• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 1999.10a
• /
• pp.323-335
• /
• 1999

In this paper, we propose a new hybrid model of artificial neural networks(ANNs) and genetic algorithm (GA) to optimal feature transformation and feature weighting. Previous research proposed several variants of hybrid ANNs and GA models including feature weighting, feature subset selection and network structure optimization. Among the vast majority of these studies, however, ANNs did not learn the patterns of data well, because they employed GA for simple use. In this study, we incorporate GA in a simultaneous manner to improve the learning and generalization ability of ANNs. In this study, GA plays role to optimize feature weighting and feature transformation simultaneously. Globally optimized feature weighting overcome the well-known limitations of gradient descent algorithm and globally optimized feature transformation also reduce the dimensionality of the feature space and eliminate irrelevant factors in modeling ANNs. By this procedure, we can improve the performance and enhance the generalisability of ANNs.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2015.11a
• /
• pp.68-69
• /
• 2015

Traditional External Insulation Finishing System(EIFS) is applied to apartment construction by performing structural framework and insulation finishing work sequentially. Separate execution of the three works increases construction cost and duration. Concurrent construction method of EIFS, which performs framework and insulation finishing work simultaneously, is introduced in order to solve these problems. However, the introduced method is exposed to the risk of construction delay caused by bottlenecks due to interacting processes and resources. Therefore, this paper presents a simulation model suitable for estimating work productivity of the concurrent construction by considering predecessor and successor processes to optimize resource allocation and minimize construction delay.

• Proceedings of the KIEE Conference
• /
• 1996.07a
• /
• pp.210-212
• /
• 1996

A nonlinear optimization method for the shape optimization of actively shielded superconducting MRI magnet is presented. The presented design method can optimize both main coil and shielding coil simultaneously by setting constraints on stray field intensity at a specified distance from the magnet center. A 1 Tesla actively shielded superconducting MRI magnet, with 30cm bore diameter, is designed using the presented method.

• Journal of Ocean Engineering and Technology
• /
• v.23 no.3
• /
• pp.1-5
• /
• 2009

It is necessary to develop an efficient optimization technique to optimize engineering structures that have given design spaces, discrete design values, and several design goals. In this study, an optimum algorithm based on the genetic algorithm was applied to the multi-object problem to obtain an optimum solution that simultaneously minimizes the structural weight and construction cost of panel blocks in ship structures. The cost model was used in this study, which includes the cost of adjusting the weld-induced deformation and applying the deformation control methods, in addition to the cost of the material and the welding cost usually included in the normal cost model. By using the proposed cost model, more realistic optimum design results can be expected.

• Proceedings of the Korean Society of Machine Tool Engineers Conference
• /
• 2002.04a
• /
• pp.614-618
• /
• 2002

In hot strip mills, Portable Roll Scanner (the portable roll surface temperature and profile measuring device) can be used to calibrate on-line Process models for strip crown and flatness by measuring the thermal expansion and wear profile of the rolls. And the surface temperature measurement can be used to optimize the roll cooling system. Portable Roll Scanner consists of the measuring device, which has two contact inductive distance transducers for roll profile measurement and one infrared Pyrometer for surface temperature measurement, and computer-based controller that is equipped with the measuring device. By the wireless data communication, the data is transferred to the memory of notebook for further analysis. After roll extraction from mills, Portable Roll Scanner measure the roll profile and surface temperature simultaneously along the work roll face and display the results in the TFT color monitor of notebook. Portable Roll Scanner is useful at mill-side and roll grinding shop.

• Journal of Electrical Engineering and Technology
• /
• v.6 no.1
• /
• pp.128-132
• /
• 2011

In this paper, two automatic mesh generation algorithms are presented. The methods seek to optimize mesh density with regard to geometries exhibiting both fine and coarse physical structures. When generating meshes, the algorithms attempt to satisfy the conditions on the maximum mesh spacing and the maximum grading ratio simultaneously. Both algorithms successfully produce non-uniform meshes that satisfy the requirements for finite-difference time-domain simulations of microwave components. Additionally, an algorithm successfully generates a minimum number of grid points while maintaining the simulation accuracy.

• The Transactions of the Korean Institute of Electrical Engineers P
• /
• v.60 no.3
• /
• pp.126-132
• /
• 2011

A conventional LDO(Low Dropout Regulator) uses one OPAMP and one signal path. This means that OPAMP's DC Gain and Bandwidth can't optimize simultaneously within usable power. This also appears that regulation property and settling time of LDO can't improve at the same time. Based on this idea, a proposed LDO uses two OPAMP and has two signal path. To improve regulation property, OPAMP where is used in the path which qualities DC gain on a large scale, bandwidth designed narrowly. To improve settling time, OPAMP where is used in the path which qualities DC gain small, bandwidth designed widely. A designed LDO used 0.5um 1P2M process and provided 200mA of output current. A line regulation and load regulation is 12.6mV/V, 0.25mV/mA, respectively. And measured settling time is 1.5us in 5V supply voltage.

• IE interfaces
• /
• v.16 no.3
• /
• pp.280-290
• /
• 2003

Many firms are trying to optimize their production and distribution functions separately, but possible savings by this approach may be limited. Nowadays, it is more important to analyze these two functions simultaneously by trading off the costs associated with the whole. In this paper, I treat a production and distribution planning problem for single-period inventory products comprised of a single production facility and multiple customers, with the aim of optimally coordinating important and interrelated decisions of production sequencing and vehicle routing. Then, I propose a hybrid genetic algorithm incorporating several local optimization techniques, HGAP, for integrated production-distribution planning. Computational results on test problems show that HGAP is effective and generates substantial cost savings over Hurter and Buer's decoupled planning approach in which vehicle routing is first developed and a production sequence is consequently derived. Especially, HGAP performs better on the problems where customers are dispersed with multi-item demand than on the problems where customers are divided into several zones based on single-item demand.

• Smart Structures and Systems
• /
• v.15 no.4
• /
• pp.1041-1062
• /
• 2015

In this paper a method of finding optimal positions for piezoelectric actuators and sensors on different structures is presented. The genetic algorithm and multi-objective genetic algorithm are selected for optimization and $H_{\infty}$ norm is defined as a cost function for the optimization process. To optimize the placement concerning the selected modes simultaneously, the multi-objective genetic algorithm is used. The optimization is investigated for two different structures: a cantilever beam and a simply supported plate. Vibrating structures are controlled in a closed loop with feedback gains, which are obtained using optimal LQ control strategy. Finally, output of a structure with optimized placement is compared with the output of the structure with an arbitrary, non-optimal placement of piezoelectric patches.

• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 2000.04a
• /
• pp.347-356
• /
• 2000

This paper compares four models of artificial neural networks (ANN) supported by genetic algorithms the prediction of stock price index. Previous research proposed many hybrid models of ANN and genetic algorithms(GA) in order to train the network, to select the feature subsets, and to optimize the network topologies. Most these studies, however, only used GA to improve a part of architectural factors of ANN. In this paper, GA simultaneously optimized multiple factors of ANN. Experimental results show that GA approach to simultaneous optimization for ANN (SOGANN3) outperforms the other approaches.