• 전자공학회논문지A
• /
• 제33A권4호
• /
• pp.27-36
• /
• 1996

It is well known that the viterbi algorithm proposed as a mthod of decoding convolutional codes is in fact maximum likelihood (ML) and therefore optimal. But, because hardware complexity grows exponentially with the constraint length, there will be severe constraints on the implementation of the viterbi decoders. In this paper, the three-layered backpropagation neural networks are proposed as an alternative in order to get sufficiently useful performance and deal successfully with the problems of the viterbi decoder. This paper shows that the neural convolutional decoder (NCD) can make a decision in the point of ML in decoding and describes simulation results. The cause of the difference between stochastic results and simulation results is discussed, and then thefuture prospect of the NCD is described on the basis of the characteristic of the transfer function.

• 디지털산업정보학회논문지
• /
• 제10권4호
• /
• pp.133-142
• /
• 2014

The least mean square(LMS) algorithm has been popular owing to its simplicity, stability, and availability to implement. But it inherently has a problem of slow convergence speed, and the presence of a transfer function in the secondary path following the adaptive controller and the error path has been shown to generally degrade the stability and the performance of the LMS algorithm in applications of acoustical noise control. In general, in order to solve these problems, the filtered-x LMS (FX-LMS) type algorithms can be used and the bi-directional Filtered-x LMS(BFXLMS) algorithm is very attractive among them, which increase the convergence speed and the performance of the controller with nearly equivalent computation complexity. In this paper, a mathematical analysis for the BFXLMS algorithm is presented. In terms of view points of time domain, frequency domain, and stochastic domain, the characteristics and stabilities of algorithm is accurately analyzed.

• 산업경영시스템학회지
• /
• 제42권4호
• /
• pp.183-193
• /
• 2019

Process mining is an analytical technique aimed at obtaining useful information about a process by extracting a process model from events log. However, most existing process models are deterministic because they do not include stochastic elements such as the occurrence probabilities or execution times of activities. Therefore, available information is limited, resulting in the limitations on analyzing and understanding the process. Furthermore, it is also important to develop an efficient methodology to discover the process model. Although genetic process mining algorithm is one of the methods that can handle data with noises, it has a limitation of large computation time when it is applied to data with large capacity. To resolve these issues, in this paper, we define a stochastic process tree and propose a tabu search-genetic process mining (TS-GPM) algorithm for a stochastic process tree. Specifically, we define a two-dimensional array as a chromosome to represent a stochastic process tree, fitness function, a procedure for generating stochastic process tree and a model trace as a string of activities generated from the process tree. Furthermore, by storing and comparing model traces with low fitness values in the tabu list, we can prevent duplicated searches for process trees with low fitness value being performed. In order to verify the performance of the proposed algorithm, we performed a numerical experiment by using two kinds of event log data used in the previous research. The results showed that the suggested TS-GPM algorithm outperformed the GPM algorithm in terms of fitness and computation time.

• 한국환경과학회지
• /
• 제8권3호
• /
• pp.349-354
• /
• 1999

The transfer function was introduced to establish the prediction method for the DO concentration at the intaking point of Kongju Water Works System. In the mose cases we analyze a single time series without explicitly using information contained in the related time series. In many forecasting situations, other events will systematically influence the series to be forecasted(the dependent variables), and therefore, there is need to go beyond a univariate forecasting model. Thus, we must bulid a forecasting model that incorporates more than one time series and introduces explicitly the dynamic characteristics of the system. Such a model is called a multiple time series model or transfer function model. The purpose of this study is to develop the stochastic stream water quality model for the intaking station of Kongju city waterworks in Keum river system. The performance of the multiplicative ARIMA model and the transfer function noise model were examined through comparisons between the historical and generated monthly dissolved oxygen series. The result reveal that the transfer function noise model lead to the improved accuracy.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1995년도 추계학술대회 논문집
• /
• pp.1186-1190
• /
• 1995

This paper is focused on the formulation of explicit closed-form functions describing the performance measures of the general flexible manufacturing system (FMS)according to the strategy of material handling system(MHS). the performance measures such as the production rate, the production lead-time and the utilization rate of the general FMS are expressed, respectively, as the explicit closed-form functions of the part processing time, the service rate of the material handling system (MHS) and the number of machine tools in the FMS. For this, the gensral FMS is presented as a generalized stochastic Petri net model, then, the moment generating function (MGF) based approach is applied to obtain the steady-state probabity formulation. Based on the steady-state formulation, the explicit closed-form functions for performance measures of the probability FMS can be obtained. Finally, the analytical results are compared with the Petri net simulation results to verify the validity of the suggested method. The paper is of significance in the sense that it provides a comprehensive formula for performance measures of the FMS even to the industry engineers and academic reademic resuarchers who have no background on Markov chain analysis method or Petrinet modeling

• International Journal of Control, Automation, and Systems
• /
• 제1권1호
• /
• pp.1-10
• /
• 2003

This paper presents a high-precision magnetically levitated (maglev) stage to meet demanding motion specifications in the next-generation precision manufacturing and nanotechnology. Characterization of dynamic behaviors of such a motion stage is a crucial task. In this paper, we address the issues related to the stochastic modeling of the stage including transfer function identification, and noise/disturbance analysis and prediction. Provided are test results on precision dynamics, such as fine settling, effect of optical table oscillation, and position ripple. To deal with the dynamic coupling in the platen, we designed and implemented a multivariable linear quadratic regulator, and performed time-optimal control. We demonstrated how the performance of the current maglev stage can be improved with these analyses and experimental results. The maglev stage operates with positioning noise of 5 nm rms in $\chi$ and y, acceleration capabilities in excess of 2g(20 $m/s^2$), and closed-loop crossover frequency of 100 Hz.

• Computers and Concrete
• /
• 제17권1호
• /
• pp.87-106
• /
• 2016

This paper deals with the reliability analysis of design formulations derived for predicting the punching shear capacity of FRP-reinforced two-way slabs. Firstly, a new design code formulation was derived by means of gene expression programming. This formulation differs from the existing ones as the slab length (L) was introduced in the equation. Next, the proposed formulation was tested for its generalization capability by a parametric study. Then, the stochastic analyses of derived and existing formulations were performed by Monte Carlo simulation. Finally, the reliability analyses of these equations were carried out based on the results of stochastic analysis and the ultimate state function of ASCE-7 and ACI-318 (2011). The results indicate that the prediction performance of new formulation is significantly higher as compared to available design equations and its reliability index is within acceptable limits.

• 전자공학회논문지C
• /
• 제34C권7호
• /
• pp.82-91
• /
• 1997

In this paper, we analyze the dynamics of langevine competitive learning neural network based on its fokker-plank equation. From the viewpont of the stochastic differential equation (SDE), langevine competitive learning equation is one of langevine stochastic differential equation and has the diffusin equation on the topological space (.ohm., F, P) with probability measure. We derive the fokker-plank equation from the proposed algorithm and prove by introducing a infinitestimal operator for markov semigroups, that the weight vector in the particular simplex can converge to the globally optimal point under the condition of some convex or pseudo-convex performance measure function. Experimental resutls for pattern recognition of the remote sensing data indicate the superiority of langevine competitive learning neural network in comparison to the conventional competitive learning neural network.

• 한국기술혁신학회:학술대회논문집
• /
• 한국기술혁신학회 2005년도 추계학술대회
• /
• pp.521-532
• /
• 2005

IT 기업의 기술적 효율성의 정도를 추정하고, 그 결정요인을 알아보기 위해 Coelli(1995)의 확률적프론티어(Stochastic Frontier model) 중 초월대수 변경생산함수(translog stochastic frontier production function)를 설정한다. 분석견과 연구개발투자, 기업의 재무구조, 설비투자효율, 노동소득분배율 등은 기업의 비효율을 감소시키는 역할을 하는 반면, 기업규모, 재고자산증가율, 자기자본증가율 등은 기업의 비효율성을 더 높여주는 것으로 나타났다. 1990년부터 2004년까지 국내 제조업 전체의 생산의 기술적 효율성은 평균 0.5311로 이는 생산효율성이 $53.11\%$임을 의미하고, 비효율성은 $46.89\%$에 달한다고 볼 수 있다. IT 기업의 기술적 효율성은 0.5337회 제조업과 비슷하지만, IT 대기업은 0.61, IT 중소기업은 0.511로 대기업과 중소기업의 격차가 크게 발생하고 있다.

• 한국항해항만학회:학술대회논문집
• /
• 한국항해항만학회 2023년도 춘계학술대회
• /
• pp.254-255
• /
• 2023

This paper discovers a robust managerial strategy for a stochastic inventory of perishable products, where the model experiences changing factors including inner parameters and an external disturbance with unknown form. An analytical solution for the optimization problem can be obtained by applying the Hamilton-Bellman-Jacobi equation, however the policy result cannot completely suppress the oscillation from the external disturbance. Therefore, an intelligent approach named Radial Basis Function Neural Networks is applied to estimate the unknown disturbance and provide a robust controller to manipulate the inventory level more effective. The final results show the outstanding performance of RBFNN controller, where both the estimation error and control error are guaranteed in the predefined limit.