. It integrates heterogeneous data collected by different sensors to ensure data consistency and timeliness, ensuring that data from various sensors can be efficiently integrated into a unified and standardized data format for subsequent processing and analysis, and ensuring low latency and high efficiency in data transmission, enabling the perception layer to quickly acquire and process real-time data.

[0016] More specifically, in addition to fixed sensors, intelligent terminals (such as smartphones, wearable devices, handheld scanning devices, etc.) are connected to the information perception layer for data supplementation and supplementary perception. For example, an operator uses an intelligent terminal to scan barcodes or QR codes to obtain detailed information about specific goods, further supplementing information such as the status and location of items in the perception network. The data of intelligent terminals is uploaded to the perception layer through wireless networks (such as Wi-Fi, Bluetooth, 5G, etc.), further enhancing the integrity of information. Intelligent terminals can provide real-time feedback to the perception network. For example, when receiving an alarm, they automatically send current operation information or production progress. Intelligent terminals can supplement specific data that sensors cannot obtain in some cases, such as manual operation information, instant geographical location information, and the status of goods during transportation. The use of intelligent terminals increases the flexibility of data collection, especially in a dynamically changing supply chain environment, enabling flexible response to on-site data collection requirements.

[0017] More specifically, the data of sensor units and intelligent terminals is uniformly integrated into the Internet of Things platform to form a complete supply chain perception network. This network covers the information perception of all links in the supply chain, can collect dynamic data of each link in real time, and is processed through the perception layer. At the perception layer, the collected data is preliminarily processed by the cloud computing or edge computing nodes of the Internet of Things platform and transmitted to subsequent levels such as the data analysis layer and the decision-making layer. Through the combination of sensors and intelligent terminals, it is possible to comprehensively monitor all links of the entire supply chain from raw material procurement, production, transportation, warehousing to sales, providing real-time and accurate supply chain data. The perception network makes the data of each link in the supply chain more transparent, and any anomalies or problems in any link can be promptly detected through the system's real-time monitoring, thus improving the response ability of the supply chain.

[0018] More specifically, the supply chain perception network analyzes the sensor data collected in real time through the Internet of Things platform to form a real-time data stream for decision-making support. Based on this data, the Internet of Things platform can automatically detect potential bottlenecks, transportation delays, production schedule lags, and other issues. The data analysis results can provide a basis for subsequent optimization algorithms, thereby influencing decisions such as production scheduling, inventory management, and transportation arrangements, and improving the overall efficiency of the supply chain.

[0019] Specifically, in step S2 of the embodiment provided by the present invention, a digital supply chain perception network model is constructed through preset sensor data, intelligent terminal data, and data streams of each link in the supply chain. This model can be built based on simulation tools or modeling platforms (such as MATLAB, Simulink, AnyLogic, etc.). The model should cover the main links of the supply chain, such as production, warehousing, transportation, inventory management, etc. In the model, a digital feedback mechanism is designed to feed the actual data back into the model to adjust the model parameters. The feedback loop can be updated in real time through the data collected in real time, sensor, and intelligent terminal information, simulating the dynamic changes in the actual operation of the supply chain network. Through digital feedback simulation, a digital twin model synchronized with the real-world supply chain network is created, which can accurately reflect the dynamic changes of the supply chain and provide a basis for subsequent decision-making and optimization. By constructing a digital feedback loop, the performance of the real supply chain under different conditions can be simulated, thereby identifying potential problems and bottlenecks in advance and avoiding mistakes in actual operations.

[0020] More specifically, the Internet of Things platform is used to continuously collect data from sensors and intelligent terminals, such as inventory data, transportation status, production progress, etc. These data are transmitted in real time to the cloud platform or edge computing nodes through wireless networks (such as Wi-Fi, 5G, LPWAN, etc.). The real-time data collected will be fed back into the supply chain perception network model to update the model parameters in a timely manner. For example, actual data such as inventory levels, transportation delays, and production capacity will affect the key parameters in the model. The system automatically adjusts the model parameters by calculating the difference between the model and the actual data, enabling the model to continuously reflect the real situation. Through real-time information collection and feedback, the high synchronization between the supply chain perception network model and actual operations is ensured, and any changes in the supply chain can be immediately reflected. The model can be adjusted and respond to changes occurring in reality at any time, providing accurate status predictions and dynamic scheduling optimization.

[0021] More specifically, based on real-time data feedback, algorithms such as adaptive control, machine learning algorithms, and Kalman filtering are used to synchronously adjust the model parameters. For example, variables such as transportation time, inventory quantity, and production progress are monitored in real time. If a delay occurs in a certain link, the model will automatically adjust the parameters of that link, such as the inventory warning threshold and transportation arrangement time. Combining real-time data, optimization algorithms such as genetic algorithms and particle swarm optimization algorithms are applied to continuously optimize the parameters of the supply chain perception network model. For example, according to changes in inventory levels, decisions on material procurement, production scheduling, and logistics routes in the supply chain are dynamically adjusted, thereby improving the overall supply chain efficiency. Through the real-time data feedback and model parameter adjustment mechanism, the supply chain model can intelligently adapt to changes in the actual environment, achieve automatic adjustment and optimization, improve the supply chain response speed and accuracy. The application of optimization algorithms can provide optimal decisions in a real-time changing environment, enabling the supply chain to more efficiently cope with uncertainties and changes, thereby reducing waste, lowering costs, and improving resource utilization rate.

[0022] More specifically, based on the supply chain perception network model after synchronous adjustment, the system can provide real-time decision support based on the model, such as predicting supply chain bottlenecks, optimizing inventory management, and adjusting transportation routes. With the help of artificial intelligence technology, the system can automatically generate optimal decision-making schemes and reduce manual intervention. In some cases, the system can directly control the operations of the supply chain according to the adjusted model, such as automatically scheduling transportation tools and production line equipment to execute the optimal supply chain operations. The real-time adjustment of model parameters makes the decision-making process more intelligent. The system can quickly react from historical data, real-time data, and feedback information and give the optimal solution. Each link of the supply chain can automatically adjust according to the changes in the real-time model, improving the response speed and automation level, thereby achieving the autonomous optimization and efficient operation of the supply chain.

[0023] More specifically, the supply chain perception network model is evaluated regularly to check the accuracy and adaptability of the model. The evaluation can be based on the results of actual operations, such as indicators like order fulfillment accuracy, inventory turnover speed, and transportation timeliness. If it is found that the model prediction error is large, the model is corrected in a timely manner. According to the evaluation results and new business requirements, the model is continuously optimized and adjusted. For example, new sensor data, new business processes, or new optimization algorithms can be introduced to continuously improve the adaptability and accuracy of the model.

[0024] Specifically, in step S3 of the embodiment provided by the present invention, the real-time operation status and key data of each link of the supply chain, such as production status, inventory level, order fulfillment progress, transportation status, etc., are captured, and key data of the process links are collected in real time through sensors, smart devices, RFID tags, Internet of Things technology (IoT), ERP systems, etc., for example, the operation status of the production line, the real-time quantity of inventory, the location and status of transportation tools, etc., including the actual completion time, production capacity, logistics route, inventory consumption rate, supplier delivery status, etc. of each link.

[0025] More specifically, the purpose of sensing the trajectory of items is to track the specific circulation path and status changes of items in the supply chain, record the current location, transportation trajectory, loading and unloading information of each item, and use RFID, GPS, barcode scanning, sensors and other technologies to track items in real time. For example, GPS tracks the location of transport vehicles, RFID tracks the storage location of goods in warehouses, and barcode scanning records product circulation information, including real-time location during transportation, estimated arrival time, whether there are delays during transportation, and the status of items in warehouses or during transportation, so as to achieve real-time and accurate monitoring of each link of the supply chain process and the trajectory of items. Through real-time data collection, the information asymmetry problem caused by traditional manual recording and data lag can be effectively avoided, and comprehensive data can be obtained through different perception dimensions (process links and item trajectories), thereby improving the diversity and depth of information.

[0026] More specifically, the collected raw data is cleaned and preprocessed to remove noise and abnormal data and fill in missing values. Data cleaning algorithms, such as filtering algorithms, interpolation methods, and standardization processing, are usually used to standardize data from different sources to ensure that data from different types of sensors and systems can be seamlessly integrated.

[0027] More specifically, data fusion technology is used to bring together information from different sensing devices and systems to form a unified data view. Kalman filtering, Bayesian networks, weighted averaging and other methods can be used to fuse information from different sources. For example, data from GPS and RFID can be combined into the real-time location and transportation path information of items.

[0028] More specifically, time series analysis is performed on information of different dimensions to identify the correlation between information. For example, the progress of the production process is associated with logistics and transportation information to ensure real-time feedback on the entire process from production to distribution. Through data fusion and preprocessing, the data differences between different devices and systems are eliminated, making the final information more accurate and consistent, achieving the synergy of multi-dimensional and multi-source data, and improving the system's understanding and response capabilities to complex supply chain operations.

[0029] More specifically, the processed real-time perception information of the process links and the perception information of the item operation trajectory are substituted into the supply chain perception network model. The supply chain perception network model usually includes dynamic information of multiple links, such as production scheduling, inventory management, logistics scheduling, etc. After substituting the information, the model will update the status of each link. The way of substituting information usually adopts technologies such as interfaces and API calls to transmit the collected real-time data to the established digital twin model or supply chain optimization model. After substituting the perception information into the model, the system will analyze the status of the supply chain in real time and provide feedback. If there is a delay or bottleneck in a certain link, the model will dynamically adjust the strategy, such as adjusting the production schedule, optimizing the inventory allocation or changing the transportation route. This process depends on the update of the real-time data stream to ensure that the model can continuously optimize itself as the real supply chain changes.

[0030] It can be understood that through the substitution of real-time information, the supply chain perception network model is updated in real time to ensure that each link in the model can reflect the real situation. The timely input of perception information provides real-time data for the decision support system, enabling it to quickly make optimization decisions based on the real-time status, such as adjusting production, allocating inventory, optimizing transportation, etc. After the perception information enters the supply chain model, the system automatically generates decision-making plans based on the output of the model. For example, based on changes in inventory levels and transportation status, the system automatically adjusts the procurement plan, production plan, etc. of the supply chain. Common optimization methods include linear programming, integer programming, genetic algorithms, etc. These methods can provide optimal or near-optimal decision-making plans in an uncertain supply chain environment. Based on the optimization results, the operation strategies of each link in the supply chain will be adjusted in real time. For example, if a certain transportation route is predicted to be congested, the system will recommend a detour and automatically adjust the corresponding transportation plan. The automation system can directly transmit the optimization plan to the execution system, such as the warehouse management system, transportation scheduling system, etc., for rapid execution.

[0031] More specifically, monitor the execution effects of each link in the supply chain and conduct real-time evaluation. The evaluation criteria can include order fulfillment rate, inventory turnover rate, transportation timeliness, etc. By comparing with the preset goals, detect the effects of the model and decisions, and use the evaluation results as new input to feedback to the perception network model. If the evaluation results indicate that some decisions are not ideal, the system will automatically correct the model and readjust the optimization strategy. Through continuous monitoring and feedback, ensure that the model is continuously adjusted and optimized, and the supply chain can adapt to market changes and demand fluctuations. The continuous evaluation and feedback mechanism can make the supply chain perception network continuously tend to the optimal state, improving the overall operation efficiency and resource utilization rate.

[0032] Specifically, in step S4 of the embodiment provided by the present invention, according to the specific needs of the supply chain, the items or dimensions that need to be analyzed are determined. For example, possible analysis items include order fulfillment efficiency, inventory management, supplier delivery performance, transportation timeliness, demand forecast accuracy, etc. For each analysis item, the key features that need to be extracted are clearly defined. For example, for order fulfillment efficiency, the features that may need to be extracted include order delivery time, order delay time, warehouse processing speed, etc. For inventory management, it may be necessary to extract features such as inventory turnover rate, inventory out-of-stock rate, and inventory holding cost. The pertinence and effectiveness of the feature extraction process are ensured to avoid overly broad or irrelevant features interfering with the analysis results. The feature definitions of different analysis items are clear and unified, which is conducive to subsequent feature extraction and model training.

[0033] More specifically, relevant real-time data is obtained from the supply chain perception network. The supply chain perception network usually includes sensor data, ERP system data, inventory data, transportation scheduling data, etc. in production, warehousing, transportation, etc. For example, the inventory status of items is obtained through RFID, the transportation status is obtained through GPS, and the production progress data is obtained through the production line monitoring system. The data is cleaned, including removing outliers, filling missing values, and deduplication. The method of use can be interpolation, mean filling, etc., to standardize data from different sources so that the data has the same measurement unit and scale for subsequent analysis. Cleaning and preprocessing ensure the accuracy and consistency of the data, eliminate interference information, and can standardize data from different sources to ensure the data integration effect in the subsequent analysis process.

[0034] More specifically, according to the analysis items determined in step 1, appropriate algorithms are used to extract corresponding features from the raw data. Common feature extraction methods include: statistical features: such as mean, variance, maximum, minimum, standard deviation, etc., which are used to describe the basic distribution characteristics of the data; time series features: such as time series trends, seasonality, and periodicity, etc., which are used to capture dynamic changes in the supply chain; frequency domain features: using methods such as Fourier transform to analyze the frequency components of time series data to help identify potential periodic changes; spatial features: such as spatial location distribution, geographic location features, etc., especially in transportation management, GPS trajectory data can be used to extract spatial features such as transportation routes, driving speeds, and dwell time.

[0035] More specifically, machine learning algorithms (such as clustering, principal component analysis (PCA), independent component analysis (ICA), etc.) are used to further mine high-order features. These algorithms can help identify potential patterns in the data, extract more abstract features. Through the combination of multiple methods, useful features can be extracted from different dimensions and levels. Not only simple statistical features are extracted, but also higher-order features can be extracted through methods such as machine learning, providing richer information for subsequent analysis.

[0036] More specifically, based on the extracted feature data, the feature distribution of each analysis item is constructed. Common methods include: Histogram: The feature values are divided according to intervals to construct the probability distribution diagram of each feature. Probability density estimation: Methods such as kernel density estimation (KDE) are used to smooth the distribution of features to obtain a more accurate feature distribution curve. Statistical distribution fitting: According to the actual distribution of features, common statistical distributions (such as normal distribution, gamma distribution, Poisson distribution, etc.) are used for fitting for subsequent prediction and analysis. Multidimensional distribution: If the features are multidimensional (for example, the comprehensive features of multiple links), joint distribution or conditional distribution can be used to represent the relationship between multiple features.

[0037] More specifically, the constructed feature distribution is analyzed to evaluate its stability and volatility. For example, whether there are obvious periodic fluctuations, whether it conforms to the expected distribution pattern, etc. Visualization tools (such as heat maps, scatter plots, box plots, etc.) are used to display the morphology of the feature distribution, facilitating analysts to understand the data features, construct accurate feature distributions, and provide reliable statistical basis for further analysis and decision-making. Through distribution modeling, the potential relationships between features can be revealed, helping analysts better understand the dynamic changes of the supply chain.

[0038] More specifically, the constructed feature distribution model is used for supply chain analysis. For example, it can be used to predict future order delivery times, predict inventory out-of-stock risks, predict supplier delivery delays, etc. The feature distribution is combined with the optimization goals of the supply chain (such as cost minimization, timeliness maximization, etc.) for intelligent optimization decisions. In practical applications, the prediction results of the model may deviate from reality. At this time, the model needs to be optimized through a feedback mechanism. By comparing with the actual results, the parameters in the feature distribution model are adjusted to make it more in line with the real situation. Based on the refined feature distribution model, the risks and demands in the supply chain can be predicted more accurately, assisting in decision-making. Through continuous model optimization, the agility and response speed of supply chain management can be improved.

[0039] Specifically, in step S5 of the embodiment provided by the present invention, the Internet of Things platform obtains data of each link in the supply chain in real time through various devices such as sensors, RFID, GPS, cameras, etc., including but not limited to: production link data: production equipment status, processing speed, production quality, etc., inventory link data: inventory quantity, inventory turnover, inventory status, etc., transportation link data: transportation route, transportation duration, cargo location, etc., sales / demand data: sales volume, market demand fluctuation, etc.

[0040] More specifically, based on the data of the Internet of Things platform, features related to each specified analysis item are extracted. For example: for the inventory management item, inventory turnover rate, out-of-stock rate, inventory cycle, etc. are extracted; for the transportation optimization item, transportation timeliness, transportation cost, distribution route efficiency, etc. are extracted; for the production scheduling item, production speed, equipment utilization rate, product qualification rate, etc. are extracted. Technologies such as principal component analysis (PCA), feature selection (such as feature importance based on tree models), etc. are used to further optimize the feature space, reduce redundancy and noise. Precise feature extraction and the optimized feature space provide high-quality input data for in-depth analysis. Redundant features are removed through feature engineering to ensure the efficiency and accuracy of the analysis process.

[0041] More specifically, based on the specific requirements of the supply chain project, different artificial intelligence models are selected for in-depth analysis. Common artificial intelligence algorithms include: supervised learning algorithms: such as regression analysis, support vector machine (SVM), decision tree, random forest, neural network, etc., which are used to train the labeled data for classification or regression prediction; unsupervised learning algorithms: such as clustering analysis, principal component analysis (PCA), autoencoder, etc., which are used to perform pattern recognition on data without labels; deep learning algorithms: such as convolutional neural network (CNN), recurrent neural network (RNN), etc., which are particularly suitable for the analysis of time series data (such as time series in inventory and transportation data).

[0042] More specifically, in the computing layer of the Internet of Things platform, historical data is used to train the artificial intelligence algorithm, and techniques such as cross-validation and hyperparameter tuning are adopted to ensure the accuracy and stability of the model. The trained artificial intelligence model can accurately predict and classify various analysis items based on the input supply chain feature data. Unsupervised learning and deep learning models can identify potential data patterns and anomalies, helping to discover potential problems in the supply chain in advance.

[0043] More specifically, various types of data collected in real time in the Internet of Things platform are input into a pre-trained artificial intelligence algorithm model for in-depth analysis. Specific analysis is carried out for each specified analysis item (such as inventory management, transportation optimization, demand forecasting, etc.), and key indicators and potential trends related to the item are identified. For example, through in-depth analysis of transportation timeliness, the relationships with factors such as transportation routes and traffic conditions are discovered; for the analysis of inventory out-of-stock risks, weak links in the supply chain are identified. The analysis results are presented through a visual interface to help managers and decision-makers better understand the data analysis results and provide specific decision support information, such as inventory replenishment suggestions, transportation route optimization suggestions, production scheduling adjustment suggestions, etc.

[0044] More specifically, the Internet of Things platform monitors the key indicators of each analysis item in real time and uses the trained artificial intelligence model for real-time prediction and anomaly detection. For example, it monitors in real time the time delays and inventory changes during transportation, and adjusts the prediction results of the model based on real-time data. Based on real-time data and newly generated data, the artificial intelligence model is continuously optimized to improve the model's adaptability to changes in the supply chain. For example, the online learning method is used to enable the model to continuously update and optimize as new data arrives.

[0045] Specifically, in step S6 of the embodiment provided by the present invention, the Internet of Things platform transmits various item analysis features (such as inventory turnover rate, transportation route efficiency, production scheduling optimization, etc.) obtained from the artificial intelligence model to the application service layer through a data channel. During the transmission process, all features are encoded and formatted according to a unified data standard to ensure that different modules within the system can correctly interpret and process these features. Each module of the Internet of Things platform (such as the data processing layer and the application service layer) exchanges data through an open API interface to ensure that the analysis results can flow to the decision-making module in a timely and accurate manner. The transmission of data from the analysis model to the application service layer is efficient and seamless, ensuring that the analysis results can be quickly transmitted and used for the next step of decision-making. The adoption of a standardized data format avoids errors or delays caused by inconsistent data formats and improves the scalability of the system.

[0046] More specifically, at the application service layer, the incoming project analysis features will be parsed and interpreted to ensure that the impact of each feature on the operation of the supply chain is fully understood. For inventory management projects, the inventory turnover rate may affect the speed of inventory replenishment, which in turn affects the material flow rate in the supply chain. For transportation projects, features such as route efficiency and timeliness will affect distribution planning and transportation scheduling. For production management projects, features such as equipment utilization rate and production capacity will affect the allocation of production tasks and the arrangement of processes. Through the accurate interpretation of the analysis features, it is ensured that the application service layer can comprehensively understand the actual meaning behind each feature, ensuring the pertinence and accuracy of decision-making. During the parsing process, a rule engine or model reasoning is adopted to avoid misunderstanding or misusing the analysis features, thereby reducing the risks in the decision-making process.

[0047] More specifically, according to the transmitted project analysis features, the supply chain operation execution algorithm models (such as inventory management algorithms, transportation scheduling algorithms, production scheduling algorithms, etc.) will synchronously adjust the relevant execution parameters. Through features such as inventory turnover rate and inventory level, the optimal replenishment quantity, inventory warning threshold, replenishment cycle, etc. are dynamically adjusted. According to features such as transportation efficiency, route, and timeliness, the transportation plan is adjusted, the best transportation route is selected, and the allocation of transportation resources is optimized. According to features such as production speed, equipment utilization rate, and production capacity, the production schedule, resource allocation, and equipment maintenance plan are adjusted. The execution algorithm of the application service layer will select the most appropriate decision path or action plan according to the optimization goal and the adjusted parameters, and make dynamic adjustments. For example, in some cases, it may be necessary to increase inventory reserves, or adjust the transportation route during certain periods to save costs. The algorithm can flexibly adjust the execution parameters according to actual needs, improving the adaptability of the supply chain. The parameter adjustment is based on the latest analysis features to ensure real-time response during the actual operation of the supply chain and avoid lagged decision-making.

[0048] More specifically, the adjusted execution parameters will be input into the supply chain decision-making model in the application service layer to generate specific execution decisions. The decisions include: inventory decisions: such as whether replenishment is needed, the quantity of replenishment, the time arrangement of replenishment, etc.; transportation decisions: such as which transportation route to choose, which transportation tasks to arrange, adjusting the transportation time, etc.; production decisions: such as adjusting the production plan, arranging the load of the production line, adjusting equipment maintenance, etc. After the decisions are output, they will be docked with relevant business systems (such as inventory management systems, transportation scheduling systems, production planning systems, etc.) through the Internet of Things platform to ensure that the decisions can be implemented in real time. At the same time, the feedback mechanism will monitor the implementation effect of the decisions and continuously optimize the decision-making model.

[0049] More specifically, the execution decisions adjusted based on the analysis features of each project are accurate and effective, capable of coping with complex supply chain challenges. The generated decisions are not only strategic but also can directly guide specific operations, ensuring the operability and efficiency of the execution plan.

[0050] More specifically, once the execution decision is issued, the application service layer will monitor the key metrics during the execution process in real time, such as inventory levels, transportation timeliness, production progress, etc. The monitoring data will be fed back into the decision-making model in real time. Based on the real-time feedback data, the execution algorithm model will make adaptive adjustments to ensure the continuous optimization of the supply chain operation. For example, if it is found that the result of a certain decision execution is not as expected, the system will automatically adjust the decision-making strategy.

[0051] The present invention provides a method for processing supply chain data of an Internet of Things platform based on artificial intelligence, having the following beneficial effects: The present invention deploys sensor units and intelligent terminals, constructs an Internet of Things information perception network for the supply chain and conducts digital feedback simulation, constructs a supply chain model and adjusts the model parameters through real-time information collection, obtains the actual situation of the supply chain process links and the item operation trajectory data and substitutes them into the model, uses a pre-trained artificial intelligence algorithm to deeply analyze the supply chain project characteristics, extracts the analysis characteristics of each analysis item and transmits them to the application service layer, adjusts the execution parameters to output supply chain optimization decisions, optimizes the supply chain process through real-time data perception and analysis, improves the overall efficiency, realizes intelligent decision-making based on the deep analysis of the AI model, reduces human errors and delays, optimizes each link of the supply chain, and solves the problem in the prior art that it is difficult to effectively process and analyze Internet of Things supply chain data to support intelligent decision-making.

[0052] Preferably, the steps of constructing the information perception level of the Internet of Things platform through a pre-deployed set of sensor units and supplementing the perception of the information perception level through intelligent terminals linked to the information perception level to obtain a supply chain perception network include: S11: Enter the information of the deployment location and preset function of each sensor unit in the pre-deployed set of sensor units respectively to obtain the perception node feature distribution of the sensor unit set; S12: Conduct supply chain process path analysis on each of the sensor units based on the perception node feature distribution of the sensor unit set to obtain the process path relationship between each of the sensor units; S13: Perform process direction connection processing between the perception nodes on the perception node feature distribution according to the process path relationship between each of the sensor units to obtain connection vectors for process path directions of each of the sensor units. Each of the sensor units and each of the connection vectors therebetween construct a supply chain perception path, and all the supply chain perception paths together form the information perception level of the Internet of Things platform corresponding to the supply chain; S14: Based on the information perception level, construct user interaction ports of several perception types. Receive data links from intelligent terminals held by various users through the user interaction ports of various perception types, obtain the information perception permissions of each of the intelligent terminals according to the data links of the intelligent terminals held by various users, and construct information perception nodes corresponding to each of the intelligent terminals based on the information perception permissions; S15: Substitute the information perception nodes of each of the intelligent terminals into the corresponding process positions of the corresponding supply chain perception paths in the information perception level to complete the construction of the supply chain perception network.

[0053] Specifically, input the deployment positions and preset functions of each sensor unit into the system. Each sensor unit has different perception functions (such as temperature, humidity, position, pressure, speed, etc.). These functions will be recorded in detail and associated with the specific geographical locations and facilities where they are deployed. Integrate the deployment positions and function information of all sensor units into the system to form the characteristic distribution of perception nodes. For example, sensors may be deployed at multiple key nodes such as warehouses, transport vehicles, production lines, etc. to achieve real-time monitoring of all links in the supply chain.

[0054] More specifically, by accurately inputting the position and function information of each sensor unit, the system can clearly map the spatial distribution of each perception node, providing an accurate basis for subsequent path analysis, data processing, and decision-making. By deploying multiple types of sensors, the system can comprehensively collect key data in all links of the supply chain, ensuring the diversity and comprehensiveness of data sources.

[0055] More specifically, analyze the supply chain process paths between these sensor units according to the functions and deployment positions of the sensor units. For example, in each link from raw material procurement to production, then to inventory management, transportation, distribution, etc., how to collect data through sensors and connect them into a complete information flow, determine the role and information flow path of each sensor unit in the supply chain, and clarify the data transfer relationship between different sensors. For example, a temperature sensor may jointly form a perception chain in the transportation link with a humidity sensor on the transportation path.

[0056] More specifically, based on the path analysis of sensor characteristics and deployment positions, an information flow relationship diagram between each node in the supply chain can be constructed, making the data interaction between each node clearer and more intuitive. The path analysis provides a basis for further supply chain optimization, can help identify bottlenecks or deficiencies in data circulation, and provides data support for subsequent decision-making.

[0057] More specifically, according to the process path relationship, the process direction connection between sensing nodes is processed. A "connection vector" is formed between each sensor unit and other sensors. These connection vectors characterize the relationship of data flowing from one node to another. On this basis, the process directions of each sensor unit are integrated into a complete supply chain sensing path, which reflects the data flow direction and interaction process from one node to another. Each sensing node forms a data flow direction through the connection vector, ensuring the coherence of the working processes of different sensors in the supply chain, avoiding information fragmentation and redundancy. The connection vector helps the system optimize the data transmission path, avoid repeated information transmission or unclear paths, and ensure the accuracy and efficiency of data transmission.

[0058] More specifically, all sensor units and their connection vectors are combined to construct a complete supply chain sensing path. These sensing paths represent the real-time monitoring and data transfer process of each link in the supply chain. The sensing network formed by each supply chain sensing path constitutes the information sensing layer of the Internet of Things platform, covering the monitoring and data collection of all links in the supply chain from raw material procurement, production and processing to logistics distribution. By constructing a complete sensing path and information layer, each link in the supply chain can achieve real-time data collection, monitoring and feedback through the Internet of Things platform. The construction of each sensing path makes each link in the supply chain process transparent and traceable, supporting the comprehensive visualization management of the supply chain.

[0059] More specifically, based on the information sensing layer, the Internet of Things platform will build multiple user interaction ports to support different types of users (such as managers, operators, customers, etc.) to interact with the supply chain system through intelligent terminals. Different user interaction ports will carry different sensing functions to meet the needs of different users. The system determines the type of sensing data that a user can receive according to the user's identity, role and permissions. For example, a warehouse manager can obtain inventory-related data, while a transportation dispatcher can obtain transportation route and status information. The user connects to the Internet of Things platform through an intelligent terminal. The system obtains the corresponding data link according to each user's identity and permissions and constructs the corresponding sensing node to provide personalized sensing services. The design of the user interaction port can meet the needs of different user groups, realize the personalized display and operation of data, and enhance the flexibility and adaptability of the platform. The construction of sensing nodes based on permission management ensures that different users can only access the sensitive data they are authorized to, guaranteeing the data security of the system.

[0060] More specifically, the smart terminal of each user is docked with the corresponding process node in the perception path through a data link to ensure that the perception data of each terminal can accurately reflect a certain link in the supply chain. By substituting the information perception nodes of the smart terminal into the information perception level, the construction of the entire supply chain perception network is completed. This process not only enables the Internet of Things platform to monitor each link in the supply chain in real time, but also enhances the user's participation and operation capabilities. The integration of smart terminals enables real-time monitoring of each link in the supply chain, improving the response speed and operation efficiency of the supply chain. After the user accesses the system through the smart terminal, they can obtain information in real time and give operation feedback, forming a dynamic and closed-loop perception and decision-making mechanism.

[0061] Preferably, the steps of performing digital feedback simulation on the supply chain perception network to obtain a supply chain perception network model and performing real-time information collection on the supply chain perception network to synchronously adjust the model parameters of the supply chain perception network model include: S21: Decompose the supply chain perception paths of the supply chain perception network, and perform digital feedback simulation on each of the supply chain perception paths in the supply chain perception network to obtain a path perception model for digital feedback simulation of each of the supply chain perception paths; S22: Perform coincidence positioning analysis on the path nodes of each of the supply chain perception paths to obtain the coincidence positioning nodes of each of the supply chain perception paths, and combine the path perception models of each of the supply chain perception paths based on the coincidence positioning nodes of each of the supply chain perception paths to obtain a supply chain perception network model for digital feedback simulation of the supply chain perception network; S23: Continuously perform real-time matching processing of the sensor units and smart terminals on the supply chain perception network to obtain the real-time information of the sensor units and smart terminals newly added to the supply chain perception network; S24: Correct the network situation of the supply chain perception network according to the real-time information of the sensor units and smart terminals newly added to the supply chain perception network, and set and connect the model nodes of the supply chain perception network model based on the results of the network situation correction, so that the supply chain perception network model follows the supply chain perception network for synchronous adjustment.

[0062] Specifically, the supply chain perception network needs to be disassembled into multiple independent perception paths. Each perception path represents the data flow process of a certain link in the supply chain (such as production, inventory, logistics, etc.). By disassembling the perception path into sub-paths, each path can be modeled and analyzed in detail separately to determine the perception objectives, data sources, data flows, etc. of each link. Disassembling the complex supply chain perception network into multiple perception paths can reduce the complexity of modeling, making it easier to implement the digital feedback simulation of each path. By disassembling the path, the system can process each perception path independently, improving the flexibility of management and optimization.

[0063] More specifically, for each disassembled supply chain perception path, a digital feedback simulation is carried out. This simulation process aims to predict the behavior and output of the perception path. For example, how to adjust the production schedule, inventory level, etc. in real time according to sensor data. Each perception path will establish a mathematical model based on the input real-time data to simulate the reaction of the path. For example, the change in environmental data collected by sensors will affect the adjustment of the production process, which in turn affects the final logistics scheduling. Through digital feedback simulation, the behavior of each perception path can be predicted and optimized more accurately, helping to identify potential supply chain bottlenecks or problems in advance. The simulation can provide data-driven decision support for decision-makers, such as production scheduling, inventory management, etc., to ensure the efficient operation of the supply chain.

[0064] More specifically, analyze the nodes between different perception paths in the supply chain to find the overlapping parts of the path nodes. Many supply chain links may share the same resources or information flows. Therefore, through coincidence location analysis, it can be determined which paths are cross-linked and which nodes are shared. Based on the analysis results, the overlapping path nodes can be merged or adjusted to avoid redundancy and ensure the efficient sharing of resources and information flows. Coincidence location analysis can help optimize the use of resources, avoid duplicate work between paths, reduce redundant data collection and processing. After merging the overlapping nodes, the supply chain perception path can be simplified, improving the efficiency and accuracy of data circulation.

[0065] More specifically, based on the aforementioned coincidence node location and combined with the model of each perception path, multiple path perception models are combined. Through this combination, an overall supply chain perception network model can be constructed, presenting the complete perception process of the entire supply chain from raw material procurement to delivery. Finally, a comprehensive supply chain perception network model is formed, reflecting the cooperation and interaction of each perception path. By combining the path perception models, the overall structure of the supply chain perception network and the interaction relationship between each link can be shown more clearly. The model combination improves the data integration degree and processing efficiency, facilitating managers to monitor and make decisions from a global perspective.

[0066] More specifically, each sensor unit in the supply chain perception network needs to be matched with the intelligent terminal in real time according to the actual situation. For example, newly added sensors may need to be paired with existing intelligent terminals for data transmission to ensure the timely collection and processing of information. Through real-time matching, the sensors start to collect data and transmit this data to the intelligent terminal. The intelligent terminal provides feedback and operations based on the real-time data. Through real-time matching, it is ensured that each newly added sensor and terminal can be immediately connected and start collecting relevant data, improving the real-time performance of the network. The system can flexibly adjust the information collection strategy and model parameters according to the actual data of the newly added devices, ensuring the accuracy of the perception network.

[0067] More specifically, according to the real-time information provided by the newly added sensor units and intelligent terminals, the existing supply chain perception network is corrected in real time. The system will be updated in real time according to the newly added data, correcting the perception path, node positions, and process sequences. The network real-time correction may involve adjusting the supply chain model parameters to ensure that the model better conforms to the current actual situation. After correction, the network model will reflect a more accurate supply chain state. The real-time correction ensures that the supply chain perception network can adapt to changes in the external environment in real time, such as changes in market demand, production delays, supply chain disruptions, etc. The network real-time correction enhances the adaptability and accuracy of the system, ensuring that the perception network reflects the latest actual operating conditions.

[0068] More specifically, after the network real-time correction, according to the correction results, the supply chain perception network model is synchronously adjusted. Specifically, adjustments need to be made to the model nodes and connection relationships to ensure that the model can reflect the real-time network state. According to the actual situation, the system may add or delete nodes in the model or adjust the connection relationships between nodes, making the model more truly reflect the information flow in the supply chain. Through synchronous adjustment, the supply chain perception network model always remains consistent with the operating state of the actual network, can accurately track the real-time changes in the supply chain. The synchronously adjusted model provides decision-makers with more accurate real-time data support to help them make better supply chain management decisions.

[0069] Preferably, the steps of obtaining the process link real-time perception information of the supply chain based on the supply chain perception network for information perception of the supply chain in terms of the real-time dimension of the supply chain process link include: S31: Based on each sensor unit and intelligent terminal of the supply chain perception network, corresponding information perception is carried out on each process link of the supply chain to obtain the process link real-time data of each process link of the supply chain at the current moment; S32: Retrieving corresponding data analysis standards according to the process links corresponding to the actual data of each process link, analyzing key features of the actual data of the process link, so as to obtain the actual perception information of the process links of each process link of each supply chain in the induction chain perception network.

[0070] Specifically, each sensor unit and intelligent terminal in the supply chain perception network are installed in each process link of the supply chain, such as production lines, warehouses, transportation, sorting, distribution and other links. Each sensor unit or intelligent terminal perceives the specific information of the process link according to its function. For example, the temperature sensor will collect ambient temperature data in real time, the position sensor will collect the location and transportation status of the items, and the RFID tag will collect inventory information. These sensing devices transmit real-time data to the central control system or cloud platform wirelessly or wiredly, and perceive the key data of each link in the supply chain in real time, ensuring that the data of each link is accurately recorded, and building a "digital twin" of the supply chain. Through intelligent terminals and sensors, it ensures real-time and comprehensive collection of on-site data from all links in the supply chain, greatly improving the efficiency and accuracy of data acquisition.

[0071] More specifically, after collecting real-time data from each process link, the system performs key feature analysis on the actual data of each link according to preset standards and process rules. These analysis standards include: production speed, inventory level, transportation status, environmental conditions, production capacity, delays, etc. Based on these standards, the system will analyze the real-time data of each process link and extract key features, such as: whether there is a production bottleneck, whether the inventory is excessive or insufficient, whether the transportation is proceeding as planned, etc.

[0072] More specifically, the most valuable information for decision-making is extracted from complex data, such as inventory turnover rate, order processing time, etc., and trend analysis is performed on the current link in combination with historical data to predict future changes. Through standard-based data analysis, the system can accurately identify key issues or potential risks in each link, such as inventory backlogs, transportation delays, etc. By conducting standardized analysis of real-time data in each process link, it provides managers with more specific and actionable decision support, helping them quickly identify links that need adjustment.

[0073] More specifically, the live data of each link is aggregated and combined with the feature analysis results to obtain the comprehensive perception information of each link in the supply chain. For example, through the real-time data of the production link, the actual production capacity can be obtained; through the real-time data of the logistics link, the timeliness of transportation can be obtained, etc. The live perception information of each link obtained can be displayed in the form of a visual dashboard or a data report for decision-makers and managers to refer to. The display content can include: process status, anomaly warning, efficiency analysis, etc. The live information of each process link can be transparently presented to the supply chain manager, reducing management blind spots and enabling real-time understanding of the actual operation status of the supply chain. Through real-time perception information, managers can timely discover potential problems and make adjustments. For example, when it is found that a certain link is lagging behind, the production rhythm or logistics route can be adjusted immediately to prevent the problem from spreading.

[0074] Preferably, the step of obtaining the item movement trajectory perception information of the supply chain based on the supply chain perception network for information perception of the supply chain in the dimension of the item movement trajectory of the supply chain includes: S33: Uniquely identify and label the item units participating in the operation of each process link of the supply chain, so that each item unit has a unit label for unique identification; S34: Statistically analyze the information of the item units available in each process link of the supply chain according to the supply chain perception network, and identify the unit labels of the statistical information to obtain the item unit list information of each process link of the supply chain; S35: Perform an association feature analysis of the time relationship and space relationship of the corresponding unit labels on the item unit list information of each process link of the supply chain to obtain the time association feature and space association feature of the item units with the same unit label in each process link of the supply chain; S36: Construct the movement trajectory of the item units with the same unit label based on the time association feature and the space association feature to obtain the item movement trajectory perception information of the item units with the same unit label.

[0075] Specifically, in each process link of the supply chain, all items participating in the operation are assigned a unique identification label, which can be in the form of an RFID tag, QR code, barcode, or other intelligent identification technologies, ensuring that each item can be uniquely identified in the supply chain. The label usually includes basic information of the item (such as item ID, production date, batch number, etc.) and dynamic logistics data (such as current location, transportation status, etc.). By assigning a unique label to each item, the problem of identification confusion of items in multiple links is avoided, ensuring that each item unit can be independently tracked. The unique identification label lays the foundation for subsequent tracking of the operation trajectory, enabling the flow of each item in the supply chain to be accurately recorded.

[0076] More specifically, in each link of the supply chain, the perception network will statistically count the item units participating in the operation in each link in real time. This statistic includes information such as the quantity, status, and location of items in each link. The unit label of each item (such as RFID or barcode) will be scanned or identified and associated with the corresponding item data to ensure that the status, location, and time point of each item in each link can be accurately recorded. By statistically counting and identifying the item units in each link in real time, the system can ensure that the item flow information in each link is not lost and is updated in real time. Through the statistical information of the item units, the transfer data of each link can be accurately obtained, further optimizing the supply chain operation.

[0077] More specifically, analyze the time attributes of the same item unit, record information such as the entry and exit time and residence time of the item in each link. Through these time characteristics, the time flow trajectory of the item in each link of the supply chain can be depicted. Analyze the flow path of the same item unit in space, record spatial characteristics such as the geographical location, transportation path, and storage location of the item in each link. Through these spatial characteristics, the spatial transfer trajectory of the item in each link can be drawn. The time and space relationship will be combined and analyzed to ensure that the flow of the item can be completely described in both the time and space dimensions. The analysis of the time and space relationship can help depict the complete dynamics of the item in each link of the supply chain, thus effectively mastering the transfer situation of the item and helping managers monitor in real time. By analyzing the flow of the item in time and space, process bottlenecks, stagnant links, and optimization spaces can be identified, providing more accurate path planning and time scheduling.

[0078] More specifically, based on the results of time and space analysis, the system will comprehensively integrate the data of each link of the item unit to construct a complete operation trajectory of the item, which includes the time flow and spatial path of the item from one link to another. The final operation trajectory perception information includes the complete flow path of the item, the residence time of each link, the transportation timeliness, etc. This information will be transmitted to the supply chain management system for decision-makers to refer to. Through in-depth analysis of the time and space relationship, the operation trajectory of each item can be accurately mapped to ensure that the flow state of each item can be traced. The operation trajectory can not only help decision-makers view the transfer path of the item in real time, but also provide customers with visual logistics information, enhancing the transparency of the supply chain. By obtaining the operation trajectory of the item in real time, supply chain managers can quickly discover potential problems, such as transportation delays, lost items, etc., and handle them in a timely manner.

[0079] Preferably, the steps of extracting the characteristics of the supply chain for the specified analysis items based on the supply chain perception network model to construct the supply chain item characteristic distribution corresponding to each specified analysis item include: S41: Obtain the project analysis information of the specified analysis items to be analyzed for the supply chain; wherein, the specified analysis items include production optimization analysis items, inventory optimization analysis items, and transportation optimization analysis items; S42: Extract the key information of production plan characteristics, sales status characteristics, and product operation characteristics from the supply chain process link perception information and item operation trajectory perception information in the supply chain perception network model according to the project analysis information of the production optimization analysis item, so as to obtain the production plan characteristics, sales status characteristics, and product operation characteristics of the supply chain; S43: Conduct a correlation analysis of the production plan effects of the production plan characteristics, sales status characteristics, and product operation characteristics of each supply chain in the supply chain perception network at each time node according to the project analysis information of the production optimization analysis item, and continuously combine the results of the correlation analysis to obtain the supply chain item characteristic distribution corresponding to the production optimization analysis item of the supply chain perception network; S44: Extract the key information of inventory status characteristics, inventory replenishment characteristics, and inventory consumption characteristics from the supply chain process link perception information and item operation trajectory perception information in the supply chain perception network model according to the project analysis information of the inventory optimization analysis item, so as to obtain the inventory status characteristics, inventory replenishment characteristics, and inventory consumption characteristics of the supply chain; S45: Conduct a correlation analysis of the inventory management effectiveness for the inventory status characteristics, inventory replenishment characteristics, and inventory consumption characteristics of each supply chain in the supply chain perception network at each time node according to the project analysis information of the inventory optimization analysis project, and continuously combine the results of the correlation analysis to obtain the supply chain project characteristic distribution of the supply chain perception network corresponding to the inventory optimization analysis project; S46: Extract the key information of the transportation path distribution characteristics, transportation efficiency characteristics, and transportation cost characteristics from the supply chain process link perception information and the item movement trajectory perception information in the supply chain perception network model according to the project analysis information of the transportation optimization analysis project, so as to obtain the transportation path distribution characteristics, transportation efficiency characteristics, and transportation cost characteristics of the supply chain; S47: Conduct a correlation analysis of the transportation path effectiveness for the transportation path distribution characteristics, transportation efficiency characteristics, and transportation cost characteristics of each supply chain in the supply chain perception network at each time node according to the project analysis information of the transportation optimization analysis project, and continuously combine the results of the correlation analysis to obtain the supply chain project characteristic distribution of the supply chain perception network corresponding to the transportation optimization analysis project.

[0080] Specifically, clarify the relevant data requirements for three specified analysis projects, namely production optimization analysis, inventory optimization analysis, and transportation optimization analysis. The project analysis information of these analysis projects includes optimization objectives, involved variables, required data types, etc. Specific projects: Production optimization analysis project: Focus on production plans, sales status, product movement, etc. Inventory optimization analysis project: Focus on inventory status, inventory replenishment and consumption. Transportation optimization analysis project: Focus on transportation paths, transportation efficiency, transportation costs, etc. After obtaining this analysis information, it can ensure the correct analysis direction, target the data extraction, avoid the interference of irrelevant data. The project analysis information provides a framework for subsequent analysis, ensuring that the information extracted from the perception network is accurate and effective.

[0081] More specifically, extract the planning information in the production process from the supply chain perception network model, including the formulation, adjustment, and execution of production plans, obtain sales data, including sales forecasts, actual sales data, market demands, etc. According to the perception information of the item movement trajectory, extract the flow characteristics of the product, such as the time stay of the item in each link, production batches, and transfer speeds. By extracting key characteristics such as production plans, sales status, and product movement from the perception network, the relationship between production plans and actual sales, and item flow can be accurately analyzed. These characteristics provide strong data support for production optimization decisions, helping supply chain managers formulate more reasonable production plans and adjust production strategies.

[0082] More specifically, in the production optimization analysis project, through the analysis of production plan characteristics, sales status characteristics, and product operation characteristics, the correlation analysis of various characteristics at time nodes is carried out. For example, the mutual influence between the production plan and the sales status is analyzed, as well as the impact of the product flow in different links on the implementation of the production plan. The results of the correlation analysis are continuously combined to form the characteristic distribution of the production optimization project, that is, based on the time series correlation analysis, the characteristic distribution is generated to show the optimization situation of the supply chain at different time nodes. The correlation analysis helps to identify the relationship between the production plan, sales, and product flow, provides accurate data support for production optimization, and helps to carry out more effective production resource scheduling. Through the characteristic distribution, production bottlenecks or links that do not meet expectations can be quickly discovered, so as to make timely adjustments.

[0083] More specifically, extraction of current inventory status characteristics: Extract information such as the quantity, distribution, warehousing situation, and item turnover rate of the current inventory. Extraction of inventory replenishment characteristics: Obtain data on inventory replenishment, including procurement cycle, inventory replenishment frequency, replenishment plan, etc. Extraction of inventory consumption characteristics: Through the item operation trajectory information of the perception network, extract the inventory consumption situation, and analyze the relationship between inventory consumption and production demand, sales situation. Through the extraction of these characteristics, clearly understand the current situation, replenishment, and consumption of inventory, so as to formulate targeted inventory optimization strategies, timely adjust the inventory replenishment strategy, optimize inventory turnover, reduce inventory backlog, and improve warehousing efficiency.

[0084] More specifically, based on the project analysis information of the inventory optimization analysis project, analyze the relationship between the current inventory status characteristics, inventory replenishment characteristics, and inventory consumption characteristics at each time node. Focus on analyzing the matching degree between inventory replenishment and inventory consumption, and the impact of the current inventory status on the overall inventory management strategy. Continuously combine the results of the correlation analysis to form the characteristic distribution of the inventory optimization project, and combine with the time series to show the changes in inventory management effects at different time nodes. The correlation analysis helps to reveal the balance between inventory replenishment and consumption, optimize the inventory management strategy, avoid excessive inventory backlog or inventory shortage, and can quickly adjust the inventory strategy according to the analysis results, improving the flexibility and response speed of the supply chain.

[0085] More specifically, according to the transportation data in the perception network, analyze the distribution of transportation routes, identify the most commonly used routes and transportation bottlenecks during transportation. Extraction of transportation efficiency characteristics: Obtain the efficiency data of each link in the transportation process, such as the time consumption, carrying capacity, and utilization rate of transportation resources in each transportation link. Extraction of transportation cost characteristics: Through the logistics data of the perception network, extract transportation cost characteristics, including the cost of each transportation link (such as transportation fees, warehousing fees, etc.).

[0086] More specifically, by extracting the distribution of transportation routes, transportation efficiency, and cost characteristics, the optimal routes and inefficient links can be identified, which helps optimize the transportation network, provides transportation cost analysis based on actual data, provides a basis for decision-makers to reduce transportation costs, and optimizes transportation expenditures in the supply chain.

[0087] More specifically, based on the project analysis information of the transportation optimization analysis project, analyze the relationships among the characteristics such as transportation route distribution, transportation efficiency, and transportation cost at each time node, focus on analyzing the impacts between route selection and transportation efficiency and cost, combine the analysis results to obtain the characteristic distribution of the transportation optimization project, and show the optimization effect of the transportation network. Through correlation analysis, the transportation routes can be optimized, bottlenecks in transportation can be reduced, and transportation efficiency can be improved. Through in-depth analysis of transportation efficiency and cost, guidance for controlling and reducing transportation costs can be provided, and the overall supply chain cost structure can be optimized.

[0088] Preferably, the steps of performing in-depth analysis of the corresponding specified analysis items on the characteristic distributions of each of the supply chain project features through an artificial intelligence algorithm model pre-trained in the data processing layer of the Internet of Things platform to obtain the project analysis features of each specified analysis item of the Internet of Things platform include: S61: Transmit the characteristic distributions of the supply chain projects of the production optimization analysis project, the inventory optimization analysis project, and the transportation optimization analysis project to the data processing layer of the Internet of Things platform, and retrieve the production optimization analysis model, inventory optimization analysis model, and transportation optimization analysis model that have been pre-trained and deployed in the data processing layer. Analyze the characteristic distributions of the supply chain projects of the production optimization analysis project, the inventory optimization analysis project, and the transportation optimization analysis project through the production optimization analysis model, the inventory optimization analysis model, and the transportation optimization analysis model respectively to obtain the corresponding project analysis features; Among them, the production optimization analysis model, inventory optimization analysis model, and transportation optimization analysis model are neural network convolution models. The neural network convolution models are trained through the training set, validation set, and test set divided from the pre-collected original data. The original data includes the characteristic distributions of the supply chain projects of the specified analysis items and the analysis feature annotations corresponding to the characteristic distributions of the supply chain projects. The analysis feature annotations correspond to the project analysis features output by the models.

[0089] Specifically, the supply chain item feature distributions extracted from the supply chain perception network model (including the feature distributions of production, inventory, and transportation optimization analysis items) are transmitted to the data processing layer of the Internet of Things platform to ensure that the format, accuracy, and timeliness of the data meet the predetermined requirements, and this data can be accepted and processed by the platform. Transmitting the supply chain feature data from different sources in a unified format to the Internet of Things platform helps subsequent unified processing and analysis. Through the real-time data transmission of the Internet of Things platform, it is ensured that the analysis data is up-to-date, providing timely decision support.

[0090] More specifically, three pre-trained deep learning models are retrieved from the data processing layer of the Internet of Things platform: a production optimization analysis model, which is used to analyze production-related features; an inventory optimization analysis model, which is used to analyze inventory-related features; and a transportation optimization analysis model, which is used to analyze transportation-related features. These three models are neural network convolutional models (CNNs). They extract features from the data through convolutional operations and conduct in-depth analysis. Through in-depth analysis of the data by the deep learning models, complex patterns and potential associations in the data can be discovered, which may be difficult to identify by traditional analysis methods. The invocation and automated analysis of the models can greatly improve the analysis efficiency and reduce manual intervention.

[0091] More specifically, through the invoked neural network convolutional model, in-depth analysis is carried out for each specified analysis item (production, inventory, transportation). The model comprehensively analyzes features such as production plans, sales status, and product operation to identify potential production bottlenecks, resource waste, or optimization opportunities. The model analyzes the current inventory status, the efficiency of inventory replenishment and consumption to help discover problems such as inventory backlogs, out-of-stock situations, and slow-moving products. The model conducts in-depth analysis on transportation routes, transportation efficiency, transportation costs, etc. to identify optimized routes and opportunities to reduce costs. In each analysis item, the model outputs the corresponding item analysis features by performing convolutional operations on the feature distribution. The neural network convolutional model can mine the deep relationships in the data through multi-level convolutional operations, providing more accurate analysis results than traditional methods. The model can quickly identify problems or optimization points in each supply chain link, such as potential bottlenecks in the production process, imbalances in inventory management, and inefficient links in transportation routes.

[0092] More specifically, relevant original data are collected in advance, including the distribution of supply chain project characteristics for specified analysis items and their corresponding analysis feature annotations. These data include: Distribution of supply chain project characteristics: namely, various characteristics of production, inventory, and transportation optimization; Analysis feature annotations: for each characteristic distribution, annotations are made to indicate which specific category or optimization feature this data belongs to (such as production bottlenecks, inventory shortages, transportation delays, etc.); Data division: The collected original data is divided into: a training set, which is the main data set for training the neural network model; a validation set, which is the data set for adjusting and optimizing model parameters; and a test set, which is an independent data set for evaluating model performance.

[0093] It can be understood that through precise annotation and scientific data division, the training effect and prediction ability of the neural network model can be ensured. The quality of the training data directly affects the accuracy of the analysis results. Through repeated training and optimization, the convolutional neural network can gradually improve the accuracy, thereby providing reliable decision-making support for supply chain optimization.

[0094] More specifically, after training is completed, the neural network convolution model can generate corresponding project analysis features based on the input distribution of supply chain project characteristics. These features include: Production optimization analysis features: for example, the accuracy of production plans, the utilization rate of production lines, production delays, etc.; Inventory optimization analysis features: for example, inventory turnover rate, inventory stockout probability, inventory backlog, etc.; Transportation optimization analysis features: for example, the optimality of transportation routes, transportation delays, transportation costs, etc. The generated project analysis features can be used for subsequent optimization decisions, scheduling plans, resource allocation, etc.

[0095] It can be understood that through in-depth analysis of the characteristics of each specified analysis item, clear and comprehensive analysis results can be provided for managers, helping to identify which links need improvement. The output of the project analysis features can form a dynamic feedback mechanism for continuous optimization and adjustment according to the changing supply chain data.

[0096] In a second aspect, the present invention provides an Internet of Things platform supply chain data processing system based on artificial intelligence for implementing the method for processing supply chain data of an Internet of Things platform based on artificial intelligence according to any one of the first aspects.

[0097] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for processing supply chain data on an Internet of Things platform based on artificial intelligence, characterized in that: include: The information perception layer of the Internet of Things platform is constructed through a set of pre-deployed sensor units, and the information perception layer is supplemented by intelligent terminals connected to the information perception layer to obtain a supply chain perception network; Constructing a supply chain perception network model corresponding to the supply chain perception network, and collecting real-time information on the supply chain perception network to synchronously adjust model parameters of the supply chain perception network model; Based on the supply chain perception network, the supply chain is perceived in terms of the actual situation of the supply chain process and the movement trajectory of the supply chain items, and the actual situation perception information of the supply chain process and the movement trajectory perception information of the items are obtained and substituted into the supply chain perception network model; Extracting features of designated analysis items of the supply chain based on the supply chain perception network model to construct supply chain item feature distribution corresponding to each designated analysis item; Performing in-depth analysis on the characteristic distribution of each supply chain item through the artificial intelligence algorithm model pre-trained in the data processing layer of the Internet of Things platform to obtain the analysis characteristics of each item; Substitute each of the project analysis features into the application service layer of the Internet of Things platform, so that the supply chain operation execution algorithm model in the application service layer outputs specific execution decisions of the supply chain according to each of the project analysis features.

2. The method for processing supply chain data of an Internet of Things platform based on artificial intelligence as claimed in claim 1, characterized in that: The steps of constructing an information perception layer of the Internet of Things platform through a set of pre-deployed sensor units, and supplementing the information perception layer through intelligent terminals linked to the information perception layer to obtain a supply chain perception network include: Entering information about the deployment location and preset function of each sensor unit in the pre-deployed sensor unit set, so as to obtain a characteristic distribution of sensing nodes of the sensor unit set; Based on the characteristic distribution of the sensing nodes of the sensor unit set, supply chain process path analysis is performed on each of the sensor units to obtain a process path relationship between each of the sensor units; According to the process path relationship between each of the sensor units, the process pointing connection processing between the sensing nodes is performed on the sensing node feature distribution to obtain the connection vector for the process path pointing of each of the sensor units. The sensor units and each of the connection vectors are used to construct a supply chain sensing path. The supply chain sensing paths together constitute the information sensing level of the Internet of Things platform corresponding to the supply chain. Based on the information perception level, several user interaction ports of perception types are constructed, data links from smart terminals held by various types of users are received through the user interaction ports of various types of perception, information perception rights of each smart terminal are obtained according to the data links of the smart terminals held by various types of users, and information perception nodes corresponding to each smart terminal are constructed based on the information perception rights; The information perception nodes of each of the smart terminals are substituted into the corresponding process positions of the corresponding supply chain perception paths in the information perception hierarchy to complete the construction of the supply chain perception network.

3. The method for processing supply chain data of an Internet of Things platform based on artificial intelligence as claimed in claim 2, characterized in that: The steps of constructing a supply chain perception network model corresponding to the supply chain perception network and collecting real-time information from the supply chain perception network to synchronously adjust model parameters of the supply chain perception network model include: Decomposing the supply chain perception network into supply chain perception paths, and performing digital feedback simulation on each supply chain perception path in the supply chain perception network, to obtain a path perception model for performing digital feedback simulation on each supply chain perception path; Performing a coincidence location analysis of the path nodes on each of the supply chain perception paths to obtain the coincidence location nodes of each of the supply chain perception paths, and combining each of the path perception models based on the coincidence location nodes of each of the supply chain perception paths to obtain a supply chain perception network model that performs digital feedback simulation on the supply chain perception network; Continuously performing real-time matching processing of sensor units and intelligent terminals on the supply chain perception network to obtain real-time information of sensor units and intelligent terminals newly added to the supply chain perception network; The supply chain perception network is corrected in real time based on the real-time information of the sensor units and smart terminals newly added to the supply chain perception network, and the model nodes of the supply chain perception network model are set and connected based on the results of the network correction, so that the supply chain perception network model can be adjusted synchronously with the supply chain perception network.

4. The method for processing supply chain data of an Internet of Things platform based on artificial intelligence as claimed in claim 1, characterized in that: The step of performing information perception of the real-time dimension of the supply chain process links based on the supply chain perception network to obtain the real-time perception information of the process links of the supply chain includes: Based on the supply chain perception network, each sensor unit and the intelligent terminal perceive the corresponding information of each process link of the supply chain to obtain the real-time data of each process link of the supply chain at the current moment; According to the process links corresponding to the actual data of each process link, the corresponding data analysis standards are retrieved, and the key features of the actual data of the process links are analyzed to obtain the actual perception information of the process links of each process link of each supply chain in the supply chain perception network model.

5. The method for processing supply chain data of an Internet of Things platform based on artificial intelligence as claimed in claim 1, characterized in that: The step of performing information perception of the supply chain item movement track dimension on the supply chain based on the supply chain perception network to obtain the perceived information of the item movement track of the supply chain includes: Uniquely identify the unit items involved in the operation of each process link of the supply chain, so that each unit item has a unit label for unique identification; Collecting information about the item units in each process link of the supply chain according to the supply chain perception network, and identifying unit labels of the collected information to obtain item unit list information in each process link of the supply chain; Performing correlation feature analysis of the time relationship and space relationship of the corresponding unit tags on the item unit list information of each process link of the supply chain, so as to obtain the time correlation feature and space correlation feature of the item unit with the same unit tag in each process link of the supply chain; Based on the time correlation feature and the space correlation feature, the operation trajectory of the item unit with the same unit tag is constructed to obtain the item operation trajectory perception information of the item unit with the same unit tag.

6. The method for processing supply chain data of an Internet of Things platform based on artificial intelligence as claimed in claim 1, characterized in that: The steps of extracting features of designated analysis items of the supply chain based on the supply chain perception network model to construct supply chain item feature distribution corresponding to each designated analysis item include: Obtaining project analysis information of designated analysis projects to be analyzed on the supply chain; wherein the designated analysis projects include production optimization analysis projects, inventory optimization analysis projects, and transportation optimization analysis projects; Extract key information of production plan characteristics, sales status characteristics, and product operation characteristics from the supply chain process link perception information and the item operation trajectory perception information in the supply chain perception network model according to the project analysis information of the production optimization analysis project, so as to obtain the production plan characteristics, sales status characteristics, and product operation characteristics of the supply chain; According to the project parsing information of the production optimization analysis project, a correlation analysis of the production plan effect is performed on the production plan characteristics, sales status characteristics and product operation characteristics of each supply chain of the supply chain perception network at each time node, and the results of the correlation analysis are continuously combined to obtain the supply chain project characteristic distribution of the supply chain perception network corresponding to the production optimization analysis project; Extract key information of inventory status characteristics, inventory replenishment characteristics and inventory consumption characteristics from the supply chain process link perception information and the item movement trajectory perception information in the supply chain perception network model according to the project parsing information of the inventory optimization analysis project, so as to obtain the inventory status characteristics, inventory replenishment characteristics and inventory consumption characteristics of the supply chain; According to the project parsing information of the inventory optimization analysis project, a correlation analysis of inventory management effects is performed on the inventory status characteristics, inventory replenishment characteristics and inventory consumption characteristics of each supply chain of the supply chain perception network at each time node, and the results of the correlation analysis are continuously combined to obtain the supply chain project characteristic distribution of the supply chain perception network corresponding to the inventory optimization analysis project; Extract key information of transportation path distribution characteristics, transportation efficiency characteristics, and transportation cost characteristics from the supply chain process link perception information and the item movement trajectory perception information in the supply chain perception network model according to the project analysis information of the transportation optimization analysis project, so as to obtain the transportation path distribution characteristics, transportation efficiency characteristics, and transportation cost characteristics of the supply chain; According to the project analysis information of the transportation optimization analysis project, a correlation analysis of the transportation path effects is performed on the transportation path distribution characteristics, transportation efficiency characteristics, and transportation cost characteristics of each supply chain of the supply chain perception network at each time node, and the results of the correlation analysis are continuously combined to obtain the supply chain project characteristic distribution of the supply chain perception network corresponding to the transportation optimization analysis project.

7. The method for processing supply chain data of an Internet of Things platform based on artificial intelligence as claimed in claim 6, characterized in that: The steps of performing in-depth analysis on the characteristic distribution of each supply chain item through the artificial intelligence algorithm model pre-trained in the data processing layer of the Internet of Things platform to obtain the analysis characteristics of each item include: The supply chain project feature distribution of the production optimization analysis project, the inventory optimization analysis project and the transportation optimization analysis project is transmitted to the data processing layer of the Internet of Things platform, and the production optimization analysis model, the inventory optimization analysis model and the transportation optimization analysis model that have been pre-trained and deployed in the data processing layer are retrieved, and the supply chain project feature distribution of the production optimization analysis project, the inventory optimization analysis project and the transportation optimization analysis project are analyzed respectively by the production optimization analysis model, the inventory optimization analysis model and the transportation optimization analysis model to obtain corresponding project analysis features; Among them, the production optimization analysis model, inventory optimization analysis model and transportation optimization analysis model are neural network convolution models, and the neural network convolution model completes model training by dividing the pre-collected original data into a training set, a validation set and a test set. The original data includes the supply chain project feature distribution of the specified analysis project and the analysis feature annotations corresponding to the supply chain project feature distribution. The analysis feature annotations correspond to the project analysis features output by the model.

8. The method for processing supply chain data of an Internet of Things platform based on artificial intelligence as claimed in claim 1, characterized in that: The steps of substituting each of the project analysis features into the application service layer of the Internet of Things platform, and making the supply chain operation execution algorithm model in the application service layer output a specific execution decision of the supply chain according to each of the project analysis features include: Performing format conversion and data integration on the analysis features of each of the projects, so as to substitute them into the supply chain operation execution algorithm model deployed in the application service layer of the Internet of Things platform; Obtain the supply chain decision tendency required by the user, and perform parameter configuration and weight adjustment of the model parameters corresponding to each of the project analysis features on the supply chain operation execution algorithm model based on the supply chain decision tendency, so as to evaluate the model adaptability of several preset decision templates through the supply chain operation execution algorithm model after parameter configuration and weight adjustment, and determine the decision template that is most adaptable to the project analysis features at the current moment according to the evaluation results; The project analysis features are analyzed for specific execution steps according to the decision template to obtain a specific execution decision for the supply chain.

9. An artificial intelligence-based Internet of Things platform supply chain data processing system, characterized in that: Used to implement an artificial intelligence-based Internet of Things platform supply chain data processing method as described in any one of claims 1-8.

Images