• Journal of Korean Institute of Industrial Engineers
• /
• v.28 no.4
• /
• pp.415-424
• /
• 2002

Semiconductor manufacturing has been emerged as a highly competitive but profitable business. Accordingly it becomes very important for semiconductor manufacturing companies to meet customer demands at the right time, in order to keep the leading edge in the world market. However, due-date oriented production is very difficult task because of the complex job flows with highly resource conflicts in fabrication shop called FAB. Due to its cyclic manufacturing feature of products, to be completed, a semiconductor product is processed repeatedly as many times as the number of the product manufacturing cycles in FAB, and FAB processes of individual manufacturing cycles are composed with similar but not identical unit processes. In this paper, we propose a production scheduling and control scheme that is designed specifically for semiconductor scheduling environment (FAB). The proposed scheme consists of three modules: simulation module, cycle due-date estimation module, and dispatching module. The fundamental idea of the scheduler is to introduce the due-date for each cycle of job, with which the complex job flows in FAB can be controlled through a simple scheduling rule such as the minimum slack rule, such that the customer due-dates are maximally satisfied. Through detailed simulation, the performance of a cycle due-date based scheduler has been verified.

• Korean Journal of Agricultural Science
• /
• v.40 no.4
• /
• pp.395-403
• /
• 2013

Plant factory has drawn attention in many countries in the world due to capability of environmental control not only for better yield and quality, but also for increase in functional and medicinal components of the products. In this paper, an experimental plant factory was constructed for various tests under different environmental conditions, and the operations were evaluated. A production room was constructed with adiabatic materials with dimensions of $6,900{\times}3,000{\times}2,500$ mm ($L{\times}W{\times}H$). Four sets of $2,890{\times}600{\times}2,320$ mm ($L{\times}W{\times}H$) production frame unit, each with 9 light-installed beds and an aeroponic fertigation system, resulting in 36 beds, were prepared. Accuracy and response were evaluated for each environmental control component with and without crops. Air temperature, humidity, $CO_2$ concentration, light intensity, frequency, and duty ratio, fertigation rate and scheduling were controllable from a main control computer through wireless communication devices. When the plant factory was operated without crop condition, the response times were 8 minutes for change in temperature from 20 to $15^{\circ}C$ and 20 minutes from 15 to $20^{\circ}C$; 7 minutes for change in humidity from 40 to 65%; and 4 minutes for change in $CO_2$ concentration from 450 to 1000 ppm. When operated for 24 hours with crop cultivation; average, maximum, and minimum values of temperatures were 20.06, 20.8, and $18.8^{\circ}C$; humidity were 66.72, 69.37, and 63.73%; $CO_2$ concentrations were 1017, 1168, and 911 ppm, respectively. Photosynthetic Photon Flux Density was increased as the distance from the light source decreased, but variability was greater at shorter distances. Results of the study would provide useful information for efficient application of the plant factory and to investigate the optimum environment for crop growth through various experiments.

• Journal of Convergence for Information Technology
• /
• v.9 no.12
• /
• pp.124-132
• /
• 2019

Scheduling problems can be classified into offline and online ones. This paper considers an online scheduling problem to minimize makespan on the identical parallel machines. For dynamically arrived jobs with their ready times, we show that the sequencing order according to the ERD (Earliest Ready Date) rule is optimal to minimize makespan. This paper suggests an algorithm by using the MIP(Mixed Integer Programming) formulation periodically to find a good periodic schedule and evaluates the required computational time and resulted makespan of the algorithm. The comparition with an offline scheduling shows our algorithm makes the schedule very fast and the makespan can be reduced as the period time reduction, so we can conclude that our algorithm is useful for scheduling the jobs under online environment even though the number of jobs and machines is large. We expect that the algorithm is invaluable one to find good schedules for the smart factory and online scheduler using the blockchain mechanism.

• The Journal of Society for e-Business Studies
• /
• v.1 no.1
• /
• pp.227-247
• /
• 1996

Presented in this paper is a control model for developing virtual factory simulator, which is being operated under the distributed environment. The control model consists of production activity plan and information flows. To incorporate elements of the characteristics of the distributed control system, we suggested a collaboration model. This model is working under the client/server architecture, and also designed for cooperative-distributed shop control(CDSC) system in order to exploit several advantages of client/server architecture. Collaboration among each agent(or client) is done through negotiation and task sharing. Based on a contract net model, the CDSC system has three kinds of agents-order agent, resource agent, and communication forwarding agent. Each agent performs shop scheduling and control through negotiation on contract net. No node in CDSC system can have authority over other node. A bidding scheme is employed far negotiation between order agent and resource agents. The CDSC system can support re-negotiation among resource agents and an algorithm for re-negotiation is also developed. Experimental results are shown to advocate the effectiveness of the CDSC system for CIM environments.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.18 no.9
• /
• pp.2297-2311
• /
• 1994

This paper presents the study on the development of a DNC, system IPIS(Interactive Plant Information System)/DNC, which can manage NC machines and robots as a distributed control method in the machine. processing factory. The IPIS/DNC system is composed of a host computer, satellites and NC machines. A set of software modules are developed on the host computer and the satellites separately. By modularizing each functions of the IPIS/DNC system and using multi-taking method, the functions such as NC program management, NC program distribution, and shop monitoring can be performed on the host computer, and the functions such as NC program transfer to the NC machines, and NC program editing can be performed on the satellites. A Relational database which is linked with job scheduling system is used for IPIS/DNC system.

• 제어로봇시스템학회:학술대회논문집
• /
• 1992.10b
• /
• pp.446-451
• /
• 1992

In this paper, we propose a method to manage production system easily for operators when either equipments or products are changed. And we explain the scheduling AI tool which realizes the proposal method. The tool's knowledge expression consists of models, rules, mathematical expression and fuzzy logic. The model expresses the relations between products and manufacture, and properties of products. The models are separated into three type, equipment model, operation model, and product model. These models are classified by applicable fields as the assembly process or continuous plant process, The model expression of each type is based on object oriented paradigm. We report systems utilizing our approach.

• Proceedings of the KIEE Conference
• /
• 2004.07d
• /
• pp.2397-2399
• /
• 2004

This paper deals with the messages scheduling of a CAN (Controller Area Network), based on the distributed control scheme to integrate actuators and sensors in a humanoid robot. In order to supply the distributed processing for a humanoid robot, each control unit should have the efficient control method, fast calculation and valid data exchange. The preliminary study has concluded that the performance of CAN is better and easier to implement than other network such as FIP (Factory Instrumentation Protocol), VAN (Vehicle Area Network), etc. Since humanoid robot has to treat the significant control signals from many actuators and sensors, the communication time limitation could be critical according to the transmission speed and data length of CAN specification. In this paper, the CAN message scheduling in humanoid robot was suggested under the presence of Jitter in the message group, the existence of high load of messages over the network and the presence of transmission errors. In addition, the response time under the worst case is compared with the simulation by using the simulation algorithm. As a result, the suggested messages scheduling can guarantee our CAN limitation, and utilized to generate the walking patterns for the humanoid.

• International Journal of Naval Architecture and Ocean Engineering
• /
• v.6 no.1
• /
• pp.162-174
• /
• 2014

Most shipbuilding scheduling research so far has focused on the load level on the dock plan. This is because the dock is the least extendable resource in shipyards, and its overloading is difficult to resolve. However, once dock scheduling is completed, making a plan that makes the best use of the rest of the resources in the shipyard to minimize any additional cost is also important. Block assignment planning is one of the midterm planning tasks; it assigns a block to the facility (factory/shop or surface plate) that will actually manufacture the block according to the block characteristics and current situation of the facility. It is one of the most heavily loaded midterm planning tasks and is carried out manually by experienced workers. In this study, a method of representing the block assignment rules using a diagram was suggested through analysis of the existing manual process. A block allocation program was developed which automated the block assignment process according to the rules represented by the diagram. The planning scenario was validated through a case study that compared the manual assignment and two automated block assignment results.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.8
• /
• pp.2993-3010
• /
• 2021

The job shop scheduling problem (JSSP) plays a critical role in smart manufacturing, an effective JSSP scheduler could save time cost and increase productivity. Conventional methods are very time-consumption and cannot deal with complicated JSSP instances as it uses one optimal algorithm to solve JSSP. This paper proposes an effective scheduler based on deep learning technology named self-supervised long-short term memory (SS-LSTM) to handle complex JSSP accurately. First, using the optimal method to generate sufficient training samples in small-scale JSSP. SS-LSTM is then applied to extract rich feature representations from generated training samples and decide the next action. In the proposed SS-LSTM, two channels are employed to reflect the full production statues. Specifically, the detailed-level channel records 18 detailed product information while the system-level channel reflects the type of whole system states identified by the k-means algorithm. Moreover, adopting a self-supervised mechanism with LSTM autoencoder to keep high feature extraction capacity simultaneously ensuring the reliable feature representative ability. The authors implemented, trained, and compared the proposed method with the other leading learning-based methods on some complicated JSSP instances. The experimental results have confirmed the effectiveness and priority of the proposed method for solving complex JSSP instances in terms of make-span.

• 제어로봇시스템학회:학술대회논문집
• /
• 1992.10a
• /
• pp.811-816
• /
• 1992

The objective of cell operation in CIM is to maximize system efficiency, not the cell itself. We introduce a cell operation scheme that pursues the direction. For specific cases, work center and AS/RS are closely investigated. Structures of their operation mechanism and methods of job scheduling are introduced along with an expert system developed for the scheme. The cell operation softwares developed are now under test at K.U.FMS, a model CIM plant.