• Proceedings of the KIEE Conference
• /
• 2006.07b
• /
• pp.759-760
• /
• 2006

In this paper proposed to Neural network PI self-tuning direct controller using Error back propagation algorithm. Proposed controller applies to speed controller and current controller. Also, this built up the interface environment to drive it simply and exactly in any kind of reference, environment fluent and parameter transaction of BLDC motor. Neural network PI self-tuning simulator using Visual C++ and Matlab Simulation is organized to construct this environment. Built-u-p interface has it's own purpose that even the user who don't have the accurate knowledge of neural network can embody operation characteristic rapidly and easily.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.2 no.3
• /
• pp.295-302
• /
• 1998

In this paper, a new auto-tuning PI controller for the speed servo system of a PMSM is designed by using limited step response characteristics. The method is proposed that gets information about auto-tuning of PI regulator by the injection of step input, called limited input, during a transient response time of control. System parameter estimation and speed control could be continuously executed. This means that in despite of system uncertainty the system information obtained by limited input can be continuously applied to the PI regulator. We demonstrate the effectiveness of the proposed auto-tuning algorithm through simulation and experiment result of the speed control for a PMSM having monotone increasing step response.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.343-347
• /
• 2022

최근 자연어처리 분야에서는 BERT, RoBERTa, 그리고 BART와 같은 사전 학습된 언어 모델 (Pre-trained Language Models, PLM) 기반 미세 조정 학습을 통하여 여러 하위 과업에서 좋은 성능을 거두고 있다. 이는 사전 학습된 언어 모델 및 데이터 집합의 크기, 그리고 모델 구성의 중요성을 보여주며 대규모 사전 학습된 언어 모델이 각광받는 계기가 되었다. 하지만, 거대한 모델의 크기로 인하여 실제 산업에서 쉽게 쓰이기 힘들다는 단점이 명백히 존재함에 따라 최근 매개변수 효율적인 미세 조정 및 Few-Shot 학습 연구가 많은 주목을 받고 있다. 본 논문은 Prompt tuning, Prefix tuning와 프롬프트 기반 미세 조정 (Prompt-based fine-tuning)을 결합한 Few-Shot 학습 연구를 제안한다. 제안한 방법은 미세 조정 ←→ 사전 학습 간의 지식 격차를 줄일 뿐만 아니라 기존의 일반적인 미세 조정 기반 Few-Shot 학습 성능보다 크게 향상됨을 보인다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.12
• /
• pp.3416-3435
• /
• 2023

Task offloading in vehicular networks is hot topic in the development of autonomous driving. In these scenarios, due to the role of vehicles and pedestrians, task characteristics are changing constantly. The classical deep learning algorithm always uses a pre-trained neural network to optimize task offloading, which leads to system performance degradation. Therefore, this paper proposes a neural network fine-tuning task offloading algorithm, combining with location prediction for pedestrians and vehicles by the Payne model of fluid dynamics and the car-following model, respectively. After the locations are predicted, characteristics of tasks can be obtained and the neural network will be fine-tuned. Finally, the proposed algorithm continuously predicts task characteristics and fine-tunes a neural network to maintain high system performance and meet low delay requirements. From the simulation results, compared with other algorithms, the proposed algorithm still guarantees a lower task offloading delay, especially when congestion occurs.

• The Transactions of the Korean Institute of Electrical Engineers B
• /
• v.48 no.3
• /
• pp.131-138
• /
• 1999

Generally, the digital controller has many advantages such as high precision, robustness to electrical noise, capability of flexible programming and fast response to the load variation. In this study, we have established proper mathematical equivalent model of Brushless DC (BLDC) motor and estimated the motor parameter by means of the back-emf measurement as being the step input to the controlled target BLDC motor. And the validity of proposed estimation method is confirmed by the test result of step response. As well, we have designed the reasonable digital controller as a consequence of the root locus method which is obtained from the open-loop transfer function of BLDC motor with hall sensor, and the determination of control gain for variable speed control. Here, revised Ziegler-Nichols tuning method is applied for the proper digital gain establishment, and the system stability is verified by the frequency domain analysis with Bode-plot and experimentation.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.11 no.2
• /
• pp.184-191
• /
• 2006

This paper deals with the HVDC system fault analysis and the tuning of the control parameter. This paper is based on the actual cable fault which was caused by the overvoltage due to commutation failure. In order to analyze the fault, the detailed PSCAD/EMTDC model which the sampling time is $50{\mu}s$, was made. Finally, the control parameter of HVDC system was tuned.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.2542-2547
• /
• 2005

In this paper, we propose a support vector machine (SVM) hyper and kernel parameter optimization method which is based on minimizing radius/margin bound which is a kind of estimation of leave-one-error. This method uses dynamic encoding algorithm for search (DEAS) and gradient information for better optimization performance. DEAS is a recently proposed optimization algorithm which is based on variable length binary encoding method. This method has less computation time than genetic algorithm (GA) based and grid search based methods and better performance on finding global optimal value than gradient based methods. It is very efficient in practical applications. Hand-written letter data of MNI steel are used to evaluate the performance.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.1
• /
• pp.77-85
• /
• 2009

With the advent of power semiconductor switching devices, power electronics relating to high frequency electromagnetic eddy current based induction heating technology have become more suitable and acceptable. This paper presents high-frequency induction heating cooking appliance circuit based on the zero current switching-PWM single ended push-pull(ZCS-PWM SEPP) resonant inverter added AC-DC converter. This inverter uses pulse-width-modulation(PWM) control method with active auxiliary quasi-resonant lossless inductor snubbers and a switched capacitor. To improved the transient performance, the PI controller is applied for this system. For the systematic parameter optimization of the PI controller, the gradient-based optimization algorithm is applied. The performance of optimized parameters is evaluated using simulation and experimental test. These results show that the proposed systematic optimal tuning method improve the transient performances of this system.

• Advances in aircraft and spacecraft science
• /
• v.4 no.2
• /
• pp.185-202
• /
• 2017

In a flexible airvehicle, an assessment of the structural coupling levels through analysis and experiments provides structural data for the design of notch filters which are generally utilized in the flight control system to attenuate the flexible response pickup. This is necessitated as during flight, closed loop control actuation driven with flexible response inputs could lead to stability and performance related problems. In the present work, critical parameters influencing servoelastic response have been identified. A sensitivity study has been carried out to assess the extent of influence of each parameter. A multi-parameter tuning approach has been implemented to achieve an enhanced analytical model for improved predictions of aircraft servoelastic response. To illustrate the model updation approach, initial and improved test analysis correlation of lateral servoelastic responses for a generic flexible airvehicle are presented.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.446-448
• /
• 2021

In training a deep learning model, it is crucial to tune various hyperparameters and gain speed and accuracy. While hyperparameters that mathematically induce convergence impact training speed, system parameters that affect host-to-device transfer are also crucial. Therefore, it is important to properly tune and select parameters that influence the data loader as a system parameter in overall time acceleration. We propose an automated framework called Num Worker Tuner (NWT) to address this problem. This method finds the appropriate number of multi-processing subprocesses through the search space and accelerates the learning through the number of subprocesses. Furthermore, this method allows memory efficiency and speed-up by tuning the system-dependent parameter, the number of multi-process spawns.