• 한국방재학회 논문집
• /
• 제2권4호
• /
• pp.123-129
• /
• 2002

본 연구에서는 3층 신경망 모형에 의해 충주호의 유입량을 예측한 결과들을 이용하여 신경망 모형의 저수지 유입량 예측 특성을 분석하였다. 신경망 모형의 적절한 입력층 및 은닉층 뉴런 개수, 학습회수를 제시하였으며, 학습 첨두유량 크기가 예측된 첨두유량보다 작을 경우 예측 값이 과소평가되는 특징을 확인하였다. 또한 뉴런 개수, 학습회수가 과다할 경우 발생 가능한 과적합 현상을 확인하였으며, 정확한 예측을 위해 필요한 최소 학습자료 기간도 제시하였다. 결과적으로 충주호의 경우 $8{\sim}10$개의 뉴런 개수 및 $1500{\sim}3000$회의 학습회수를 이용한 신경망 모형이 적합한 것으로, 학습자료 기간 수는 최소한 600개 이상의 자료를 적용하여야 정확한 예측이 가능한 것으로 결과되었다.

• 제어로봇시스템학회논문지
• /
• 제20권4호
• /
• pp.449-455
• /
• 2014

This paper presents a method for developing a TM (Traversability Map) from a DTM (Digital Terrain Model) collected by remote sensors of autonomous mobile robots. Such a map can be used to plan traversable paths and estimate navigation speed quantitatively in real time for robots capable of performing autonomous tasks over rough terrain environments. The proposed method consists of three parts: a DTM partition module which divides the DTM into equally spaced patches, a terrain information module which extracts the slope and roughness of the partitioned patches using the curve fitting and the fractal-based triangular prism method, and a traversability analysis module which assesses traversability incorporating with extracted terrain information and fuzzy inference to construct a TM. The potential of the proposed method is validated via simulation works over a set of fractal DTMs.

• Smart Structures and Systems
• /
• 제10권4_5호
• /
• pp.459-470
• /
• 2012

This paper summarizes a study for the modal analysis and model updating conducted using the monitoring data obtained from the Canton Tower of 610 m tall, which was established as an international benchmark problem by the Hong Kong Polytechnic University. Modal properties of the tower were successfully identified using frequency domain decomposition and stochastic subspace identification methods. Finite element model updating using the measurement data was further performed to reduce the modal property differences between the measurements and those of the finite element model. Over-fitting during the model updating was avoided by using an optimization scheme with a regularization term.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제9권4호
• /
• pp.1424-1440
• /
• 2015

It is a challenging problem to search the intended images from a large number of candidates. Content based image retrieval (CBIR) is the most promising way to tackle this problem, where the most important topic is to measure the similarity of images so as to cover the variance of shape, color, pose, illumination etc. While previous works made significant progresses, their adaption ability to dataset is not fully explored. In this paper, we propose a similarity learning method on the basis of probabilistic generative model, i.e., probabilistic latent semantic analysis (PLSA). It first derives Fisher kernel, a function over the parameters and variables, based on PLSA. Then, the parameters are determined through simultaneously maximizing the log likelihood function of PLSA and the retrieval performance over the training dataset. The main advantages of this work are twofold: (1) deriving similarity measure based on PLSA which fully exploits the data distribution and Bayes inference; (2) learning model parameters by maximizing the fitting of model to data and the retrieval performance simultaneously. The proposed method (PLSA-FK) is empirically evaluated over three datasets, and the results exhibit promising performance.

• Asian Pacific Journal of Cancer Prevention
• /
• 제15권22호
• /
• pp.9731-9737
• /
• 2014

Large scale secular registry or surveillance systems have been accumulating vast data that allow mathematical modeling of cancer incidence and mortality rates. Most contemporary models in this regard use time series and APC (age-period-cohort) methods and focus primarily on predicting or analyzing cancer epidemiology with little attention being paid to implications for designing cancer registry, surveillance or evaluation initiatives. This research models age-specific cancer incidence rates using logistic growth equations and explores their performance under different scenarios of data completeness in the hope of deriving clues for reshaping relevant data collection. The study used China Cancer Registry Report 2012 as the data source. It employed 3-parameter logistic growth equations and modeled the age-specific incidence rates of all and the top 10 cancers presented in the registry report. The study performed 3 types of modeling, namely full age-span by fitting, multiple 5-year-segment fitting and single-segment fitting. Measurement of model performance adopted adjusted goodness of fit that combines sum of squred residuals and relative errors. Both model simulation and performance evalation utilized self-developed algorithms programed using C# languade and MS Visual Studio 2008. For models built upon full age-span data, predicted age-specific cancer incidence rates fitted very well with observed values for most (except cervical and breast) cancers with estimated goodness of fit (Rs) being over 0.96. When a given cancer is concerned, the R valuae of the logistic growth model derived using observed data from urban residents was greater than or at least equal to that of the same model built on data from rural people. For models based on multiple-5-year-segment data, the Rs remained fairly high (over 0.89) until 3-fourths of the data segments were excluded. For models using a fixed length single-segment of observed data, the older the age covered by the corresponding data segment, the higher the resulting Rs. Logistic growth models describe age-specific incidence rates perfectly for most cancers and may be used to inform data collection for purposes of monitoring and analyzing cancer epidemic. Helped by appropriate logistic growth equations, the work vomume of contemporary data collection, e.g., cancer registry and surveilance systems, may be reduced substantially.

• East Asian mathematical journal
• /
• 제36권1호
• /
• pp.81-90
• /
• 2020

Many compartment models assume a kinetically homogeneous amount of materials that have well-stirred compartments. However, based on observations from such processes, they have been heuristically fitted by exponential or gamma distributions even though biological media are inhomogeneous in real environments. Fractional differential equations using a specific kernel in Pharmacokinetic/Pharmacodynamic (PKPD) model are recently introduced to account for abnormal drug disposition. We discuss a tumor growth inhibition (TGI) model using fractional-order derivative from it. This represents a tumor growth delay by cytotoxic agents and additionally show variations in the equilibrium points by the change of fractional order. The result indicates that the equilibrium depends on the tumor size as well as a change of the fractional order. We find that the smaller the fractional order, the smaller the equilibrium value. However, a difference of them is the number of concavities and this indicates that TGI over time profile for fitting or prediction should be determined properly either fractional order or tumor sizes according to the number of concavities shown in experimental data.

• Journal of Forest and Environmental Science
• /
• 제34권4호
• /
• pp.313-320
• /
• 2018

Tectona grandis (teak) is one of the most important timber species worldwide and India is one of the major teak growing countries. Though some volume equations were developed for teak in India but the models developed were neither evaluated using robust statistical criteria nor validated. Hence, the objective of this study was to develop statistically tested appropriate volume equation to predict total wood volume (over- and under-bark) for teak trees in Gujarat. A total of 41 trees with age varying from 15 to 33 years and diameter at breast height (dbh) from 7.3 to 30.8 cm were felled for the purpose. Linear and non-linear equations were used to model the relationship of the total wood volume with respect to dbh and total height. The equations tested mostly fitted well to the data. Model evaluation and validation indicated that models should be calibrated with local data for greater accuracy in the prediction.

• ETRI Journal
• /
• 제11권2호
• /
• pp.109-119
• /
• 1989

Based on the 2-dimensional charge-control simulation[4], a purely analytical model for MODFET's is proposed. In this model, proper treatment of the diffusion effect in the 2-DEG transport due to the gradual channel opening along the 2-DEG channel was made to explain the enhanced mobility and increased thershold voltage. The channel thickness and gate capacitance are experssed as functions of gate vlotage including subthreshold characteristics of the MODFET's analytically. By introducing the finite channel opening and an effective channel-length modulation, the slope of the saturation region of the I-V curves was modeled. The smooth transition of the I-V curves from linear-to-saturation region of the I-V curves was possible using the continuous Troffimenkoff-type of field-dependent mobility. Furthermore, a correction factor f was introduced to account for the finite transtition section forming between the GCA and the saturated section. This factor removes the large discrepanicies in the saturation region fo the I-V curves presicted by existing 1-dimensional models. The fitting parameters chosen in our model were found to be predictable and vary over relatively small range of values.

• 한국생산제조학회지
• /
• 제21권1호
• /
• pp.76-84
• /
• 2012

This paper presents the measurement system of arrow's point of impact using laser scan camera and describes the image calibration method. The calibration process of distorted image is primarily divided into explicit and implicit method. Explicit method focuses on direct optical property using physical camera and its parameter adjustment functionality, while implicit method relies on a calibration plate which assumed relations between image pixels and target positions. To find the relations of image and target position in implicit method, we proposed the performance criteria based polynomial theorem model that overcome some limitations of conventional image calibration model such as over-fitting problem. The proposed method can be verified with 2D position of arrow that were taken by SICK Ranger-D50 laser scan camera.

• Computers and Concrete
• /
• 제31권4호
• /
• pp.349-358
• /
• 2023

This paper describes an in-depth analysis on flexural strength of a girder-deck system experiencing a strand debonding damage with various strengthening systems, based on finite element software ABAQUS. A detailed finite element analysis (FEA) model was developed and verified against the relevant experimental data performed by other researchers. The proposed analytical model showed a good agreement with experimental data. Based on the verified FE model, over a hundred girder-deck systems were investigated with the consideration of following variables: 1) debonding level, 2) span-to-depth ratio (L/d), 3) strengthening type, 4) strengthening material thickness. Based on the data above, a new detailed analytical model was developed and proposed for estimating residual flexural strength of the strand-debonding damaged girder-deck system with strengthening systems. It was demonstrated that both finite element model and analysis model could be used to predict flexural behaviors for debonding damaged prestressed girder-deck systems. Since the strands are debonding from surrounding concrete over a certain zone over the length of the beam, the increase of strain in strands can be linked with a ratio ψ, which is L_p/c. The analytical model was proposed and developed regarding the ratio ψ. By conducting procedure of calculating ψ, the ψ value varies from 9.3 to 70.1. Multiple nonlinear regression analysis was performed in Software IBM SPSS Statistics 27.0.1 to derive equation of ψ. ψ equation was curved to be an exponential function, and the independent variable (X) is a linear function in terms of three variables of debonding level (λ), span length (L), and amount of strengthening material (A_s). The coefficient of determinate (R²) for curve fitting in nonlinear regression analysis is 0.8768. The developed analytical model was compared to the ultimate capacities computed by FEA model.