• Journal of the Korean Operations Research and Management Science Society
• /
• v.22 no.3
• /
• pp.163-173
• /
• 1997

Although numerous studies demonstrate that one technique outperforms the others for a given data set, it is hard to tell a priori which of these techniques will be the most effective to solve a specific problem. It has been suggested that the better approach to classification problem might be to integrate several different forecasting techniques by combining their results. The issues of interest are how to integrate different modeling techniques to increase the predictive performance. This paper proposes the post-model integration method, which tries to find the best combination of the results provided by individual techniques. To get the optimal or near optimal combination of different prediction techniques, Genetic Algorithms (GAs) are applied, which are particularly suitable for multi-parameter optimization problems with an object function subject to numerous hard and soft constraints. This study applies three individual classification techniques (Discriminant analysis, Logit model and Neural Networks) as base models for the corporate failure prediction. The results of composite predictions are compared with the individual models. Preliminary results suggests that the use of integrated methods improve the performance of business classification.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1996.10a
• /
• pp.31-34
• /
• 1996

The extensive researches have compared the performance of neural networks(NN) with those of various statistical techniques for the classification problem. The empirical results of these comparative studies have indicated that the neural networks often outperform the traditional statistical techniques. Moreover, there are some efforts that try to combine various classification methods, especially multivariate discriminant analysis with neural networks. While these efforts improve the performance, there exists a problem violating robust assumptions of multivariate discriminant analysis that are multivariate normality of the independent variables and equality of variance-covariance matrices in each of the groups. On the contrary, cluster analysis alleviates this assumption like neural networks. We propose a new approach to classification problems by combining the cluster analysis with neural networks. The resulting predictions of the composite model are more accurate than each individual technique.

• Wind and Structures
• /
• v.4 no.1
• /
• pp.19-30
• /
• 2001

This paper concerns computational response predictions when a tuned mass damper is intended to be used for the suppression of vortex shedding induced vibrations of e.g., a bridge deck. A general frequency domain theory is presented and its application is exemplified on a suspension bridge (where vortex shedding vibrations have been observed and where such an installation is a possible solution). Relevant load data are taken from previous wind tunnel tests. In particular, the displacement response statistics of the tuned mass damper as well as the bridge deck are obtained from time domain simulations, showing that after the installation of a TMD peak factors between three and four should be expected.

• Proceedings of the IEEK Conference
• /
• 1999.11a
• /
• pp.732-735
• /
• 1999

A GA-based optimization technique is adopted in the paper to obtain optimal future control inputs for predictive control systems. For reliable future predictions of a process, we identify the underlying process with an NNARX model structure and investigate to reduce the volume of neural network based on the Lipschitz index and a criterion. Since most industrial processes are subject to their constraints, we deal with the input-output constraints by modifying some genetic operators and/or using a penalty strategy in the GAPC. Some computer simulations are given to show the effectiveness of the GAPC method compared with the adaptive GPC algorithm.

• International Journal of Precision Engineering and Manufacturing
• /
• v.1 no.2
• /
• pp.67-75
• /
• 2000

virtual computer numerical control(VCNC) arises from the concept that one can experience pseudo-real machining with a computer-numerically-controlled(CNC) machine before actually cutting an object. To achieve accurate VCNC, it is important to determine abnormal behavior, such as chatter, before cutting. Detecting chatter requires an understanding of the dynamic cutting force model. In general, the cutting process is a closed loop system the consists of structural and cutting dynamic. Machining instability, namely chatter, results from the interaction between these two dynamics. Several previous reports have predicted stability for a single path, using a simple cutting force model without run out and penetration effects. This study considers both tool run out and penetration effects, using experimental modal analysis, to obtain predictions that are more accurate. The machining stability during a corner cut, which is a typical transient cut, was assessed from an evaluation of the cutting configurations at the corner.

• Journal of the Korean Wood Science and Technology
• /
• v.17 no.4
• /
• pp.35-43
• /
• 1989

Nailed joints, which are commonly used in Wooden structures, transmit loads from one member to another and induce partial composite actions between members. Long-term loads induce creep slip in nailed joints and affect load sharing and partial composite action, which may reduce joint stiffness. Two theoretical viscous-viscoelastic models were developed for nailed joints to predict creep behavior under long-term variable loads. Those models were also used to predict stiffness changes under long-term variable loads. The stiffness of nailed joint is defined as a Secant modulus which is called the joint modulus or slip modulus. Input data for the models are the results of constant load tests under three different load levels. To verify the models, nailed joints were also tested under two long-term variable load functions. The predictions of the models were very close to the experimental data. Therefore, the theoretical viscous-viscoelastic models and procedures developed in this study can be applied to predict creep slip and the changes in joint moduli of nailed joints under long-term variable loads.

• Bulletin of the Korean Chemical Society
• /
• v.3 no.1
• /
• pp.30-33
• /
• 1982

Photochemical and thermal solvolysis of 2,3,4-picolyl chlorides (2,3,4-PC) were studied in amine solvents and the results were correlated with the electronic structures calculated by PPP-SCF-MO CI method. Activation parameters show that the thermal solvolysis of PC is $S_N2$ type rcaction. The rates of thermal reaction in pyridine or t-butylamine solvent decrease in the order of 2-PC > 3-PC > 4-PC. These results are consistent with the predictions based on the electron densities of picolyl chlorides. In photosolvolysis, the same products as those of thermal reactions were obtained. The results indicate that photochemical solvolysis undergoes through heterolytic cleavage. Relative quantum yields of photosolvolysis of 2,3,4-picolyl chlorides in t-butylamine solvent were determined to be 0.73, 1, and 0.50 respectively. These results are in good agreement with the electron densities of the excited triplet state of picolyl chlorides.

• International conference on construction engineering and project management
• /
• 2013.01a
• /
• pp.122-128
• /
• 2013

Computer simulations designed to predict technical and financial returns of wind turbine installations are used to make informed investment decisions. These simulations used fixed values to represent real-world variables, while the actual projects can be highly uncertain, resulting in predictions that are less accurate and less useful. In this article, by modifying a popular wind power simulation sourced from the American Wind Energy Association to use Monte Carlo techniques in its calculations, the authors have proposed a way to improve simulation usability by producing probability distributions of likely outcomes, which can be used to draw broader, more useful conclusions about the simulated project.

• Korea-Australia Rheology Journal
• /
• v.12 no.2
• /
• pp.125-133
• /
• 2000

The present work is aimed at performing injection molding filling simulation of fiber suspension in polymer based matrix. The numerical simulation incorporates the coupling effect between the flow field and the fiber orientation state together with in-plane velocity gradient effect with recently proposed closure approximations. Predicted orientation components are compared with available experimental data of a film-gated strip and a center-gated disk. Predictions with IBOF closure approximation show excellent behaviors with regard to accuracy and numerical efficiency. However, predicted results seem to have consistent errors in comparison with experimental data. Diffusivity term which accounts for fiber-fiber interaction might have to be modified.

• 제어로봇시스템학회:학술대회논문집
• /
• 1991.10a
• /
• pp.99-104
• /
• 1991

An adaptive predictive control method for SISO and MIMO plants is proposed. In this method, future predictions of process output based on a bilinear CARIMA model are used to calculate the control input. Also, a classical recursive adaptation algorithm, equation error method, is used to decrease the uncertainty of the process model. As a result of the application on distillation process, the ability of the set-point tracking and the disturbance rejection is acceptable to apply to the industrial distillation processes.