Lean production training platform based on production big data analysis

Abstract

The application relates to the technical field of intelligent manufacturing teaching and training equipment, and discloses a lean production training platform based on production big data analysis, which comprises an entity production line module, a multi-source heterogeneous data acquisition system, an edge computing and control hub and a lean production training terminal. The entity production line module is used for simulating a production operation process in a real discrete manufacturing environment. The multi-source heterogeneous data acquisition system is distributedly arranged on the entity production line module and is used for collecting multi-dimensional data in a production process in real time. Through cooperation of an industrial sensor array, a machine vision unit, an RFID (Radio Frequency Identification) unit and a data aggregation terminal, multi-dimensional data such as equipment states, process parameters and material tracking are collected in real time, millisecond-level perception of whole-factor data such as people, machines, materials, methods and environments in the production process is realized, and the problems of single data dimension and rough perception granularity of a traditional training platform are fundamentally solved, so that a rich and accurate data basis is provided for lean analysis.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing teaching and training equipment technology, specifically a lean production training platform based on production big data analysis. Background Technology

[0002] Lean manufacturing, as a core management philosophy of modern manufacturing, aims to maximize efficiency through eliminating waste and continuous improvement. Currently, with the advancement of Industry 4.0, lean manufacturing is transforming and upgrading from traditional "on-site improvement" to "data-driven" approaches. However, lean manufacturing training methods for universities and enterprises still lag significantly behind industrial practice, mainly due to problems such as limited training scenarios, limited data dimensions, and outdated analytical methods. The limited training scenarios are manifested in their disconnect from the real industrial environment; existing training often uses sand table simulations or simple assembly line teaching aids, preventing trainees from perceiving the complex dynamic factors such as equipment vibration, cycle time fluctuations, and quality defects in a real production environment, making it difficult to develop a true understanding of the complexity of production systems. The data dimensions are limited, resulting in a lack of multi-source heterogeneous sensing capabilities. Existing training platforms typically limit data acquisition to PLC switching signals, failing to comprehensively collect multi-dimensional data such as equipment status, process parameters, visual quality, and material tracking. Core indicators required for lean improvement, such as OEE and work-in-process inventory, often rely on manual recording, leading to low data accuracy and poor real-time performance. The outdated analytical methods are reflected in the difficulty of quantifying the improvement effects. When trainees draw value stream maps and perform bottleneck analysis, they often use manual time measurement and Excel statistics, which cannot handle massive amounts of real-time data. Improvement solutions are difficult to quickly verify on the physical production line, and there is a lack of systematic comparison of key indicators before and after improvement.

[0003] In summary, there is an urgent need to develop a training platform that deeply integrates production big data collection, edge intelligence analysis, and lean production theory to solve the problems of scene distortion, data uniformity, analysis lag, and difficulty in quantifying improvement in existing technologies. Therefore, a lean production training platform based on production big data analysis is proposed. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a lean production training platform based on production big data analysis. It has advantages such as full-element data perception, real-time edge analysis, digital twin synchronous mapping, multi-role collaborative training, and automatic distribution of improvement solutions. It solves the technical problems of traditional lean training platforms, such as single scenario, insufficient data dimensions, outdated analysis methods, and difficulty in quantifying improvement effects.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a lean production training platform based on production big data analysis, comprising a physical production line module, a multi-source heterogeneous data acquisition system, an edge computing and control center, and a lean production training terminal; The physical production line module is used to simulate the production operation process in a real discrete manufacturing environment. The multi-source heterogeneous data acquisition system is deployed in a distributed manner on the physical production line module and is used to collect multi-dimensional data in the production process in real time. The edge computing and control center is communicatively connected to the multi-source heterogeneous data acquisition system and the physical production line module, respectively, and is used to receive and process data, perform real-time edge analysis, and generate control commands. The lean production training terminal is interconnected with the edge computing and control hub network to run lean production big data analysis software for lean production diagnosis, analysis and improvement verification.

[0006] Preferably, the physical production line module includes at least two functionally independent workstation units, an automatic conveying unit, and a programmable logic controller array. The workstation units include at least two types of assembly workstations, inspection workstations, and processing workstations. Each workstation unit has a quick-change base and a standard interface panel integrating pneumatic circuits, electrical circuits, and communication interfaces at its bottom for rapid workstation reconfiguration. The automatic conveying unit is connected between each of the workstation units for workpiece transfer. The programmable logic controller array is electrically connected to both the workstation units and the automatic conveying unit to control their execution actions.

[0007] Preferably, the multi-source heterogeneous data acquisition system includes an industrial sensor array, a machine vision unit, an RFID radio frequency identification unit, and a data aggregation terminal. The industrial sensor array is distributed and installed on the workstation unit and the automatic conveying unit, and includes at least three of the following: photoelectric sensors, pressure sensors, torque sensors, and vibration sensors. The machine vision unit is mounted above the key workstation unit and is used to collect product appearance and assembly quality data. The RFID radio frequency identification unit includes an RFID tag embedded in the workpiece tray and an RFID reader / writer installed at the workstation entrance, and is used to track material location and batch information. The data aggregation terminal is electrically connected to each of the above acquisition units and is used to convert multi-source signals into digital signals with a unified timestamp and upload them.

[0008] Preferably, the edge computing and control hub includes an industrial edge server, a real-time stream processing engine, a digital twin driving engine, and a data platform. The industrial edge server is communicatively connected to the data aggregation terminal and is used for preprocessing and edge caching of real-time data. The real-time stream processing engine runs inside the industrial edge server and is used for filtering, aligning, and extracting features from multi-source data. The digital twin driving engine interacts with the real-time stream processing engine to drive the virtual production line model and the physical production line to run synchronously based on real-time data, with a virtual-physical synchronization delay of less than milliseconds. The data platform is communicatively connected to the industrial edge server and is used to store historical data and provide a data API interface.

[0009] Preferably, the industrial edge server has a pre-installed equipment health prediction model, which is used to predict the equipment status and generate early warning signals based on the real-time data collected by the vibration sensors and current sensors in the industrial sensor array.

[0010] Preferably, the lean production training terminal is one or more computer devices, on which the lean production big data analysis software includes a value stream mapping automated generation module, a bottleneck identification and simulation module, a dynamic Kanban management module, and a continuous improvement closed-loop management module. The value stream mapping automated generation module interacts with the data platform to automatically read production data and generate a value stream map of the current state. The bottleneck identification and simulation module calculates the real-time utilization rate of each workstation and identifies bottlenecks, supporting virtual simulation verification of improvement plans. The dynamic Kanban management module generates electronic Kanban instructions based on the production plan and actual consumption, and sends them to the physical production line module through the edge computing and control center to drive production. The continuous improvement closed-loop management module records improvement proposals, tracks the implementation process, and automatically compares key performance indicators before and after improvement.

[0011] Preferably, the lean production training terminal further includes an augmented reality auxiliary operation unit, which includes AR glasses and is connected to the digital twin drive engine via a wireless network to overlay standard operating procedures or virtual instructions onto the trainee's field of vision.

[0012] Preferably, it also includes an automatic improvement instruction distribution gateway, which is communicatively connected to the lean production training terminal and the programmable logic controller array, respectively, and is used to automatically convert the improvement scheme parameters verified by simulation into control programs that conform to the IEC standard, and distribute them to the corresponding data blocks of the programmable logic controller array to realize the automated reconfiguration of the physical production line.

[0013] Preferably, it also includes a multi-dimensional assessment and evaluation system for trainees, which is communicatively connected to the lean production training terminal and the data platform, and is used to give a comprehensive score based on the trainees' operational data, analysis reports and improved production line performance during the training process.

[0014] Preferably, the lean production training terminal supports multiple users online simultaneously and assigns different roles. The roles include at least three of the following: production planner, material scheduler, quality manager, equipment maintenance engineer, and lean improvement specialist. Each role shares the same data source but has different operating permissions.

[0015] (III) Beneficial Effects Compared with existing technologies, this invention provides a lean production training platform based on production big data analysis, which has the following beneficial effects: 1. This lean production training platform based on big data analysis of production, through the coordinated setup of industrial sensor arrays, machine vision units, RFID radio frequency identification units and data aggregation terminals, collects multi-dimensional data such as equipment status, process parameters, and material tracking in real time. It achieves millisecond-level perception of all elements of the production process, including people, machines, materials, methods and environment, fundamentally solving the problems of single data dimensions and coarse perception granularity of traditional training platforms, and providing a rich and accurate data foundation for lean analysis.

[0016] 2. This lean production training platform based on big data analysis of production processes uses the collaborative work of industrial edge servers, real-time stream processing engines and digital twin driving engines to perform real-time filtering, alignment and feature extraction on massive amounts of sensor data, driving the virtual production line to achieve millisecond-level synchronous mapping. This allows trainees to intuitively observe the real-time operating status of the production line through the digital twin interface, solving the pain points of data separation from the physical world and analysis lag in traditional training.

[0017] 3. This lean production training platform based on big data analysis of production, through the coordinated setup of value stream mapping module, bottleneck identification module and dynamic Kanban module in the lean production training terminal, upgrades the time-consuming manual measurement and paper record in traditional lean teaching to an automated and visualized data analysis mode. Students can identify bottlenecks in real time and conduct virtual simulation verification, which significantly improves teaching efficiency and learning effect.

[0018] 4. This lean production training platform based on big data analysis of production, through the coordinated setup of an automatic instruction distribution gateway and a programmable logic controller array, enables trainees to complete the simulation verification of improvement schemes on the lean production training terminal. After completing the simulation verification, the parameters can be converted into PLC executable code with one click and distributed. The physical production line immediately runs according to the new scheme, realizing a zero-delay closed loop from "paper improvement" to "on-site improvement".

[0019] 5. This lean production training platform based on big data analysis of production supports multiple people to be assigned different roles online at the same time. Each role shares the same data source but has different operating permissions, which fully simulates the organizational structure and collaborative decision-making process of a real manufacturing enterprise, enabling trainees to cultivate cross-departmental communication and collaborative improvement capabilities in a simulated organizational environment. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall system architecture of a lean production training platform based on production big data analysis proposed in this invention. Figure 2 This is a schematic diagram of the data acquisition and processing flow in a lean production training platform based on production big data analysis proposed in this invention.

[0021] Figure 3 This is a schematic diagram of the functional modules of the lean production training terminal software in a lean production training platform based on production big data analysis proposed in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1-3 This invention provides a lean manufacturing training platform based on big data analysis of production data. The core modules are specifically structured as follows: The physical production line module 100 simulates a small reducer assembly and testing line, consisting of three workstation units 110: Workstation 1 is for housing and gear press-fitting, equipped with a pneumatic press and servo tightening gun; Workstation 2 is for rotational performance testing, equipped with a torque sensor and vibration test bench; Workstation 3 is for appearance inspection and marking, equipped with a machine vision unit 220 and a laser marking machine. Workstation units 110 are connected by a double-speed chain automatic conveyor unit 120. Each workstation unit 110 has a quick-change base with positioning pins and a standard interface panel integrating pneumatic circuitry, power supply, and Profinet communication interfaces, allowing for production line layout reconfiguration within 30 minutes. Each workstation operating position has a three-color indicator light and an Andon call button, connected to a Siemens S7-1200 series programmable logic controller array 130.

[0024] The multi-source heterogeneous data acquisition system 200 uses an NI cRIO-9045 controller as the data aggregation terminal 240. The industrial sensor array 210 is deployed as follows: a SICK WL4-3 photoelectric sensor is installed at the entrance and exit of each workstation; an HBM PW12 pressure sensor is integrated into the tail of the cylinder at the press-fitting station; a FUTEK TFF torque sensor is integrated into the output of the tightening gun; and an ADI ADXL1002 vibration sensor is mounted on the base of the inspection station. The machine vision unit 220 uses a Basler acA2440-75um industrial camera with a ring light source and communicates with the data aggregation terminal 240 via a GigE interface. The RFID radio frequency identification unit 230 uses a Siemens RF68x series reader and RF620T tag; the reader is installed on the side of the entrance to each workstation. The data aggregation terminal 240 performs 24-bit ADC conversion on various sensor signals and adds nanosecond-level timestamps to all data based on the IEEE 1588 protocol. The data packets are periodically sent to the edge computing and control hub 300 via Profinet IRT.

[0025] The edge computing and control hub 300 is based on the Advantech MIC-770 V2 edge computing platform, equipped with an Intel Core i7 processor and 32GB of memory, running Windows 10 IoT Enterprise and Codesys Runtime. The real-time stream processing engine 320 uses Kafka to perform sliding window filtering and anomaly removal on over 5000 sensor data points per second. The digital twin drive engine 330, based on Unity Reflect, subscribes to real-time data via OPC UA to drive the status of every motion axis and indicator light in the virtual production line model, with a virtual-real synchronization latency of less than 100ms. The data platform 340 stores historical data based on TimescaleDB and provides a GraphQL API interface. The industrial edge server 310 internally deploys an XGBoost-based equipment health prediction model, trained using historical vibration and current data, and outputs real-time predictions of the remaining lifespan of the equipment.

[0026] The Lean Production Training Terminal 400 consists of three Dell Precision 7920 tower workstations, equipped with NVIDIA RTX A6000 graphics cards, and connected to the Edge Computing and Control Hub 300 via a 10 Gigabit fiber optic cable. The software functional modules deployed on it include: a value stream map automated generation module 410, which reads data from the data platform 340 via a RESTful API and automatically generates a current state value stream map containing data bins, timelines, and inventory triangles; a bottleneck identification and simulation module 420, which accesses data such as the utilization rate and queue length of each workstation in real time, highlights bottlenecks with color blocks in the digital twin scenario, and supports trainees to adjust parameters for virtual simulation; a dynamic Kanban management module 430, which generates electronic Kanban instructions based on the master production plan and real-time consumption using a "pull-push" algorithm, and writes them to the PLC array 130 via OPC UA to execute pull production; a continuous improvement closed-loop management module 440, which has a built-in improvement proposal workflow, automatically associates data before and after improvement, and generates benefit reports; and an augmented reality auxiliary operation unit 450, including Microsoft HoloLens 2 AR glasses, which communicates in real time with the digital twin drive engine 330 via WiFi 6, overlaying standard operating procedures in the trainee's field of vision in the form of animation.

[0027] The improvement instruction automatic distribution gateway 500, based on OPC UA, converts the improvement parameters verified through simulation into structured text conforming to the IEC 61131-3 standard and writes it into the corresponding data block of PLC array 130. The trainee multi-dimensional assessment and evaluation system 600 records every step of the trainee's operation in the background, correlates it with the actual operating performance of the production line, and generates a comprehensive score.

[0028] System Workflow: Trainees log in to the Lean Manufacturing Training Terminal 400 and select a training task. The multi-source heterogeneous data acquisition system 200 continuously collects production line data and uploads it to the edge computing and control center 300. The digital twin drive engine 330 drives the virtual production line and the physical production line to run synchronously. Trainees observe the real-time bottleneck location through the bottleneck identification module 420, analyze sensor data, and propose improvement solutions. After successful simulation verification in the virtual environment, the improvement command is automatically sent to the gateway 500, which converts the improvement parameters into PLC code and sends it out. The physical production line immediately runs according to the new solution. The continuous improvement module 440 automatically records key indicators before and after improvement and generates an improvement report. The entire process is recorded and scored by the assessment system 600.

[0029] Specific Implementation Example 1: Application in an Industrial Engineering Course at a University Two sets of the platform of this invention were deployed in the "Lean Manufacturing" course of the Industrial Engineering major at a university. In the production line balancing training session, students used the bottleneck identification module (420) to find that the average waiting queue length in front of the pressing station was 5.6 pieces and the average changeover time was 8.2 minutes. Students proposed a solution of "standardized changeover operation + quick fixture improvement". The simulation in the virtual environment showed that the changeover time could be shortened to 3.5 minutes and the production line balancing rate increased from 72% to 86%. Students received standardized operation guidance through AR glasses and carried out on-site operation. The actual changeover time was reduced to 4.1 minutes and the production line balancing rate reached 85.3%. After the improvement, the output per shift increased by 22% and the work-in-process inventory decreased by 35%. After the course, the evaluation report generated by the student assessment system (600) showed that more than 90% of the students mastered the data-driven lean analysis method.

[0030] Specific Implementation Example 2: Application in Employee Training at an Automotive Parts Manufacturing Company An automotive parts manufacturing company introduced the platform of this invention for training front-line team leaders. The company imported real historical data from its own production lines into a data platform (340) to construct a training case highly similar to the company's actual scenario. Trainees were divided into groups to play different roles such as production planners, quality managers, and lean improvement specialists, and carried out lean improvement around "reducing the defect rate of a certain model of housing production line". Through the analysis of historical image data by the machine vision unit (220), it was found that 80% of the defects were concentrated in "missing seals"; through RFID data traceability, it was found that the batch of seals came from a new supplier. Trainees proposed a "visual error prevention" solution. After verifying its feasibility in the simulation module, the visual inspection program parameters were written into the field PLC through the improvement instruction distribution gateway (500). After actual operation, this type of defect was reduced by 95%. Within 3 months after the training, the trainees submitted more than 40 effective improvement proposals.

[0031] In summary, this lean production training platform based on production big data analysis, through the deep integration of physical production line modules (100), data acquisition system (200), edge computing and control center (300), and lean production training terminal (400), constructs a lean production training environment covering the entire chain of "perception-analysis-decision-feedback", effectively cultivating lean management talents who can meet the needs of the intelligent manufacturing era.

[0032] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0033] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A lean manufacturing training platform based on production big data analysis, characterized in that, It includes a physical production line module (100), a multi-source heterogeneous data acquisition system (200), an edge computing and control hub (300), and a lean production training terminal (400). The physical production line module (100) is used to simulate the production operation process in a real discrete manufacturing environment; The multi-source heterogeneous data acquisition system (200) is distributed on the physical production line module (100) and is used to collect multi-dimensional data in the production process in real time. The edge computing and control hub (300) is communicatively connected to the multi-source heterogeneous data acquisition system (200) and the physical production line module (100) respectively, and is used to receive and process data, perform real-time analysis on the edge side, and generate control commands; The lean production training terminal (400) is interconnected with the edge computing and control center (300) to run lean production big data analysis software and perform lean production diagnosis, analysis and improvement verification.

2. The lean production training platform based on production big data analysis according to claim 1, characterized in that, The physical production line module (100) includes at least two functionally independent workstation units (110), an automatic conveying unit (120), and a programmable logic controller array (130). The workstation unit (110) includes at least two of the following: assembly workstation, inspection workstation, and processing workstation. The bottom of the workstation unit (110) is provided with a quick-change base and a standard interface panel integrating air circuits, circuits, and communication interfaces to enable rapid reconfiguration of the workstation. The automatic conveying unit (120) is connected between each of the workstation units (110) to enable workpiece transfer. The programmable logic controller array (130) is electrically connected to the workstation unit (110) and the automatic conveying unit (120) respectively to control their execution actions.

3. The lean production training platform based on production big data analysis according to claim 1, characterized in that, The multi-source heterogeneous data acquisition system (200) includes an industrial sensor array (210), a machine vision unit (220), an RFID radio frequency identification unit (230), and a data aggregation terminal (240). The industrial sensor array (210) is distributed and installed on the workstation unit (110) and the automatic conveying unit (120), including at least three of photoelectric sensors, pressure sensors, torque sensors, and vibration sensors. The machine vision unit (220) is mounted above the key workstation unit (110) and is used to collect product appearance and assembly quality data. The RFID radio frequency identification unit (230) includes an RFID tag embedded in the workpiece tray and an RFID reader installed at the workstation entrance, which is used to track material location and batch information. The data aggregation terminal (240) is electrically connected to the above acquisition units and is used to convert multi-source signals into digital signals with a unified timestamp and upload them.

4. The lean production training platform based on production big data analysis according to claim 1, characterized in that, The edge computing and control hub (300) includes an industrial edge server (310), a real-time stream processing engine (320), a digital twin driving engine (330), and a data platform (340). The industrial edge server (310) is communicatively connected to the data aggregation terminal (240) and is used to preprocess and cache real-time data. The real-time stream processing engine (320) runs inside the industrial edge server (310) and is used to filter, align, and extract features from multi-source data. The digital twin driving engine (330) interacts with the real-time stream processing engine (320) to drive the virtual production line model and the physical production line to run synchronously according to the real-time data, with a virtual-physical synchronization delay of less than 100ms. The data platform (340) is communicatively connected to the industrial edge server (310) and is used to store historical data and provide a data API interface.

5. The lean production training platform based on production big data analysis according to claim 1, characterized in that, The industrial edge server (310) has a pre-installed equipment health prediction model, which is used to predict the equipment status and generate early warning signals based on the real-time data collected by the vibration sensor and current sensor in the industrial sensor array (210).

6. The lean production training platform based on production big data analysis according to claim 1, characterized in that, The lean production training terminal (400) is one or more computer devices. The lean production big data analysis software deployed on it includes a value stream map automatic generation module (410), a bottleneck identification and simulation deduction module (420), a dynamic Kanban management module (430), and a continuous improvement closed-loop management module (440). The value stream map automatic generation module (410) interacts with the data platform (340) to automatically read production data and generate the current state value stream map. The bottleneck identification and simulation deduction module (420) is used to calculate the real-time utilization rate of each workstation and identify bottlenecks, and supports virtual simulation verification of improvement plans. The dynamic Kanban management module (430) is used to generate electronic Kanban instructions based on the production plan and actual consumption, and sends them to the physical production line module (100) through the edge computing and control center (300) to execute pull production. The continuous improvement closed-loop management module (440) is used to record improvement proposals, track the implementation process, and automatically compare key performance indicators before and after improvement.

7. The lean production training platform based on production big data analysis according to claim 1, characterized in that, The lean production training terminal (400) also includes an augmented reality auxiliary operation unit (450), which includes AR glasses and is connected to the digital twin drive engine (330) via a wireless network to overlay standard operating procedures or virtual instructions in the trainee's field of vision.

8. The lean production training platform based on production big data analysis according to claim 1, characterized in that, It also includes an automatic improvement instruction distribution gateway (500), which is communicatively connected to the lean production training terminal (400) and the programmable logic controller array (130) respectively. It is used to automatically convert the improvement scheme parameters verified by simulation into a control program that conforms to the IEC 61131-3 standard and distribute it to the corresponding data block of the programmable logic controller array (130) to realize the automated reconfiguration of the physical production line.

9. A lean production training platform based on production big data analysis according to claim 1, characterized in that, It also includes a multi-dimensional assessment and evaluation system for trainees (600), which is connected to the lean production training terminal (400) and the data platform (340) for comprehensive scoring based on the trainees' operational data, analysis reports and improved production line performance during the training process.

10. A lean production training platform based on production big data analysis according to claim 1, characterized in that, The lean production training terminal (400) supports multiple users online simultaneously and assigns different roles. The roles include at least three of the following: production planner, material scheduler, quality manager, equipment maintenance engineer, and lean improvement specialist. Each role shares the same data source but has different operating permissions.

Images