Production manufacturing management method and system based on digital twinning

Abstract

The invention discloses a production and manufacturing management method and system based on digital twinning, which are used in the field of production and manufacturing management, and the method comprises the following steps: collecting logistics management data in logistics storage and operation data of a production line, and constructing a logistics digital twinning processing model; simulating a logistics management process based on a logistics digital twin processing model, and tracking the positions of goods and transportation tools in real time by using a positioning and identification technology; utilizing a demand prediction model to predict a material demand and a product demand based on the data in the logistics digital twinborn model updated in real time, and adjusting the stock of logistics storage according to a prediction result; and calculating an optimal transportation path by using a path optimization algorithm, making a production and manufacturing plan of logistics storage, and monitoring the production and manufacturing process in real time. According to the bidirectional fast random search tree, a new path can be quickly recalculated, and the accuracy and adaptability of decision making are improved by comprehensively considering various transportation and operation constraints.

Description

Technical Field

[0001] The present invention relates to the field of production and manufacturing management. Specifically, it particularly relates to a production and manufacturing management method and system based on digital twin. Background Art

[0002] Logistics warehousing refers to facilities used for storing and protecting goods. Goods are temporarily stored at a certain point in the supply chain until they are transported to their final destination or further processed. Logistics warehousing usually includes warehouses, distribution centers, and other types of storage facilities. They are an important part of supply chain management and are responsible for ensuring the smooth flow of products between production and consumption.

[0003] Digital twin technology, also known as digital mirror or virtual copy, is a technology that creates digital copies of physical objects, systems, or processes in the real world for simulation, analysis, and optimization in a virtual environment. In production and manufacturing management, digital twin technology can create virtual copies of factories, production lines, or individual machines, enabling managers to monitor, analyze, and optimize the actual production process in a virtual environment.

[0004] With the rise of Industry 4.0, the manufacturing industry is developing towards a highly automated, networked, and intelligent direction. Digital twin technology is one of the key technologies for realizing intelligent manufacturing. It combines the physical world with the virtual world to help enterprises achieve precise control and optimization of the production process. Digital twin technology relies on a large amount of real-time data support, which comes from various sensors, devices, and systems. The development of the Internet of Things provides a powerful data foundation for digital twin, making real-time monitoring and analysis possible.

[0005] Traditional logistics systems often fail to react to market changes and supply chain status in real time due to untimely data updates, making it difficult to effectively respond to rapidly changing market demands and supply conditions. Previous demand forecasting methods rely on historical data and simple statistical models, often resulting in large deviations between the forecast results and actual demands due to the lack of real-time data support and the application of advanced algorithms, increasing the risk of inventory overflow or shortage. Traditional logistics path planning often fails to optimize transportation costs and time, resulting in resource waste and high operating costs. Moreover, safety measures and risk management are often passive, relying on manual monitoring and empirical judgment, lacking data analysis and prevention mechanisms. In complex supply chain management, traditional decision support systems often cannot comprehensively integrate and analyze multi-source data, resulting in a lack of data-driven in-depth support in the decision-making process, affecting the efficiency and accuracy of decision-making.

[0006] Regarding the problems in the related technologies, no effective solutions have been proposed yet. Summary of the Invention

[0007] To overcome the above problems, the present invention aims to provide a production manufacturing management method and system based on digital twin, aiming to solve the problems that traditional logistics path planning often fails to optimize transportation costs and time, resulting in waste of resources and high operating costs, and safety measures and risk management are often passive, relying on manual monitoring and empirical judgment, lacking data analysis and prevention mechanisms. In complex supply chain management, traditional decision support systems often cannot comprehensively integrate and analyze multi-source data, resulting in a lack of data-driven in-depth support in the decision-making process, affecting the efficiency and accuracy of decision-making.

[0008] Therefore, the specific technical solutions adopted by the present invention are as follows:

[0009] According to one aspect of the present invention, there is provided a production manufacturing management method based on digital twin, and the production manufacturing management method includes the following steps:

[0010] S1. Collect logistics management data in the logistics warehouse and operation data of the production line, and construct a logistics digital twin processing model based on the logistics management data and the operation data of the production line;

[0011] S2. Simulate the logistics management process based on the logistics digital twin processing model, and use positioning and identification technologies to track the positions of goods and transportation tools in real time, and at the same time update the digital twin model in real time to reflect the real-time status of the logistics warehouse and the production line;

[0012] S3. Use a demand forecasting model to predict material requirements and product requirements based on the data in the logistics digital twin model updated in real time, and adjust the inventory level of the logistics warehouse according to the prediction results;

[0013] S4. Based on the adjusted inventory level and predicted demand, use a path optimization algorithm to calculate the optimal transportation path, and calculate eigenvalue and demand value during the path optimization process, formulate a production manufacturing plan for the logistics warehouse based on the eigenvalue and demand value, and monitor the production manufacturing process in real time;

[0014] Among them, based on the adjusted inventory level and predicted demand, using a path optimization algorithm to calculate the optimal transportation path, and calculating eigenvalue and demand value during the path optimization process, formulating a production manufacturing plan for the logistics warehouse based on the eigenvalue and demand value, and monitoring the production manufacturing process in real time includes the following steps:

[0015] S41. Initialize the search space parameters of the rapidly-exploring random tree, and set the starting node and the target node, construct a bidirectional rapidly-exploring random tree based on the starting node and the target node, and search for the path between the starting node and the target node by the random sampling method;

[0016] S42. Evaluate the transportation time and cost of the candidate paths generated by the bidirectional fast random search tree, and analyze the impact of each path on the customer satisfaction demand value;

[0017] S43. During the path optimization process, calculate the eigenvalue and demand value of each candidate path according to the real-time updated inventory and forecast demand;

[0018] S44. Considering comprehensively the eigenvalue and demand value of all candidate paths and the real-time updated inventory, formulate the production and manufacturing plan for the logistics warehouse;

[0019] S45. Monitor the production progress and logistics status of the goods in real time to respond to any changes in the supply chain and adjust the production and transportation plans in a timely manner.

[0020] Optionally, simulate the logistics management process based on the logistics digital twin processing model, and use the positioning and identification technology to track the positions of the goods and transportation tools in real time. At the same time, update the digital twin model in real time to reflect the real-time status of the logistics warehouse and the production line, including the following steps:

[0021] S21. Use the two-dimensional code technology to uniquely identify the goods in the logistics warehouse to ensure that each good has a unique identification code;

[0022] S22. Collect the logistics data in the logistics warehouse in real time, and use the correlation coefficient matching algorithm to process the real-time collected logistics data, and extract the position and status information of the goods and transportation tools in the logistics data;

[0023] S23. Associate the identified position with the unique identification of the goods, establish the correspondence between the goods and the logistics data, and at the same time associate the position and status data of the transportation tool with the preset transportation task and goods;

[0024] S24. Integrate the associated goods position and the status data of the transportation tool into the digital twin model, and update the logistics digital twin processing model according to the integrated data.

[0025] Optionally, collect the logistics data in the logistics warehouse in real time, and use the correlation coefficient matching algorithm to process the real-time collected logistics data, and extract the position and status information of the goods and transportation tools in the logistics data, including the following steps:

[0026] S221. In the logistics warehouse environment, collect the scenario data including goods and transportation tools in real time;

[0027] S222. Select the area potentially containing the target features in the collected scenario data, and use the two-dimensional array window sliding technology to calculate the normalized correlation coefficient of the feature values within the window;

[0028] S223. Compare the calculated normalized correlation coefficient with the threshold value of the preset sequential similarity detection algorithm. If the normalized correlation coefficient exceeds the threshold value, it indicates that the current position contains the features of the target goods or transportation vehicle;

[0029] S224. Implement an interval point calculation strategy between the matched feature points and skip the feature points at the positions where the normalized correlation coefficient exceeds the threshold value;

[0030] S225. Extract the position and status information of the goods and transportation vehicle from the scene area according to the comparison result.

[0031] Optionally, select an area potentially containing the target features from the collected scene data, and calculate the normalized correlation coefficient of the feature values within the window using the two-dimensional array window sliding technique, including the following steps:

[0032] S2221. Screen the areas containing the target goods or transportation vehicle from the large amount of scene data collected in real time;

[0033] S2222. Set a two-dimensional array window for the selected area and define the size and shape of the block for calculating the normalized correlation coefficient;

[0034] S2223. Apply the window sliding technique to move the window from one end of the area to the other end;

[0035] S2224. Calculate the normalized correlation coefficient for each area covered by the window.

[0036] Optionally, use a demand prediction model to predict the material demand and product demand based on the data in the real-time updated logistics digital twin model, and adjust the inventory level of the logistics warehouse according to the prediction result, including the following steps:

[0037] S31. Collect historical data related to the material demand and product demand, and divide the historical data into a training set, a validation set, and a test set;

[0038] S32. Build the architecture of the demand prediction model based on the characteristics of the historical data and the predicted demand;

[0039] S33. Use the training set data to train the demand prediction model (BP neural network model). During the training process, optimize the weights and biases of the demand prediction model through the backpropagation algorithm;

[0040] S34. After the model training is completed, use the test set data to evaluate the prediction accuracy of the demand prediction model;

[0041] S35. If the prediction accuracy of the test set does not meet the preset requirements, adjust the hyperparameters (including the learning rate, the number of neurons in the hidden layer, etc.) of the demand prediction model according to its performance on the validation set;

[0042] S36. Integrate real-time data into the logistics digital twin model, utilize the trained BP neural network model, and predict the material demand and product demand based on the latest data in the logistics digital twin model. Meanwhile, adjust the inventory level according to the prediction results.

[0043] Optionally, train the demand prediction model using the training set data. During the training process, optimize the weights and biases of the demand prediction model through the backpropagation algorithm, including the following steps:

[0044] S331. Initialize the weights and biases of the initial demand prediction model;

[0045] S332. Input the data of the training set into the demand prediction model. After being received by the input layer of the demand prediction model, the data is passed to the hidden layer;

[0046] S333. Apply the activation function in the hidden layer, calculate the difference between the prediction result and the actual label in the output layer, and quantify the difference using the loss function;

[0047] S334. Use the backpropagation algorithm to update the weights and biases according to the result of the loss function.

[0048] Optionally, initialize the search space parameters of the rapidly-exploring random tree (RRT), set the start node and the goal node, construct a bidirectional rapidly-exploring random tree based on the start node and the goal node, and search for the path between the start node and the goal node through the random sampling method, including the following steps:

[0049] S411. Determine the search space boundary of the rapidly-exploring random tree according to the geographical scope of the supply chain network;

[0050] S412. Set the parameter data of the algorithm based on the search space boundary;

[0051] S413. Clearly set the positions of the start node and the goal node within the search space based on the parameter data;

[0052] S414. Start constructing rapidly-exploring random trees from the start node and the goal node respectively. Use random sampling to expand each tree, and try to connect the nearest node of the current tree to the current node;

[0053] S415. During the process of expanding the rapidly-exploring random tree, take into account the constraints including transportation cost, time window, and vehicle transportation capacity;

[0054] S416. As the two quick random search trees grow, monitor the growth of the quick random search trees and find the connection points between the two quick random search trees;

[0055] S417. When the two quick random search trees meet, form a complete path and check whether the complete path meets the constraint conditions including transportation cost, time window, and transportation capacity of the vehicle. If not, continue the search until the optimal path is found.

[0056] Optionally, during the process of path optimization, calculating the eigenvalue and demand value of each candidate path according to the real-time updated inventory and predicted demand includes the following steps:

[0057] S431. Obtain the current inventory data and predicted demand;

[0058] S432. Calculate the transportation time according to the distance of the candidate path and the road speed limit;

[0059] S433. Calculate the transportation cost based on the transportation distance, type of transportation tool, and fuel cost;

[0060] S434. Use the transportation time as the time eigenvalue of the path;

[0061] S435. Use the transportation cost as the cost eigenvalue of the path;

[0062] S436. By measuring the impact of transportation time on customer satisfaction, calculate the satisfaction score using a linear or non-linear model;

[0063] S437. Considering comprehensively the transportation time and cost eigenvalue, customer satisfaction, and delivery time demand value, score each path.

[0064] Optionally, the formula for scoring each path is:

[0065] E = w t ·T + + w c ·C + w s ·S;

[0066] In the formula, T represents the transportation time eigenvalue;

[0067] C represents the transportation cost eigenvalue;

[0068] S represents the customer satisfaction score;

[0069] w t represents the weight coefficient of the transportation time;

[0070] w c represents the weight coefficient of the transportation cost;

[0071] ws The weight coefficient representing customer satisfaction.

[0072] According to another aspect of the present invention, there is also provided a production manufacturing management system based on digital twin, which system includes: a model construction module, a model simulation module, a demand prediction and inventory adjustment module, and a path optimization and management module;

[0073] The model construction module, the model simulation module, the demand prediction and inventory adjustment module, and the path optimization and management module are sequentially connected.

[0074] The model construction module is used to collect logistics management data in the logistics warehouse and operation data of the production line, and construct a logistics digital twin processing model based on the logistics management data and the operation data of the production line.

[0075] The model simulation module is used to simulate the logistics management process based on the logistics digital twin processing model, and use positioning and identification technologies to track the positions of goods and transportation tools in real time, and at the same time update the digital twin model in real time to reflect the real-time states of the logistics warehouse and the production line.

[0076] The demand prediction and inventory adjustment module is used to utilize a demand prediction model, based on the data in the logistics digital twin model updated in real time, predict material demand and product demand, and adjust the inventory level in the logistics warehouse according to the prediction results.

[0077] The path optimization and management module is used to calculate the optimal transportation path based on the adjusted inventory level and predicted demand by using a path optimization algorithm, and calculate eigenvalue and demand value during the path optimization process, formulate a production manufacturing plan for the logistics warehouse based on the eigenvalue and demand value, and at the same time monitor the production manufacturing process in real time.

[0078] Compared with the prior art, the present application has the following beneficial effects:

[0079] 1. Through real-time data collection and synchronous update of the digital twin model in the present invention, an enterprise can monitor the accurate positions and states of each piece of goods and transportation tool in real time. The digital twin model provides a highly accurate virtual representation, which can simulate operations and scenarios in the real physical environment. By using algorithms to analyze and process real-time data, it is possible to more effectively plan logistics paths and transportation plans, reduce transportation costs, shorten delivery times. The real-time update and precise logistics management allow an enterprise to provide more reliable delivery time predictions, enhance customer trust. Through comprehensive monitoring and high-level control of the logistics system, it is possible to better identify potential safety problems and risks, take measures in a timely manner to prevent accidents from occurring, and ensure the safety of personnel and goods.

[0080] 2. Based on the data in the real-time updated logistics digital twin model, the demand forecasting model can more accurately predict the future demand for materials and products. Through real-time data integration and demand forecasting, enterprises can quickly respond to market changes, timely adjust inventory strategies, meet consumer demands, enhance market competitiveness. By using historical data and BP neural network models, the accuracy of demand forecasting can be improved, and inventory management problems caused by inaccurate forecasting can be reduced. By continuously training and validating the model, enterprises can continuously improve the accuracy of demand forecasting and adapt to the changing market trends and consumer behaviors. The demand forecasting model provides strong data support for management, helping them make more informed inventory management and supply chain decisions, reducing subjective speculation and uncertainty.

[0081] 3. By constructing two search trees simultaneously from the starting point and the target point, the bidirectional rapidly-exploring random tree (RRT) can find an effective path faster than the unidirectional RRT, especially in complex or high-dimensional search spaces. The bidirectional RRT allows path search considering multiple constraints, can effectively balance transportation costs and time, and find the most cost-effective logistics path. Since the RRT algorithm can handle complex obstacles and irregular terrains, it is particularly suitable for supply chain networks with complex geographical features, such as urban traffic networks, mountain roads, etc. Under the influence of changing environmental conditions or emergencies, the bidirectional RRT can quickly recalculate a new path to ensure the continuity and efficiency of logistics. By comprehensively considering various transportation and operation constraints, the bidirectional RRT supports a data-driven complex decision-making process, improving the accuracy and adaptability of decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] With the following description of the embodiments, the above characteristics, features, and advantages of the present invention and their implementation manners and methods become more understandable. The embodiments are described in detail in conjunction with the drawings. Shown herein in schematic diagrams:

[0083] Figure 1 is a flowchart of a digital-twin-based production and manufacturing management method according to an embodiment of the present invention;

[0084] Figure 2 is a schematic block diagram of a digital-twin-based production and manufacturing management system according to an embodiment of the present invention;

[0085] Figure 3 is a schematic structural diagram of a computer device according to an embodiment of the present invention.

[0086] In the figure:

[0087] 1. Model construction module; 2. Model simulation module; 3. Demand forecasting and inventory adjustment module; 4. Path optimization and management module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0088] To enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0089] According to an embodiment of the present invention, a production manufacturing management method and system based on digital twin are provided.

[0090] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. As Figure 1 shown, according to an embodiment of the present invention, a production manufacturing management method based on digital twin is provided. The production manufacturing management method includes the following steps:

[0091] S1. Collect the logistics management data in the logistics warehouse and the operation data of the production line, and construct a logistics digital twin processing model based on the logistics management data and the operation data of the production line.

[0092] It should be explained that devices such as sensors, RFID tags, and GPS trackers are installed to collect information such as the position, moving speed, and storage conditions (such as temperature and humidity) of goods in the warehouse. In addition, inventory management data such as inventory levels and inbound / outbound records are collected, and sensors are installed on the production line to monitor equipment status (such as temperature and vibration), production rate, fault records, etc.

[0093] Clean the collected data to remove errors and outliers, format the data to ensure that all data follows a unified standard and format, and integrate the data to merge data from different sources and formats into a unified data warehouse.

[0094] Select a suitable digital twin modeling platform or software, such as Siemens MindSphere, PTC ThingWorx, etc., determine the structure of the digital twin model, including the digital representation of physical entities, the interaction relationships between entities, and the influence of the external environment, map the preprocessed data into the digital twin model to ensure that the model can accurately reflect the state of the actual physical system, enable the digital twin model to simulate the dynamic behavior of the actual physical system through algorithms and simulation technologies, verify the digital twin model using historical data to ensure the accuracy of the model, and adjust the model parameters according to the verification results to optimize the model performance.

[0095] S2. Simulate the logistics management process based on the logistics digital twin processing model, and use positioning and identification technologies to track the positions of goods and transportation tools in real time. At the same time, update the digital twin model in real time to reflect the real-time status of the logistics warehouse and the production line.

[0096] Preferably, simulating the logistics management process based on the logistics digital twin processing model, and using positioning and identification technologies to track the positions of goods and transportation tools in real time, while updating the digital twin model in real time to reflect the real-time status of the logistics warehouse and the production line includes the following steps:

[0097] S21. Use two-dimensional code technology to uniquely identify the goods in the logistics warehouse to ensure that each good has a unique identification code.

[0098] It should be noted that professional two-dimensional code generation tools or systems, such as Adobe Illustrator plugins, online two-dimensional code generators, etc., are used to design the two-dimensional code content, which usually includes information such as the ID, name, warehousing date, batch number of the goods. According to the designed content, generate the two-dimensional code to ensure that the generated two-dimensional code has sufficient data redundancy and error correction capabilities to adapt to various challenges in the warehouse environment (such as dust, moisture, etc.). Considering the warehouse environment, select durable label materials that can adapt to different environmental conditions, such as waterproof and high-temperature resistant materials. Use a high-quality printer (such as a thermal transfer printer) to print the two-dimensional code to ensure the clarity and scanning and recognition efficiency of the two-dimensional code. Ensure that the surface of the goods is clean, flat, without oil stains or dust so that the label can be firmly pasted. Precisely attach the printed two-dimensional code label to the specified position of the goods, and the position should be easy to scan, usually on the side or above of the goods.

[0099] S22. Real-time collect the logistics data in the logistics warehouse, and use the correlation coefficient matching algorithm to process the real-time collected logistics data to extract the position and status information of the goods and transportation tools in the logistics data.

[0100] Preferably, real-time collecting the logistics data in the logistics warehouse, and using the correlation coefficient matching algorithm to process the real-time collected logistics data to extract the position and status information of the goods and transportation tools in the logistics data includes the following steps:

[0101] S221. In the logistics warehouse environment, real-time collect the scene data including goods and transportation tools;

[0102] S222. Select the area potentially containing the target features in the collected scene data, and use the two-dimensional array window sliding technology to calculate the normalized correlation coefficient of the feature values within the window;

[0103] S223. Compare the calculated normalized correlation coefficient with the threshold value of the preset sequential similarity detection algorithm. If the normalized correlation coefficient exceeds the threshold value, it indicates that the current position contains the characteristics of the target goods or transportation vehicle.

[0104] S224. Implement an interval point calculation strategy between the matched feature points and skip the feature points at the positions where the normalized correlation coefficient exceeds the threshold value.

[0105] S225. Extract the position and status information of the goods and transportation vehicle from the scene area according to the comparison result.

[0106] It should be noted that various sensors (such as RFID readers, cameras, GPS trackers) are installed on the warehouse and transportation vehicle to collect data such as the position of goods, transportation status, and environmental conditions in real time. The collected data is transmitted to the data processing center in real time through a wireless network to ensure the timeliness and integrity of the data.

[0107] Determine the window size for analyzing the data. This usually depends on the resolution of the data and the expected size of the target object. By sliding the window in the data array, calculate the normalized correlation coefficient of the data in each window, locate the features in the data that match the preset pattern in the data, use the normalized correlation coefficient to measure the similarity between the data in the window and the predefined template, which helps to identify the target features (such as goods or transportation vehicle labels) in the data. Determine a threshold value based on historical data and experiments for identifying valid matches. Compare the normalized correlation coefficient of each window with the threshold value. If the correlation coefficient exceeds the threshold value, it is considered that this position may contain the target feature. After determining the matching points, adopt an interval point calculation strategy to ignore adjacent high-correlation points to avoid duplicate counting and optimize the processing efficiency. For continuous high-correlation coefficient points, only process the first point or process at a set interval to reduce unnecessary computational load. Connect the identified feature point position information with the actual logistics management system, extract the detailed position and status information of the corresponding goods or transportation vehicle, and update the goods and transportation vehicle information in the logistics management system to achieve real-time asset tracking and management.

[0108] Preferably, select the area potentially containing the target feature from the collected scene data, and calculate the normalized correlation coefficient of the feature values in the window using the two-dimensional array window sliding technology, including the following steps:

[0109] S2221. Screen the area containing the target goods or transportation vehicle from the large amount of scene data collected in real time.

[0110] S2222. Set a two-dimensional array window for the selected area and define the size and shape of the block for calculating the normalized correlation coefficient.

[0111] S2223. Apply window sliding technology to move the window from one end of the area to the other end;

[0112] S2224. For each area covered by the window, calculate the normalized correlation coefficient.

[0113] It should be noted that first, use preliminary screening conditions (such as basic features like color, shape, size, etc.) to screen out areas that may contain the target goods or transportation tools from the real-time collected scene data, and further analyze these pre-screened areas to confirm whether they actually contain the target object. A trained model can be used to identify specific cargo labels or characteristics of transportation tools, determine the size and shape of the window, which is usually based on the expected size and shape of the target object. The window size needs to be large enough to cover the target features to ensure effective calculation. Design the window to maximize the capture of target features, such as adjusting the width and height of the window to adapt to different types of goods and transportation tools. The window starts from one end and gradually moves to the other end, moving a certain number of pixels (step size) each time. This step size can be adjusted as needed to balance the speed and accuracy of the analysis. During the window sliding process, continuous data capture and processing are carried out to ensure that no potential matching areas are missed. For each area covered by the window, use the normalized correlation coefficient calculation method to measure the similarity between this area and a predefined template (such as a feature template of a specific cargo or transportation tool).

[0114] The normalized correlation coefficient is calculated by the following formula:

[0115]

[0116] In the formula, a i represents the observed value at the i-th position in the first dataset;

[0117] represents the average value of the observed values in the first dataset;

[0118] b i represents the observed value at the i-th position in the second dataset;

[0119] represents the average value of the observed values in the second dataset;

[0120] n represents the total number of data points in the dataset.

[0121] S23. Associate the identified position with the unique identifier of the goods to establish the correspondence between the goods and the logistics data. At the same time, associate the position and status data of the transportation tool with the preset transportation tasks and goods.

[0122] S24. Integrate the associated cargo location and the status data of the transportation vehicle into the digital twin model, and update the logistics digital twin processing model according to the integrated data.

[0123] It should be noted that the cargo location data identified in step S22 is associated with the QR code identification code generated in step S21, which is usually implemented through a database management system (such as an SQL database). The cargo location data and the QR code identification code are stored as related fields in the same record. When the cargo location changes, the location information in the database is updated in real time through a scanning device to ensure the timeliness and accuracy of the data. According to the preset transportation tasks, the location and status data of the transportation vehicle are associated with specific cargo, the transportation vehicle is allocated according to the transportation tasks, and the information is recorded in the database. The location and status of the transportation vehicle are monitored in real time through a GPS tracker and other sensors. The corresponding relationship data stream established in S23 is integrated into the digital twin model, which usually involves the invocation of an API (Application Programming Interface) to transmit the real-time data stream to the digital twin model to ensure that the data in the digital twin model is synchronized with the state of the actual physical system, including the cargo location, the status of the transportation vehicle, etc. The digital twin model is updated according to the integrated data to reflect the latest logistics status, reflecting the new cargo location and the status of the transportation vehicle.

[0124] S3. Use the demand prediction model to predict the material demand and product demand based on the data in the real-time updated logistics digital twin model, and adjust the inventory level of the logistics warehouse according to the prediction results.

[0125] Preferably, using the demand prediction model to predict the material demand and product demand based on the data in the real-time updated logistics digital twin model and adjusting the inventory level of the logistics warehouse according to the prediction results includes the following steps:

[0126] S31. Collect historical data associated with the material demand and product demand, and divide the historical data into a training set, a validation set, and a test set;

[0127] S32. Build the architecture of the demand prediction model based on the characteristics of the historical data and the predicted demand;

[0128] S33. Use the training set data to train the demand prediction model (BP neural network model). During the training process, optimize the weights and biases of the demand prediction model through the backpropagation algorithm;

[0129] S34. After the model training is completed, use the test set data to evaluate the prediction accuracy of the demand prediction model;

[0130] S35. If the prediction accuracy of the test set does not meet the preset requirements, adjust the hyperparameters (including the learning rate, the number of neurons in the hidden layer, etc.) of the demand prediction model according to its performance on the validation set;

[0131] S36. Integrate real-time data into the logistics digital twin model, use the trained BP neural network model, and predict the material demand and product demand based on the latest data in the logistics digital twin model. At the same time, adjust the inventory level according to the prediction results.

[0132] It should be noted that relevant historical data is collected from sales records, warehouse management systems, etc. The historical data includes sales volume, seasonal variations, promotional activities, market trends, etc. The collected data is cleaned and preprocessed, such as removing outliers, filling in missing values, data normalization, etc. The data division includes dividing the processed data into a training set, a validation set, and a test set, determining the number of neurons in the input layer, hidden layer, and output layer of the model, as well as the type of activation function. Use the training set data to train the BP neural network model through the backpropagation algorithm. Backpropagation is an optimization algorithm used to adjust the weights and biases in the model according to the error. Monitor the loss function of the model during the training process to ensure that the model gradually converges during the learning process. Use independent test set data to evaluate the prediction accuracy of the model, which is carried out by calculating the error between the predicted value and the actual value (such as the mean squared error MSE). If the prediction accuracy of the model does not reach the expected goal, the hyperparameters (such as the learning rate, the number of neurons in the hidden layer) can be adjusted to optimize the model. Use the validation set to test the effects of different hyperparameter configurations and select the best-performing configuration.

[0133] Integrate the data collected in real time from the logistics system into the logistics digital twin model, use the trained BP neural network model, and predict the future material demand and product demand based on the latest data in the digital twin model. Adjust the inventory level according to the prediction results to optimize the inventory holding cost and meet future demands.

[0134] Preferably, use the training set data to train the demand prediction model. During the training process, optimizing the weights and biases of the demand prediction model through the backpropagation algorithm includes the following steps:

[0135] S331. Initialize the weights and biases of the demand prediction model;

[0136] S332. Input the data of the training set into the demand prediction model. After being received at the input layer of the demand prediction model, the data is passed to the hidden layer;

[0137] S333. Apply the activation function in the hidden layer, calculate the difference between the prediction result and the actual label at the output layer, and quantify the difference using the loss function;

[0138] S334. Update the weights and biases according to the results of the loss function using the backpropagation algorithm.

[0139] It should be noted that at the beginning of training, the weights and biases of the model are randomly initialized. The initial values are usually randomly selected from a standard distribution (such as a Gaussian distribution). The data of the training set is input into the input layer of the model. The data may be time series data, market trends, seasonal factors, etc., which are features related to demand forecasting. After the input layer receives the data, the data is passed to the hidden layer through the weights. In the hidden layer, the data will be linearly transformed according to the weights and biases. An activation function (such as Sigmoid, ReLU, etc.) is applied in the hidden layer to introduce non-linearity, which helps the model learn more complex patterns. After the data passes through the hidden layer, it reaches the output layer. The output layer calculates the prediction results, which are usually the predicted values of the target variable. Calculate the difference between the prediction results and the actual labels, and use a loss function (such as mean squared error MSE, cross-entropy loss, etc.) to quantify this difference. Calculate the gradient of the loss function with respect to each weight, which usually involves the calculation of the chain rule and partial derivatives. Update the weights according to the gradient and the learning rate (a hyperparameter that controls the step size of model updates). The goal of weight update is to reduce the value of the loss function. Similarly, calculate the gradient of the loss function with respect to each bias and update the bias. Repeat this process, updating the weights and biases in each iteration until the performance of the model on the training set reaches a satisfactory level.

[0140] S4. Based on the adjusted inventory level and predicted demand, use a path optimization algorithm to calculate the optimal transportation path, and calculate the eigenvalue and demand value during the path optimization process. Based on the eigenvalue and demand value, formulate a production and manufacturing plan for logistics warehousing, and monitor the production and manufacturing process in real time;

[0141] Among them, based on the adjusted inventory level and predicted demand, using a path optimization algorithm to calculate the optimal transportation path, and calculating the eigenvalue and demand value during the path optimization process, formulating a production and manufacturing plan for logistics warehousing based on the eigenvalue and demand value, and monitoring the production and manufacturing process in real time includes the following steps:

[0142] S41. Initialize the search space parameters of the rapidly-exploring random tree, and set the starting node and the target node. Based on the starting node and the target node, construct a bidirectional rapidly-exploring random tree, and search for the path between the starting node and the target node through the random sampling method.

[0143] Preferably, initializing the search space parameters of the rapidly-exploring random tree, setting the starting node and the target node, constructing a bidirectional rapidly-exploring random tree based on the starting node and the target node, and searching for the path between the starting node and the target node through the random sampling method includes the following steps:

[0144] S411. Determine the search space boundary of the rapidly exploring random tree according to the geographical scope of the supply chain network;

[0145] S412. Set the parameter data of the algorithm based on the search space boundary;

[0146] S413. Clearly set the positions of the starting node and the target node within the search space based on the parameter data;

[0147] S414. Start constructing the rapidly exploring random tree from the starting node and the target node respectively, use random sampling to expand each tree, and try to connect the nearest node of the current tree to the current node;

[0148] S415. During the process of expanding the rapidly exploring random tree, take into account the constraints including transportation cost, time window, and vehicle transportation capacity;

[0149] S416. As the two rapidly exploring random trees grow, monitor the growth of the rapidly exploring random trees and search for the connection point between the two rapidly exploring random trees;

[0150] S417. When the two rapidly exploring random trees meet, form a complete path, check whether the complete path meets the constraints including transportation cost, time window, and vehicle transportation capacity, and if not, continue the search until the best path is found.

[0151] It should be noted that, first, analyze the geographical scope of the supply chain network to determine the spatial boundaries applicable to path search, including the terrain, transportation network, and permitted driving areas of a specific region. Set the boundaries of the search space according to the analysis results to ensure that all search activities are restricted within this defined space. Set the parameters required for the algorithm to run, such as the maximum number of iterations, branch length, specifications for collision detection, etc. Prepare the data structure for storing the nodes of the RRT tree, which usually includes the position of the node, a reference to the parent node, and any path-specific cost evaluation. Clearly mark the specific positions of the starting node and the target node within the search space. The nodes are usually based on the actual geographical location or predefined network nodes. Confirm the accessibility and rationality of the selected nodes to ensure their key positions in the supply chain network. Initialize two RRT trees respectively from the starting node and the target node. Generate new points within the search space through random sampling techniques and attempt to connect these points to the nearest tree nodes. The expansion process takes into account avoiding obstacles and meeting specific motion constraints. During the expansion of the tree, consider constraints such as transportation costs, time windows, and vehicle transportation capabilities. These constraints affect node selection and path generation. Adjust the search strategy according to real-time feedback to meet the changing external conditions and internal goals. Continuously monitor the growth of the two trees to find possible connection points. The core of the bidirectional rapidly-exploring random tree aims to effectively shorten the path search time. When the two trees approach, attempt to create connection points to form a complete path from the starting node to the target node. After the two trees are connected to form a complete path, check whether the path meets all transportation and operational constraints. If the preliminary path does not meet the conditions, continue the search and optimize the path until the best path that meets all conditions is found.

[0152] S42. Evaluate the transportation time and cost of the candidate paths generated by the bidirectional rapidly-exploring random tree, and analyze the impact of each path on the customer satisfaction demand value.

[0153] It should be noted that, first, necessary data is collected from the candidate paths output by the two-way fast random search tree algorithm, which includes the length of the path, the estimated transit time, and possible delays or obstacles. Using the path data and combining factors such as the average speed and fuel efficiency of the transportation vehicle, the estimated transportation time and cost of each path are calculated. This usually involves specific mathematical models or calculation formulas, such as considering oil prices, wages, maintenance costs, etc., to establish an evaluation model to predict the impact of different transportation paths on customer satisfaction. This model may include factors such as on-time delivery rate, cargo condition, service quality, etc. Analyze how the data of different paths affects customer satisfaction, including performing regression analysis, classification tasks, or clustering analysis to identify which path characteristics are closely related to high customer satisfaction. Comprehensive evaluation of the transportation time, cost, and predicted customer satisfaction of each path, and use decision support tools such as multi-criteria decision analysis (MCDM) to determine the optimal path. Based on the analysis results, select the path that best meets cost-effectiveness, efficiency, and customer satisfaction as the implementation plan, and implement logistics operations according to the selected best path, including dispatching transportation vehicles, arranging cargo loading, etc.

[0154] S43. During the process of path optimization, calculate the eigenvalue and demand value of each candidate path according to the real-time updated inventory and predicted demand.

[0155] Preferably, during the process of path optimization, calculating the eigenvalue and demand value of each candidate path according to the real-time updated inventory and predicted demand includes the following steps:

[0156] S431. Obtain the current inventory data and predicted demand;

[0157] S432. Calculate the transportation time according to the distance of the candidate path and the road speed limit;

[0158] S433. Calculate the transportation cost based on the transportation distance, transportation vehicle type, and fuel cost;

[0159] S434. Use the transportation time as the time eigenvalue of the path;

[0160] S435. Use the transportation cost as the cost eigenvalue of the path;

[0161] S436. Calculate the satisfaction score using a linear or non-linear model by measuring the impact of transportation time on customer satisfaction;

[0162] S437. Score each path by comprehensively considering the transportation time and cost eigenvalues, customer satisfaction, and delivery time requirement values.

[0163] Preferably, the formula for scoring each path is:

[0164] E = w t ·T + w c ·C + w s ·S;

[0165] Wherein, T represents the eigenvalue of transportation time;

[0166] C represents the eigenvalue of transportation cost;

[0167] S represents the customer satisfaction score;

[0168] w t represents the weight coefficient of transportation time;

[0169] w c represents the weight coefficient of transportation cost;

[0170] w s represents the weight coefficient of customer satisfaction.

[0171] It should be explained that real-time inventory data is obtained through an integrated ERP system or supply chain management system. The ERP system or supply chain management system usually has the function of real-time data update and can provide accurate inventory status. Using the established demand forecasting model, the product demand forecast for a future period is obtained. The product demand forecast is calculated based on historical sales data, market trends, seasonal factors, etc. Using a geographic information system (GIS) or similar tools, the actual road distance of the candidate route is calculated. Considering information such as road type and speed limit, the average driving speed of the entire route is estimated, and then the total transportation time is calculated. Considering factors such as transportation distance, the type of transportation tool used (such as trucks, trains, etc.) and fuel cost, and combining the fuel efficiency and fuel price of the transportation tool, the cost of the entire transportation process is estimated. Determine the importance of transportation time for the overall route evaluation and use it as a key time eigenvalue. Similarly, determine the weight of transportation cost in route selection and use it as a cost eigenvalue. Apply a linear or non-linear model to estimate the specific impact of different transportation times on customer satisfaction. Combining transportation time, cost eigenvalue, customer satisfaction and delivery time demand value, use a multi-factor decision analysis method to calculate a comprehensive score for each route, compare the scores of different routes, and select the route with the highest score to optimize supply chain efficiency and customer satisfaction.

[0172] S44. Considering comprehensively the eigenvalues, demand values of all candidate routes and the real-time updated inventory, formulate the production and manufacturing plan of logistics warehousing.

[0173] S45. Monitor the production progress and logistics status of goods in real time to respond to any changes in the supply chain and adjust the production and transportation plans in a timely manner.

[0174] It should be noted that first, all relevant data (including the eigenvalue and demand value of the candidate path, real-time inventory, etc.) need to be integrated into a decision support, analyze the advantages and disadvantages of each candidate path and the current inventory status, determine which products need to be produced or replenished first, and based on the above analysis, formulate a detailed production and manufacturing plan, including deciding which products to produce, which raw materials to use, the quantity to be produced, and the production schedule, to ensure that the resources required in the production plan (such as machines, manpower, raw materials) have been reasonably allocated and scheduled to support the implementation of the plan, establish a real-time monitoring system, which can track and report the production progress and logistics status, usually including installing sensors, cameras and other monitoring devices, collect real-time data such as data on the production line and location information during the logistics transportation process through the monitoring devices, and build an automated response mechanism according to the collected data. For example, if production delays or logistics anomalies are detected, the system will automatically trigger an early warning and send notifications to relevant personnel, and adjust the production and transportation plans in a timely manner according to the real-time data and system analysis. For example, if it is found that the supply of raw materials is insufficient, it may be necessary to adjust the production plan or find alternative suppliers.

[0175] Suppose a company manufacturing consumer electronics faces the challenge of large demand fluctuations and needs to flexibly adjust its production line and supply chain to adapt to changes in market demand, while ensuring cost efficiency and customer satisfaction;

[0176] 1) Establish a logistics digital twin model:

[0177] Equipment installation: Install RFID tags and sensors in the warehouse and production line to collect real-time data, including the location of items, moving speed, environmental conditions (such as temperature, humidity), and the status of production equipment (such as running speed, failure rate).

[0178] Data integration: Transmit the collected data to the central database, and through data cleaning and formatting, eliminate outliers to ensure data consistency.

[0179] Model construction: Establish a digital twin model on the Siemens MindSphere platform, map physical entities and processes, and conduct dynamic simulation.

[0180] 2) Simulate the logistics management process:

[0181] QR code application: Print and attach QR codes containing detailed product information to each product in the warehouse for tracking and management.

[0182] Real-time data collection: Use RFID and GPS technologies to real-time monitor the location of goods and transportation tools, and update this data to the digital twin model in real-time.

[0183] 3) Demand forecasting and inventory adjustment:

[0184] Requirement analysis: Based on historical sales data and market analysis, use the BP neural network model to predict future product requirements.

[0185] Inventory optimization: Adjust the inventory level according to the demand predicted by the model to minimize the holding cost and meet the market demand.

[0186] 4) Route optimization and production adjustment:

[0187] Route calculation: Use the bidirectional fast random search tree algorithm to optimize the transportation route from the supplier to the warehouse, considering cost, time, and customer satisfaction.

[0188] Production plan: According to the optimized route and real-time inventory data, quickly adjust the work plan of the production line to cope with changes in demand.

[0189] 5) Real-time monitoring and response mechanism:

[0190] Monitoring system: Install cameras and sensors to monitor the status of the production line and the logistics transportation process around the clock.

[0191] Automatic adjustment: The system analyzes the monitoring data, automatically identifies and responds to problems in production or logistics, such as machine failures or transportation delays, and adjusts the relevant plans in a timely manner.

[0192] According to another embodiment of the present invention, as Figure 2 shown, a production manufacturing management system based on digital twin is further provided, and the system includes: a model construction module 1, a model simulation module 2, a demand prediction and inventory adjustment module 3, and a route optimization and management module 4;

[0193] The model construction module 1, the model simulation module 2, the demand prediction and inventory adjustment module 3, and the route optimization and management module 4 are sequentially connected.

[0194] The model construction module 1 is used to collect logistics management data in the logistics warehouse and operation data of the production line, and construct a logistics digital twin processing model based on the logistics management data and the operation data of the production line;

[0195] The model simulation module 2 is used to simulate the logistics management process based on the logistics digital twin processing model, use positioning and identification technologies to track the positions of goods and transportation tools in real time, and update the digital twin model in real time to reflect the real-time status of the logistics warehouse and the production line;

[0196] The demand prediction and inventory adjustment module 3 is used to use the demand prediction model to predict material requirements and product requirements based on the data in the logistics digital twin model updated in real time, and adjust the inventory level of the logistics warehouse according to the prediction results;

[0197] A path optimization and management module 4 is configured to calculate an optimal transportation path by using a path optimization algorithm based on the adjusted inventory and predicted demand, calculate eigenvalue and demand value during the path optimization process, formulate a production and manufacturing plan for logistics warehousing based on the eigenvalue and demand value, and monitor the production and manufacturing process in real time.

[0198] Figure 3 An embodiment of a computer device according to the present invention is shown. The computer device may be a server. The computer device includes a processor, a memory, and a network interface connected by a system bus. Among them, the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store static information and dynamic information data. The network interface of the computer device is configured to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0199] Those skilled in the art can understand that Figure 3 the structure shown in is only a block diagram of some structures related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0200] In addition, the present invention also provides a computer device, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the steps in the above method embodiments are implemented.

[0201] In addition, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0202] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0203] In summary, by means of the above technical solutions of the present invention, through real-time data collection and synchronous update of the digital twin model, an enterprise can monitor the accurate positions and statuses of each cargo and transportation vehicle in real time. The digital twin model provides a highly accurate virtual representation that can simulate operations and scenarios in the real physical environment. By using algorithms to analyze and process real-time data, it is possible to more effectively plan logistics routes and transportation plans, reduce transportation costs, shorten delivery times. The real-time update and precise logistics management allow enterprises to provide more reliable delivery time predictions, enhancing customer trust. Through comprehensive monitoring and high-level control of the logistics system, potential safety issues and risks can be better identified, and measures can be taken in a timely manner to prevent accidents and ensure the safety of personnel and goods. Based on the data in the real-time updated logistics digital twin model of the present invention, the demand forecasting model can more accurately predict the future demand for materials and products. Through real-time data integration and demand forecasting, enterprises can quickly respond to market changes, adjust inventory strategies in a timely manner to meet consumer demands, and enhance market competitiveness. By using historical data and the BP neural network model, the accuracy of demand forecasting can be improved, and inventory management problems caused by inaccurate forecasting can be reduced. By continuously training and validating the model, enterprises can continuously improve the accuracy of demand forecasting to adapt to the changing market trends and consumer behaviors. The demand forecasting model provides strong data support for management, helping them make more informed inventory management and supply chain decisions, reducing subjective speculation and uncertainty. By constructing two search trees simultaneously from the starting point and the target point, the bidirectional rapidly-exploring random tree (RRT) can find an effective path faster than the unidirectional RRT, especially performing outstandingly in complex or high-dimensional search spaces. The bidirectional RRT allows path search considering multiple constraints, can effectively balance transportation costs and time, and find the most cost-effective logistics path. Since the RRT algorithm can handle complex obstacles and irregular terrains, it is particularly suitable for supply chain networks with complex geographical features such as urban traffic networks and mountain roads. Under the influence of changing environmental conditions or emergencies, the bidirectional RRT can quickly recalculate a new path to ensure the continuity and efficiency of logistics. By comprehensively considering various transportation and operation constraints, the bidirectional RRT supports a data-driven complex decision-making process, improving the accuracy and adaptability of decisions.

[0204] Although the present invention has been disclosed above with preferred embodiments, the embodiments are only for the purpose of illustration and exemplification, and are not intended to limit the present invention. Those skilled in the art can make several modifications and refinements without departing from the spirit and scope of the present invention. The scope of protection claimed by the present invention shall be subject to what is described in the claims.

Claims

1. A production manufacturing management method based on digital twin, characterized in that The production and manufacturing management method includes the following steps: S1. Collect the logistics management data in the logistics warehouse and the operation data of the production line, and build a logistics digital twin processing model based on the logistics management data and the operation data of the production line; S2. Simulate the logistics management process based on the logistics digital twin processing model, and use positioning and identification technologies to track the positions of goods and transportation tools in real time, and at the same time update the digital twin model in real time to reflect the real-time status of the logistics warehouse and the production line; S3. Use the demand forecasting model to predict the material demand and product demand based on the data in the logistics digital twin model updated in real time, and adjust the inventory level of the logistics warehouse according to the prediction results; S4. Based on the adjusted inventory level and the predicted demand, use the path optimization algorithm to calculate the optimal transportation path, and calculate the eigenvalue and demand value during the path optimization process. Make a production and manufacturing plan for the logistics warehouse based on the eigenvalue and demand value, and at the same time monitor the production and manufacturing process in real time; Among them, the step of calculating the optimal transportation path using the path optimization algorithm based on the adjusted inventory level and the predicted demand, calculating the eigenvalue and demand value during the path optimization process, making a production and manufacturing plan for the logistics warehouse based on the eigenvalue and demand value, and at the same time monitoring the production and manufacturing process in real time includes the following steps: S41. Initialize the search space parameters of the rapidly-exploring random tree, and set the starting node and the target node. Build a bidirectional rapidly-exploring random tree based on the starting node and the target node, and search for the path between the starting node and the target node by the random sampling method; S42. Evaluate the transportation time and cost of the candidate paths generated by the bidirectional rapidly-exploring random tree, and analyze the impact of each path on the customer satisfaction demand value; S43. During the path optimization process, calculate the eigenvalue and demand value of each candidate path according to the inventory level and predicted demand updated in real time; S44. Make a production and manufacturing plan for the logistics warehouse by comprehensively considering the eigenvalue and demand value of all candidate paths and the inventory level updated in real time; S45. Monitor the production progress and logistics status of the goods in real time to respond to any changes in the supply chain and adjust the production and transportation plans in a timely manner.

2. The manufacturing management method based on digital twin according to claim 1, wherein The step of simulating the logistics management process based on the logistics digital twin processing model, using positioning and identification technologies to track the positions of goods and transportation tools in real time, and at the same time updating the digital twin model in real time to reflect the real-time status of the logistics warehouse and the production line includes the following steps: S21. Use the two-dimensional code technology to uniquely identify the goods in the logistics warehouse to ensure that each good has a unique identification code; S22. Collect the logistics data in the logistics warehouse in real time, and use the correlation coefficient matching algorithm to process the real-time collected logistics data to extract the position and status information of the goods and transportation tools in the logistics data; S23. Associate the identified position with the unique identification of the goods, establish the corresponding relationship between the goods and the logistics data, and at the same time associate the position and status data of the transportation tool with the preset transportation task and goods; S24. Integrate the associated goods position and the status data of the transportation tool into the digital twin model, and update the logistics digital twin processing model according to the integrated data.

3. The method for production manufacturing management based on digital twin according to claim 2, wherein, The real-time collection of logistics data in a logistics warehouse, using a correlation coefficient matching algorithm to process the real-time collected logistics data, and extracting the position and status information of goods and transportation tools in the logistics data includes the following steps: S221. In the logistics warehouse environment, collect scene data including goods and transportation tools in real time; S222. Select an area potentially containing target features from the collected scene data, and use the two-dimensional array window sliding technique to calculate the normalized correlation coefficient of the feature values within the window; S223. Compare the calculated normalized correlation coefficient with the threshold of the preset sequential similarity detection algorithm. If the normalized correlation coefficient exceeds the threshold, it indicates that the current position contains the features of the target goods or transportation tools; S224. Implement an interval point calculation strategy between the matched feature points and skip the feature points at the positions where the normalized correlation coefficient exceeds the threshold; S225. According to the comparison result, extract the position and status information of goods and transportation tools from the scene area.

4. The production manufacturing management method based on digital twin according to claim 3, wherein, The step of selecting an area potentially containing target features from the collected scene data and using the two-dimensional array window sliding technique to calculate the normalized correlation coefficient of the feature values within the window includes the following steps: S2221. Screen the areas containing target goods or transportation tools from a large amount of real-time collected scene data; S2222. Set a two-dimensional array window for the selected area, and define the size and shape of the block for calculating the normalized correlation coefficient; S2223. Apply the window sliding technique to move the window from one end of the area to the other end; S2224. For each area covered by the window, calculate the normalized correlation coefficient.

5. The production manufacturing management method based on digital twin according to claim 1, characterized in that, The step of using a demand prediction model to predict material demand and product demand based on the data in the real-time updated logistics digital twin model and adjusting the inventory level of the logistics warehouse according to the prediction result includes the following steps: S31. Collect historical data related to material demand and product demand, and divide the historical data into a training set, a validation set, and a test set; S32. Based on the characteristics of the historical data and the predicted demand, construct the architecture of the demand prediction model; S33. Use the training set data to train the demand prediction model. During the training process, optimize the weights and biases of the demand prediction model through the backpropagation algorithm; S34. After the model training is completed, use the test set data to evaluate the prediction accuracy of the demand prediction model; S35. If the prediction accuracy of the test set does not meet the preset requirements, adjust the hyperparameters of the demand prediction model according to the performance on the validation set; S36. Integrate the real-time data into the logistics digital twin model, use the trained BP neural network model, and predict the material demand and product demand based on the latest data in the logistics digital twin model, and at the same time adjust the inventory level according to the prediction result.

6. The manufacturing management method based on digital twin according to claim 5, wherein, The step of using the training set data to train the demand prediction model and optimizing the weights and biases of the demand prediction model through the backpropagation algorithm during the training process includes the following steps: S331. Initialize the weights and biases of the demand prediction model; S332. Input the data of the training set into the demand prediction model. After the data is received at the input layer of the demand prediction model, it is passed to the hidden layer; S333. Apply an activation function in the hidden layer, calculate the difference between the predicted result and the actual label in the output layer, and quantify the difference using a loss function; S334. Use the backpropagation algorithm to update the weights and biases according to the result of the loss function.

7. A production manufacturing management method based on digital twin according to claim 1, characterized in that The steps of initializing the search space parameters of the fast random search tree, setting the start node and the target node, constructing a bidirectional fast random search tree based on the start node and the target node, and searching for the path between the start node and the target node by the random sampling method include the following steps: S411. Determine the search space boundary of the fast random search tree according to the geographical scope of the supply chain network; S412. Set the parameter data of the algorithm based on the search space boundary; S413. Clearly set the positions of the start node and the target node within the search space based on the parameter data; S414. Construct fast random search trees starting from the start node and the target node respectively, use random sampling to expand each tree, and try to connect the nearest node of the current tree to the current node; S415. During the process of expanding the fast random search tree, take into account the constraints including transportation cost, time window, and vehicle transportation capacity; S416. As the two fast random search trees grow, monitor the growth of the fast random search trees and find the connection points between the two fast random search trees; S417. When the two fast random search trees meet, form a complete path, check whether the complete path meets the constraints including transportation cost, time window, and vehicle transportation capacity. If not, continue to search until the best path is found.

8. A production manufacturing management method based on digital twin according to claim 7, characterized in that, The steps of calculating the eigenvalue and demand value of each candidate path according to the real-time updated inventory level and predicted demand during the path optimization process include the following steps: S431. Obtain the current inventory data and predicted demand; S432. Calculate the transportation time according to the distance of the candidate path and the road speed limit; S433. Calculate the transportation cost based on the transportation distance, transportation tool type, and fuel cost; S434. Use the transportation time as the time eigenvalue of the path; S435. Use the transportation cost as the cost eigenvalue of the path; S436. Calculate the satisfaction score using a linear or non-linear model by measuring the impact of transportation time on customer satisfaction; S437. Considering the transportation time and cost eigenvalues, customer satisfaction, and delivery time demand value comprehensively, score each path.

9. A production manufacturing management method based on digital twin according to claim 8, characterized in that, During the path optimization process, the formula for scoring each path is: E = w t ·T + w c ·C + w s ·S; In the formula, T represents the transportation time eigenvalue; C represents the transportation cost eigenvalue; S represents the customer satisfaction score; w t The weight coefficient representing the transportation time; w c represents the weight coefficient of transportation cost; w s Represents the weight coefficient of customer satisfaction.

10. A digital twin-based production manufacturing management system for implementing the digital twin-based production manufacturing management method according to any one of claims 1-9, characterized in that, The system includes: a model construction module, a model simulation module, a demand prediction and inventory adjustment module, and a path optimization and management module; The model construction module, the model simulation module, the demand prediction and inventory adjustment module, and the path optimization and management module are connected in sequence; The model construction module is used to collect the logistics management data in the logistics warehouse and the operation data of the production line, and construct a logistics digital twin processing model based on the logistics management data and the operation data of the production line; The model simulation module is used to simulate the logistics management process based on the logistics digital twin processing model, and use positioning and identification technologies to track the positions of goods and transportation tools in real time, while updating the digital twin model in real time to reflect the real-time status of logistics warehousing and production lines; The demand forecasting and inventory adjustment module is used to utilize the demand forecasting model, based on the data in the logistics digital twin model updated in real time, to forecast material demand and product demand, and adjust the inventory levels in the logistics warehouse according to the forecasting results; The route optimization and management module is used to calculate the optimal transportation route using the route optimization algorithm based on the adjusted inventory levels and forecasted demands, and calculate the eigenvalue and demand value during the route optimization process, formulate the production and manufacturing plan for the logistics warehouse based on the eigenvalue and demand value, and monitor the production and manufacturing process in real time.

Images