• The Transactions of the Korean Institute of Electrical Engineers P
• /
• v.58 no.2
• /
• pp.164-171
• /
• 2009

This paper presents new learning algorithm of dynamic Bayesian networks (DBN) by means of constrained least square (LS) estimation algorithm and gradient descent method. First, we propose constrained LS based parameter estimation for a Markov chain (MC) model given observation data sets. Next, a gradient descent optimization is utilized for online estimation of a hidden Markov model (HMM), which is bi-linearly constructed by adding an observation variable to a MC model. We achieve numerical simulations to prove its reliability and superiority in which a series of non stationary random signal is applied for the DBN models respectively.

• Journal of Korean Society of Transportation
• /
• v.27 no.6
• /
• pp.147-156
• /
• 2009

We present a methodology for modeling and solving the transit frequency design problem with variable demand. The problem is described as a bi-level model based on a non-cooperative Stackelberg game. The upper-level operator problem is formulated as a non-linear optimization model to minimize net cost, which includes operating cost, travel cost and revenue, with fleet size and frequency constraints. The lower-level user problem is formulated as a capacity-constrained stochastic user equilibrium assignment model with variable demand, considering transfer delay between transit lines. An efficient algorithm is also presented for solving the proposed model. The upper-level model is solved by a gradient projection method, and the lower-level model is solved by an existing iterative balancing method. An application of the proposed model and algorithm is presented using a small test network. The results of this application show that the proposed algorithm converges well to an optimal point. The methodology of this study is expected to contribute to form a theoretical basis for diagnosing the problems of current transit systems and for improving its operational efficiency to increase the demand as well as the level of service.

• The KSFM Journal of Fluid Machinery
• /
• v.16 no.1
• /
• pp.11-16
• /
• 2013

A numerical study was conducted to improve the performance of an impeller of centrifugal compressor. Nine design variables were chosen with constraints. Only meridional contours and blade profile were adjusted. ANN (Artificial Neural Net) was adopted as a main optimization algorithm with PSO (Particle Swarm Optimization) in order to reduce the optimization time. At first, ANN was learned and trained with the design variable sets which were obtained using DOE (Design of Experiment). This ANN was continuously improved its accuracy for each generation of which population was one hundred. New design variable set in each generation was selected using a non-gradient based method of PSO in order to obtain the global optimized result. After $7^{th}$ generation, the prediction difference of efficiency and pressure ratio between ANN and CFD was less than 0.6%. From more than 1,200 design variable sets, a pareto of efficiency versus pressure ratio was obtained and an optimized result was selected based on the multi-objective function. On this optimized impeller, the efficiency and pressure ratio were improved by 1% and 9.3%, respectively.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.23 no.11
• /
• pp.973-981
• /
• 2013

The main role of an acoustic diffuser is to diffuse reflected sound field spatially. Since the pioneering work of Schroeder, there have been investigations to improve its performance by using shape/sizing optimization methods. In this paper, a gradient-based topology optimization algorithm is newly presented to find the optimal distribution of reflecting materials for maximizing diffuser performance. Time-harmonic acoustic analysis in a two-dimensional acoustic domain is carried out where the domain is discretized by finite elements. Perfectly matched layers are placed to surround the domain to simulate non-reflecting boundary conditions. Design variables are assigned to each element of which material properties are interpolated between those of air and those of a rigid body. An approach to extract the reflected field from the total acoustic field is employed. To validate the effectiveness of the proposed method, design problems are solved at different frequencies. The performance of the optimized diffusers obtained by the proposed method is compared against that of the conventional Schroeder diffusers.

• 제어로봇시스템학회:학술대회논문집
• /
• 1995.10a
• /
• pp.315-318
• /
• 1995

The optimum design problem of a coat-hanger die is solved by the inverse formulation. The flow in the die is analyzed using three-dimensional model. The new model for the manifold geometry is developed for the inverse formulation. The inverse problem for the optimum die geometry is formed as the optimization problem whose objective function is the linear combination of the square sum of pressure gradient deviation at die exit and the penalty function relating to the measure of non-smoothness of solution. From the several iterative solutions of the optimization problem, the optimum solution can be obtained automatically while producing the uniform flow rate distribution at die exit.

• Proceedings of the KSME Conference
• /
• 2000.04b
• /
• pp.144-147
• /
• 2000

A numerical method for the solution of one-dimensional inverse heat conduction problem is established and its performance is demonstrated with computational results. The present work introduces the maximum entropy method in order to build a robust formulation of the inverse problem. The maximum entropy method finds the solution that maximizes the entropy functional under given temperature measurement. The philosophy of the method is to seek the most likely inverse solution. The maximum entropy method converts the inverse problem to a non-linear constrained optimization problem of which constraint is the statistical consistency between the measured temperature and the estimated temperature. The successive quadratic programming facilitates the maximum entropy estimation. The gradient required fur the optimization procedure is provided by solving the adjoint problem. The characteristic feature of the maximum entropy method is discussed with the illustrated results. The presented results show considerable resolution enhancement and bias reduction in comparison with the conventional methods.

• Proceedings of the KSME Conference
• /
• 2003.11a
• /
• pp.1430-1435
• /
• 2003

Satellite upper platform is optimized by response surface method which has non-gradient, semi-glogal, discrete and fast convergency characteristics. Sampling points are extracted by design of experiments using Central Composite Method and Factorial Design. Also response surface is generated by the various regression functions. Structure analysis is execuated with regard for static and dynamic environment in launching stage. As a result response surface method is superior to other optimization method with respect to optimum value and cost of computation time. Also a confidence is varified in the various regression models.

• Journal of the Korean Society for Precision Engineering
• /
• v.23 no.11 s.188
• /
• pp.78-84
• /
• 2006

The objective of this study is to design the optimal die profile that can prevent material fracture in the tube drawing process for automobile steering input shaft. First, the CDV(Critical Damage Value) of material is obtained by the compression test and FE-analysis. The occurrence of fracture is estimated by the FE-analysis considering the CDV. In order to achieve the objective of this study, optimization technique and FE-analysis are applied. FPS(Flexible Polyhedron Search) method, which is one of the non-gradient optimization techniques often used in engineering, is used to search optimal die profile. The drawing die profile is represented by Bezier-curve to generate all the possible die profile. Using FPS method and FE-analysis the optimal drawing die profile is determined. To verify tile effectiveness of the redesigned optimal die, the tube drawing experiment is performed. In the experimental result, it is possible to produce sound product without material fracture using the redesigned optimal die.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.324-329
• /
• 2005

Dynamic Encoding Algorithm for Searches (DEAS) which is known as a fast and reliable non-gradient optimization method, was proposed [1]. DEAS reaches local or global optimum with binary strings (or binary matrices for multi-dimensional problem) by iterating the two operations; bisectional search (BSS) and unidirectional search (UDS). BSS increases binary strings by one digit (i.e., 0 or 1), while UDS performs increment or decrement of binary strings in the BSS' result direction with no change of string length. Because the interval of UDS exponentially decreases with increment of bit string length (BSL), DEAS is difficult to escape from local optimum when DEAS falls into local optimum. Therefore, this paper proposes hopping UDS (HUDS) which performs UDS by hopping as many as BSL in the final point of UDS process. HUDS helps to escape from local optimum and enhances a probability searching global optimization. The excellent performance of HUDS will be validated through the well-known benchmark functions.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.57 no.6
• /
• pp.915-923
• /
• 2008

This paper focuses on optimal tuning of nonlinear parameters of a dual-input power system stabilizer(dual-input PSS), which can improve the system damping performance immediately following a large disturbance. Until recently, various PSS models have developed to bring stability and reliability to power systems, and some of these models are used in industry applications. However, due to non-smooth nonlinearities from the interaction between linear parameters(gains and time constants of linear controllers) and nonlinear parameters(saturation output limits), the output limit parameters cannot be determined by the conventional tuning methods based on linear analysis. Only ad hoc tuning procedures('trial and error' approach) have been used. Therefore, the steepest descent method is applied to implement the optimal tuning of the nonlinear parameters of the dual-input PSS. The gradient required in this optimization technique can be computed from trajectory sensitivities in hybrid system modeling with the differential-algebraic-impulsive-switched(DAIS) structure. The optimal output limits of the dual-input PSS are evaluated by time-domain simulation in both a single machine infinite bus(SMIB) system and a multi-machine power system in comparison with those of a single-input PSS.