• Journal of Institute of Control, Robotics and Systems
• /
• v.22 no.11
• /
• pp.925-931
• /
• 2016

The BLDC motor is used widely in industry due to its controllability and freedom from maintenance because there is no mechanical brush in the BLDC motor. Furthermore, it is suitable for high-speed applications, such as compressors and air blowers. For instance, for a compressor with a small impeller due to miniaturizing, the BLDC motor has to rotate at a very high speed to maintain the compression ratio of the compressor. Typically, to reach an ultra-high speed, airfoil bearings must be used in place of ball bearings because of their friction. Unfortunately, the characteristics of airfoil bearings change drastically depending on the revolution speed. In this paper, a BLDC motor with airfoil bearings is controlled with a PID controller. To analyze and determine the PID coefficients, the relay-feedback method is used. Additionally, for adaptive control, a fuzzy logic controller is used. Furthermore, the auto-tuning and self-tuning techniques are combined to control the BLDC motor. The proposed method is able to control the airfoil-bearing BLDC motor efficiently.

• Transactions on Control, Automation and Systems Engineering
• /
• v.3 no.1
• /
• pp.43-50
• /
• 2001

In this paper, we propose the design procedure of advance Polynomial Neural Networks(PNN) architecture for optimal model identification of complex and nonlinear system. The proposed PNN architecture is presented as the generic and advanced type. The essence of the design procedure dwells on the Group Method of Data Handling(GMDH). PNN is a flexible neural architecture whose structure is developed through learning. In particular, the number of layers of the PNN is not fixed in advance but is generated in a dynamic way. In this sense, PNN is a self-organizing network. With the aid of three representative numerical examples, compari-sons show that the proposed advanced PNN algorithm can produce the model with higher accuracy than previous other works. And performance index related to approximation and generalization capabilities of model is evaluated and also discussed.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.10
• /
• pp.2239-2246
• /
• 2011

In image signal processing, image signal is corrupted by various noises and caused the degradation phenomenon. And Images often corrupted by AWGN(additive white gaussian noise) and impulse noise which called mixed noise. In this paper, the algorithm is proposed to remove mixed noise while keeping edge information. The proposed algorithm first classifies the noise type, if the classify result is AWGN, then the mean of the output after using self-adaptive weighted mean filter and median value will be the outfiltering value. And if the noise type is impulse noise, then the noise is removed by a modified nonlinear filter. Also we compare existing methods through the simulation and using PSNR(peak signal to noise ratio) as the standard of judgement of improvement effect. The result of computer simulation on test images indicates that the proposed method is superior to traditional filtering algorithms.

• International Journal of Control, Automation, and Systems
• /
• v.6 no.3
• /
• pp.453-459
• /
• 2008

The neural network is currently being used throughout numerous control system fields. However, it is not easy to obtain an input-output pattern when the neural network is used for the system of a single feedback controller and it is difficult to obtain satisfactory performance with when the load changes rapidly or disturbance is applied. To resolve these problems, this paper proposes a new mode to implement a neural network controller by installing a real object for control and an algorithm for this, which can replace the existing method of implementing a neural network controller by utilizing activation function at the output node. The real plant object for controlling of this mode implements a simple neural network controller replacing the activation function and provides the error back propagation path to calculate the error at the output node. As the controller is designed using a simple structure neural network, the input-output pattern problem is solved naturally and real-time learning becomes possible through the general error back propagation algorithm. The new algorithm applied neural network controller gives excellent performance for initial and tracking response and shows a robust performance for rapid load change and disturbance, in which the permissible error surpasses the range border. The effect of the proposed control algorithm was verified in a test that controlled the speed of a motor equipped with a high speed computing capable DSP on which the proposed algorithm was loaded.

• Structural Monitoring and Maintenance
• /
• v.1 no.4
• /
• pp.427-449
• /
• 2014

A hybrid optimization method for the identification of state-space models is presented in this study. Hybridization is succeeded by combining the advantages of deterministic and stochastic algorithms in a superior scheme that promises faster convergence rate and reliability in the search for the global optimum. The proposed hybrid algorithm is developed by replacing the original stochastic mutation operator of Evolution Strategies (ES) by the Levenberg-Marquardt (LM) quasi-Newton algorithm. This substitution results in a scheme where the entire population cloud is involved in the search for the global optimum, while single individuals are involved in the local search, undertaken by the LM method. The novel hybrid identification framework is assessed through the Monte Carlo analysis of a simulated system and an experimental case study on a shear frame structure. Comparisons to subspace identification, as well as to conventional, self-adaptive ES provide significant indication of superior performance.

• 한국정보통신설비학회:학술대회논문집
• /
• 2009.08a
• /
• pp.382-386
• /
• 2009

Online or sequential learning is one of the most basic and powerful method to train neuron network, and it has been widely used in disease detection, weather prediction and other realistic classification problem. At present, there are many algorithms in this area, such as MRAN, GAP-RBFN, OS-ELM, SVM and SMC-RBF. Among them, SMC-RBF has the best performance; it has less number of hidden neurons, and best efficiency. However, all the existing algorithms use signal normal distribution as kernel function, which means the output of the kernel function is same at the different direction. In this paper, we use multi-variable normal distribution as kernel function, and derive EKF learning formulas for multi-variable normal distribution kernel function. From the result of the experience, we can deduct that the proposed method has better efficiency performance, and not sensitive to the data sequence.

• Journal of Korea Multimedia Society
• /
• v.21 no.4
• /
• pp.471-477
• /
• 2018

As an important key technology using on electronic devices, face recognition has become one of the hottest technology recently. The traditional BP Neural network has a strong ability of self-learning, adaptive and powerful non-linear mapping but it also has disadvantages such as slow convergence speed, easy to be traversed in the training process and easy to fall into local minimum points. So we come up with an algorithm based on BP neural network but also combined with the PCA algorithm and other methods such as the elastic gradient descent method which can improve the original network to try to improve the whole recognition efficiency and has the advantages of both PCA algorithm and BP neural network.

• Nonlinear Functional Analysis and Applications
• /
• v.27 no.1
• /
• pp.121-140
• /
• 2022

In this paper, we study the strong convergence of new methods for solving classical variational inequalities problems involving semistrictly quasimonotone and Lipschitz-continuous operators in a real Hilbert space. Three proposed methods are based on Tseng's extragradient method and use a simple self-adaptive step size rule that is independent of the Lipschitz constant. The step size rule is built around two techniques: the monotone and the non-monotone step size rule. We establish strong convergence theorems for the proposed methods that do not require any additional projections or knowledge of an involved operator's Lipschitz constant. Finally, we present some numerical experiments that demonstrate the efficiency and advantages of the proposed methods.

• Nuclear Engineering and Technology
• /
• v.55 no.10
• /
• pp.3515-3524
• /
• 2023

A reinforcement learning framework is proposed for the control problem of outlet steam pressure of the once-through steam generator(OTSG) in this paper. The double-layer controller using Proximal Policy Optimization(PPO) algorithm is applied in the control structure of the OTSG. The PPO algorithm can train the neural networks continuously according to the process of interaction with the environment and then the trained controller can realize better control for the OTSG. Meanwhile, reinforcement learning has the characteristic of difficult application in real-world objects, this paper proposes an innovative pretraining method to solve this problem. The difficulty in the application of reinforcement learning lies in training. The optimal strategy of each step is summed up through trial and error, and the training cost is very high. In this paper, the LSTM model is adopted as the training environment for pretraining, which saves training time and improves efficiency. The experimental results show that this method can realize the self-adjustment of control parameters under various working conditions, and the control effect has the advantages of small overshoot, fast stabilization speed, and strong adaptive ability.

• Advances in Computational Design
• /
• v.2 no.4
• /
• pp.313-331
• /
• 2017

In this study, teaching-learning based optimization (TLBO) is improved by incorporating model of multiple teachers, adaptive teaching factor, self-motivated learning, and learning through tutorial. Modified TLBO (MTLBO) is applied for simultaneous topology, shape, and size optimization of space and planar trusses to study its effectiveness. All the benchmark problems are subjected to stress, displacement, and kinematic stability constraints while design variables are discrete and continuous. Analyses of unacceptable and singular topologies are prohibited by seeing element connectivity through Grubler's criterion and the positive definiteness. Performance of MTLBO is compared to TLBO and state-of-the-art algorithms available in literature, such as a genetic algorithm (GA), improved GA, force method and GA, ant colony optimization, adaptive multi-population differential evolution, a firefly algorithm, group search optimization (GSO), improved GSO, and intelligent garbage can decision-making model evolution algorithm. It is observed that MTLBO has performed better or found nearly the same optimum solutions.