Dynamic Management Operations Scheduling Strategy for Hybrid Manufacturing Production Line Based on Data Twin and Robotics Technology
Pubblicato online: 17 mar 2025
Ricevuto: 29 set 2024
Accettato: 26 gen 2025
DOI: https://doi.org/10.2478/amns-2025-0313
Parole chiave
© 2025 Jianjia Qi, published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
In the Chinese industry gradually from man-machine cooperation production and manufacturing to technology-intensive, full automation direction of transformation and upgrading, automated production line technology has been rapidly popularized in the automotive, machinery manufacturing, food, electronic supplies and other industries [1-3]. The use of robotics to replace the highly repetitive and labor-intensive tasks in the production process can significantly improve production efficiency on the basis of reducing human labor intensity [4-5]. From the application of robotics in the production line, foreign countries in the high-end manufacturing industry in the robot penetration rate is high, and already has the typical characteristics of intelligent manufacturing, China, although the relevant technology research and popularization of late, but in the “Made in China 2025” and other national strategies to promote the implementation of China’s automated production line of technological sophistication continues to The robot technology in the degree of adaptation, performance, production efficiency and other aspects have been improved significantly [6-9].
At this stage in the manufacturing line, industrial robots show a diversified trend of application, according to the mechanical structure and the form of work movement, robots can be divided into right-angle coordinates, cylindrical coordinates, spherical coordinates, joints, etc., and according to the classification of the program input method can also be divided into programming input type, demonstration of teaching input type and other types [10-12]. In addition to the common robotic arm robots, the robots commonly used in production lines include linear robots, SCARA robots, parallel robots, etc [13-14].
On the other hand, with the integration and development of advanced technologies such as the Internet of Things (IoT) and Artificial Intelligence (AI) in the manufacturing field, which contributed to a new round of industrial revolution, intelligent manufacturing has become the main direction of the development of manufacturing industry [15-16]. The traditional digital production line has the problems of long on-site commissioning cycle, poor interaction between physical production line entity and virtual 3D model, poor visualization ability, high cost of on-site status monitoring, weak data transmission ability, and production data out of synchronization, which makes it difficult to realize real-time visualization and remote management of production line information [17-18]. In recent years, with the use of digital twin, it breaks the barrier that hinders the interaction between physical space and information space, and provides technical support for the realization of production line digitalization, intelligence, and networking. As one of the important solution ways to realize the interaction between physical entities and virtual space, digital twin technology has been widely concerned by related researchers [19-21].
This paper firstly introduces the hardware composition, software design and physical composition system of the robot assembly line system in industrial manufacturing, i.e., control system, conveyor system and handling system. For the production process of the robot assembly line, the key technology of production scheduling system and scheduling optimization model based on digital twin is studied, and a two-layer progressive production scheduling strategy is designed with job sequencing layer (static scheduling) and job control layer (dynamic scheduling). In the robot production line scheduling optimization model, with the help of the strategy iteration algorithm under the Markov performance potential theory, the model is solved to realize the optimization of the robot production line scheduling objective. Through simulation experiments, the effectiveness and practicality of the method can be verified.
Industrial robot is one of the members of the robot family, it is the most mature development of a kind of robot, at the same time, the most widely used.IOS definition of industrial robot is: “a kind of automatic control operation and mobile function, can complete a variety of operations of the programmable operator”. Industrial robots are defined in our standard as a multi-functional, multi-degree-of-freedom operator that can be automatically positioned and controlled, and can be reprogrammed.It is capable of handling materials, parts, or clamping various tools to perform various operations.As an industrial robot actuator operator’s motion accuracy, flexibility, and dynamic characteristics directly affect the quality of work of the industrial robot system.The robot’s dimensional graphic simulation technology is a combination of emerging disciplines that began developing in the 1980s, integrating robotics, computer graphics, and computer technology.Along with the development of robotics and computer graphics, the simulation of robotic systems has progressed from initial two-dimensional simulations to three-dimensional graphical simulations.
With the growing maturity of industrial robotics technology, industrial robot production lines have replaced traditional industrial production lines in the industrial and various equipment manufacturing industries. In the implementation of a production line based on industrial robots, the use of an industrial robot production line simulation system to simulate the layout of industrial robots in the production line is particularly important. After completing the program design of the production line, the simulation of the production line simulation system can be proposed to optimize the reasonableness of the program design, so that, to a certain extent, it saves the cost of the enterprise.
The physical system of the production line designed in this paper consists of a control system, a conveyor system, and a handling system. The final realization is the completion of the process flow. The workpiece 3 is taken from one side of the storage turntable 4 by the first robot 1 and placed on the conveyor 2, following the movement of the conveyor 2. At the same time, the first robot 1 returns to the storage table 4 and grabs the next workpiece 3. When the workpiece 3 passes the photoelectric switch 5, it is clamped by the second robot 7 and placed on the other side of the rotary table 4. After the placement is completed, the second robot 7 then grabs the next workpiece 3 coming from the conveyor 2. The cycle is repeated, and the entire handling process is automated until all the workpieces 3 on the storage turntable 4 are picked up and placed on the other side.The second robot 7 is mounted with a horizontal moving guide 6 at the bottom of the second robot, increasing its working range.
The PLC serves as a control system and a bridge for the two KUKA industrial robots to communicate with each other.The PLC mainly consists of a CPU, memory, I/O interfaces, etc., and is capable of providing real-time tracking and feedback of the feedback information from the hardware system. Specifically, it receives the signals transmitted during the work of the two KUKA industrial robots and outputs the signals after digital-to-analog conversion. Output signals for controlling the robot’s hand claw action, starting and stopping the conveyor system, and displaying system status, etc.Conveyor system refers to the device used for conveying materials. This assembly line system specifically refers to the linear conveyor and storage turntable. In this loading and unloading system, the linear conveyor enables long-distance transportation and saves energy and power. Additionally, the conveyor is equipped with a photoelectric switch that detects the position of the workpiece. For efficient storage of workpieces in this handling system, a 360° rotatable storage table is provided. The rotary table is divided into two sides and is designed to allow the workpiece to be returned to its original ungripped state after a complete process has been completed by a simple rotation.
The handling system is the part of the entire loading and unloading system that is responsible for the handling tasks of the workpieces and consists of the KUKA KR6R900 robot and the KUKA LBR IIWA 7 R800 robot. In this handling system, in order to increase the operating range of the KUKA KR6 R900, a horizontal sliding guide is attached to the bottom of the robot, which corresponds to an additional degree of freedom for the robot, i.e., an additional axis for the KR6.The KUKA LBR IIWA 7 R800 robot is a new seven-degree-of-freedom sensitive robot from KUKA, which has the advantage of powerful sensors at each joint that can Programming languages such as Python, Java, and C++ support the motion control of the KUKA LBR IIWA robot, and the rich variety of languages can satisfy all kinds of scholars to program and control the robot according to the programming language they are familiar with.
The digital twin-oriented production scheduling system is driven by the production operation plan and is the executor of the operation plan, which is mainly used to simulate and evaluate the production capacity of the production line to optimize the daily production activities of the production line to regulate and control, and the whole process of regulating and controlling the production line is a real-time feedback and dynamic adjustment process. On the one hand, the production line needs to receive the parts processing tasks issued by the upper planning system, such as the number of orders, types of parts, parts process, the number of processes, delivery time and the number of equipment resources and other information to prepare the production line operation plan, which can be specifically and reasonably allocate the production tasks to the various processing equipment. On the other hand, according to the implementation of the actual production line production and the emergence of abnormalities to equalize production, such as the emergence of various types of disturbances, at this time you need to amend the operation plan arrangements to ensure that the processing schedule, if necessary, can be re-scheduled, and thus the production line operations for effective scheduling scheduling.
The actual production scheduling hierarchical framework shown in Figure 1. This paper adopts a two-level progressive production scheduling strategy, so that the scheduling results can be consistent with the actual production activities of the production line. Production line production scheduling mainly includes two aspects: the job sequencing layer and the job control layer, i.e., the static scheduling and dynamic scheduling of jobs, which are complementary to each other.
The hierarchy of production scheduling
When the job planning is refined to the work processes, the job sequencing is performed based on the imported basic data such as process routes, equipment information, material resources, tooling resources, etc. The purpose of doing so is to determine the processing order of all workpieces on the equipment according to the scheduling strategy, as well as the planned start time and planned end time of each process process processed by the equipment, and to meet the delivery time of each workpiece as much as possible. Because it determines the sequence of workpiece processing, process start time, and end time based on the available production data information without processing, the process can be called a static scheduling process.
The main task of the job control layer is to carry out scheduling and control of the processing operations that are being processed and have been queued up in the system. The use of real-time processing information data collected to track the workpiece processing progress, and the need to deal with abnormal production line such as urgent parts, equipment failure, rework, overrun and other disturbing events, according to the size of the impact of the disturbing events to determine the need for re-scheduling. In fact, in the actual processing process will produce a variety of disturbing events, the beginning of the production plan can not be carried out smoothly, such as urgent insertion of orders, which leads to a change in the order of the order processing, due to the production of certain equipment failures and thus delaying the normal processing tasks. All these series of disturbances will lead to the original optimized pre-scheduling program becoming a non-optimized scheduling program. At this point in order to ensure the smooth progress of the production task needs to be re-scheduling of the remaining workpiece scheduling, change the processing order at the beginning, which is a dynamic, real-time process, so it can also be understood as a dynamic scheduling process.
Job control level to the production and processing layer, its basic task is to ensure the smooth progress of the job production, real-time control of the processing situation, and timely access to data information on the production site so that the production information feedback to the relevant departments. When encountering various types of disturbing events, the first need to query the state of equipment resources and workpiece process status, and based on these status information to determine whether manual adjustment, such as arranging for employees to work overtime or transfer the process, or even re-scheduling, re-generation of scheduling scheduling plan.
In the industrial manufacturing process, the digital twin technology is integrated into the robot production line, the simulation model is built for the optimization problem of the production process, and the interaction between the physical entity and the virtual entity is completed, which can accurately realize the information to guide the production and improve the efficiency of the assembly process. The workpieces will be distributed in the intelligent robot production line according to the Poisson distribution with parameter λ. If each beat is a cycle, all the movements of the intelligent robot are on the beat point. The model can be constructed:
Where,
Where,
where
In the field of industrial manufacturing, in order to achieve the goal of intelligent robot production line scheduling optimization, after completing the construction of the production line scheduling model, the constructed model should be optimized and solved. In the model solving process, the Markov performance potential theory is mainly used as the basis, and with the help of the strategy iteration algorithm based on the Markov performance potential theory, the relevant smooth strategy performance potential vector is defined for the
In the actual solving session, the performance criterion will be directly affected by the value of the discount factor, when the discount factor takes the value of 1, it is the average performance criterion; when the discount factor takes the value of 0~1, it is the discount performance criterion. When the intelligent robot production line scheduling optimization, the binary group
Where Ω denotes the set of policies, and
The solution session, specifically, will go through the following process: (1) Initialize the processing scheduling optimization policy, and set the discount factor at the same time. (2) Obtain
Robot assembly line is the main equipment to realize a digital workshop and intelligent factory. The emergence of assembly lines has brought a revolution in industrial production methods and is widely used in electronics manufacturing, automobile manufacturing, processing and packaging, and goods sorting industries. In particular, the application of machine vision technology to the production process can make the entire process more flexible, greatly improving its level of intelligence and automation.Robot operating lines are a representative operating line, which is widely used in the fields of cargo handling, parcel sorting, and parts extraction. In a robotic production line, one or more pickup robots are generally configured as specific actuators. One or more conveyors to transport workpieces and boxes of packages. One industrial vision system is used to perform operations such as positioning, identification, and dimensional measurement of workpieces. Along with the rapid development of modern industrial production, its complexity is increasing and its life span is decreasing.
Optimization of data twin and robotics-oriented production line scheduling problems. In order to verify the optimization effect, the optimized scheduling strategy is applied to a data twin and robotics production line in an industrial manufacturing plant, and the effect before and after the optimization is compared. In the research process, the actual automated production line is used as the basis for setting the relevant parameters, and the principle is to make the arrival rate of the production line and the average productivity of the robot reach a kind of equilibrium, in order to avoid a large number of blockages and losses of workpieces caused by too large a rate of arrival of workpieces, or heat generation and losses of the robot due to the excessive speed of the robot’s movement. Setting parameters for the data twin and robotics production line. Workpiece gripping time 1/s, workpiece placement time 1.5/s, workpiece size 3/cm, conveyor speed 0.03/(m-s-1), robot running speed 1.0/(m-s-1).
After completing the preparatory work, the two production lines (pre-optimization line and post-optimization line) start running at the same time, and at the end of the run, the time consumption of each link of workpiece production is recorded, and the results are recorded as shown in Table 1. Analysis of the data recorded in the table shows that the application of the pre-optimization scheduling strategy is significantly more time-consuming than the post-optimization scheduling strategy in all aspects of workpiece production. For example, in the workpiece conveying link, 26.32s before optimization and 11.14s after optimization, it can be seen that the application of the optimized scheduling strategy can significantly improve the efficiency of workpiece production and the efficiency of the whole production line, which can bring higher economic benefits for the industrial manufacturing plant.
The production of artifacts is time-consuming and time-consuming (s)
| Link classification | Time consuming before optimization | Optimized after time |
|---|---|---|
| Workpiece | 26.32 | 11.14 |
| Workpiece handling and marking | 12.47 | 6.34 |
| Workpiece to specify position (position) | 38.73 | 18.67 |
| Placement of workpiece | 5.37 | 2.24 |
| Workpiece to specify position (position) | 55.42 | 24.36 |
| Grab the workpiece to the processing center | 28.75 | 13.28 |
The production line scheduling problem, as an indispensable part of modern industrial production, is an important research direction in the field of operations research.Its essence lies in the reasonable allocation of limited resources to multiple tasks under the condition of satisfying constraints, in order to optimize one or more performance metrics or meet specific demands. That is, in the production workshop, for a variety of production and processing tasks and processing resources (such as equipment, workers, materials, etc.) to rationally arrange the scheduling program, in order to maximize the production efficiency and resource utilization, and at the same time to meet the production demand and process constraints, and ultimately for each processing task to allocate the processing equipment and to determine the start time and the end time of the processing task.
In order to verify the effectiveness of this paper’s algorithm in the workshop production scheduling problem, this section selects examples in common databases for experimental simulation, and compares and analyzes them with other algorithms. In the experiment, the population size is set to popu_size=400, the maximum number of iterations of the genetic algorithm is Max_Iter=50, the number of iterations of reinforcement learning Iter=300, the random coefficient
To evaluate the performance of the algorithm in this paper, it is first compared with the standard algorithm. The mk01 benchmark algorithm in the Brandimate dataset was chosen for the experiment to test the algorithm’s optimization finding ability. The Gantt chart can effectively show the details of the scheduling scheme, and the Gantt charts of the traditional GA algorithm and this paper’s algorithm for solving the optimal scheduling solution for the mk01 shop floor are shown in Fig. 2. Among them, the horizontal axis represents the processing time axis, the vertical axis represents different processing equipment, the numbers labeled in the squares in the figure indicate the workpiece numbers currently being processed by the equipment, and the sequence of the same numbers from left to right represents the workpiece processing process. For example, the first device
Workshop production scheduling optimization problem solving
The two algorithms’ convergence performance curves for solving the mk01 problem are depicted in Fig. 3. Where the horizontal coordinate represents the number of iterations and the vertical coordinate represents the completion time of the optimal solution after each iteration. From the figure, it can be seen that under the same dataset, with the increase of the number of iterations, the algorithm proposed in this paper has faster convergence speed and better convergence results compared with the GA algorithm, and can solve the workshop scheduling problem more efficiently.
Comparison of algorithm convergence performance
The example problem simulated in this paper comes from a project on modeling and control strategy generation based on data twin and robotics in cooperation with a company, and one of the objectives of this project research is to optimize the modeling and scheduling of beer packaging production line using data twin and robotics at system level, so the research object of the reconfigurable production line scheduling simulation in this paper is the beer packaging production line of a company.
Beer packaging production line consists of multiple robotic automatic processing equipment, conveyor belts, various types of sensors and detection systems, processing equipment is the object of the production line scheduling, including unloading machines, bottle washing machines, bottle inspection machines, filling machines, sterilizers, labeling machines, packaging machines and so on.
From the beer packaging production process, it can be seen that the depalletizer and the case unloader are the starting point of the production line, the palletizer is the end point of the production line, and the bottle washer-bottle inspection machine-filling machine-sterilizer-labeling machine-packaging machine is the main process of beer packaging, so the object of the shop floor scheduling is only at the system level level for the above six types of manufacturing equipment, with the case as the basic processing unit. Four products are selected to form the pending processing order as the data input for the production line scheduling simulation, and the maximum completion time is taken as the only optimization objective. The processing order of all four products is: bottle washer, bottle inspection machine, filling machine, sterilizer, labeling machine, and packaging machine.The scheduling problem involves a flow shop scheduling problem, which is a distinct type of job shop scheduling problem.The processing time for each of the four products, based on the processing order, is as follows (unit: s):
Product 1: 324s-732s-1256s-180s-193s-205s Product 2: 324s-276s-586s-322s-594s-639s Product3:324s-118s-118s-612s-236s-736s Product 4: 239s-242s-72s-729s-765s-739s
The processing information for the four above-mentioned products is transformed into a dataset, and the optimal scheduling policy generated for this order is shown in Fig. 4.
Optimal scheduling strategy
In this paper, a production line composed of a control system, conveying system, and handling system is designed based on industrial robotics.Combined with the digital twin-oriented production scheduling system and production scheduling model, it enables dynamic scheduling optimization for the robot production line. The results of simulation experiments show that in the workpiece conveying link, the time consumed before optimization is 26.32s, and the time consumed after optimization is 11.14s, and the time consumed in each link of the workpiece production is significantly reduced after applying the scheduling strategy of this paper’s method. This paper’s algorithm reduces the idle time of the equipment and the utilization rate of the production line equipment is more balanced in the performance test of algorithm optimization.
220124038, Heilongjiang Institute of Technology Horizontal Research Project, Development of a Web-based Product Selection Platform for S Enterprise’s Gear Reducers.