Internet of Things intelligent logistics system for cement industry

Abstract

The invention provides a cement industry internet-of-things intelligent logistics system, which comprises the following steps of: according to standardized sales order data and delivery order data, judging the relevance between the sales order data and the delivery order data by adopting a decision tree algorithm, establishing a mapping relationship, and realizing information connection between sales and logistics links; vehicle positioning data and loading and unloading confirmation information are obtained, a real-time tracking record of the whole logistics process is generated through data fusion processing, and a visual track is generated in combination with map information; for mass logistics data, a distributed data processing platform is established, the execution efficiency of logistics tasks is analyzed by adopting a clustering algorithm, abnormal tasks are identified, and intelligent early warning signals are generated; and according to the intelligent early warning signal, the priority of the logistics task and the vehicle scheduling strategy are adjusted, the distribution of logistics resources is optimized, and the overall logistics efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a cement industry Internet of Things smart logistics system. Background Art

[0002] The core technical problem faced by intelligent logistics management is how to achieve efficient and accurate full-process collaboration in a complex and ever-changing business environment. Traditional logistics management has information islands, and it is difficult to connect internal and external data, resulting in inefficient resource scheduling. At the same time, the real-time processing and analysis of massive heterogeneous data has become a bottleneck, affecting the timely discovery and handling of abnormal situations. In addition, the dynamic and uncertain nature of logistics tasks makes it difficult for static scheduling plans to adapt to actual needs, resulting in waste of transportation capacity and delivery delays. While improving overall efficiency, how to balance the interests of various links is also a major challenge. For example, the pursuit of low costs by the procurement department may conflict with the timely supply needs of the production department, and the volatility of sales orders brings difficulties to logistics planning. On the other hand, the service quality of external logistics providers varies, and how to establish a scientific evaluation mechanism and achieve optimal configuration is also an urgent problem to be solved. In general, intelligent logistics management needs to build a system that can adapt, self-learn, and self-optimize on the premise of ensuring data security to cope with increasingly complex logistics scenarios and changing market demands. Summary of the invention

[0003] The present invention provides a cement industry Internet of Things smart logistics system, which mainly includes: Obtain the original data of purchase orders, production plans, sales orders and delivery notes, parse the original data, extract key field information, and convert the format according to the preset data standardization protocol to obtain standardized data; based on the standardized purchase order data and production plan data, use the association rule algorithm to determine the correlation between the two, establish a mapping relationship, and realize the information connection between the procurement and production links; based on the standardized sales order data and delivery note data, use the decision tree algorithm to determine the correlation between the two, establish a mapping relationship, and realize the information connection between the sales and logistics links; obtain vehicle positioning data and loading and unloading confirmation information, generate real-time tracking records of the entire logistics process through data fusion processing, and generate visual trajectories in combination with map information; for massive logistics data, establish a distributed data processing platform, use clustering algorithms to analyze the execution efficiency of logistics tasks, identify abnormal tasks and generate intelligent early warning signals; according to the intelligent early warning signals, adjust the priority of logistics tasks and vehicle scheduling strategies, optimize the allocation of logistics resources, and improve overall logistics efficiency.

[0004] Furthermore, after obtaining the original data of the purchase order, production plan, sales order and delivery note, it also includes: preprocessing the original data, including data cleaning, data conversion and data integration, ensuring data quality, and converting the data into an input format suitable for the machine learning algorithm.

[0005] Furthermore, the association rule algorithm is used to determine the association between the purchase order data and the production plan data, including: setting minimum support and minimum confidence thresholds, mining frequent item sets and association rules between the purchase order data and the production plan data, calculating the support and confidence of the rules, and screening out strong association rules.

[0006] Furthermore, the use of a decision tree algorithm to determine the correlation between sales order data and delivery note data includes: using the amount, commodity category, customer level, etc. of the sales order as feature variables, whether there is a corresponding delivery note as the target variable, using ID3, C4.5 or CART algorithm to generate a decision tree, and pruning and optimizing the decision tree.

[0007] Furthermore, after obtaining the vehicle positioning data and loading and unloading confirmation information, it also includes: using a Kalman filter algorithm to remove noise and smooth the trajectory of the vehicle positioning data to improve positioning accuracy and stability; analyzing the video surveillance data of the loading and unloading scene through a machine vision algorithm to automatically identify the loading and unloading status.

[0008] Furthermore, the clustering algorithm is used to analyze the execution efficiency of logistics tasks for massive logistics data, including: clustering tasks using the K-means clustering algorithm according to the characteristic attributes of logistics tasks; calculating efficiency indicators such as completion time, timeliness, and cargo damage for tasks in each cluster, and obtaining efficiency distribution characteristics through statistical analysis; setting an abnormal threshold based on the efficiency distribution characteristics, identifying abnormal tasks below the threshold, and triggering an early warning mechanism.

[0009] Furthermore, the priority of logistics tasks and vehicle scheduling strategies are adjusted according to intelligent warning signals, including: dynamically adjusting task priorities according to the type and urgency of the warning signals using a decision tree algorithm; solving the optimal scheduling solution through a vehicle scheduling optimization model combining task priority and vehicle status; and continuously optimizing scheduling strategies according to real-time logistics operation data using a reinforcement learning algorithm.

[0010] Furthermore, the system also includes: analyzing historical logistics data through machine learning algorithms, predicting future logistics demand, and generating optimized purchase orders and production plans; returning actual logistics data to the historical database, regularly updating the data set and retraining the prediction model to form a business closed loop and continuously optimize the accuracy of demand forecasting and plan generation.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects: The present invention discloses an intelligent logistics management method, which processes purchase orders, production plans, sales orders and bills of lading through data standardization to achieve seamless connection of internal business links. For the standardized bill of lading data, a clustering algorithm is used to establish an association relationship with external logistics providers to achieve complete connection of internal and external logistics information. The present invention also forms a large logistics data set through distributed data collection and storage, classifies logistics tasks using the K-means clustering algorithm, calculates efficiency indicators and sets thresholds, and generates intelligent early warning signals. The characteristics of abnormal tasks are analyzed through an association rule mining algorithm to provide decision support for logistics scheduling optimization. Finally, the present invention dynamically adjusts the priority of logistics tasks according to the early warning signal, optimizes the vehicle scheduling plan, and improves the overall logistics efficiency. The present invention realizes the intelligent management of the entire logistics process and improves the logistics operation efficiency and abnormality handling capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a flow chart of a cement industry Internet of Things smart logistics system of the present invention.

[0013] Figure 2 This is a schematic diagram of a cement industry Internet of Things smart logistics system of the present invention.

[0014] Figure 3 This is another schematic diagram of a cement industry Internet of Things smart logistics system of the present invention.

[0015] Figure 4 This is another schematic diagram of a cement industry Internet of Things smart logistics system of the present invention.

[0016] Figure 5 This is another schematic diagram of a cement industry Internet of Things smart logistics system of the present invention.

[0017] Figure 6 This is another schematic diagram of a cement industry Internet of Things smart logistics system of the present invention.

[0018] Figure 7 This is another schematic diagram of a cement industry Internet of Things smart logistics system of the present invention.

[0019] Figure 8 This is another schematic diagram of a cement industry Internet of Things smart logistics system of the present invention.

[0020] Fig. 9 This is another schematic diagram of a cement industry Internet of Things smart logistics system of the present invention. DETAILED DESCRIPTION

[0021] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant inventions, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0022] like Figure 1-9 In this embodiment, a cement industry Internet of Things smart logistics system may specifically include: Step S101, obtain the original data of the purchase order, production plan, sales order and delivery note, parse the original data according to the pre-established data standardization protocol, extract the key field information, convert the format of the extracted key field information according to the data standardization protocol, and obtain the standardized purchase order, production plan, sales order and delivery note data.

[0023] According to the pre-established data standardization protocol, the original data of purchase orders, production plans, sales orders and bills of lading are obtained, and the original data is parsed to extract key field information, including order number, product name, quantity, amount, etc. The extracted key field information is converted into a format according to the data standardization protocol, the format of data from different sources is unified, and the information islands caused by format differences are eliminated, so as to obtain standardized purchase order, production plan, sales order and bill of lading data. The association rule algorithm is used to determine whether there is a correlation between the standardized purchase order data and the production plan data. If there is a correlation, a mapping relationship between the two is established. By establishing a mapping relationship between the purchase order and the production plan, the information connection between the procurement link and the production link is realized, and a data link running through the two links is obtained. The decision tree algorithm is used to determine whether there is a correlation between the standardized sales order data and the bill of lading data. If there is a correlation, a mapping relationship between the two is established. By establishing a mapping relationship between the sales order and the bill of lading, the information connection between the sales link and the logistics link is realized, and a data link running through the two links is obtained. Integrate the mapping relationships between purchase orders and production plans, sales orders and bills of lading, obtain a complete data link that runs through procurement, production, sales, logistics and other links, and form a seamless connection between various business links within the enterprise. Use a clustering algorithm to group data based on the dimensions of the pickup address, transportation method, etc. in the standardized bill of lading data, determine the external logistics providers corresponding to different bills of lading, and establish the association between bills of lading and external logistics. Send the data on the association between bills of lading and external logistics to the corresponding logistics providers, and receive the waybill return data from the logistics providers. By comparing the information between the two, seamless connection between internal bills of lading and external logistics waybills is achieved, and ultimately the internal and external logistics information of the enterprise is completely connected.

[0024] Exemplarily, according to the pre-established data standardization protocol, the original data of the purchase order, production plan, sales order and delivery note are obtained, such as the purchase order data including the order number PO20230101, the commodity name A product, the quantity 1000 pieces, the amount 100,000 yuan, etc., and the original data is parsed to extract the key field information. The extracted key field information is converted into a format according to the data standardization protocol, such as unifying the date format to YYYY-MM-DD, and the amount format to retain two decimal places, to obtain the standardized data. The Apriori association rule algorithm is used to judge the correlation between the standardized purchase order data and the production plan data with the parameters of minimum support 0.05 and minimum confidence 0.8. It is found that the correlation between the purchase order PO20230101 and the production plan PP20230201 is 0.9, which exceeds the minimum confidence threshold, and a mapping relationship between the two is established. Through the mapping relationship, the quantity, delivery date and other information of the purchase order are automatically connected to the production plan. The C4.5 decision tree algorithm is used, and the information gain ratio is used as the attribute selection metric to judge the correlation between the standardized sales order data and the delivery order data. It is found that the correlation between the sales order SO20230301 and the delivery order DO20230401 is 0.95. A mapping relationship is established between the two to realize the automatic transmission of the sales order's delivery quantity, time and other information to the delivery order. The mapping relationship established between the purchase order and the production plan, and between the sales order and the delivery order is integrated. Based on key attributes such as order number and product name, a complete data link running through the procurement, production, sales, logistics and other links is generated. K-means cluster analysis is performed on the standardized delivery order data, and the delivery address, transportation method and other clustering dimensions are used. When the number of clusters k=3, three delivery order clusters are obtained, corresponding to the three external logistics companies A, B and C, respectively, and the correlation between the delivery order and external logistics is established. The associated relationship data is sent to the corresponding logistics provider through the API interface, and the waybill feedback data of the logistics provider is received through WebService. The order numbers, cargo information, etc. of the two are matched and compared to achieve automatic docking and synchronization of internal and external logistics information, and finally form a full-link connection of enterprise logistics data.

[0025] Step S102, using the association rule mining algorithm to analyze the correlation between the standardized purchase order data and the production plan data, if there is a correlation, a mapping relationship is established; using the decision tree algorithm to analyze the correlation between the standardized sales order data and the delivery note data, if there is a correlation, a mapping relationship is established to achieve seamless connection between various internal business links.

[0026] According to the pre-established data standardization protocol, the original data of purchase orders and production plans are obtained, and the original data is parsed to extract key field information, including order number, product name, quantity, etc. The extracted key field information is converted into a format according to the data standardization protocol to unify the format of data from different sources, eliminate information islands caused by format differences, and obtain standardized purchase order and production plan data. The Apriori association rule algorithm is used, with the minimum support set to 0.5 and the minimum confidence set to 0.8, to mine the association rules between the standardized purchase order data and the production plan data, determine the correlation between the two, and establish a mapping relationship if there is a correlation. According to the established mapping relationship between the purchase order and the production plan, a complete data link running through the procurement, production and other links is obtained, and the information connection between the procurement and production links is realized through the establishment of the data link. According to the pre-established data standardization protocol, the original data of sales orders and delivery orders are obtained, and the original data is parsed to extract key field information, including order number, product name, quantity, amount, etc. The extracted key field information is converted into a format according to the data standardization protocol, the format of data from different sources is unified, and the information islands caused by format differences are eliminated to obtain standardized sales order and delivery order data. The decision tree algorithm is adopted, and the information gain ratio is set as the division standard. Based on the key fields of the standardized sales order data and delivery order data, such as order number, product name, quantity, etc., a decision tree model is constructed. The correlation between the sales order and the delivery order is judged through the decision tree classification. If there is a correlation, a mapping relationship is established. According to the established mapping relationship between the sales order and the delivery order, a complete data link running through the sales, logistics and other links is obtained, and the information connection between the sales and logistics links is realized through the establishment of the data link. The mapping relationship established between the purchase order and the production plan, and between the sales order and the delivery order is integrated to build an end-to-end data link covering the procurement, production, sales, logistics and other links, and the seamless connection between the various business links within the enterprise is realized through the connection of the data link. For the standardized bill of lading data, the K-means clustering algorithm is used, with key fields such as the pickup address and transportation method as clustering dimensions. The optimal cluster number K is determined through iterative optimization, and the bill of lading data is divided into K clusters. According to the clustering results, the external logistics providers corresponding to different bills of lading are determined, and the association between bills of lading and external logistics is established. The data on the association between bills of lading and external logistics is sent to the corresponding logistics providers, and the waybill return data from the logistics providers is received. By comparing the information between the two, the seamless connection between the internal bill of lading and the external logistics waybill is achieved, and finally the internal and external logistics information of the enterprise is completely connected.

[0027] For example, according to the pre-established data standardization protocol, the original data of the purchase order and production plan are obtained, such as the order number of the purchase order is PO20230101, the product name is product A, and the quantity is 1000 pieces; the plan number of the production plan is PP20230101, the product name is product A, and the quantity is 1000 pieces. The original data is parsed to extract the key field information, and the extracted key field information is converted into the format of order number: string, product name: string, quantity: integer according to the data standardization protocol, and the information island caused by the format difference is eliminated to obtain the standardized purchase order and production plan data. The Apriori association rule algorithm is used, and the minimum support is set to 0.5 and the minimum confidence is set to 0.8. The association rules are mined in the standardized 1000 purchase order data and 800 production plan data. It is found that there is a strong correlation between the order of product A in the purchase order and the plan of product A in the production plan, with a support of 0.6 and a confidence of 0.9, and a mapping relationship between the two is established. According to the established mapping relationship, a complete data link running through procurement, production and other links is obtained to achieve information docking between procurement and production links. Similarly, the original data of sales orders and delivery orders are obtained, key fields are extracted, and format conversion is performed to obtain standardized data. Using the decision tree algorithm, with the information gain ratio as the division standard, a decision tree model is constructed based on the standardized 5,000 sales order data and 3,000 delivery order data. The correlation between sales orders and delivery orders is judged by decision tree classification, with an accuracy rate of 95%. A mapping relationship between the two is established to achieve information docking between sales and logistics links. The mapping relationship established between purchase orders and production plans, and between sales orders and delivery orders is integrated to build an end-to-end data link covering procurement, production, sales, logistics and other links, and seamless docking between various business links within the enterprise is achieved through the connection of data links. For the standardized 10,000 delivery order data, the K-means clustering algorithm is used, with key fields such as delivery address and transportation method as clustering dimensions. The optimal cluster number K is determined to be 5 through the elbow rule, and the delivery order data is divided into 5 clusters. According to the clustering results, the external logistics providers corresponding to different delivery orders are determined, and the association between delivery orders and external logistics is established. The association data is sent to the corresponding logistics provider, and the waybill return data of the logistics provider is received at the same time. Through information comparison, the seamless connection between the internal delivery order and the external logistics waybill is realized, and finally the internal and external logistics information of the enterprise is completely connected.

[0028] Step S103, for the standardized delivery note data, a clustering algorithm is used to group them according to dimensions such as the delivery address and mode of transportation, determine the external logistics providers corresponding to different delivery notes, establish the association between the delivery note and the external logistics, send the association data to the corresponding logistics provider, and receive the waybill return data from the logistics provider at the same time, and through information comparison, realize the complete connection of internal and external logistics information.

[0029] According to the standardized delivery order data, a clustering algorithm is used to group the delivery orders according to the dimensions of pick-up address, transportation method, etc., and the classification results of external logistics providers corresponding to different delivery orders are obtained. Through the classification results of the delivery order and the external logistics provider, the association relationship between the delivery order and the external logistics provider is established to determine which external logistics provider is responsible for the transportation of each delivery order. The established association data between the delivery order and the external logistics provider is sent to the corresponding logistics provider, and the waybill data returned by the logistics provider is received at the same time. The received logistics provider waybill data is parsed to extract key field information such as waybill number, pick-up date, delivery date, and receipt status. The delivery order data is compared with the logistics provider waybill data, and the corresponding relationship between the two is determined by matching key fields such as order number, pick-up date, and delivery date. If the key fields of the delivery order and the logistics provider waybill data are completely matched, they are regarded as the same logistics task, and the mapping relationship between the internal delivery order and the external logistics waybill is established. According to the established mapping relationship between internal and external logistics documents, the real-time status information of external logistics waybills, such as in transit, received, etc., is obtained, and these statuses are synchronously updated to the internal delivery note. Through the seamless connection and status synchronization of internal and external logistics documents, the real-time tracking and management of the entire logistics process is realized, and the logistics efficiency and accuracy are improved. The internal and external logistics information after the connection is integrated into the enterprise's information management system, providing real-time and accurate logistics status query and tracking services for sales, customer service and other related departments, and improving customer satisfaction.

[0030] Exemplarily, first, the K-means clustering algorithm is used to cluster the bill of lading according to the two dimensions of the pick-up address and the mode of transportation. By calculating the Euclidean distance between each bill of lading and the cluster center, the cluster center is iteratively updated until the clustering result converges, and the bill of lading grouping of different logistics companies is obtained. Then, according to the clustering result, the association matrix between the bill of lading and the logistics company is established. The elements in the matrix represent the association strength between the bill of lading and the logistics company, and the value range is 0 to 1. Then, the association matrix is ​​sent to the corresponding logistics company through the API interface, and the waybill data returned by the logistics company is received in real time through WebSocket. For the received waybill data, regular expressions are used to extract key fields such as waybill number, pick-up date, delivery date and receipt status. Then, the key fields of the bill of lading and the waybill are sequenced through the dynamic programming algorithm, and the longest common subsequence of the two is calculated. If the similarity exceeds 90%, it is considered that the two correspond to the same logistics task, and the mapping relationship between internal and external logistics documents is established. Based on the mapping relationship, the distributed message queue Kafka is used to synchronize the status change events of the external waybill to the internal delivery note in real time, realizing real-time tracking of the entire logistics process. Finally, the logistics information is integrated into the enterprise's information management system through the RESTful API, providing a query interface for the logistics status for sales, customer service and other departments, with a response time of less than 50 milliseconds to ensure service quality.

[0031] Step S104, through the distributed data acquisition component, massive logistics task execution data is obtained from each logistics node and stored in the distributed storage system to form a logistics big data set; the logistics big data is pre-processed, including data cleaning, data conversion and data integration to ensure data quality, and the data is converted into a format suitable for cluster analysis.

[0032] By deploying data collection components at each logistics node, such as installing sensors and cameras in warehouses, transport vehicles, and distribution centers, real-time data on logistics task execution is collected, including in-and-out time, transportation routes, and delivery status, and stored in the Hadoop distributed file system HDFS through network transmission. The amount of data collected every day can reach 100GB. The collected logistics big data is processed by ETL, and Spark's data cleaning library such as SparkSQL is used to filter out noise data such as missing values ​​and outliers, and data from different sources and formats are converted into a unified Parquet column storage format to improve data analysis efficiency. Based on the cleaned and converted logistics data, data integration technology is used to associate data scattered in different nodes according to a unified primary key such as the waybill number to form a complete logistics task execution record, which is stored in the Hive data warehouse. By analyzing the characteristic attributes of logistics tasks, such as task type, origin, destination, and transportation method, One-Hot encoding is used to convert categorical features into numerical vectors, and the numerical features are normalized to convert the data into a format suitable for clustering analysis. According to the converted logistics task feature vector, the K-means clustering algorithm is used to cluster the logistics tasks. By calculating the Euclidean distance between tasks, the cluster center is iteratively optimized until the clustering results converge, and similar tasks are divided into the same cluster. For each task cluster, the efficiency indicators such as the average completion time, punctuality rate, and cargo damage rate of the tasks in the cluster are calculated to obtain the efficiency distribution characteristics of the cluster. Based on the normal distribution assumption, the upper and lower efficiency thresholds are determined according to the 3σ principle. If the efficiency index of a logistics task exceeds the threshold range of the cluster, it is judged as an abnormal task, and an intelligent warning signal is generated, and the trigger warning is pushed to the logistics scheduling system. The Aprior association rule mining algorithm is used to analyze the correlation between the transportation path, weather conditions, cargo type and other factors of the abnormal task and the occurrence of the abnormality, and obtain the strong association rules of the abnormal cause, such as "cargo type = fragile goods ∧ transportation path = mountainous area → abnormal occurrence". The cluster analysis results, abnormal task warning signals, and abnormal cause association rules are integrated into the logistics monitoring dashboard built by Tableau. Real-time monitoring of logistics task execution efficiency and abnormal warning are achieved through graphical methods, and it is interconnected with the logistics scheduling system to provide data support for scheduling optimization.

[0033] For example, data collection components are deployed at each logistics node, such as installing 50 RFID sensors in the warehouse, 20 GPS locators in the transport vehicles, and 100 high-definition cameras in the distribution center, to collect logistics task execution data in real time, once every 5 seconds on average, and the amount of data collected every day can reach 120GB. SparkSQL, a data cleaning library of Spark, is used to filter out noise data by setting the missing value threshold to 10% and the outlier range to 3σ, and convert data in formats such as JSON and CSV to Parquet column storage, with a data compression rate of 75%. Data integration technology is used, with the waybill number as the primary key, to associate data scattered in the warehouse, transportation, distribution and other nodes to form a complete logistics task execution record, which is stored in the Hive data warehouse, and the table association degree reaches 95%. By One-Hot encoding the characteristic attributes of logistics tasks, such as encoding the task type as

[100] ,

[010] ,

[001] , the numerical features are normalized to the

[01] interval and converted into a 10-dimensional feature vector. The K-means clustering algorithm was used, K=20 was set, 500 iterations were performed, and the convergence threshold was 0.01. One million logistics tasks were divided into 20 task clusters, and the average silhouette coefficient reached 0.85. For each task cluster, the efficiency indicators such as average completion time, punctuality rate, and cargo damage rate were calculated. Based on the normal distribution assumption, the efficiency threshold was determined by 3σ above and below. Warning signals were generated for abnormal tasks that exceeded the threshold and pushed to the logistics scheduling system. The Aprior association rule mining algorithm was used, with the minimum support set to 01 and the minimum confidence set to 0.8. The influencing factors of abnormal tasks were analyzed, and 10 strong association rules were obtained. Finally, a logistics monitoring dashboard was built through Tableau to integrate and display clustering results, abnormal warnings, association rules, etc., to achieve real-time monitoring and scheduling optimization.

[0034] Step S105, clustering the logistics tasks using the K-means clustering algorithm based on the characteristic attributes of the logistics tasks, such as task type, starting point, destination, mode of transportation, etc., and dividing similar tasks into the same cluster; for the logistics tasks in each cluster, calculating their execution efficiency indicators, such as task completion time, transportation timeliness, cargo damage rate, etc., and obtaining the efficiency distribution characteristics of the cluster through statistical analysis.

[0035] According to the characteristic attributes of logistics tasks, such as task type, origin, destination, mode of transportation, etc., a multidimensional feature vector is constructed as the input of the clustering algorithm. The K-means clustering algorithm is used to calculate the Euclidean distance between the task feature vectors, and the tasks with high similarity are divided into the same cluster to obtain K task clusters. For each logistics task in the task cluster, the execution efficiency-related data, such as task completion time, transportation timeliness, cargo damage rate and other indicators, are obtained from HDFS. By statistically analyzing the efficiency indicators of the tasks in the cluster, the efficiency mean and standard deviation of the cluster are calculated to obtain the efficiency distribution characteristics of the cluster. According to the efficiency distribution characteristics, the normal range threshold of the cluster efficiency is determined. If the efficiency indicator of a task exceeds the threshold, it is determined to be an abnormal task. For tasks determined to be abnormal, an intelligent early warning signal is generated, and the abnormal cause analysis process is triggered. The task feature attributes are associated with the early warning signal and stored. The association rule mining algorithm is used to analyze the association between abnormal task characteristics and abnormal causes, and the association rules of abnormal causes are obtained. The clustering results, abnormal task distribution, association rules and other information are stored in the database, and a logistics monitoring dashboard is generated through visualization components. The dashboard is interconnected with the logistics scheduling system to display real-time monitoring information on logistics task execution efficiency, and provide decision support for scheduling optimization based on abnormal warnings and association rules.

[0036] Exemplarily, the characteristic attributes of logistics tasks include task type (such as distribution, transshipment, etc.), latitude and longitude coordinates of the starting point and destination, and transportation methods (such as roads, railways, etc.), which are converted into multidimensional feature vectors. The K-means clustering algorithm is used, K=10 is set, and 10 task clusters are obtained by calculating the Euclidean distance between task feature vectors and iteratively optimizing the cluster center. For each cluster, the efficiency indicators such as completion time and punctuality of the corresponding task are obtained from HDFS, and the mean and standard deviation of the efficiency of the tasks within the cluster are calculated. Based on the normal distribution assumption, the efficiency threshold is determined by "mean ± 3 times the standard deviation", and an early warning signal is generated for abnormal tasks that deviate from the threshold. The Aprior association rule mining algorithm is used to mine the association rules between abnormal task features and abnormal causes with parameters of minimum support 0.05 and minimum confidence 0.8, such as "cargo type = fragile goods and transportation route = mountainous area → abnormal occurrence". The clustering results, anomaly distribution, and association rules are stored in the MySQL database, and Tableau is used to draw a logistics monitoring dashboard to display the efficiency distribution and anomaly warning information of the task cluster in real time. It also communicates with the logistics scheduling system to achieve real-time monitoring and optimization of the entire process.

[0037] Step S106: For each cluster, determine the threshold of the normal efficiency range according to its efficiency distribution characteristics. If the efficiency index of a task exceeds the threshold, it is judged as an abnormal task and an intelligent warning signal is generated. Use the association rule mining algorithm to analyze the association between the characteristic attributes of the abnormal task and the abnormal cause, and obtain the association rules of the abnormal cause to guide the processing of abnormal tasks.

[0038] For each task cluster, the efficiency indicators such as the average completion time, punctuality rate, and cargo damage rate of the tasks in the cluster are calculated. Based on the normal distribution assumption, the upper and lower efficiency thresholds are determined by the 3σ principle. If the efficiency indicator of a task exceeds the threshold, it is determined as an abnormal task, triggering an early warning and pushing it to the logistics scheduling system to generate an intelligent early warning signal. The Aprior association rule mining algorithm is used to analyze the correlation between the transportation path, weather conditions, cargo type and other factors of the abnormal task and the occurrence of abnormalities. By calculating the support and confidence of each factor combination, a strong association rule with a minimum support of 0.5 and a minimum confidence of 0.8 is screened out. According to the mined association rules, the association relationship between the characteristic attributes of the abnormal task and the abnormal cause is determined to obtain the association rules of the abnormal cause. The abnormal cause association rules are converted into judgment statements in the form of IF-THEN, such as "IF cargo type = fragile goods AND transportation path = mountain area THEN abnormal occurrence". For the newly collected logistics task data, by applying the abnormal cause association rules, it is possible to judge in real time whether the task has abnormal risks and estimate the possible abnormal causes. The warning information of abnormal tasks and the estimated abnormal causes are pushed to the logistics dispatch system and task executors to guide the handling of abnormal tasks. Through visualization tools such as Tableau, the distribution of abnormal tasks and the associated network of abnormal causes are displayed in the logistics monitoring dashboard for managers to make decisions and analysis.

[0039] For example, for each task cluster, the efficiency indicators such as the average completion time, punctuality rate, and cargo damage rate of the tasks in the cluster can be calculated. For example, the average completion time is 24 hours, the punctuality rate is 95%, and the cargo damage rate is 1%. Then, based on the normal distribution assumption, the upper and lower thresholds of efficiency are determined according to the 3σ principle, that is, the normal range is (μ-3σ, μ+3σ), and the corresponding completion time threshold is (18,30) hours. If the completion time of a task exceeds this range, such as 36 hours, it is determined as an abnormal task, triggering an early warning and pushing it to the logistics scheduling system. The Aprior association rule mining algorithm is used to analyze the correlation between the transportation path, weather conditions, cargo type and other factors of the abnormal task and the occurrence of the abnormality. By calculating the support and confidence of each factor combination, a strong association rule with a minimum support of 0.5 and a minimum confidence of 0.8 is selected, such as "Cargo type = fragile goods ∧ Transportation path = mountainous area → abnormal occurrence", with a confidence of 0.85. According to the mined association rules, the association between the characteristic attributes of abnormal tasks and the abnormal causes is determined, and converted into judgment statements in the form of IF-THEN. For newly collected logistics task data, by applying the abnormal cause association rules, it is judged in real time whether the task has abnormal risks and the possible abnormal causes are estimated, such as "cargo type = fragile goods, transportation route = mountainous area, then the abnormal risk is 85%, and the possible cause is cargo damage". The warning information and estimated causes are pushed to the logistics scheduling system and task executors. Finally, through visualization tools such as Tableau, the distribution of abnormal tasks and the association network of abnormal causes are displayed in the logistics monitoring dashboard. For example, abnormal task points are marked in the task cluster efficiency distribution diagram, and linked with the association rule network diagram for managers to make decisions and analysis.

[0040] Step S107, integrating the cluster analysis results, abnormal task warning signals, and abnormal cause association rules into the logistics monitoring dashboard to achieve real-time monitoring of logistics task execution efficiency and abnormal warning, and provide decision support for logistics scheduling optimization.

[0041] According to the characteristic attributes of logistics tasks, the K-means clustering algorithm is used to cluster logistics tasks, and similar tasks are divided into the same cluster. For each logistics task in each cluster, its execution efficiency index is calculated, such as task completion time, transportation timeliness, cargo damage rate, etc., and the efficiency distribution characteristics of the cluster are obtained through statistical analysis. For each cluster, the threshold of the normal efficiency range is determined according to its efficiency distribution characteristics. If the efficiency index of a task exceeds the threshold, it is determined as an abnormal task and an intelligent early warning signal is generated. The association rule mining algorithm is used to analyze the association between the characteristic attributes of abnormal tasks and the abnormal causes, and the association rules of abnormal causes are obtained. Through visualization tools such as Tableau, a logistics monitoring dashboard is designed to graphically display the efficiency distribution, abnormal task distribution, association rules and other information of the task cluster. The logistics monitoring dashboard is interconnected with the logistics scheduling system to realize the closed loop of real-time monitoring, early warning and scheduling optimization. When the logistics monitoring dashboard detects an abnormal task warning signal, it automatically triggers the task priority adjustment and vehicle rescheduling process of the logistics scheduling system. The logistics scheduling system dynamically adjusts the priority of the affected logistics tasks according to the severity of the warning signal and the association rules of the abnormal cause, and re-optimizes the vehicle scheduling. The optimized logistics task priorities and vehicle scheduling plans are fed back to the logistics monitoring dashboard in real time, forming a dynamic closed loop of monitoring, early warning and scheduling optimization, and continuously improving logistics operation efficiency.

[0042] For example, by installing sensors and cameras in warehouses, transport vehicles, distribution centers, etc., real-time data on logistics task execution, including in-and-out time, transport routes, delivery status, etc., is collected and stored in the Hadoop distributed file system HDFS through network transmission. The amount of data collected every day can reach 100GB. Based on the multi-dimensional attributes of logistics tasks, the K-means clustering algorithm is used to iteratively optimize the clustering center by calculating the Euclidean distance between tasks until the clustering results converge to form K task clusters. For each task cluster, the efficiency indicators such as the average completion time, punctuality rate, and cargo damage rate of the tasks in the cluster are calculated. Based on the normal distribution assumption, the upper and lower thresholds of efficiency are determined by the 3σ principle. For abnormal tasks that deviate from the threshold, an early warning is triggered and pushed to the logistics scheduling system. The Aprior association rule mining algorithm is used to analyze the correlation between factors such as the transportation route, weather conditions, and cargo type of abnormal tasks and the occurrence of abnormalities, and generate strong association rules with a minimum support of 0.5 and a minimum confidence of 0.8, such as "cargo type = fragile goods ∧ transportation route = mountainous area → abnormal occurrence", to guide the analysis and processing of abnormal causes. Through visualization tools such as Tableau, a logistics monitoring dashboard is designed to graphically display information such as the efficiency distribution of task clusters, abnormal task distribution, and association rules, and communicate with the logistics scheduling system. When the logistics monitoring dashboard detects an abnormal task warning signal, it automatically triggers the task priority adjustment and vehicle rescheduling process of the logistics scheduling system. According to the severity of the warning signal and the association rules of the abnormal cause, the priority of the affected logistics tasks is dynamically adjusted, and the vehicle scheduling is re-optimized. The optimized logistics task priority and vehicle scheduling plan are fed back to the logistics monitoring dashboard in real time, forming a dynamic closed loop of monitoring warning and scheduling optimization, and continuously improving the efficiency of logistics operations.

[0043] Step S108, obtain the intelligent warning signal, determine the severity and impact range of the warning signal, and determine the logistics tasks and vehicles that need to be adjusted; according to the type of warning signal and the urgency of the logistics task, use the decision tree algorithm to dynamically adjust the priority of the logistics task; through the vehicle scheduling strategy optimization model, combined with the logistics task priority and vehicle status, generate the optimal vehicle scheduling plan, optimize the allocation of logistics resources, and improve the overall logistics efficiency.

[0044] Obtain intelligent early warning signals, determine the severity and impact range of the early warning signals, and determine the logistics tasks and vehicles that need to be adjusted. According to the type of early warning signals and the urgency of logistics tasks, the decision tree algorithm is used to dynamically adjust the priority of logistics tasks to obtain the adjusted priority of logistics tasks. The optimal vehicle scheduling plan is generated through the vehicle scheduling strategy optimization model, combined with the adjusted logistics task priority and vehicle status. Based on the resource allocation model, the spatial distribution of logistics resources is optimized by comprehensively considering the adjusted logistics tasks, the optimal vehicle scheduling plan and the layout of storage outlets, and the optimized logistics resource allocation plan is obtained. Using the reinforcement learning algorithm, according to real-time logistics operation data and efficiency feedback, the resource allocation strategy and vehicle scheduling strategy are continuously optimized to obtain a dynamically optimized logistics operation strategy. The optimized logistics task priority, vehicle scheduling plan and resource allocation strategy are sent to each logistics node to guide logistics operations and ensure that logistics tasks are efficiently executed according to priority. Real-time monitoring of logistics operation status, evaluation of the execution effect of the optimization strategy, and judgment of whether the expected logistics efficiency improvement goal is achieved. If the logistics efficiency does not reach the expected goal, the monitored logistics operation status data is used as feedback, and the process returns to step 5 to continue optimizing the logistics operation strategy. If the logistics efficiency reaches the expected target, the optimized strategy will be solidified to form a standardized logistics emergency optimization process, and the logistics operation status will be continuously monitored to achieve intelligent and dynamic logistics operation optimization and comprehensively improve logistics efficiency.

[0045] For example, when the system obtains the intelligent warning signal of a serious traffic accident, by analyzing the accident location, the number of accident vehicles and other information, it is judged that the severity of the accident is level 8 (level 10 is the highest), and the impact range is 5 kilometers. It is determined that the logistics tasks of 3 distribution centers and 50 distribution points in the surrounding area and 200 distribution vehicles need to be adjusted. According to the type of warning signal as a traffic accident, the system analyzes the order information of the affected logistics tasks, customer importance and other factors, uses the decision tree algorithm to calculate the urgency of each logistics task, and dynamically generates the adjusted logistics task priority. Then, the system uses the vehicle scheduling strategy optimization model to comprehensively consider the logistics task priority, vehicle location, load, fuel consumption and other states, and runs the genetic algorithm for 500 iterations to obtain the optimal vehicle scheduling plan under the current situation. At the same time, based on the resource allocation model, the system analyzes the distribution of logistics tasks, the optimal vehicle scheduling plan, the inventory of storage outlets and other data, and uses the ant colony algorithm to optimize the distribution ratio of logistics resources in each outlet to improve the overall operation efficiency of the logistics network. During the logistics operation process, the system continuously collects real-time data such as the GPS trajectory and delivery progress of each vehicle, and uses the deep reinforcement learning algorithm DQN to update the policy network every 5 minutes to continuously optimize resource allocation and vehicle scheduling strategies. The optimized logistics task priority, vehicle scheduling plan, and resource allocation strategy are automatically sent to the WMS and TMS systems of each logistics node through the API interface to guide front-line employees to carry out logistics operations. The system monitors the KPI data such as the outbound time, in-transit time, and delivery on-time rate of each logistics node in real time, and evaluates the actual effect of the optimization strategy by comparing with the historical data of the same period. If the expected logistics efficiency improvement target is not achieved, a new round of strategy iteration optimization will be automatically triggered. When the system monitors that the logistics efficiency is stable at the target level for three consecutive cycles, the optimization strategy will be solidified and continuously monitored to form a standardized intelligent logistics emergency response process to comprehensively improve the logistics service level under extreme circumstances.

[0046] The above description is merely a preferred embodiment of one or more embodiments of the present specification and is not intended to limit one or more embodiments of the present specification. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present specification shall be included in the scope of protection of one or more embodiments of the present specification.

Claims

1. A cement industry Internet of Things smart logistics system, characterized in that: include: Obtain the original data of purchase orders, production plans, sales orders and delivery orders, parse the original data, extract key field information, and convert the format according to the preset data standardization protocol to obtain standardized data; Based on the standardized purchase order data and production plan data, the association rule algorithm is used to determine the correlation between the two, establish a mapping relationship, and realize the information connection between the procurement and production links; Based on the standardized sales order data and delivery note data, the decision tree algorithm is used to determine the correlation between the two, establish a mapping relationship, and realize the information connection between sales and logistics links; Obtain vehicle positioning data and loading and unloading confirmation information, generate real-time tracking records of the entire logistics process through data fusion processing, and generate visual trajectories in combination with map information; For massive logistics data, a distributed data processing platform is established, clustering algorithms are used to analyze the execution efficiency of logistics tasks, identify abnormal tasks and generate intelligent early warning signals; According to the intelligent early warning signals, the priority of logistics tasks and vehicle scheduling strategies are adjusted to optimize the allocation of logistics resources and improve the overall logistics efficiency.

2. The system according to claim 1, characterized in that After obtaining the original data of the purchase order, production plan, sales order and bill of lading, it also includes: Preprocess the raw data, including data cleaning, data conversion, and data integration, to ensure data quality and convert the data into an input format suitable for machine learning algorithms.

3. The system according to claim 1, characterized in that The method of using an association rule algorithm to determine the association between purchase order data and production plan data includes: Set the minimum support and minimum confidence thresholds, mine the frequent item sets and association rules between purchase order data and production plan data, calculate the support and confidence of the rules, and filter out strong association rules.

4. The system according to claim 1, characterized in that The use of a decision tree algorithm to determine the correlation between the sales order data and the bill of lading data includes: The amount of sales orders, product categories, customer levels, etc. are used as feature variables, and whether there is a corresponding delivery note is used as the target variable. The decision tree is generated using the ID3, C4.5 or CART algorithm, and the decision tree is pruned and optimized.

5. The system according to claim 1, wherein: After obtaining the vehicle positioning data and the loading and unloading confirmation information, the method further includes: The Kalman filter algorithm is used to remove noise and smooth the trajectory of vehicle positioning data to improve positioning accuracy and stability; The video surveillance data of the loading and unloading scenes is analyzed through machine vision algorithms to automatically identify the loading and unloading status.

6. The system according to claim 1, characterized in that The clustering algorithm is used to analyze the execution efficiency of logistics tasks for massive logistics data, including: According to the characteristic attributes of logistics tasks, K-means clustering algorithm is used to cluster tasks; For each task in each cluster, calculate efficiency indicators such as completion time, timeliness, and cargo damage, and obtain efficiency distribution characteristics through statistical analysis; According to the efficiency distribution characteristics, set the abnormal threshold, identify abnormal tasks below the threshold, and trigger the early warning mechanism.

7. The system according to claim 1, characterized in that The method of adjusting the priority of logistics tasks and vehicle dispatching strategies according to the intelligent early warning signals includes: According to the type and urgency of the warning signal, the decision tree algorithm is used to dynamically adjust the task priority; Through the vehicle scheduling optimization model, the optimal scheduling solution is solved by combining task priority and vehicle status; Utilize reinforcement learning algorithms to continuously optimize scheduling strategies based on real-time logistics operation data.

8. The system of claim 1, wherein: The system further comprises: Analyze historical logistics data through machine learning algorithms, predict future logistics needs, and generate optimized purchase orders and production plans; The actual logistics data is fed back into the historical database, the data set is updated regularly and the prediction model is retrained to form a business closed loop and continuously optimize the accuracy of demand forecasting and plan generation.

Images