• Journal of Institute of Control, Robotics and Systems
• /
• v.8 no.7
• /
• pp.556-562
• /
• 2002

This paper studies the identification of a combustion model, which is used to design a linear controller of a steam generation quantity and harmful exhaust gases of a Refuse Incineration Plant(RIP). Even though the RIP has strong nonlinearities and complexities, it is identified as a MIMO parametric ARX model from experimental input-output data sets. Unknown model parameters are decided from experimental input-output data sets, using system identification algorithm based on Instrumental Variables(IV) method. It is shown that the identified model well approximates the input-output combustion characteristics.

• International Journal of Korean Welding Society
• /
• v.1 no.1
• /
• pp.44-50
• /
• 2001

A genetic algorithm was applied to the arc welding process as to determine the near-optimal settings of welding process parameters that produce the good weld quality. This method searches for optimal settings of welding parameters through the systematic experiments without the need for a model between the input and output variables. It has an advantage of being capable to find the optimal conditions with a fewer number of experiments rather than conventional full factorial designs. A genetic algorithm was applied to the optimization of the weld bead geometry. In the optimization problem, the input variables were wire feed rate, welding voltage, and welding speed. The output variables were the bead height bead width, and penetration. The number of levels for each input variable is 16, 16, and 8, respectively. Therefore, according to the conventional full factorial design, in order to find the optimal welding conditions,2048 experiments must be performed. The genetic algorithm, however, found the near optimal welding conditions in less than 40 experiments.

• Journal of the Korean Society of Safety
• /
• v.34 no.5
• /
• pp.46-54
• /
• 2019

Even though the fire performance-based design concept has been introduced for various structures and buildings, which have their own specific fire performance level, the uncertainties of input parameters always exist and, then, could reduce significantly the reliability of the fire modeling. Sensitivity analysis was performed with three limited input parameters, HRRPUA, type of combustible materials, and mesh size, which are significantly important for fire modeling. The output variables are limited to the maximum HRR, the time reaching the reference temperature($60^{\circ}C$), and that to reach limited visible distance(5 m). In addition, correlation coefficient analysis was attempted to analyze qualitatively and quantitatively the degree of relation between input and output variables above. Finally, the relationship among the three variables is also analyzed by the principal component analysis (PCA) to systematically analyze the input data bias. Sensitivity analysis showed that the type of combustible materials is more sensitive to maximum HRR than the ignition source and mesh size. However, the heat release parameter of the ignition source(HRR) is shown to be much more sensitive than the combustible material types and mesh size to both time to reach the reference temperature and that to reach the critical visible distance. Since the derived results can not exclude the possibility that there is a dependency on the fire model applied in this study, it is necessary to generalize and standardize the results of this study for the fire models such as various buildings and structures.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.17 no.5
• /
• pp.79-84
• /
• 2017

This paper is a multi-input multi-path (Multiple-input multiple-output: MIMO) using a space-alternating generalized expectation maximization(SAGE) algorithm in the parameter channel and determine the channel estimation performance. Estimated by the algorithm, SAGE time-varying channel environment, the channel parameters estimated from the parameters of the channel measured in the island region 781 of the band in order to compare the performance and compares the original data. This allows you to check the performance of the algorithm SAGE and is highly stable to delay spread (Delay Spread), the diffusion angle of arrival (Arrive of Angular Spread) performance in terms of accuracy down through the SAGE algorithm for estimating a more general calculation parameters.

• 제어로봇시스템학회:학술대회논문집
• /
• 1993.10b
• /
• pp.1-6
• /
• 1993

This report proposes a novel method of identification of continuous time-delay systems from sampled input-output data. By the aid of a digital pre-filter, an approximated discrete-time estimation model is first derived, in which the system parameters remain in their original form and the time delay need not be an integral multiple of th sampling period. Then an identification method combining the common linear least squares(LS) method or the instrumental variable(IV) method with the genetic algorithm(GA) is proposed. That is, the time-delay is selected by the GA, and the system parameters are estimated by the LS or IV method. Furthermore, the proposed method is extended to the case of multi-input multi-output systems where the time-delays in the individual input channels may differ each other. Simulation resutls show that our method yields consistent estimates even in the presence of high measurement noises.

• Journal of Power Electronics
• /
• v.19 no.1
• /
• pp.89-98
• /
• 2019

This paper analyzes the input side performance of a conventional three-phase to single-phase matrix converter (3-1MC). It also presents the input-side waveform quality under this topology. The suppression of low-frequency input current harmonics is studied using the 3-1MC plus capacitance compensation unit. The constraint between the modulation function of the output and compensation sides is analyzed, and the relations among the voltage utilization ratio and the output compensation capacitance, filter capacitors and other system parameters are deduced. For a 3-1MC without large-capacity energy storage, the system performance is susceptible to input voltage imbalance. This paper decouples the inner current of the 3-1MC using a Lyapunov function in the input positive and negative sequence bi-coordinate axes. Meanwhile, the outer loop adopts a voltage-weighted synthesis of the output and compensation sides as a cascade of control objects. Experiments show that this strategy suppresses the low-frequency input current harmonics caused by input voltage imbalance, and ensures that the system maintains good static and dynamic performances under input-unbalanced conditions. At the same time, the parameter selection and debugging methods are simple.

• Proceedings of the KIEE Conference
• /
• 2000.07d
• /
• pp.2545-2547
• /
• 2000

In this paper, a hybrid fuzzy controller using genetic algorithm based on parameter estimation mode to obtain optimal control parameter is presented. First, The control input for the system in the HFC is a convex combination of the FLC's output in transient state and PID's output in steady state by a fuzzy variable, namely, membership function of weighting coefficient. Second, genetic algorithms is presented to automatically improve the performance of hybrid fuzzy controller utilizing the conventional methods for finding PID parameters and estimation mode of scaling factor. The algorithms estimates automatically the optimal values of scaling factors, PID parameters and membership function parameters of fuzzy control rules according to the rate of change and limitation condition of control input. Computer simulations are conducted to evaluate the performance of proposed hybrid fuzzy controller. ITAE, overshoot and rising time are used as a performance index of controller.

• Journal of Korean Society on Water Environment
• /
• v.21 no.1
• /
• pp.21-28
• /
• 2005

The training and prediction modeling using an artificial neural network was implemented to predict the turbidity of treated water as well as to estimate the optimized feed concentration of polyaluminium chloride (PACl) in a water treatment plant. The parameters used in the input layers were pH, temperature, turbidity and alkalinity, while those in output layers were PACl and turbidity of treated water. Levenberg-Marquadt method of feedforward back-propagation perceptron in the neural network toolbox of MATLAB program was used in this study. Correlation coefficients of the training data with the measured data were 0.9997 for PACl and 0.6850 for turbidity and those of the testing data with measured data were 0.9140 for PACl and 0.3828 for turbidity, when four parameters at input layer, 12-12 nodes each at both the first and the second hidden layers, and two parameters(PACl and turbidity) at output layer were used. Although the predictability of PACl was improved, compared to that of the previous studies to use the only coagulant dose as output layer, turbidity in treated water could not be predicted well. Acquisition of more data through several years obtained with the advanced on-line measuring system could make the artificial neural network useful and practical in actual water treatment plants.

• Korean Management Science Review
• /
• v.15 no.2
• /
• pp.125-142
• /
• 1998

The aggregation on linear large-scale dynamic systems is examined in this paper and a "two-step" approach is proposed. In this procedure, the aggregated system consists of two subsystems. The first subsystem represents aggregation through the retainment of dominant eigenvalues of the original system, leading to a first approximation of the desired output of the original system. The purpose of augmenting it with a second subsystem is to provide an estimation of the error on the first approximation, thus permitting a second correction to the output approximation and resulting in an output approximation of greater accuracy. Optimization techniques are discussed for the determination of unknown parameters in the aggregated system. These techniques use minimization principles of certain suitable performance indices and are developed for both single input-single output and multiple input-multiple output system. Numerical examples illustrating these procedures are given and the results are compared with those obtained using existing methods. Finally, a pharmacokinetics problem is studied from the aggregation point of view.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.2
• /
• pp.159-167
• /
• 1996

In this paper, a new competitive learning algorithm called dynamic competitive learning (DCL) is presented. DCL is a supervised learning mehtod that dynamically generates output neuraons and nitializes weight vectors from training patterns. It introduces a new parameter called LOG (limit of garde) to decide whether or not an output neuron is created. In other words, if there exist some neurons in the province of LOG that classify the input vector correctly, then DCL adjusts the weight vector for the neuraon which has the minimum grade. Otherwise, it produces a new output neuron using the given input vector. It is largely learning is not limited only to the winner and the output neurons are dynamically generated int he trining process. In addition, the proposed algorithm has a small number of parameters. Which are easy to be determined and applied to the real problems. Experimental results for patterns recognition of remote sensing data and handwritten numeral data indicate the superiority of dCL in comparison to the conventional competitive learning methods.