• Advances in environmental research
• /
• 제5권3호
• /
• pp.153-167
• /
• 2016

We applied multilayer perceptron (MLP) and radial basis function (RBF) neural network in upstream and downstream water quality stations of the Karaj Reservoir in Iran. For both neural networks, inputs were pH, turbidity, temperature, chlorophyll-a, biochemical oxygen demand (BOD) and nitrate, and the output was dissolved oxygen (DO). We used an MLP neural network with two hidden layers, for upstream station 15 and 33 neurons in the first and second layers respectively, and for the downstream station, 16 and 21 neurons in the first and second hidden layer were used which had minimum amount of errors. For learning process 6-fold cross validation were applied to avoid over fitting. The best results acquired from RBF model, in which the mean bias error (MBE) and root mean squared error (RMSE) were 0.063 and 0.10 for the upstream station. The MBE and RSME were 0.0126 and 0.099 for the downstream station. The coefficient of determination ($R^2$) between the observed data and the predicted data for upstream and downstream stations in the MLP was 0.801 and 0.904, respectively, and in the RBF network were 0.962 and 0.97, respectively. The MLP neural network had acceptable results; however, the results of RBF network were more accurate. A sensitivity analysis for the MLP neural network indicated that temperature was the first parameter, pH the second and nitrate was the last factor affecting the prediction of DO concentrations. The results proved the workability and accuracy of the RBF model in the prediction of the DO.

• Journal of Biosystems Engineering
• /
• 제38권1호
• /
• pp.48-54
• /
• 2013

Purpose: This study evaluated the feasibility of using a discrete wavelet transform (DWT) method as a preprocessing tool for visible/near-infrared spectroscopy (VIS/NIRS) with a spectroscopic transmittance dataset for predicting the internal quality of cherry tomatoes. Methods: VIS/NIRS was used to acquire transmittance spectrum data, to which a DWT was applied to generate new variables in the wavelet domain, which replaced the original spectral signal for subsequent partial least squares (PLS) regression analysis and prediction modeling. The DWT concept and its importance are described with emphasis on the properties that make the DWT a suitable transform for analyzing spectroscopic data. Results: The $R^2$ values and root mean squared errors (RMSEs) of calibration and prediction models for the firmness, sugar content, and titratable acidity of cherry tomatoes obtained by applying the DWT to a PLS regression with a set of spectra showed more enhanced results than those of each model obtained from raw data and mean normalization preprocessing through PLS regression. Conclusions: The developed DWT-incorporated PLS models using the db5 wavelet base and selected approximation coefficients indicate their feasibility as good preprocessing tools by improving the prediction of firmness and titratable acidity for cherry tomatoes with respect to $R^2$ values and RMSEs.

• 한국수자원학회논문집
• /
• 제57권2호
• /
• pp.73-85
• /
• 2024

하천과 저수지의 수질을 예측하는 것은 수자원관리를 위해 필요하다. 높은 정확도의 수질 예측을 위해 많은 연구들에서 인공신경망이 활용되었다. 기존 연구들은 매개변수를 탐색하는 인공신경망의 연산자인 옵티마이저로 경사하강법 기반 옵티마이저를 사용하였다. 그러나 경사하강법 기반 옵티마이저는 지역 최적값으로의 수렴 가능성과 해의 저장 및 비교구조가 없다는 단점이 있다. 본 연구에서는 인공신경망을 이용한 수질 예측성능을 향상시키기 위해 개량형 옵티마이저를 개발하여 경사하강법 기반 옵티마이저의 단점을 개선하였다. 본 연구에서 제안한 옵티마이저는 경사하강법 기반 옵티마이저 중 학습오차가 낮은 Adaptive moments (Adam)과 Nesterov-accelerated adaptive moments (Nadam)를 Harmony Search(HS) 또는 Novel Self-adaptive Harmony Search (NSHS)와 결합한 옵티마이저이다. 개량형 옵티마이저의 학습 및 예측성능 평가를 위해 개량형 옵티마이저를 Long Short-Term Memory (LSTM)에 적용하여 국내의 다산 수질관측소의 수질인자인 수온, 용존산소량, 수소이온농도 및 엽록소-a를 학습 및 예측하였다. 학습결과를 비교하면, Nadam combined with NSHS (NadamNSHS)를 사용한 LSTM의 Mean Squared Error (MSE)가 0.002921로 가장 낮았다. 또한, 각 옵티마이저별 4개 수질인자에 대한 MSE 및 R²에 따른 예측순위를 비교하였다. 각 옵티마이저의 평균 순위를 비교하면, NadamNSHS를 사용한 LSTM이 2.25로 가장 높은 것을 확인하였다.

• 소성∙가공
• /
• 제25권3호
• /
• pp.182-188
• /
• 2016

Flexibly-reconfigurable roll forming (FRRF) is a novel sheet metal forming technology conducive to producing multi-curvature surfaces by controlling the strain distribution along longitudinal direction. In FRRF, a sheet metal is shaped into the desired curvature by using reconfigurable rollers and gaps between the rollers. As FRRF technology and equipment are under development, a simulation model corresponding to the physical FRRF would aid in investigating how the shape of a sheet varies with input parameters. To facilitate the investigation, the current study exploits regression analysis to construct a predictive model for the longitudinal curvature of the sheet. Variables considered as input parameters are sheet compression ratio, radius of curvature in the transverse direction, and initial blank width. Samples were generated by a three-level, three-factor full factorial design, and both convex and saddle curvatures are represented by a quadratic regression model with two-factor interactions. The fitted quadratic equations were verified numerically with R-squared values and root mean square errors.

• 대한전자공학회논문지
• /
• 제26권3호
• /
• pp.160-166
• /
• 1989

공간영역 예측기의 계수를 계산하기 위한 적응 알고리듬이 제안되었다. 제안된 방법은 LMS 알고리듬을 사용하여 TDL(tapped-delay-line)과 ESC(escalator) 구조를 갖는 공간영역 예측기의 계수를 계산한다. 기종존의 일반적인 예측기와 다른점은 순방향과 역방향 예측 오차의 평균 자승값의 합을 최소화하며 예측기의 계수를 계산함으로 향상된 선형예측 공간 스펙트럼을 얻을 수 있다. 제안된 방법을 선형으로 배열된 센서에 의하여 얻어진 협대역신호의 입사각 추정문제에 적용시켜 기존의 적응예측 알고리듬과 컴퓨터 시뮬레이션을 통하여 성능을 비교하였다.

• Genomics & Informatics
• /
• 제19권1호
• /
• pp.11.1-11.8
• /
• 2021

For the novel coronavirus disease 2019 (COVID-19), predictive modeling, in the literature, uses broadly susceptible exposed infected recoverd (SEIR)/SIR, agent-based, curve-fitting models. Governments and legislative bodies rely on insights from prediction models to suggest new policies and to assess the effectiveness of enforced policies. Therefore, access to accurate outbreak prediction models is essential to obtain insights into the likely spread and consequences of infectious diseases. The objective of this study is to predict the future COVID-19 situation of Korea. Here, we employed 5 models for this analysis; SEIR, local linear regression (LLR), negative binomial (NB) regression, segment Poisson, deep-learning based long short-term memory models (LSTM) and tree based gradient boosting machine (GBM). After prediction, model performance comparison was evelauated using relative mean squared errors (RMSE) for two sets of train (January 20, 2020-December 31, 2020 and January 20, 2020-January 31, 2021) and testing data (January 1, 2021-February 28, 2021 and February 1, 2021-February 28, 2021) . Except for segmented Poisson model, the other models predicted a decline in the daily confirmed cases in the country for the coming future. RMSE values' comparison showed that LLR, GBM, SEIR, NB, and LSTM respectively, performed well in the forecasting of the pandemic situation of the country. A good understanding of the epidemic dynamics would greatly enhance the control and prevention of COVID-19 and other infectious diseases. Therefore, with increasing daily confirmed cases since this year, these results could help in the pandemic response by informing decisions about planning, resource allocation, and decision concerning social distancing policies.

• 자원환경지질
• /
• 제56권3호
• /
• pp.331-341
• /
• 2023

지구통계 기법을 기반으로 토양오염지도를 작성하는 경우 예측 오차가 발생하며 이에 영향을 미치는 다양한 원인이 존재한다. 본 연구에서는 정규 크리깅을 활용하여 폐광산지역의 토양 내 중금속 농도 샘플링 데이터로부터 격자형 기반의 토양오염지도를 작성하였다. 해당 지도의 예측 오차에 영향을 미친다고 판단된 5개 인자를 선정하고, Leave-one-out 기법을 기반으로 인자의 옵션과 설정값의 변화에 따른 예측값과 실측값 간의 평균제곱근오차(root mean square error, RMSE) 변화를 분석하였다. 이후 머신러닝 알고리즘을 이용하여 RMSE에 영향을 미치는 상위 3개 인자를 도출하였다. 그 결과, Standard interpolation에서는 Variogram Model, Minimum Neighbors, Anisotropy 인자가 RMSE에 가장 큰 영향을 미치는 것으로 분석되었다. 베리오그램 모델에서는 Spherical 모델이 가장 낮은 RMSE를 보였으며, Minimum Neighbors는 3에서 최젓값을 보인 후 값이 증가함에 따라 증가하였다. Anisotropy의 경우 이방성을 고려하지 않는 것이 더 적합한 것으로 나타났다. 본 연구에서는 지구통계와 머신러닝의 복합 활용을 통해 지역 규모에서 높은 신뢰성을 갖는 토양오염지도를 작성할 수 있었고, 적은 수의 토양 샘플링 데이터의 보간 작업 시 어떠한 요인들이 큰 영향을 미치는지 파악할 수 있었다.

• Environmental Engineering Research
• /
• 제21권4호
• /
• pp.333-340
• /
• 2016

An Artificial Neural Network including a Radial Basis Function (RBF) and a Time Delay Neural Network (TDNN) was used to predict total dissolved solid (TDS) in the river Zayanderud. Water quality parameters in the river for ten years, 2001-2010, were prepared from data monitored by the Isfahan Regional Water Authority. A factor analysis was applied to select the inputs of water quality parameters, which obtained total hardness, bicarbonate, chloride and calcium. Input data to the neural networks were pH, $Na^+$, $Mg^{2+}$, Carbonate ($CO{_3}^{-2}$), $HCO{_3}^{-1}$, $Cl^-$, $Ca^{2+}$ and Total hardness. For learning process 5-fold cross validation were applied. In the best situation, the TDNN contained 2 hidden layers of 15 neurons in each of the layers and the RBF had one hidden layer with 100 neurons. The Mean Squared Error and the Mean Bias Error for the TDNN during the training process were 0.0006 and 0.0603 and for the RBF neural network the mentioned errors were 0.0001 and 0.0006, respectively. In the RBF, the coefficient of determination ($R^2$) and the index of agreement (IA) between the observed data and predicted data were 0.997 and 0.999, respectively. In the TDNN, the $R^2$ and the IA between the actual and predicted data were 0.957 and 0.985, respectively. The results of sensitivity illustrated that $Ca^{2+}$ and $SO{_4}^{2-}$ parameters had the highest effect on the TDS prediction.

• Communications for Statistical Applications and Methods
• /
• 제16권1호
• /
• pp.103-113
• /
• 2009

표본조사에서 무응답은 비 표본추출오차를 발생시키는 중요한 원인 중 하나이다. 단위무응답이 발생하는 경우 무응답에 의한 편향을 줄이는 동시에 추정의 정도를 향상시키기 위해 단위무응답 조정층을 구성해 무응답 가중치 조정을 하는 것이 일반적이다. 본 연구에서는 무응답 조정층 구성과 관련된 기존의 이론들을 정리하고 어업총조사 자료를 이용한 실증적인 모의실험을 통해 효과적으로 무응답 조정층을 구성하는 방안에 대해 살펴본다. 모의실험결과 응답성향에 따른 조정층 구성보다는 예측평균을 기준으로 한 조정층 구성이 효율성 측면에서 효과적인 것으로 나타났으며, 아울러 다른 관심변수에도 적용될 수 있는 로버스트한 조정층 구성을 위해서는 예측평균만을 고려하는 것보다 응답성향과 예측평균을 모두 고려한 조정층 구성방법이 효과적인 것으로 나타났다. 한편 무응답 조정을 위한 응답률 산출에 있어서 설계가중치의 적용 필요성에 대해 살펴본 결과 설계가중치 적용 여부는 추정결과에 거의 영향을 주지 않는다는 사실을 확인할 수 있었다.

• 대한영양사협회학술지
• /
• 제13권2호
• /
• pp.157-168
• /
• 2007

The purposes of this study were to measure the resting metabolic rate(RMR) of 30-40 year old women and to compare it with values predicted using published equations. Body weight, height and body fat of subjects were measured. RMR was measured by two indirect calorimeter(method 1 and method 2). RMR was predicted using various equations. Average height, weight and body fat(%) of subjects were 158.6cm, 59.1kg and 30.9%, respectively. The RMR(1621.2$\pm$301.5 kcal/day) measured by portable indirect calorimeter(method 2) was significantly higher than RMR(1447.4$\pm$223.6 kcal/day) measured by typical indirect calorimeter(method 1). Comparison of measured RMR with predicted RMRs suggested that there was a least difference in RMR predicted by equation of Cunningham. According to RMSPEs(Root Mean Squared Prediction Errors), equations of Cunningham and body surface area were found to predict measured RMR(by method 1) most accurately (within 239.1kcal/day and 232.9kcal/day, respectively). The fat free mass and fat mass - adjusted correlation showed that measured RMR(by method 1) had negative relationships with muscle mass(r = -0.873) and fat free mass(r = -0.866). The equations of Cunningham and body surface area provide relatively accurate estimates of RMR when determining energy needs of 30-40 aged women. There are needs for development of RMR predicted equations that are derived from large samples of Korean.