• 디지털융복합연구
• /
• 제18권2호
• /
• pp.261-268
• /
• 2020

기계학습은 주어진 디지털 신호 Data로부터 비용함수를 만들고, 그 비용함수를 최소화함으로 학습이 이루어진다. 비용함수는 디지털 신호 Data의 양과 인공신경망의 구조에 따라 비용함수에 부분 최솟값(local minimum)들이 생기게 된다. 비용함수의 부분 최솟값들은 학습을 방해하는 요소가 된다. 이러한 방법을 해결하는 여러 방법 중 우리의 제안 방법은 학습률(Step-size)을 변화시키는 방법이다. 학습률을 고정된 상수로 이용하는 기존의 방법과는 다르게 비용함수를 이용한 다변수함수를 이용함으로써 불필요한 기계학습이 이루어지는 것을 방지할 수 있으며, 최솟값으로 가는 최적의 길을 찾을 수 있다. 수치적 실험을 통하여 기존의 방법보다 우리가 제안하는 방법을 이용하여 약 3%(88.8%→91.5%)의 성능이 향상하는 결과를 얻었다.

• 자원ㆍ환경경제연구
• /
• 제15권4호
• /
• pp.713-740
• /
• 2006

본 논문에서는 석회석광산 개발사업의 사례를 통해 자원개발사업의 비용-편익분석시 고려되어야 할 비용 및 편익 항목이 무엇이며 이의 정확한 계산을 위해서는 어떤 점을 주의해야 하는지 등을 살펴보았다. 본 논문의 결과에 따르면 임산불 생산가치, 수원함양가능, 대기정화 가능 등은 산림훼손에 따른 환경비용으로 간주되어야 하나 통상적으로 환경비용무로 간주되어 왔던 산림휴양 기능, 산림정수 기능 등은 개별 단위의 석회석 개발 사업시는 환경비용으로 간주하기는 어려운 것으로 나타났다. 비용-편익분석 결과는 할인율에 따라 달랐으나 모두 순편익이 있는 것으로 나타났다. 순편익의 규모는 6,899 억 원에서 1조 784억 원인 것으로 나타났다.

• Computational Structural Engineering : An International Journal
• /
• 제1권1호
• /
• pp.59-69
• /
• 2001

This study is intended to propose a systematic approach for reliability-based assessment of life cycle cost (LCC) effectiveness and economic efficiency for cost-effective seismic upgrading of existing bridges. The LCC function is expressed as the sum of the upgrading cost and all the discounted life cycle damage costs, which is formulated as a function of the Park-Ang damage index and structural damage probability. The damage costs are expressed in terms of direct damage costs such as repair/replacement costs, human losses and property damage costs, and indirect damage costs such as road user costs and indirect regional economic losses. For dealing with a variety of uncertainties associated with earthquake loads and capacities, a simulation-based reliability approach is used. The SMART-DRAIN-2DX, which is a modified version of the well-known DRAIN-2DX, is extended by incor-porating LCC analysis based on the LCC function developed in the study. Economic efficiencies for optimal seismic upgradings of the continuous PC segmental bridges are assessed using the proposed LCC functions and benefit-cost ratio.

• 대한전기학회논문지
• /
• 제37권5호
• /
• pp.272-281
• /
• 1988

The problem of optimal transmission system planning is to find the most economical locations and time of transmission line construction under the various constraints such as available rights-of-way, finances, the technical characteristics of power system, and the reliability criterion of power supply, and so on. In this paper the constraint of right-of-way is represented as a finite set of available rights-of-way. And the constructed for a unit period. The electrical constraints are represented in terms of line overload and steady state stability margin. And the reliability criterion is dealt with the suppression of failure cost and with single-contingency analysis. In general, the transmission planning problem requires integer solutions and its objective function is nonlinear. In this paper the objective function is defined as a sum of the present values of construction cost and the minimum operating cost of power system. The latter is represented as a sum of generation cost and failure cost considering the change of yearly load, economic dispatch, and the line contingency. For the calculation of operating cost linear programming is adopted on the base of DC load flow calculation, and for the optimization of main objective function nonlinear Branch-and-Bound algorithm is used. Finally, for improving the efficiency of B & B algorithm a new sensitivity analysis algorithm is proposed.

• 경영과학
• /
• 제20권2호
• /
• pp.179-196
• /
• 2003

Function point model is the international standard method to measure the software size which is one of the most important factors to determine the software development cost. Function point model can successfully be applied only when the detailed specification of users' requirements is available. In the domestic public sector, however, the budgeting for software projects is carried out before the requirements of softwares ere specified in detail. Therefore, an efficient function point estimation method is required to apply function point model at the early stage of software development projects. The purpose of this paper is to compare various function point estimation methods and analyse their accuracies in domestic software projects. We consider four methods : NESMA model, ISBSG model, the simplified function point model and the backfiring method. The methods are applied to about one hundred of domestic projects, and their estimation errors are compared. The results can used as a criterion to select an adequate estimation model for function point counts.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1997년도 하계학술대회 논문집 B
• /
• pp.679-681
• /
• 1997

A $H_{\infty}$ adaptive controller is designed by using polynomial approach. The $H_{\infty}$ robust controllers for adaptive system were designed by Grimble. But they did not minimize the mixed sensitivity ra cost function which is the re sum of weighted sensitivity and complementary sensitivity terms Moreover pole placement is dependent of cost function. In this paper, the mixed sensitivity re cost function is minimized by employing the Youla parameterization and polynomial approach at the same time. And pole plaement is independent of weighting function.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2003년도 춘계학술대회논문집
• /
• pp.652-657
• /
• 2003

This paper proposes a method to generate virtual intensity field in space. The sound field of a zone enclosing the listener position is controlled to have maximum acoustic intensity to the desired direction. In order to control acoustic intensity of a zone, space-averaged active intensity is introduced. The ratio of space-averaged active intensity and control effort is defined as a cost function and expressed as a function of source control signals. It is shown that the cost function represents radiation efficiency of multiple sources. The control signals maximizing the cost function is found through eigenvalue analysis. The proposed method is verified by numerical simulations performed in free field condition, and the results provide a relation between wavelength and the size of controllable intensity field.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1996년도 춘계공동학술대회논문집; 공군사관학교, 청주; 26-27 Apr. 1996
• /
• pp.147-150
• /
• 1996

Human posture prediction and motion simulation methods try to solve inverse kinematic problems using the optimization technique based on the concept of minimum principle. It is very important to select a cost function which relfects the human posture acurately. In this study, lifting postures were predicted using the five biomechanical cost functions and compared with real human postures in order to evaluate the predictivities of the cost functions. The result showed that all the biomechanical cost functions used in this study could not predict lifting postures accurately. The cost function which minimizes the sum of joint moments showed the smallest mean prediction error, while the one which minimizes the MUR showed statistically better performance.

• 한국국방경영분석학회지
• /
• 제6권1호
• /
• pp.79-92
• /
• 1980

In cost-only incentive contract the emphasis is the attainment of efficient and effective cost control. In contrast multiple incentives contract correlates contractor's profit motives with the generalized government objective function, the decision variables of which are performance or quality (technical progress), time or schedule (timely development and delivery) and the cost (efficient and effective cost control) Under multiple incentive structure, it is essential to formulate the trade-off curves between cost and performance, which are called iso-fee curves. Trade-off curves depict the combination of cost and performance achievement for which the contractor will be rewarded with the same fee. The basic function of trade-off curves is to show he the contractor will be motivated by incentive arragement to trade off or sacrifice the achievement in one incentive element for the acnievment in another.

• 대한가정학회지
• /
• 제41권7호
• /
• pp.157-168
• /
• 2003

The purposes of this study were to explore the rural women's labor by measuring labor value economically. To achieve the purpose, this study applied four methods : replacement cost method individual function, replacement cost method generalist, opportunity cost method and shadow wage method. The results of this study were as follows: 1) Time used for agricultural work and household work by with women were 5.3 hours and 3.8 hours each. 2) According to the methods used, the amounts of valuing rural women's labor were varied and ranged from 23,000 won to 43,000 won per day. This study might help recognize the degree of rural women's labor contribution to the household income of farm households and improve the socio-economic status of rural women through showing the productivity of the rural women's labor.