• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.47 no.2
• /
• pp.134-146
• /
• 2024

This study introduces and experimentally validates a novel approach that combines Instruction fine-tuning and Low-Rank Adaptation (LoRA) fine-tuning to optimize the performance of Large Language Models (LLMs). These models have become revolutionary tools in natural language processing, showing remarkable performance across diverse application areas. However, optimizing their performance for specific domains necessitates fine-tuning of the base models (FMs), which is often limited by challenges such as data complexity and resource costs. The proposed approach aims to overcome these limitations by enhancing the performance of LLMs, particularly in the analysis precision and efficiency of national Research and Development (R&D) data. The study provides theoretical foundations and technical implementations of Instruction fine-tuning and LoRA fine-tuning. Through rigorous experimental validation, it is demonstrated that the proposed method significantly improves the precision and efficiency of data analysis, outperforming traditional fine-tuning methods. This enhancement is not only beneficial for national R&D data but also suggests potential applicability in various other data-centric domains, such as medical data analysis, financial forecasting, and educational assessments. The findings highlight the method's broad utility and significant contribution to advancing data analysis techniques in specialized knowledge domains, offering new possibilities for leveraging LLMs in complex and resource-intensive tasks. This research underscores the transformative potential of combining Instruction fine-tuning with LoRA fine-tuning to achieve superior performance in diverse applications, paving the way for more efficient and effective utilization of LLMs in both academic and industrial settings.

• Journal of Advanced Marine Engineering and Technology
• /
• v.39 no.10
• /
• pp.1031-1036
• /
• 2015

Proportional-integral-derivative (PID) controllers are widely used in industrial sites. Most tuning methods for PID controllers use an empirical and experimental approach; thus, the experience and intuition of a designer greatly affect the tuning of the controller. The representative methods include the closed-loop tuning method of Ziegler-Nichols (Z-N), the C-C tuning method, and the Internal Model Control tuning method. There has been considerable research on the tuning of PID controllers for single-input single-output systems but very little for multi-input multi-output systems. It is more difficult to design PID controllers for multi-input multi-output systems than for single-input single-output systems because there are interactive control loops that affect each other. This paper presents a tuning method for the PID controller for a two-input two-output system. The proposed method uses a real-coded genetic algorithm (RCGA) as an optimization tool, which optimizes the PID controller parameters for minimizing the given objective function. Three types of objective functions are selected for the RCGA, and each PID controller parameter is determined accordingly. The performance of the proposed method is compared with that of the Z-N method, and the validity of the proposed method is examined.

• 제어로봇시스템학회:학술대회논문집
• /
• 1997.10a
• /
• pp.704-707
• /
• 1997

In this paper, we present a fuzzy logic based tuner for continuous on-line tuning of PID controllers. The essential idea of the scheme is to parameterize a Ziegler-Nichols-like tuning formula by a singler parameter .alpha., then to use an on line fuzzy logic to self-tune the parameter. The adaptive scaling makes the controller robust against large variations in parametric and dynamics uncertainties in the plant model. New self-tuning controller has the ability to decide when to use PI or PID control by extracting process dynamics from relay experiments. These scheme lead to improved performance of the transient and steady state behavior of the closed loop system, including processes with nonminimum phase processes.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.3E
• /
• pp.112-118
• /
• 2002

An unconstrained tuning fork with a 3-D model has been numerically analyzed by Finite Element Method (FEM) and Boundary Element Method (BEM). The first three natural frequencies were calculated by the FEM modal analysis. Then the trend of the change of the modal frequencies was examined with the variation of the tuning fork length and width. An formula for the natural frequencies-tuning fork length relationship were derived from the numerical analysis results. Finally the BEM was used for the sound pressure field calculation from the structural displacement data.

• Proceedings of the Korean Society of Machine Tool Engineers Conference
• /
• 1996.10a
• /
• pp.150-155
• /
• 1996

This paper presents a new approach to the design of self-tuning adaptive control system that is robust to the changing dynamic configuration as well as to the load variation factors using digital signal processors for robot manipulators. TMS3200C50 is used in implementing real-time adaptive control algorithms provide advanced performance for robot manipulator. In this paper an adaptive control scheme is proposed in order to design the pole-placement self-tuning controller which can reject the offset due to any load disturbance without a detailed description of robot dynamics. parameters of discrete-time difference model are estimated by the recursive least-square identification algorithm and controller parameters are detemined by the pole-placement method. Performance of self-tuning adaptive controller is illusrated by the simulation and experiment for a SCARA robot.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.8
• /
• pp.1119-1126
• /
• 1995

The construction of rule-base for a nonlinear time-varying system, becomes much more complicated because of model uncertainty and parameter variations. Furthemore, FLC does not have an ability of adjusting rule- base in responding to some sudden changes of control environments. To cope with these problems, an auto-tuning method of the fuzzy rule-base is required. In this paper, the GA-based Fuzzy-Neural control system combining Fuzzy-Neural control theory with the genetic algorithm(GA), which is known to be very effective in the optimization problem, will be proposed. The tuning of the proposed system is performed by two tuning processes(the course tuning process and the fine tuning/adaptive learning process). The effectiveness of the proposed control system will be demonstrated by computer simulations using a two degree of freedom robot manipulator.

• Journal of the Korean Society for Precision Engineering
• /
• v.29 no.11
• /
• pp.1207-1213
• /
• 2012

Typically commercial controllers do not give data of the controller gains. Therefore, it is very hard to determine the optimal controller gain even though the dynamic model is derived. In this case, design of experiment (DOE) methodology can be a powerful tool for gain tuning. In this research, gain tuning process is proposed based on the DOE. Micro parallel mechanism platform with 3 degrees-of-freedom (DOF) is used for the experiments. Controller gains are measured indirectly from the voltages of adjustable resistors. The controller gains of three actuators are optimized by two or three steps, respectively. The correlations of the controller gains are also analyzed. The process and methodology can be adopted in gain tuning of other mechanical systems.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1996.11a
• /
• pp.754-758
• /
• 1996

This paper presents a new approach to the design of self-tuning adaptive control system that is robust to the changing dynamic configuration as well as to the load variation factors using Digital signal processors for robot manipulators. TMS320C50 is used in implementing real-time adaptive control algorithms to provide advanced performance for robot manipulator, In this paper, an adaptive control scheme is proposed in order to design the pole-placement self-tuning controller which can reject the offset due to any load disturbance without a detailed description of robot dynamics. Parameters of discrete-time difference model are estimated by the recursive least-square identification algorithm, and controller parameters we determined by the pole-placement method. Performance of self-tuning adaptive controller is illusrated by the simulation and experiment for a SCARA robot.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2003.05a
• /
• pp.1201-1204
• /
• 2003

An unconstrained tuning fork with a 3-D model has been numerically analyzed by Finite Element Method(FEM) and Boundary Element Method (BEM). The first three natural frequencies were calculated by the FEM modal analysis. Then the trend of the change of the modal frequencies was examined with the variation of the tuning fork length and width. An formula for the natural frequencies-tuning fork length relationship were derived from the numerical analysis results. Finally the BEM was used fur the sound pressure field calculation from the structural displacement data.

• Proceedings of the Korea Society for Simulation Conference
• /
• 2005.11a
• /
• pp.165-170
• /
• 2005

원전 시뮬레이터의 노심모델(REMARK)은 원자로에서 발생하는 1차원 열원을 정확히 모사하고, 정상상태 및 과도상태의 반응도 변화에 따른 중성자속의 거동을 실제 원자로와 유사하게 모사할 수 있어야 한다. 모사의 정확성을 높이기 위해 Tuning 작업이 필수적이나 그 작업 단계가 매우 복잡하여 많은 시간과 노력이 필요하기 때문에 이를 간소화하면서 정확성을 높일 수 있는 Tuning 지원 프로그램을 개발하였다. 개발된 프로그램의 사용결과 신속하고 정확한 Tuning이 가능하였다.