KR20220072311A - Method for designing intelligent integrated logistics platform - Google Patents

Abstract

An intelligent comprehensive logistics platform design method is disclosed. According to an embodiment of the present invention, a method for designing an intelligent integrated logistics platform includes: managing a core verification field task for any one step of reception, classification, transportation, delivery, and complex in postal logistics; and designing a platform capable of integrated operation by using the step-by-step core demonstration field task.

Description

How to design an intelligent integrated logistics platform

The present invention relates to a method and apparatus for designing an intelligent comprehensive logistics platform for building and demonstrating an intelligent comprehensive logistics platform using 5G edge cloud.

In particular, in the present invention, an intelligent comprehensive logistics platform for managing the entire core verification field tasks (total of 7 detailed tasks) for each stage such as reception, classification, transportation, delivery, and compounding in postal logistics, and designing a platform capable of integrated operation A design method and apparatus are provided.

As the world's first 5G commercialization service for 5G communication technology started on April 3, 2019, related industries such as autonomous vehicles, robots, and VR are expected to grow significantly. 5G has three main characteristics: 'ultra-speed', 'ultra-low latency', and 'super-connection'. The maximum transmission speed of 5G is 20 Gbps, which is 20 times faster than 4G LTE (Long Term Evolution) (it takes 16 seconds for LTE and 0.8 seconds for 5G to download about 2GB of HD movie). In the case of ultra-low latency, which is the most important feature of autonomous vehicles, LTE has a latency of 10 to 20 ms, while 5G has a latency of 1 ms, which can be improved up to 10 times or more. As a feature of hyper-connectivity, 5G allows up to 1 million devices to be connected simultaneously within a radius of 1 km.

The characteristics of 5G such as ultra-high speed, ultra-low latency, and ultra-connection are the era of wireless operation, control, and monitoring of vehicles, logistics automation equipment, transport equipment (forklifts, AGVs, etc.), cooperative logistics robots, etc. to come

With the advantages of 5G, the technologies applicable to the postal logistics field are autonomous vehicles, autonomous delivery delivery droids (robots), and autonomous driving loading and unloading devices. 5G communication enables various information such as spatial information, roads, traffic, and objects in roads, distribution centers, and delivery sites to communicate and exchange information with lower latency in connection with all surrounding objects and devices. The introduction of 5G technology will allow vehicles and robots (devices, droids, etc.) that can autonomously drive more safely by receiving data at high speed from surrounding roads, traffic information, surrounding vehicles, location information, pallets in distribution centers, people, and mechanical devices. to operate, etc. As high-speed data transmission, ultra-low latency, and hyper-connection enable responses to emergencies, vehicles, robots, and droids can safely autonomously drive.

The 5G edge cloud-based intelligent comprehensive logistics platform requires an environment that can manage the 5G communication-applied autonomous vehicle, loading and unloading equipment, etc. in 'receipt → transport → sorting → delivery' throughout the entire postal logistics process. In order to operate autonomous mobile post offices, autonomous delivery delivery droids (robots), and various autonomous driving robots and equipment in logistics centers, an intelligent logistics information system that can command, cancel, and monitor work status is required. . The 5G edge cloud-based intelligent comprehensive logistics platform must establish a mobile edge cloud (MEC) to create an intelligent logistics platform environment that can collect and process data from each terminal in the postal logistics process. 5G edge cloud-based logistics for three core functions, namely, standardization linkage with each terminal for each logistics process (step), large-capacity data collection, and terminal data collection and processing such as various vehicles and devices for building and demonstrating an intelligent comprehensive logistics platform It is necessary to establish a management platform and establish an integrated management system to demonstrate service linkage in the entire postal distribution cycle.

Accordingly, there is a need to advance postal logistics and public service by establishing an intelligent comprehensive postal logistics management platform that can manage various 5G-based logistics equipment applied to the entire postal logistics cycle.

An embodiment of the present invention aims to build and demonstrate an intelligent comprehensive logistics platform using 5G-based logistics ICT data.

In addition, an embodiment of the present invention aims to enable the establishment of an integrated management platform for key verification fields such as reception, classification, transportation, delivery, and complex step-by-step.

In addition, an embodiment of the present invention provides monitoring, failure response, emergency situation response, call center video connection, etc. The purpose is to develop this possible control technology.

In addition, the embodiment of the present invention aims to enable performance analysis such as business connection verification promotion and work performance capability (work throughput, improvement compared to existing equipment, etc.) at the postal logistics site of the technologies developed in each detailed task. do.

In addition, an embodiment of the present invention aims to collect data such as self-driving postal vehicles, parcel delivery droids, and smart logistics centers that are active in the entire postal logistics process from postal logistics reception to delivery in real time and utilize it as a comprehensive control system. do it with

In addition, an embodiment of the present invention aims to identify the real-time status based on data collected from autonomous vehicles, droids, robots, various IoT devices, distribution centers, etc. and apply it to comprehensive control through dashboard monitoring status. .

In addition, an embodiment of the present invention aims to provide a data-based insight necessary for quantity or operation by establishing a decision support system based on data collected in the control system.

In addition, an embodiment of the present invention aims to develop various statistical algorithms and prediction techniques based on data accumulated in an intelligent logistics platform and utilize them as an auxiliary tool for business decision-making.

In addition, the embodiment of the present invention aims to utilize it in discovering new business models through the accumulation of smart technologies such as autonomous vehicle technology, heavy-duty goods transport technology, and small fulfillment logistics center.

In addition, an embodiment of the present invention aims to contribute to the advancement of the existing postal service and the advancement of public service through linkage with legacy systems such as PostNet and various postal related application services.

In addition, an embodiment of the present invention aims to prepare 5G edge cloud-based integrated operation and monitoring technology for the detailed task of strengthening ICT convergence competitiveness by developing postal logistics integrated control platform technology.

In addition, an embodiment of the present invention aims to apply the autonomous driving logistics equipment to the postal logistics site by applying 5G to the postal logistics site.

In addition, an embodiment of the present invention aims to secure an integrated operation function by developing technology for controlling and controlling autonomous driving-based logistics devices such as autonomous vehicles and droids on a platform.

In addition, an embodiment of the present invention aims to provide an ICT-based safety technology application system that identifies, avoids, and suppresses safety risk factors in the entire postal distribution process.

In addition, an embodiment of the present invention aims to establish a safe working environment for postal logistics field workers through the introduction of automated safety technology.

In addition, an embodiment of the present invention aims to improve the intensity of labor-intensive work by deriving a work improvement plan that can satisfy field workers through demonstration of new ICT convergence logistics technology for solving postal logistics field problems.

In addition, an embodiment of the present invention aims to enable productivity improvement through postal logistics efficiency by demonstrating ICT convergence logistics technology in the postal logistics field.

In addition, an embodiment of the present invention aims to improve system efficiency and productivity in the existing post office logistics center by introducing smart logistics center functions such as sorter automation, loading and unloading automation, and pallet collection automation.

In addition, an embodiment of the present invention aims to reduce the time required for business processing and improve the efficiency of the divisional delivery business through convenient functions, intelligence, and automation technology centered on postal logistics workers.

According to an embodiment of the present invention, a method for designing an intelligent integrated logistics platform includes: managing a core verification field task for any one step of reception, classification, transportation, delivery, and complex in postal logistics; and designing a platform capable of integrated operation by using the step-by-step core demonstration field task.

According to an embodiment of the present invention, it is possible to build and demonstrate an intelligent comprehensive logistics platform using 5G-based logistics ICT data.

In addition, according to the present invention, it is possible to build an integrated management platform for core verification fields such as reception, classification, transportation, delivery, and complex.

In addition, according to the present invention, the movement path monitoring, failure response, emergency situation response, call center video connection, etc. Possible control technology can be developed.

In addition, according to the present invention, it is possible to perform performance analysis such as business linkage verification promotion and work performance capability (work throughput, improvement compared to existing equipment, etc.) at the postal logistics field of the technologies developed in each detailed task.

In addition, according to the present invention, it is possible to collect data such as self-driving postal vehicles, parcel delivery droids, and smart logistics centers that are active in the entire postal logistics process from postal logistics reception to delivery in real time and utilize it as a comprehensive control system.

In addition, according to the present invention, it is possible to grasp the real-time status based on data collected from autonomous vehicles, droids, robots, various IoT devices, distribution centers, etc., and apply it to comprehensive control through the dashboard monitoring status.

In addition, according to the present invention, it is possible to provide a data-based insight necessary for quantity or operation by establishing a decision support system based on data collected in the control system.

In addition, according to the present invention, various statistical algorithms and prediction techniques can be developed based on the data accumulated in the intelligent logistics platform and utilized as an auxiliary tool for business decision-making.

In addition, according to the present invention, it can be utilized for discovering new business models through the accumulation of smart technologies such as autonomous vehicle technology, heavy-duty transport technology, and small fulfillment logistics center.

In addition, according to the present invention, it is possible to contribute to the advancement of the existing postal service and the advancement of public service through linkage with legacy systems such as PostNet and various postal-related application services.

In addition, according to the present invention, it is possible to prepare 5G edge cloud-based integrated operation and monitoring technology for the detailed task of strengthening ICT convergence competitiveness by developing the postal logistics integrated control platform technology.

In addition, according to the present invention, by applying 5G to the postal logistics site, autonomous driving logistics equipment can be applied to the postal logistics site.

In addition, according to the present invention, it is possible to secure an integrated operation function by developing a technology for controlling and controlling autonomous driving-based logistics devices such as autonomous vehicles and droids on the platform.

In addition, according to the present invention, it is possible to prepare an ICT-based safety technology application system that identifies, avoids, and suppresses safety risk factors in the entire postal distribution process.

In addition, according to the present invention, it is possible to establish a safe working environment for postal logistics field workers through the introduction of automated safety technology.

In addition, according to the present invention, it is possible to improve the intensity of labor-intensive work by deriving a work improvement plan that can satisfy field workers through the demonstration of new ICT convergence logistics technology for solving postal logistics field problems.

In addition, according to the present invention, it is possible to improve productivity through postal logistics efficiency by demonstrating ICT convergence logistics technology in the postal logistics field.

In addition, according to the present invention, system efficiency and productivity improvement in the existing post office logistics center can be expected by introducing smart logistics center functions such as sorter automation, loading and unloading automation, and pallet collection automation.

In addition, according to the present invention, it is possible to reduce the time required for business processing and improve the efficiency of the divisional delivery business through convenient functions, intelligence, and automation technology centered on postal logistics workers.

1 is a diagram for explaining a key verification field for each logistics step according to the present invention.
2 is a diagram for explaining the goals and objectives of the data linkage management system.
3 is a block diagram of a data linkage management system.
4 is a diagram for explaining the principle of the data linkage management system.
5 is a diagram for explaining a data linkage management system process.
6 is a diagram showing an example of data processing.
7 is a diagram for explaining the structure of a message queue.
8 is a diagram for explaining a configuration diagram of a real-time information processing agent pipeline.
9 is a functional structural diagram of an intelligent integrated control system.
10 is a data distribution storage structure diagram of a large-capacity IoT high-speed search module based on distributed processing.
11 is an exemplary diagram of an open source dashboard for real-time visualization.
12 is a diagram to summarize the technical contents according to an embodiment for the technology of the intelligent comprehensive logistics platform design device.
13 is a conceptual diagram of a data linkage structure.
14 is a diagram for explaining a batch data collection method.
15 is a diagram for explaining a real-time data collection method.
16 is a diagram illustrating divided data stream processing.
17 is a diagram for explaining supported types of databases.
18 is a diagram illustrating an outlier.
19 is a diagram for explaining an outlier detection classification.
20 is a diagram for explaining types of identification prevention technology.
21 is a diagram illustrating application of identification prevention technology.
22 is a diagram for illustrating an ARX de-identification process.
23 is a diagram for illustrating a 5G real-time data diagnosis technology.
24 is a diagram for explaining the configuration of a data pipeline environment.
25 is a diagram illustrating a configuration diagram of a producer and consumer connection.
26 is a diagram for illustrating a data section encryption method.
27 is a diagram illustrating a distributed processing large-capacity message queue (Message) monitoring process.
28 is a diagram for illustrating data transmission monitoring.
29 is an intelligent integrated control structure diagram.
30 is a diagram illustrating a distributed processing-based large-capacity IoT high-speed search module architecture.
31 is a diagram for explaining the configuration of real-time monitoring of logistics and delivery information.
32 is a diagram for explaining the configuration of the distribution center information monitoring.
33 is a configuration diagram of an intelligent efficiency analysis technology.
34 is a diagram for explaining the basic structure of an autoencoder.
35 is a diagram for explaining a reinforcement learning structure.
36 is a diagram for explaining hyperparameter optimization.
37 is a diagram illustrating a Bayesian Optimization process.
38 is another diagram for illustrating a Bayesian Optimization process.
39 is a configuration diagram of an intelligent quantity prediction technology.
40 is a diagram for explaining the principle of the learning process of the GAN.
41 is a configuration diagram of an intelligent abnormality diagnosis technology.
42 is a diagram for illustrating HDBSCAN classification.
43 is a configuration diagram of an intelligent performance analysis technology.
44 is a diagram for illustrating variable importance calculation using a random forest.
45 is a diagram for illustrating correlation analysis.
46 is a diagram for explaining the configuration of a postal distribution information system (PostNet).
47 is a diagram for explaining a configuration (To-Be) of a postal distribution information system (PostNet).
48 is a layout diagram of the Daedeok Research Complex and the Korea Advanced Institute of Science and Technology in the area (plan) to be promoted.
49 is a diagram to summarize the technical contents according to another embodiment for the technology of the intelligent comprehensive logistics platform design device.
50 is a diagram for explaining a type of a Data Imputation model.
51 is a diagram for explaining the configuration of a customer center connection module.
52 is a diagram for illustrating a Video Stream Consumer architecture.
53 is a diagram for explaining the configuration of autonomous driving freight vehicle/courier droid movement path monitoring.
54 is a block diagram of module status check monitoring and integrated control.
55 is a diagram for explaining the technology development of each element of intelligent logistics analysis.
56 is a diagram for explaining interworking of an intelligent integrated control system.
57 is a diagram for explaining advancement of intelligent logistics analysis technology.
58 is a diagram for explaining the configuration (As-Is) of the postal distribution information system (PostNet).

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. However, since various changes may be made to the embodiments, the scope of the patent application is not limited or limited by these embodiments. It should be understood that all modifications, equivalents and substitutes for the embodiments are included in the scope of the rights.

The terms used in the examples are used for the purpose of description only, and should not be construed as limiting. The singular expression includes the plural expression unless the context clearly dictates otherwise. In this specification, terms such as "comprise" or "have" are intended to designate that a feature, number, step, operation, component, part, or a combination thereof described in the specification exists, but one or more other features It should be understood that this does not preclude the existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiment belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

In addition, in the description with reference to the accompanying drawings, the same components are given the same reference numerals regardless of the reference numerals, and the overlapping description thereof will be omitted. In describing the embodiment, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the embodiment, the detailed description thereof will be omitted.

1 is a diagram for explaining a key verification field for each logistics step according to the present invention.

As shown in Figure 1, the intelligent integrated logistics platform design device develops a 5G edge cloud-based intelligent logistics management platform to provide an operating system that enables integrated control of the entire logistics process.

The intelligent integrated logistics platform design device is an integrated management platform system for key verification fields such as reception, classification, transportation, delivery, and complex.

The intelligent integrated logistics platform design device develops technologies for data collection, analysis, management and monitoring in logistics devices such as autonomous vehicles, delivery devices, and automatic loading and unloading devices developed in the ICT convergence logistics technology development task.

The intelligent comprehensive logistics platform design device utilizes all data generated from the platform to predict and analyze the receipt volume and delivery time based on artificial intelligence and machine learning prediction methodology, and supports intelligent performance analysis through automatic reporting of the contents. .

The intelligent integrated logistics platform design device can develop control technologies such as autonomous driving vehicles and droids, etc. movement path monitoring, failure response, emergency response, and video connection with call centers.

The intelligent integrated logistics platform design device provides an integrated control platform that responds in real time to customer requests by utilizing a mobile unmanned post office and autonomous driving droid.

The intelligent comprehensive logistics platform design device promotes the business connection demonstration of R&D service developed technologies through connection simulation with the existing postal logistics system.

The intelligent integrated logistics platform design device can utilize ICT convergence logistics technology to reduce the work of postal logistics field workers and prevent safety accidents.

The intelligent integrated logistics platform design device can build a platform capable of integrated operation by introducing unmanned and autonomous logistics devices for reducing the work of postmen and field workers and preventing safety accidents.

The intelligent integrated logistics platform design device can perform comprehensive management tasks that can be demonstrated at the postal logistics site by synthesizing ICT convergence logistics technology tasks (7 cases).

The intelligent integrated logistics platform design device will promote the development of technologies that can demonstrate technological performance in the field through communication with the 7 project organizers and participating organizations by addressing the needs for ICT convergence technology at the postal logistics site of the Postal Service Headquarters. can

The intelligent integrated logistics platform design device can establish a standardized data linkage management system.

2 is a diagram for explaining the goals and objectives of the data linkage management system.

The intelligent integrated logistics platform design device requires a data linkage management system that introduces a new standardization strategy to actively respond to the 4th industrial revolution era and lead the future ICT technology and standardization.

The intelligent integrated logistics platform design device develops a standardization response strategy based on hyper-connection, hyper-realistic, and super-intelligent technology to respond to future ICT environmental changes.

The purpose of the data linkage management system is to aim for the use of quality data without interruption through data flow and tracking management, and linkage standard management functions. The data linkage management system supports the selection of CTQ (Critical to Quality) that can measure and analyze consumer needs for consumer satisfaction in terms of data utilization.

The data linkage management system manages data transmission and reception for linkage management of all stages of the data life cycle (transmission, reception, use), and improves and manages linkage by designing a process based on the PDCA (Plan-Do-Check-Act) cycle. Strengthen your competencies. The data linkage management system provides flow information such as utilization analysis using information on the source of use to support activation of use and quality improvement activities, and provides tracking information such as impact and cause analysis using source information. The data linkage management system carries out the collection, storage, analysis/processing process by applying the big data life cycle to the internal linkage to reflect the characteristics of big data.

The data linkage management system provides a standardized code conversion module configured by using an open source based on DCAT, an international metadata standard, for efficient linkage management according to standards.

3 is a block diagram of a data linkage management system.

4 is a diagram for explaining the principle of the data linkage management system.

The principle of data linkage management system manages data transmission and reception and tracking based on standardized linkage metadata and PDCA (Plan-Do-Check-Act) cycle in order to utilize high-quality data without interruption by both data providers and users. It is to contribute to the promotion of data use vitalization, support for quality improvement activities, and national data linkage activities, and it consists in deriving it into 6 principles and 12 detailed principles.

The data linkage management system management function subdivides the linkage management function by applying the PDCA (Plan-Do-Check-Act) cycle to the three linkage management functions. The data linkage management system management function defines 12 linkage management processes from the subdivided linkage management function.

5 is a diagram for explaining a data linkage management system process.

The intelligent integrated logistics platform design device can develop 5G edge cloud-based real-time information processing agent technology.

The intelligent comprehensive logistics platform design device is a '5G system that processes and processes large-capacity ICT logistics data generated by devices (droids, robots, autonomous vehicles, etc.) Develop edge cloud-based real-time information processing agent.

5G edge cloud-based real-time information processing agent reduces data load. 5G edge cloud-based real-time information processing agent can reduce the network load for sending and receiving data to the cloud for processing of large-scale data, that is, more than 30 billion data. The 5G edge cloud-based real-time information processing agent can reduce the computing load because the cloud can collect only the necessary result values after pre-processing the data or play a role of filtering and sending the data. 5G edge cloud-based real-time information processing agent can improve service quality by reducing data transmission rate and service waiting time.

1ms) 이용으로 인하여 실시간 서비스의 지원이 가능하다.5G edge cloud-based real-time information processing agent guarantees real-time service. Because the 5G edge cloud-based real-time information processing agent transmits and receives data within a short-distance network, the quality of service can be improved by ensuring real-time processing of data. 5G edge cloud-based real-time information processing agent has ultra-low latency characteristics (

1ms), real-time service support is possible.

5G edge cloud-based real-time information processing agent improves security. The 5G edge cloud-based real-time information processing agent can improve data security by completing the collection and processing of data within a short-distance network.

5G edge cloud-based real-time information processing agent handles high-speed/large-capacity data transmission. 5G edge cloud-based real-time information processing agent can transmit and process large amounts of data to 5G edge servers at high speed due to the expansion of data bandwidth (up to 20 Gbps) according to the use of 5G communication.

The 5G edge cloud-based real-time information processing agent develops a real-time data collection module by composing the Stream Layer, which processes data collection one by one when it arrives at the streaming system, and the Batch Layer, which processes batches of time period by grouping them into units that can be processed.

Streaming processing includes Time agnostic, Filtering, Inner Join, and Windowing methods, and it can be more complicated than arrangement because the order in which data arrives to the system is not sequential as the Skew varies greatly depending on the environment. Time agnostic processes data without time attribute in the order they come in. Filtering saves only specific data among incoming data after filtering. Inner Join distinguishes values by comparing and matching incoming values from two unlimited data sets. Windowing processes streaming data at regular time intervals.

Batch is a method of collecting and processing streaming data in a certain time unit, and although it is simple to implement, it is inferior to real-time processing because it is processed after data collection.

The 5G edge cloud-based real-time information processing agent can diagnose the collected ICT logistics data and develop a technology that can maximize the use of data in control and analysis through data cleansing techniques such as refining, shaping, and correcting outliers and missing values. . The 5G edge cloud-based real-time information processing agent can pre-process data using open source Spark, etc., and can correct outliers and missing values using Spark Streaming and Kafka Stream.

6 is a diagram showing an example of data processing.

The 5G edge cloud-based real-time information processing agent prevents data loss and builds a large-capacity real-time data processing pipeline from collection to storage based on distributed processing message queues. A message queue is a communication method used to exchange data between processes (programs) and refers to a Message Oriented Middleware (MOM) implementation system. When exchanging messages, message queues use AMQP (Advanced Message Queuing Protocol) to communicate with applications using other AMQPs, and data transmission between different networks is possible.

The structure of the message queue is such that the producer puts a message in the queue, and the consumer gets the message from the queue and processes it.

7 is a diagram for explaining the structure of a message queue.

A message queue can be processed later because it is put in a queue (asynchronous), can be separated from the application (decoupling), and when some parts fail, the whole is not affected (resilience) , it can be re-executed in case of failure (Redundancy), it can be confirmed that the operation has been processed (Guarantees), and multiple processes can send messages to the queue (Scalable). has a

A message queue can send and receive data from APIs in other places, asynchronous communication in various applications, sending e-mails and uploading documents, and processing a large amount of processes.

Data pipeline construction utilizes Spark Streaming, Logstash, Fluentd, etc. in the collection part, connects to RabbitMQ and Apache Kafka from Message Queue, a data connection pipeline, and Prometheus, InfluxDB, Prometheus, InfluxDB, and It is stored in OpenTSDB and Elastic Search.

8 is a diagram for explaining a configuration diagram of a real-time information processing agent pipeline.

The intelligent integrated logistics platform design device can develop intelligent integrated control and monitoring technology.

As the issue of big data and social perspectives change, ICT integrated control, which manages ICT services and develops business control, beyond the existing simple IT infrastructure management/control, is drawing attention for large-capacity ICT logistics data generated in the field.

It is necessary to develop an intelligent integrated control monitoring technology that can analyze/control data flow in real time beyond simple infrastructure management.

Intelligent integrated control monitoring technology delivers and manages data such as mail delivery information, autonomous driving mobile post office station status, logistics center loading status, unmanned cargo vehicle operation status, and droid operation status, to the management system and manages the data. Comprehensive monitoring technology.

Intelligent integrated control and monitoring technology, postal logistics mobile reception/delivery', 'logistics collection and distribution work on pallet loading/unloading of logistics center', 'labor load on/off logistics center', 'postal logistics cargo transportation', 'individual parcel delivery service It aims to analyze and monitor in real time the flow of large-scale ICT logistics data that occurs in the field, such as 'delivery', 'support for heavy-duty mobile delivery by postman', and 'automated storage and warehousing using small loading space'.

9 is a functional structural diagram of an intelligent integrated control system.

The control system is developed in the G-Cloud environment for system scalability and data connection with the results of detailed ICT convergence logistics technology tasks (vehicles, droids, loading and unloading equipment, etc.), postal logistics big data centers, etc.

The intelligent integrated control monitoring technology develops a distributed processing-based high-capacity IoT high-speed search module for real-time data flow analysis. Intelligent integrated control monitoring technology utilizes open sources such as Elasticsearch to form a technology environment for processing large-scale data based on distributed processing. The intelligent integrated control and monitoring technology configures a distributed processing-based large-capacity data processing technology environment, so server expansion is easy and high availability is guaranteed. The intelligent integrated control monitoring technology is used to distribute and store large amounts of data subject to ICT logistics management at a high speed.

10 is a data distribution storage structure diagram of a large-capacity IoT high-speed search module based on distributed processing.

The intelligent integrated control monitoring technology is a technology for receiving and monitoring data in real time from a large-capacity IoT high-speed search module based on distributed processing.

The intelligent integrated control monitoring technology utilizes open sources such as Grafana, Kibana, and React to visualize distributed and stored ICT logistics data in real time.

The intelligent integrated control monitoring technology can configure time-series data as a dashboard by connecting to a large-capacity IoT high-speed search module based on distributed processing by using the open source above.

The intelligent integrated control monitoring technology can display indicators of multiple metric systems at the same time on one dashboard, and real-time data such as metrics can be visualized.

The intelligent integrated control monitoring technology provides the abnormal situation analysis result generated through the intelligent logistics analysis technology to provide the abnormal situation analysis result so that it can be referred to for failure response.

11 is an exemplary diagram of an open source dashboard for real-time visualization.

The intelligent integrated logistics platform design device can develop intelligent logistics analysis technology that analyzes and provides accumulated data using the control system.

Intelligent logistics analysis technology can provide intelligent efficiency analysis, intelligent quantity prediction, intelligent anomaly diagnosis, and intelligent performance analysis information by combining analysis technologies such as GAN, deep learning, reinforcement learning, and hyperparameter optimization.

The intelligent logistics analysis technology includes intelligent efficiency analysis technology that analyzes whether the logistics flow is efficiently distributed and feeds back the analyzed contents.

The intelligent logistics analysis technology includes intelligent quantity prediction technology to predict the quantity to occur in a specific area among the data accumulated in the control system and utilize it for scheduling, logistics operation simulation, and maximum input quantity of the logistics transport robot.

The intelligent logistics analysis technology includes intelligent abnormality diagnosis technology that analyzes whether the logistics flow operates smoothly and analyzes the section where sensor abnormalities and overload problems occur.

Intelligent logistics analysis technology includes intelligent performance analysis technology that can be used for additional analysis by providing summarized information and predicting average delivery time using statistical-based analysis of data collected from the control system.

Intelligent logistics analysis technology includes process-based logistics analysis technology that considers the specialized logistics processing environment and equipment for each detailed task.

The intelligent logistics analysis technology includes specialized logistics analysis elements and common KPIs considering the logistics processing method of the logistics unit process for each detailed task and the characteristics of various logistics processing equipment.

Intelligent logistics analysis technology includes intelligent logistics analysis technology for various logistics processing equipment such as logistics transport vehicles, loading and unloading support robots, and delivery droids, and analysis elements identified according to logistic unit work types.

Intelligent logistics analysis technology includes various logistics visualization and reporting functions to provide effective analysis results for logistics processing equipment and analysis elements specialized for logistics business characteristics.

Intelligent logistics analysis technology includes integrated logistics process linkage and intelligent logistics analysis technology for agent tasks such as autonomous vehicles, logistics robots, and delivery droids.

The intelligent logistics analysis technology develops an agent-based logistics process model for the analysis of differentiated logistics environments such as logistics centers and external delivery spaces, which are closed spaces.

The intelligent logistics analysis technology includes intelligent logistics analysis technology using process mining and data mining based on the integrated logistics process in which the logistics unit processes for each detailed task are organically linked.

The intelligent integrated logistics platform design device can build demonstrations and services.

The demonstration and service technology can design a G-Cloud-based system architecture and build a security environment that can reliably perform large-scale traffic processing and multiple connections.

The demonstration and service technology is a data transmission design based on 5G edge computing, and it is necessary to link 7 core field specific tasks, data standards, protocols, etc.

Demonstration and service technology can be linked through the integrated introduction of ROS (Robot Operating System). ROS may be a concept that includes the role of an operating system running on robot hardware from a robot development software platform. ROS can support from hardware design visualization to development libraries and debugging tools.

Demonstration and service technology can define collection data for each detailed task and establish a data collection standardization plan.

Demonstration and service technology can define monitoring necessary items and identify data that must be visually expressed.

Demonstration and service technology can collect requirements definitions and related data for analysis expectations.

Demonstration and service technology can test linkage simulation with linkage with legacy systems such as postal and logistics information systems in mind.

The intelligent comprehensive logistics platform design device is implemented through the G-cloud-based system, and is designed to be linked through the 5G edge cloud server and MEC system linked to each individual technology development task, so whether the information system through the cloud infrastructure works or is implemented can be the main demonstration area.

The intelligent comprehensive logistics platform design device normally collects monitoring data on the operation status collected through the link agent in individual technology development tasks from the server, and it is necessary to check whether central monitoring is possible.

The demonstration and service technology checks whether it is linked with the simulated legacy system.

Demonstration and service technology examines whether log records can be checked.

The verification and service technology checks whether the status data collected through the 5G data pipeline is normally collected, processed, and monitored in the detailed task.

The demonstration and service technology checks the feedback record on whether the predefined control functions such as emergency stop are normally delivered and operated when necessary, such as when an abnormal situation occurs during central monitoring.

The verification and service technology checks whether separate data inquiry, call, and control functions that can be processed like a call center are provided.

The verification and service technology verifies whether performance analysis, trend analysis, and prediction functions using the collected data operate normally.

The intelligent integrated control system provides functions such as dashboard, logistics process efficiency analysis, quantity forecasting, abnormal diagnosis, performance analysis, demand forecasting, and statistical management.

12 is a diagram for organizing the technical contents according to an embodiment for the technology of the intelligent comprehensive logistics platform design device.

1. Standardized data linkage management system

① Analysis and definition of data linkage management system management requirements

Investigate and define requirements related to linked data to reflect requirements for data linkage management system configuration and related detailed tasks, define requirements for analysis expected items, and collect related data

Data linkage process definition and data linkage model development for data linkage management system development

Analysis of data-linked domestic organization standardization trend

② Data link module and standardized code conversion verification module design

Define data details linked to each stage of logistics processing such as autonomous driving mobile post office, logistics center, transportation, and delivery

Design of transformation verification module for linking with intelligent logistics analysis and control system that will use the data collected for each step of logistics processing

13 is a conceptual diagram of a data linkage structure.

2. 5G edge cloud-based information processing agent technology

① Real-time/batch large-capacity data collection technology development

The real-time data collection method consists of two collection methods: batch and streaming.

Batch data collection method

A method of collecting data coming in through streaming at a certain time unit and processing it in batches

14 is a diagram for explaining a batch data collection method.

Stream data collection method

Collect large amounts of data by supporting streaming processing in real time

15 is a diagram for explaining a real-time data collection method.

Developing a real-time data collection module using a distributed data stream processing framework

Provides features such as scalability and fault tolerance

Large data throughput can be increased by adding cluster nodes

Split data stream processing through batches of very small fixed-size time intervals (micro-batches)

Main functions of real-time data collection module

Supports various data stream sources

Provides APIs for various languages

Partition throughput limit per second

Dynamically adjusting the number of input records according to the processing time of the cluster.

16 is a diagram illustrating divided data stream processing.

② 5G real-time data diagnosis technology development

Development of data preprocessing technology

Developed a technology that can refine and format data by receiving data from the data collection module and modifying and extracting data before passing the data to the message queue.

Features of real-time data diagnosis technology

column selection

row filtering

Column aggregation

Join data sets

CSV. Supports various file formats such as JSON, Parquet, ORC

Support for storing various types of data

The performance of real-time data diagnostic technology

Reduce disk I/O

Partitioning function

Column-based storage storage

In-memory column caching

skip low

Query optimization

17 is a diagram for explaining supported types of databases.

Development of missing value detection technology

When real-time data comes in, we developed a technology for detecting missing values using filtering functions such as NooneType, NULL, Blank, Empty String, etc.

Development of outlier detection technology

Anomaly is the opposite of normal, and anomaly detection refers to a method of finding ‘odd’ values that are opposite to normal.

There are three types of anomalies

Point anomaly

Collective anomaly

contextual anomaly

18 is a diagram illustrating an outlier.

There are various types of anomaly detection depending on the type of anomaly.

19 is a diagram for explaining an outlier detection classification.

A family of algorithms for detecting outliers

regression series

time-series series

decision tree series (random forest, Monte Carlo)

bayesian network family

Developed outlier correction technology based on data types and various outlier detection methods

Personal and Sensitive Information Non-Identification Technologies

De-identification refers to a set of measures that make it impossible to identify a specific individual by deleting/replacing some or all of the information or preventing it from being easily combined with other information.

Identifiers: Refers to attributes that can identify an individual (all information that can be matched 1:1)

Quasi-Identifiers (QI): properties that are not identifiers by themselves, but that can be used to indirectly infer a specific individual through combination with other data.

Sensitive Attributes (SA): Refers to attributes that can reveal an individual's privacy (target attributes that are mainly measured during data analysis)

De-identification exists as two principles: anti-identification (removal of identifiers) and anti-inference (according to the privacy model).

Privacy model refers to a methodology that probabilistically/quantitatively limits the degree of personal information inference risk for various inference attacks.

Development of technology to prevent personal identification using general techniques

Pseudonymization: Substituting other values for key identification elements in personal information

Total processing: By showing the total of the data, the value of individual data is not shown.

Data deletion: Delete unnecessary values or important values for personal identification among the values configured in the data set

Data categorization: transforms the values of data into values of categories

Data masking: In combination with public information, etc., it is processed so that the key personal identifier that is highly likely to contribute to identifying an individual is hidden.

20 is a diagram for explaining types of identification prevention technology.

21 is a diagram illustrating application of identification prevention technology.

Development of anti-inference technology using privacy protection model

k-anonymity: A privacy protection model proposed to defend against vulnerabilities such as connection attacks on public data

l-diversity: A model for defending against homogeneity attacks and background knowledge attacks on k-anonymity. Each block must have at least l different types of sensitive information.

t-closeness: A model that protects privacy by making the difference between the distribution of sensitive information in undistinguished records in the data set and the distribution of sensitive information in the entire data set less than t

In de-identification, k-anonymity is used as a basic technique, and if sensitive information is included, the methodologies of l-diversity (diversity) and t-closeness (proximity) should also be considered.

Development of de-identification module using identification element prevention technology and inference prevention technology

Development of de-identification modules using open source programs such as ARX that perform de-identification

22 is a diagram for illustrating an ARX de-identification process.

Configure: The step of importing raw data to set the data transformation model and privacy model

Explore: A step that schematically shows all possible transformations that satisfy the set model.

Analyze: The stage of determining whether to export or not by analyzing the risk level such as the possibility of re-identification

23 is a diagram for illustrating a 5G real-time data diagnosis technology.

③ Development of distributed processing and monitoring technology for large-capacity data

Development of large-capacity data distributed processing technology

Technology that delivers real-time data received from real-time/batch large-capacity data collection module to the broker software through the producer, and the broker stores the message in the topic and stores the topic in the distributed processing-based high-capacity IoT high-speed search module through the consumer develop

24 is a diagram for explaining the configuration of a data pipeline environment.

Development of producer technology that transmits data transmitted from real-time/batch large-capacity data collection module to message queue

Development of consumer technology to receive topic data of message queue from distributed processing-based high-capacity IoT high-speed search module

It is composed of open source message queue broker software such as Apache Kafka and RabbitMQ, and refers to an environment in which the roles of producer and consumer are separable, with high capacity and high efficiency, real-time log integration, non-stop (device failure), compatibility with heterogeneous types.

The producer creates a message of a specific topic and delivers the message to the broker (Kafka, RabbitMQ). If the broker classifies the received messages by topic and stacks them up, consumers subscribing to the topic take the messages and process them.

25 is a diagram illustrating a configuration diagram of a producer and consumer connection.

Developed end-to-end data encryption technology-based section encryption technology when data is transmitted from Message Queue to a distributed processing-based high-capacity IoT high-speed search module of another network

There are methods for configuring KMS (Key Management Server) using the local key storage and the KMS function of Hadoop.

The local keystore provides from the store, which is provided as a local file to Producers and Consumers, which contains all keys related to the creation/subject.

Hadoop KMS provides envelope encryption for keys stored in encrypted form with data using a central key management server.

Envelope encryption means encrypting data with a data encryption key (DEK) and then wrapping the DEK with a fully manageable root key.

26 is a diagram for illustrating a data section encryption method.

Development of large-capacity data monitoring technology

Cluster monitoring operation of large-capacity distributed message queue is possible by using MBean (managed bean) monitoring method using JMX interface and offset

JMX stands for Java Management eXtensions, which refers to a Java API that provides monitoring tools for applications written in Java.

Monitoring technology that can check the status of data transmitted to the message queue, message queue status, and consumer status from the message queue from the producer who builds the data pipeline by using the offset information. Development

27 is a diagram for illustrating distributed processing large-capacity message queue (Message) monitoring processing.

Large data monitoring function is provided as follows

Processing status of each topic

Brokers and Consumers

cluster status, etc.

28 is a diagram for illustrating data transmission monitoring.

3. Intelligent integrated control and monitoring technology

① Real-time data visualization-based technology development

Data generated in the field is stored in a large-capacity IoT high-speed search module based on distributed processing through a 5G edge cloud-based real-time information processing agent, and a monitoring function to analyze the data in real time is developed.

Development of integrated control monitoring module without delay/interruption based on open sources such as Grafana, Kibana, React, and D3.js that enables effective visualization of incoming data in real time

29 is an intelligent integrated control structure diagram.

High-capacity IoT high-speed search module based on distributed processing is configured based on a distributed open source search and analysis engine for storing all types of data, such as text, numbers, location-based information, structured and unstructured data, such as ElasticSearch

Distributed processing-based high-capacity IoT high-speed search module can be configured with multiple clusters for processing, so even if one server is abnormally terminated, data is automatically replicated to other clustered servers to eliminate SPOF (Single Point Of Failure)

SPOF (Single Point Of Failure) is a single point of failure, and when a failure occurs in the system, the entire system stops working.

In addition, since data is distributed and processed in parallel, real-time search and analysis can be performed, and it is designed to grow horizontally.

30 is a diagram illustrating a distributed processing-based large-capacity IoT high-speed search module architecture.

② Development of integrated control and monitoring technology

Logistics delivery information real-time monitoring technology

31 is a diagram for explaining the configuration of real-time monitoring of logistics and delivery information.

Real-time monitoring of logistics delivery information from receipt of logistics to completion of delivery with the postal information system data stored in the distributed processing-based high-capacity IoT high-speed search module

Logistics shipment location is tracked based on logistic receipt number and Geographic Information System (GIS) data

GIS is a comprehensive information system designed to create and manage maps and geographic information using a computer, collect, analyze and process data based on the geographic information obtained from it, and apply it to all terrain-related fields.

Real-time monitoring of the operational status of logistics support from receipt of logistics to completion of delivery using the postal information system data stored in the high-speed search module for the utilization status of automated equipment and robot resources that perform logistics center, transportation, and delivery tasks

Logistics center information monitoring technology development

Data such as dock plate and vehicle (plate) docking/undocking information and logistics loading/unloading information through large-capacity IoT high-speed search module and logistics center information monitoring consumer module for data generated from the logistics center automation system that is loaded/unloaded without an operator monitoring in real time

32 is a diagram for explaining the configuration of the distribution center information monitoring.

③ Standardized data linkage information system design

Standardized data linkage and standardized code conversion verification expression module design

System structure design of functions to collect linkage/verification status of standardized data processed in related systems

System structure design for the function of expressing standardized data linkage status information

4. Intelligent logistics analysis technology

① Development of intelligent efficiency analysis technology

33 is a configuration diagram of an intelligent efficiency analysis technology.

Autoencoder-based feature extraction

Integrated data load

Derived variable generation technology development

Autoencoder model creation

Development of feature extraction technology using autoencoder

Development of input variable selection technology using feature extraction technology

Optimal time calculation using integrated data

Develop learned autoencoder save and load skills

When making predictions using machine learning and deep learning algorithms, the data features used as input values influence algorithm learning accordingly.

Development of feature extraction technology using autoencoder to extract high-impact data

34 is a diagram for explaining the basic structure of an autoencoder.

Autoencoder is a structure mainly used in unsupervised learning, and it learns the output value as an approximate function of the input value.

Same structure as general MLP (Multi-Layer Perceptron) except that the number of input and output layers is the same.

It has a method of extracting features from the input data through the encoder and reconstructing the original data through the decoder.

In the encoding process, only the key characteristic information of the input data is learned from the hidden layer, and the rest of the information is lost.

When the output value of the hidden layer is extracted during the decoding process, it is an approximation of the input value, not a perfect copy of the value.

The principle of autoencoder is to make it possible to extract features well by tuning the output value to be the same as the input value as much as possible.

Selection of input variables to be used in quantity prediction through feature extraction technology using autoencoder

Applied methods such as stacked autoencoder, denoising autoencoder, sparse autoencoder, and variational autoencoder

Efficient distribution based on reinforcement learning

DDDQN (Double Dueling Deep Q Network) Reinforcement learning internal environment, reward, reward rules, etc.

Creating a DDDQN reinforcement learning model

Development of efficient distribution technology using DDDQN reinforcement learning

Develop skills for storing and loading trained models

Distribution technology development using loaded models

Efficient distribution technology is a method of adjusting numerical values to derive optimal performance by inputting various data

It operates by adjusting the figures with simple rules, statistical analysis, user experience, etc.

Developing skills to learn a variety of data and adjust numerical values using reinforcement learning algorithms

35 is a diagram for explaining a reinforcement learning structure.

The reinforcement learning algorithm, which is widely known to the public and used in various fields recently, is widely used in the field of optimal control of AlphaGo announced by Google.

Recently, reinforcement learning has been applied to classic games and Go, demonstrating that self-learning of policies is possible, and human-level decision-making has been demonstrated even for problems with a large number of cases.

As a representative example, Google optimizes power consumption using reinforcement learning to reduce unnecessary power consumption.

Reinforcement learning makes sequential decision making on a state through a policy, and evaluates the judgment as a value of reward for the appropriateness of the action.

Through repetition of this process, we explore and establish an optimal policy that can select the behavior that maximizes the value in a given state.

Reinforcement learning does not require prediction of state changes or interpretation of the relationship between states and behaviors, so unlike existing methods, models are not required.

Reinforcement learning in the past defines the probability of changing from the current state to the next state (Markov Decision Problem), expresses the relationship between the present state and the next state value as a formula (Bellman Equation), and calculates the value of each state through iteration. Using a dynamic programming approach to update

Applied methods such as DQN (Deep Q Network), Double Q Network, Dueling Double Q Network, Actor Critic, etc.

Hyperparameter Optimization-based Parameter Optimization Technology Development

Reinforcement Learning-based Efficient Distribution Module Parameter Definition

Bayesian Optimization Technology Development

Optimization application technology development for defined parameters

Learning and storing reinforcement learning-based efficient distribution module as an optimal performance parameter

Hyperparameter Optimization is a problem of searching for optimal values of hyperparameters, which are values that must be set in advance to perform learning.

Hyperparameters have been set with a priori knowledge of researchers such as learning rate, cost function, regularization parameter, batch size, etc., but hyperparameter optimization technology is used to find hyperparameters that derive the optimal model and performance.

36 is a diagram for explaining hyperparameter optimization.

Derive a combination of preprocessing, model, and parameter that derives the best performance using Bayesian Optimization

Bayesian Optimization is a method that can systematically perform the entire search process while sufficiently reflecting ‘prior knowledge’ when conducting research on new hyperparameter values.

The purpose of Bayesian Optimization is to find an optimal solution that maximizes the function value f(x) by supposing an unknown objective function f that receives an input value x.

Define other setting values such as input value and objective function

Select by randomly sampling the first n input values within the set search target section

After learning the deep learning model by setting each of the selected n input values as the learning rate value, the performance result of the trained model is calculated using the validation dataset.

Perform probabilistic estimation with Surrogate Model for each input value-function value

Repeat the above process until the number of points of the investigated input value-function value reaches the specified total number of N points.

Final selection of the optimal solution that maximizes the average function based on the result of the objective function estimated probabilistically for a total of N input value-function value points

If the deep learning model is trained using the optimal solution value as the learning rate, a model with maximal performance can be obtained.

37 is a diagram illustrating a Bayesian Optimization process.

38 is another diagram for illustrating a Bayesian Optimization process.

② Development of intelligent quantity prediction technology

39 is a configuration diagram of an intelligent quantity prediction technology.

Autoencoder-based feature extraction technology development

Integrated data load

Derived variable generation technology development

Autoencoder model creation

Development of feature extraction technology using autoencoder

Development of input variable selection technology using feature extraction technology

Calculation of cargo volume using integrated data

Develop learned autoencoder save and load skills

Development of GAN (Generative Adversarial Network) + Deep Learning-based quantity prediction technology

Autoencoder-based feature extraction Loading input variables

Data division (training data, verification data, test data)

Deep learning model creation (DNN (Deep Neural Network), LSTM (Long Short Term Memory), etc.)

GAN model creation

GAN + Deep Learning Model Convergence

Development of GAN + deep learning-based quantity prediction technology

Develop skills to store and load trained models

Developing predictive technology using loaded models

Integrate data produced in each logistics process to predict quantity

Features extracted through feature extraction technology using autoencoder are used as input data

The algorithm used for volume prediction is GAN, which predicts the volume generated in the region.

GAN consists of two neural networks, one neural network called a generator creates new data instances and the other discriminator evaluates the authenticity of the data.

40 is a diagram for explaining the principle of the learning process of the GAN.

Hyperparameter Optimization-based Parameter Optimization Technology Development

GAN + Deep Learning-based Quantity Prediction Module Parameter Definition

Bayesian Optimization Technology Development

Optimization application technology development for defined parameters

Learning and storing GAN + deep learning-based quantity prediction module as an optimal performance parameter

Development of load analysis and demand forecasting technology based on cargo volume

Development of statistical analysis technology (transportation pattern EDA and load analysis, etc.)

Development of logistics resource demand forecasting model

③ Development of intelligent abnormal diagnosis technology

41 is a configuration diagram of an intelligent abnormality diagnosis technology.

Development of outlier data generation technology based on GAN + deep learning

Integrated data load

Derived variable generation and input variable selection technology development

Autoencoder model creation

Development of outlier data generation technology using autoencoder

Develop learned autoencoder save and load skills

Creation of outlier data and development of integrated technology with existing integrated data

Machine learning and deep learning algorithms show excellent performance when training a sufficient amount of data, but if the ratio between each category of training data is uneven, the performance is significantly reduced.

For outlier data, the ratio between categories is uneven, so to solve the problem, a method is used to alleviate the problem of imbalanced data using the GAN (Generative Adversarial Network) technique.

The GAN technique is an unsupervised learning-based model in which a generator model and a discriminator model compete against each other and learn to improve learning performance, and the structure is as follows.

The generator is a model that generates data similar to the original data, and the discriminator consists of a model that determines the original data and the generated data.

As the learning progresses, the generator generates data similar to the original data so that it is difficult for the discriminator to determine it.

After sufficient learning, the discriminator is generated to the extent that it is difficult for the discriminator to distinguish the close data from the original data.

This method prevents data loss and overfitting, and also solves the problem of generating noise data by generating data similar to the original data.

The outlier data is relatively composed of a ratio of 98:2 compared to normal data, so the outlier data is generated using the GAN technique.

Development of hybrid clustering-based anomaly detection technology

Invention of K-meas, K Nearest Neighbors (KNN), and Hierarchical Density-based Spatial Clustering of Applications with Noise (HDBSCAN) clustering generation technology

Development of abnormal/normal cluster generation technology

Developing Learned Clustering Store and Load Skills

Development of anomaly detection technology using clustering technique for real-time data abnormality diagnosis

Anomaly detection refers to the part that degrades speed and performance in the entire process.

First, we need a way to distinguish normal data from abnormal data.

Since there is not enough abnormal data, data is generated using the GAN technique.

Anomaly detection technology uses a clustering method to distinguish normal data from abnormal data

Among the clustering techniques, HDBSCAN algorithm, which improved the representative DBSCAN algorithm among the density based clustering methodologies, and the representative K-means and KNN clustering algorithms are used.

Although each clustering algorithm has its pros and cons, it is used in a way to compensate for the shortcomings by fusion in a hybrid format.

Anomaly detection is performed by weighting the results of each algorithm and fusion

42 is a diagram for illustrating HDBSCAN classification.

Hyperparameter Optimization-based Parameter Optimization Technology Development

GAN + Deep Learning-based Outlier Data Generation Module Parameter Definition

Hybrid clustering based anomaly detection module parameter definition

Bayesian Optimization Technology Development

Optimization application technology development for defined parameters

GAN + deep learning-based outlier data generation module learning and storage as optimal performance parameters

Learning and storing hybrid clustering-based anomaly detection module as optimal performance parameter

④ Development of intelligent performance analysis technology

43 is a configuration diagram of an intelligent performance analysis technology.

Statistical-based analysis technology development

Integrated data load

Development of Derived Variable Confirmation Technology

Variable trend analysis technology development

Variable importance analysis technology development

Variable correlation analysis technology development

Statistical analysis result integration and input variable selection technology development

You can check the characteristics and brief description of the data through the statistical analysis function

The following statistical functions were developed to check the data produced by various logistics processes at a glance.

Statistical analysis function is largely defined as four functions: Derived Variable Check, Variable Trend Analysis, Variable Importance Analysis, and Variable Correlation Analysis.

Each function is a comprehensive concept, and detailed statistical analysis techniques are included in each category

Derived variable check function was developed using the following method

In the derived variable check function, you can check the moving average and momentum of time series data.

The moving average line is the most basic indicator used in data analysis, and you can check short-term and long-term trends in time series data.

In addition, most studies use moving averages for time series prediction.

In general, moving averages can be divided into short-term, medium-term and long-term.

Similarly, the momentum indicator of time series data indicates the direction and speed of the data trend and is mainly used to calculate the direction of the trend.

In addition, the momentum indicator can be used to compare and analyze the strength and direction of a trend through data comparison between the current period and a certain period.

Variable trend analysis function is developed using the following method

Seasonal and Trend decomposition using Loess

STL is an analytical technique that can be used in a variety of contexts and can be decomposed into seasonality, trend, and residual to handle any kind of seasonality, including monthly, quarterly, and monthly data.

The variable importance analysis function is developed using the following method

In general, deep learning models learn by giving more weight to the input variable that has the most influence on the prediction result among many input variables.

However, the use of such “black box” input variables can provide excellent model performance, but cannot explain the analysis results sufficiently and clearly.

Therefore, before learning the deep learning model, it is necessary to sufficiently analyze the characteristics of input variables and reasonable reasons.

As the most representative algorithm, the input variable importance calculation using the Random Forest algorithm is done as follows.

The random forest algorithm is a voting-based algorithm and generally shows high prediction performance because the final decision is made by referring to the ratio of output results.

Variables with high importance scores have higher ranking importance than variables with small values.

Random Forest can be used to evaluate the importance of variables and to select variables to use for modeling.

In a random forest, the importance of a variable is measured by how much it contributes to improving accuracy and node impurity.

This obtained variable importance can be used to select variables for use in other models (e.g., linear regression).

44 is a diagram for illustrating variable importance calculation using a random forest.

Variable correlation analysis function is developed using the following method

Correlation analysis and regression analysis techniques are used

45 is a diagram for illustrating correlation analysis.

Development of average delivery time prediction technology based on GAN + deep learning

Autoencoder-based feature extraction Loading input variables

Calculate average delivery time using integrated data

Data division (training data, verification data, test data)

Deep learning model creation (DNN (Deep Neural Network), LSTM (Long Short Term Memory), etc.)

GAN model creation

GAN + Deep Learning Model Convergence

Development of average delivery time prediction technology based on GAN + deep learning

Develop skills for storing and loading trained models

Developing predictive technology using loaded models

Hyperparameter Optimization-based Parameter Optimization Technology Development

GAN + Deep Learning-based Average Delivery Time Prediction Module Parameter Definition

Bayesian Optimization Technology Development

Optimization application technology development for defined parameters

Learning and storing the GAN + deep learning-based average delivery time prediction module as an optimal performance parameter

⑤ Process-based logistics analysis and visualization module design considering specialized logistics processing environment and equipment for each detailed task

Deduction of specialized logistics analysis elements and common KPIs considering the logistics processing method of the logistics unit process for each detailed task and the characteristics of various logistics processing equipment

Identification of logistical analysis elements in consideration of logistics processing equipment and business characteristics for each stage of reception, classification, transportation, and delivery, which are major steps in the postal logistics process

Development of common KPIs for each logistics processing step, such as mail throughput and processing time per package, which are common logistics processing targets

Definition of standardized logistics process for logistics reception/delivery, pallet loading and unloading, etc.

Define the analysis function for the analysis elements identified according to the type of logistics processing equipment and logistics unit

Movement distance, movement speed, movement time, processing time and logistics throughput and resource utilization rate for various logistics processing entities such as autonomous transport vehicles, logistics robots and delivery droids

Systematic logistics processing statistical analysis and logistics performance analysis from the perspective of logistics unit tasks, including agent tasks performed by logistics processing equipment

Logistics efficiency analysis for optimal resource allocation according to various logistics processing environments such as volume distribution by unit task and delivery density

Design of various logistics visualization and reporting functions to provide effective analysis results for logistics processing equipment and analysis elements specialized for logistics business characteristics

Various visualization functions such as temporal visualization, distribution visualization, relational visualization, and spatial visualization to effectively provide specialized analysis results according to various logistics processing subjects

Various types of visualization-based reporting function for logistic analysis results for each step of logistics processing according to the flow of the postal logistics process

5. Demonstration and Service

① Derivation of legacy system linkage requirements

In order to apply the postal service of ICT convergence technology through linkage with the Postal Logistics Information System (PostNet), the requirements for system linkage and data transmission/reception are discovered.

System linkage plan

Measures to ensure data security and stability in the process of linking

Definition of data transmission standards and formats

Distinguishing between data requiring real-time provision and statistical data

Derivation of requirements for connection with call center CS system

System linkage plan

Definition of accessible and controllable parts in the call center area

Establishment of call center business processing scenarios for each task

Customer response manual

46 is a diagram for explaining the configuration of a postal distribution information system (PostNet).

② System-linked agent development

In order not to interfere with the system load due to the linkage with the postal logistics information system and existing business processing, the method of sending and receiving data through the linkage agent is promoted.

Includes all functions for batch processing such as real-time data linkage and statistical data

Design progress in a direction that can minimize the problem of available throughput and load of the postal logistics information system received in the data transmission process

③ Execution of simulation in connection with the implementation stage system

Currently, the postal logistics information system is in the development stage of an advanced system, and it is in progress with a plan to complete development and perform stabilization work in 2020.

The implemented platform system aims to link with the developed advanced system, but it is unclear whether the stabilization will be completed in the actual demonstration stage, so in 2020, the first year, we simulate testing with the linked system to derive improvement plans

The advanced postal logistics information system is equipped with an intelligent control function, and it is expected that higher synergies will be created by linking with the system developed in this project.

47 is a diagram for explaining a configuration (To-Be) of a postal distribution information system (PostNet).

Expand the data analysis base through data linkage with the postal logistics big data center and build a highly usable data production base

Prepare a data transmission plan that meets the standardization rules for the big data center

Establishment of security systems such as VPN and firewall for data interworking

Establishment of data de-identification and pseudonymization processing measures to protect personal information

④ Development of demonstration scenarios linked to other projects

48 is a layout diagram of the Daedeok Research Complex and the Korea Advanced Institute of Science and Technology in the area (plan) to be promoted.

Promotion of postal service application of new postal service centering around Daedeok Research Complex and Korea Advanced Institute of Science and Technology

The infrastructure environment without 5G communication shadow areas has been completed, and KT's 5G robot operation demonstration is in progress (September `19), suitable for technology demonstration

Due to the establishment of the Central Distribution Center (IMC), it is relatively easy to create space for the distribution center, so it is suitable for demonstration of loading, unloading, collection, and distribution technologies.

Cost advantage because the National Information Resources Service is nearby when working with the dedicated network linkage with the postal service system of the comprehensive control system and 5G edge cloud system

Demonstration scenarios for each task are established at the working committee level, carried out in accordance with the demonstration plan, and advanced through the process of collecting opinions from field workers and stakeholders

In the demonstration scenario for each task, a plan is established in accordance with the service aspect and operation aspect

49 is a diagram to summarize the technical contents according to another embodiment of the technology of the intelligent comprehensive logistics platform design device.

1. Standardized data linkage management system

① Data linkage and standardization code conversion verification module development

Development of 5G edge cloud-based information processing agent and intelligent logistics analysis and control system and data linkage verification module

Development of standardized code conversion verification module for linking data between logistics processing data received through 5G edge cloud-based information processing agent and intelligent analysis and control systems

② System-linked test and domestic and international standardization

Linkage test with related systems such as intelligent integrated control system

Promoting domestic and international standardization according to system-linked test results

Submission of domestic group standards* contributions (terms, interface technology, etc.) and analysis/participation in international standardization group trends

* Domestic group standard: TTA PG 1003/1004 (planned)

2. 5G edge cloud-based information processing agent technology

① Advancement of large-capacity data diagnosis and improvement technology

Development of automatic correction technology for missing values

Developed a technology to detect and automatically correct missing values and outliers in real-time data

The types of missing values can be divided into MCAR (missing completely at random), MAR (missing at random), MNAR (missing not at random), etc.

MCAR (missing completely at random): A missing value representing the highest level of randomness, which occurs when the value of that property or the values of other properties of the object are not affected.

MAR (missing at random): A missing value representing a moderate level of randomness, which occurs when the value of another attribute of the subject is affected but not affected by the value of that attribute.

MNAR (missing not at random): A missing value representing the lowest level of randomness, which occurs when the value of the attribute is affected.

The algorithm used to correct missing values (missing values) in the data is as follows.

50 is a diagram for explaining a type of a Data Imputation model.

Imputation technology development by applying various Data Imputation Models such as various Single Imputation Models and Multiple Imputation Models based on data shape and pattern

Development of a module to compensate for missing data using the domain of data and MCAR (missing completely at random) mechanism

Development of outlier automatic correction technology

When an outlier is detected, machine learning develops a technology that can automatically correct the outlier using two methods of correcting the outlier.

How to correct outliers

Winsorizing method: Correcting the value beyond the boundary value as the boundary value

Trim, truncation method: remove all values beyond the boundary value

Boundary values are determined by upper and lower n% (±1) or by z-score (±3) based on observed values.

3. Intelligent integrated control and monitoring technology

① Development of image linkage technology for civil complaints customer center

Development of customer center connection module to handle customer complaints that occur in the field

Video/audio data generated in the field, such as mobile unmanned post offices and autonomous driving droids, are stored in a distributed processing-based high-capacity IoT high-speed search module, and video/voice is connected to a customer center counselor in real time through Video Stream Consumer

51 is a diagram for explaining the configuration of a customer center connection module.

Video Stream Consumer consists of libraries such as OpenCV that can process image and video files, and libraries such as GraphX and Spark Streaming for graph processing.

GraphX and Spark Streaming express streaming data in a discrete stream (DStream) method, and it is a real-time streaming data processing framework in which the data introduced within the last time window is composed of RDD.

52 is a diagram for illustrating a Video Stream Consumer architecture.

② Advancement of integrated control and monitoring technology

Development of autonomous vehicle/droid movement path monitoring technology

IOT sensor data generated in the field, such as mobile unmanned post offices and autonomous driving droids, are monitored in real time by inquiring sensor data such as GIS (Geographic Information System) and GPS (Global Positioning System)* from the distributed processing-based high-capacity IoT high-speed search module.

* GPS is a satellite navigation system that measures the exact time and distance from three or more satellites and calculates the location of three different distances by triangulation. It has data such as latitude, longitude, and altitude.

The autonomous vehicle/droid movement path displays the robot's current location in real time along with the logistics delivery movement path on an open source-based map such as OpenStreetMap.

53 is a diagram for explaining the configuration of autonomous driving freight vehicle/courier droid movement path monitoring.

Module status check monitoring and integrated control technology development

Check operation/non-operation status of robots/equipment such as mobile unmanned post offices and autonomous driving droids, and develop real-time monitoring and notification modules for abnormalities

54 is a block diagram of module status check monitoring and integrated control.

5G edge cloud-based real-time information processing agent health actuator module development

The Health Actuator module checks and checks the monitoring, metric information, and traffic information of agents existing in other servers in real time, and enables remote control of various control technologies such as agent version update and data collection target change. provided

The Health Actuator module consists of a cloud-type config server, manages IP and PORT information of agents installed in different servers, and communicates with each agent in a RESTful manner.

REST is an abbreviation of the term Representational State Transfer. It is an architecture that can take full advantage of the advantages of the web. It specifies a resource through HTTP URI and applies CRUD operation to the resource through HTTP method (GET, POST, PUT, DELETE). means that

RESTful is an application development interface that provides access to resources based on REST.

③ Development of standardized data linkage status information system

Development of standardized data linkage and standardized code conversion verification expression module

Implement a function to collect standardized data linkage/verification status information

Implement the function to display the current status information in the related system

4. Development of intelligent logistics analysis technology

① Development of technology interlocking technology for each element of intelligent logistics analysis

Intelligent Logistics Analysis Analysis proceeds based on the results of each element

Derivation of supplementary contents through interlocking for each function

Development of interworking technology between intelligent efficiency analysis, intelligent quantity prediction, logistics resource demand prediction, intelligent abnormal diagnosis, and intelligent performance analysis technology

55 is a diagram for explaining the technology development of each element of intelligent logistics analysis.

② Development of intelligent integrated control monitoring technology and interlocking technology

Development of technology to deliver intelligent logistics analysis results to intelligent integrated control platform

Organize results for each analysis technology and design a linkage plan

Test design to provide real-time analytics technology

Development of technology linked with intelligent integrated control and monitoring technology

56 is a diagram for explaining interworking of an intelligent integrated control system.

③ Advancement of intelligent logistics analysis technology

Technology supplementation and advancement through demonstration of intelligent logistics analysis

Advancement through technology such as modification of input derivative variables and modification of algorithm structure

Developing technologies that can further train existing models using the added data

Development of intelligent logistics analysis technology automation technology

57 is a diagram for explaining advancement of intelligent logistics analysis technology.

④ Logistics unit process analysis and integrated logistics process visualization technology development

Development of agent-based logistics process model to analyze differentiated logistics environment such as closed space logistics center and external delivery space

Logistics process modeling for logistics processing inside the logistics center, which is a closed space (logistics transport vehicles and robot equipment that supports loading and unloading, etc.)

Last-mile delivery process modeling for external delivery business processing (mobile reception/delivery autonomous vehicle, autonomous logistics transportation vehicle (central-central office), delivery droid, etc.)

Development of intelligent logistics analysis technology based on the integrated logistics process that organically links the logistic unit process for each detailed task, and standardization and standardization of the postal logistics process

Deduction of detailed logistic unit process map through event log analysis based on collected data for each detailed task using various Process Discovery algorithms

Development of logistics processing and quantity flow analysis technology based on the integrated logistics process in which each detailed unit process is linked using process mining

Local bottleneck for each detailed unit process scope and global bottleneck section identification and cause analysis environment implementation according to the logistics flow of each detailed link section

Development of multiple presentation layer visualization technology for each logistic step that can support self-analysis (R, Python, etc.) through on-demand access for each detail

Establishment of standardization plan for 5G edge cloud-based postal logistics internal process (postal logistics receipt-delivery, pallet loading and unloading, cargo transportation, etc.)

Analysis of expected effects of logistics unit process innovation such as postman work reduction when introducing an intelligent comprehensive logistics platform

5. Demonstration and Service

① Perform system linkage work

The postal logistics information system is currently in the process of upgrading the system, and is expected to be developed in 2020 and stabilized by the end of 2020.

System linkage is promoted in the direction of linkage to the new To-Be system, but the complementary points derived from the simulation results of the first year are applied.

Compliance with security requirements for connection to postal distribution systems and Internet post offices

The area connected to each task consists of an internal dedicated network cut off from the external network.

58 is a diagram for explaining the configuration (As-Is) of the postal distribution information system (PostNet).

② System linkage with DW and call center

Linking statistical and monitoring information to the Postal Data Warehouse (DW)

Check the requirements for transmitting data information to the data warehouse linked to the OLAP system used in the postal service, and link the system

Since the related DB is linked with the postal logistics information system, direct access is highly likely to be difficult.

Linked work with call center

Configure the dedicated page used by call center agents according to the consultation scenario

Considered in terms of UI/UX design in a direction that can secure the convenience of consulting work

Consider providing a high-quality service by establishing a method for video confirmation consultation or related content search

Establishment of linkage for data transmission to big data specialized center

Refining related standardization measures such as DCAT, NiFi, and Kafka for data transmission to the big data center in line with the standardization guidelines of the big data center

Prepare a way to set the sensor data collected in each task into a dataset

Consideration of data commercialization of high-precision maps built for autonomous driving

Implement relevant measures such as encryption and de-identification at the data transmission stage

③ System-linked performance test

Conducting tests on performance indicators related to transmission and reception such as data transmission and control information with individual systems connected through system linkage work

If it does not meet the appropriate level, the problem is identified and improvements are made so that a smooth demonstration environment can be established within the project execution period.

When connecting to an existing legacy system, take special care so that there is no disruption to the existing business, and proceed in consideration of the system load and line usage.

④ Control Scenario Demonstration Test

Based on the demonstration scenario developed in the first year, the verification test is performed in the actual postal service or similar environment in the verification target area

Conduct long-term empirical tests, not single-shot tests, and derive areas to be advanced and areas to be improved

Revision of the demonstration test scenario by listening to opinions of business people such as postal service workers and postal officials in charge of postal service

<Demonstration of autonomous driving technology>

(Vehicle autonomous driving demonstration) Promoting demonstration of ICT convergence autonomous driving technology and service in connection with postal service in technology demonstration areas

To demonstrate unmanned reception and delivery technology and autonomous driving convergence technology, service is provided by cyclically driving and stopping the designated route inside the restricted area

In order to demonstrate the logistics and freight transportation problem solving task, the autonomous vehicle performs logistics transportation work between the demonstration area control office (logistics center) and the post office in the same unit

(Demonstration of autonomous driving droid) Technology demonstration within limited space of individual delivery droids and delivery support droids that have a variable driving environment compared to vehicles

In the case of individual delivery delivery droid, technology development and demonstration are mainly focused on delivering parcels in limited space and providing services linked to shared delivery boxes.

Demonstration of delivery assistance tasks such as tracking function and mail delivery function of mobile intermediate water supply and delivery and delivery collaboration droids according to the predefined pickup and delivery routes

(Demonstration plan) Carry out preliminary work necessary for demonstration such as temporary operation permission and high-precision map production for self-driving vehicles in the demonstration area

(Control, operation) Self-driving vehicles prepare their own control system to monitor service status, respond to emergency situations, and provide smooth services, and perform linked work with the integrated platform, which is the overall task

(Step-by-step promotion) Promoting driving demonstration for self-driving vehicles first, and then applying a postal service that faces users in a phased way

<Demonstration of logistics center technology>

(Demonstration area) Designate a logistics center that can create an idle space in the technology demonstration area, and perform the demonstration project technology development and testing in one place

(Demonstration plan) Construct a demonstration environment to maximize the effect of technology demonstration by making the postal transport vehicle and baggage used for the demonstration identical to the actual postal logistics site

Check ways to use aging vehicles, etc. through the logistics support group

(Control, operation) Prepare a control system that allows a large number of transport robots to operate simultaneously with workers, and perform linkage with the platform integrated control system

(Linking by task) Promoting demonstration of linkage plans for each task so that the task of transporting pallets that have been unloaded from the vehicle to the sorter can be collectively automated

(Deduction of standardization plan) Prepare a plan for efficient operation of introduced resources by deriving work standardization plans according to various types or structures, such as logistics centers and centralized stations, and logistics processing processes through technology demonstration

<Demonstration of small smart logistics center>

(Demonstration plan) Developing and operating a micro-fulfillment logistics center by composing a step-by-step scenario from the approach of the logistics vehicle to the loading of the delivered goods

(Demonstration area) In consideration of business application after technology development, some space in the post office or distribution center demonstration area that can utilize idle space is used for demonstration

<Integration, control demonstration>

(Experimental composition) Implementation of status monitoring and control functions based on data collected from 7 tasks, and development and demonstration of data analysis and utilization platforms

Demonstration of ICT safety technology that can detect risks through monitoring of safety risk factors

(Demonstration environment) The integrated control system is composed of G-Cloud-based cloud infrastructure to ensure flexible connection, expansion, and operation environment.

The collected data is used for integrated data analysis in connection with the postal logistics big data center

(Step-by-step verification) Promote smooth introduction into the work environment by linking the integrated control and monitoring functions with the postal distribution system

Step-by-step verification is carried out so as not to affect the provision of postal services through simulation-type verification before connection with the actual business system.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or apparatus, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

As described above, although the embodiments have been described with reference to the limited drawings, those skilled in the art may apply various technical modifications and variations based on the above. For example, the described techniques are performed in an order different from the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims (7)

managing the core verification field tasks for each stage of reception, classification, transportation, delivery, and complex in postal logistics; and
Designing a platform capable of integrated operation using the above step-by-step core demonstration field tasks
Including, intelligent comprehensive logistics platform design method. According to claim 1,
Any of (1) standardized data linkage management system, (2) 5G edge cloud-based information processing agent technology development, (3) intelligent integrated control technology development, (4) intelligent logistics analysis technology development, and (5) demonstration and service linkage Step of setting one as a key demonstration field task for each step
A method of designing an intelligent comprehensive logistics platform further comprising a. 3. The method of claim 2,
The (1) standardized data linkage management system is,
Setting up standardization data linkage processing management requirements analysis, standardization data linkage and code conversion verification module design, data linkage and standardization code conversion verification module development, and system linkage testing and domestic and international standardization as key verification field tasks for each stage
An intelligent comprehensive logistics platform design method. 3. The method of claim 2,
(2) 5G edge cloud-based information processing agent technology development,
Real-time/batch large-capacity data collection technology development, 5G real-time data diagnosis technology development, large-capacity data distributed processing and monitoring technology development, 5G edge cloud-based information processing agent technology advancement, and outlier/missing value automatic correction technology development set as a task
An intelligent comprehensive logistics platform design method. 3. The method of claim 2,
The above (3) intelligent integrated control technology development,
Real-time data visualization-based technology development, integrated control monitoring technology development, standardized data connection status information system design, civil complaint customer center image connection technology development, integrated control monitoring technology advancement, and standardized data connection status information system development, key demonstration areas for each stage set as a task
An intelligent comprehensive logistics platform design method. 3. The method of claim 2,
The above (4) development of intelligent logistics analysis technology,
Development of intelligent efficiency analysis technology, development of intelligent quantity prediction technology, development of intelligent anomaly diagnosis technology, development of intelligent performance analysis technology, development of statistical analysis technology (transportation pattern EDA and load analysis), development of logistics resource demand forecasting model, postal logistics unit process by detailed type Analysis and visualization module design, technology linkage for each analysis element (including logistics resource demand forecasting model) technology development, intelligent integrated control technology and linkage technology development, intelligent logistics analysis technology advancement, postal logistics unit process analysis and integrated postal logistics process visualization technology Setting development as a key demonstration field task for each stage
An intelligent comprehensive logistics platform design method. 3. The method of claim 2,
The above (5) verification and service linkage,
Derivation of requirements for linkage with legacy systems, development of system linkage agents, conducting simulations linked to the implementation stage system, development of demonstration scenarios linked to detailed tasks for ICT convergence logistics technology development, system linkage work, system linkage work with DW, call center, system linkage performance test , and setting the control scenario demonstration test as a key demonstration field task for each stage
An intelligent comprehensive logistics platform design method.

Images