• 대한산업공학회지
• /
• 제23권1호
• /
• pp.197-213
• /
• 1997

It is not easy to find a good production schedule which can be used in practice. Therefore, production scheduling simulation with a simple dispatching rule or a set of dispatching rules is used. However, a simple dispatching rule may not create a robust schedule, for the same rule is blindly applied to all internal production processes. The presumption is that there might be a specific combination of appropriate rules that can improve the efficiency of a total production system for a certain type of orders. In order to acquire a better set of dispatching rules, simulation is used to examine the performance of various combinations of dispatching rule sets. There are innumerable combination of rule sets. Hence it takes too much computer simulation time to find a robust set of dispatching rule for a specific production system. Therefore, we propose a concept of the knowledge based simulation to circumvent the problem. The knowledge based simulation consists of knowledge bases, an inference engine and a simulator. The knowledge base is made of rule sets that is extracted from both simulation and human intuition obtained by the simulation studies. For a certain type of orders, the proposed system provides several sets of dispatching rules that are expected to generate better results. Then the scheduler tries to find the best by simulating all proposed set of rules with the simulator. The knowledge-based simulator armed with the acquired knowledge has produced improved solutions in terms of time and scheduling performance.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2009년도 춘계학술대회 논문집
• /
• pp.1057-1058
• /
• 2009

• 한국국방경영분석학회지
• /
• 제26권2호
• /
• pp.55-74
• /
• 2000

A knowledge-based with production rules is a representation of static knowledge of an expert. On the other hand, a real system such as the stock market is dynamic in nature. Therefore we need a method to reflect the dynamic nature of a system when we make inferences with a knowledge-based system. This paper suggests a strategy of dynamic inference that can be used to take into account the dynamic behavior of decision-making with the knowledge-based system consisted of fuzzy production rules. A degree of match(DM) between actual input information and a condition of a rule is represented by a value [0,1]. Weights of relative importance of attributes in a rule are obtained by the AHP(Analytic Hierarchy Process) method. Then these weights are applied as exponents for the DM, and the DMs in a rule are combined, with the Min operator, into a single DM for the rule. In this way, the importance of attributes of a rule, which can be changed from time to time, can be reflected in an inference with fuzzy production systems.

• 한국산업정보학회논문지
• /
• 제15권1호
• /
• pp.47-57
• /
• 2010

다품종 소량생산 공정에서는 동일 특성을 가지는 제품의 제작 개수가 절대적으로 적기 때문에 전통적인 공정제어 기법인 통계적 공정관리(Statistical Process Control)를 적용하기에는 어려움이 많이 존재한다. 그러므로 통계적인 접근법과 아울러 다양한 제품 특성을 규정짓기 위한 다양한 조건의 조합으로 이루어지는 SPEC규칙, 그리고 엔지니어의 경험에 기반한 노하우가 응집되어 있는 KNOWHOW규칙을 유연하게 설정하여 공정을 제어할 수 있는 규칙기반 공정관리 기술의 접목이 필요하다. 본 연구는 다품종 소량생산 공정에 적용 가능한 규칙기반 공정관리(Rule-based Process Control) 시스템을 제안하고, 이 시스템을 실제 반도체 생산 공정에 적용하여 그 성과를 검증하였다.

• 한국경영과학회지
• /
• 제25권4호
• /
• pp.81-95
• /
• 2000

A knowledge-based system with fuzzy production rules is a representation of static knowledge of an expert. On the other hand, a real system such as the stock market is dynamic in nature. Therefore we need a strategy to reflect the dynamic nature of real system when we make inferences with a knowledge-based system. This paper proposes a strategy of dynamic inferencing for a knowledge-based system with fuzzy production rules. The strategy suggested in this paper applies weights of attributes of conditions of a rule in the knowledge-base. A degree of match(DM) between actual input information and a condition of a rule is represented by a value [0,1]. Weights of relative importance of attributes in a rule are obtained by AHP(Analytic Hierarcy Process) method. Then these weights are applied as exponents for the DM, and the DMs in a rule are combined, with MIN operator, into a single DM for the rule. In this way, overall DM for a rule changes depending on the importance of attributes of the rule. As a result, the dynamic nature of a real system can be incorporated in an inference with fuzzy production rules.

• 한국통신학회논문지
• /
• 제12권1호
• /
• pp.50-59
• /
• 1987

많은 production rule(혹은 간단히 production)로부터 현 상태에 만족되는 rule을 빨리 찾아내기 위하여 현재까지는 RAM(Ramdom Access Memory)에 탕을 둔 필터(filter)사용등의 여러 방법에 제시되었으나, 본 연구에서는 보다 효율적인 CAM(Content Addressable Memory)의 이용을 제시하고, 이를 위해 CAM의 각 bit에 따라, 용도에 다른 구분 및 데이터 구조를 설계하고, 이를 컴퓨터 시뮬레이션을 통해 기존 RAM을 사용했을 경우와 비교하였다.

• Industrial Engineering and Management Systems
• /
• 제9권3호
• /
• pp.196-203
• /
• 2010

Make-to-forecast production is a way to realize high customization and fast responsiveness. This study firstly investigates the effect of introducing a common component in a make-to-forecast production environment. The common component can eliminate a modification step, which is a major cost component in make-to-forecast production. It is illustrated, however, that introducing a versatile component that merely covers several variants is unattractive, and thus adding values to the common component is inevitable in this environment. Secondly, an order-matching rule under the condition that two partially overlapped delivery lead time intervals exist is proposed. The rule considers the effect of matching orders to units that can cover both intervals. An alternative re-matching rule is also developed and examined. Numerical experiments clarify that the proposed rule generally realizes higher contribution ratio and lower percentages of orphans and rejected orders. The proposed re-matching rule increases the average contribution ratio at the expense of increased orphans and order rejections.

• 한국정보처리학회논문지
• /
• 제5권11호
• /
• pp.2897-2907
• /
• 1998

학습 예(training examples)로 부터 규칙을 생성하는 문제는 큰 탐색 공간상에서 많은 지역최소치를 가지고 있는 최적화 문제로 귀결되므로 복잡하고 어려운 문제로 알려져 있다. 이러한 생성규칙을 만들기 위한 여러 가지 학습방법들이 제안되었으며, 그 중 한가지 학습방법이 유전자알고리즘을 연산모델로 사용하는 것이다. 그러나 전통적인 유전자알고리즘은 전역해 부근에서 수렴속도가 떨어지고, 추출된 규칙의 효율성에 문제가 있다. 본 논문에서는 유전자알고리즘의 학습과정에서 포착되는 염색체의 스키마를 분석하여 탐색공간을 부분해(subsolution)를 구할 수 있는 공간들로 분할함으로써, 보다 일반화된 분류 규칙집합을 찾는 방법을 제안하였다. 또한, 실험을 통하여 기존의 기계학습 방법을 사용한 경우와 효율을 상호 비교하여 제안한 방법을 타당성을 입증하였다.

• 한국산업정보학회논문지
• /
• 제16권2호
• /
• pp.73-83
• /
• 2011

생산 시스템에 있어서 일정계획은 많은 비용이 투자된 생산설비 및 기계를 효율적으로 운영한다는 의미에서 매우 중요하다. 계획과 실행에서 오는 차이를 줄이기 위한 생산 시스템인 제조실행 시스템(Manufacturing Execution System : MES)의 효율성을 높이기 위해서 일정계획은 대상이 되는 작업의 특성에 적합한 일정계획 할당규칙 (Dispatching rule)에 대한 선정방법이 필요하다. 따라서 본 논문에서는 공정계획의 데이터에서 얻어지는 작업특정을 기반으로 시뮬레이션 (Simulation)을 진행하여 일정계획 할당규칙을 선정한다. 시뮬레이션을 통해 얻어진 정보를 지식기반 사례추론 방법론의 하나인 온톨로지(Ontology)로 구성함으로써 작업할당규칙 선정에 의한 적응형 일정계획 플랫폼을 개발한다. 구현하는 플랫폼은 특정 작업장에만 국한되지 않고 다양한 작업장에 적용시킬 수 있는 일정계획 시스템이라는 장점이 있다. 본 시스템 개발을 위해 온톨로지 추론 부분은 RacerPro와 Prot$\acute{e}$g$\acute{e}$를 이용하였으며, 일정계획 시뮬레이션은 JAVA와 FlexChart를 사용하였다.

• 대한산업공학회지
• /
• 제1권2호
• /
• pp.27-38
• /
• 1975

This paper explores a quantitative decision-making system for planning production, inventories and work-force in a multi-item production system. The Multi-item Parametric Decision Rule (MPDR) model, which assumes the existence of two types of linear feed-back rules, one for work-force level and one for production rates, is basically an extension of the existing method of Parametric Production Planning (PPP) proposed by C.H. Jones. The MPDR model, however, explicitly considers the effect of manufacturing progress and other factors such as employee turn-over, difference in work-days between month etc., and it also provides decision rules for production rates of individual items. First, the cost relations of the production system are estimated in terms of mathematical functions, and then decision rules for work-force level and production rates of individual items are establised based upon the estimated objective cost function. Finally, a direct search technique is used to find a set of parameters which minimizes the total cost of the objective function over a specified planning horizon, given estimates of future demands and initial values of inventories and work-force level. As a case problem, a hypothetical decision rule is developed for a particular firm (truck assembly factory).