• 지능정보연구
• /
• 제1권1호
• /
• pp.135-148
• /
• 1995

Neural Networks have been advocated as an alternative to statistical forecasting methods. However, the empirical evidences are not consistent. In the present experiments, multi-layered perceptron (MLP) are adopted as approximator to the time series generating processes. To prevent the MLP from being overfitted to the given time series, the information obtained from ARMA modeling is used to determine the architecture of MLP. The proposed approach was tested empirically using the subsamples of the 111 time series used in the first Markridakis Competition. The forecasting results were analyzed to find out the factors that affect the performance of MLP. The experimental results show that the proposed approach outperforms ARMA models in terms of fitting and forecasting accuracy. In addition, it is found that the use of deseasonalized data improves the forecasting accuracy of MLP.

• Journal of Electrical Engineering and Technology
• /
• 제12권1호
• /
• pp.64-71
• /
• 2017

Considering the volatility, intermittent and random of photovoltaic (PV) generation systems, accurate forecasting of PV power output is important for the grid scheduling and energy management. In order to improve the accuracy of short-term power forecasting of PV systems, this paper proposes a prediction model based on environmental factors and support vector machine optimized by genetic algorithm (GA-SVM). In order to improve the prediction accuracy of this model, weather conditions are divided into three types, and the gray correlation coefficient algorithm is used to find out a similar day of the predicted day. To avoid parameters optimization into local optima, this paper uses genetic algorithm to optimize SVM parameters. Example verification shows that the prediction accuracy in three types of weather will remain at between 10% -15% and the short-term PV power forecasting model proposed is effective and promising.

• 자원ㆍ환경경제연구
• /
• 제14권1호
• /
• pp.201-220
• /
• 2005

최근 들어, 우리는 유례 없는 국제 유가의 급등현상을 목격하고 있다. 이러한 시점에서, 의문점은 유가에 대한 예측 가능성과 이의 정확도에 관한 것이다. 본 연구에서는 전문가 예측시스템과 비교하여 선물가격의 상대적인 예측력에 관하여 통계적으로 분석하고자 한다. 이를 위해, 미국 텍사스 중질유(WTI)의 현물가격과 선물가격을 활용하여, 예측 정확도에 관한 단순한 형태의 통계적 분석과 함께 분석수단별 예측오차 차이의 유의성에 관한 체계적 분석을 시도하였다. 통계적 검정결과에 따르면, WTI 선물시장을 활용한 예측은 미국 에너지정보기구(EIA)의 예측과 비교하여 뒤지지 않는 것으로 판명되었다. 결과적으로, 석유 생산자와 소비자 모두가 WTI 선물시장을 유가 예측의 유용한 수단으로 활용할 수 있고, 이로써 효율적인 자원배분 측면에서도 유익할 것으로 판단된다.

• 한국항만경제학회지
• /
• 제36권2호
• /
• pp.53-62
• /
• 2020

해운항만물류산업은 세계 경제활동과 밀접한 관계를 가지고 있으며, 특히 무역의존도가 높은 우리나라의 항만 시설은 중요한 사회간접자본시설이다. 부산항은 우리나라 최대의 항만으로 우리나라 컨테이너 운송의 75%가 부산항을 통해 운송되고 있으며, 국가 경쟁력 측면에서 그 중요성은 매우 크다. 항만 물동량 예측은 항만 개발 및 운영 전략에 영향을 미치며, 정확도 높은 컨테이너 물동량 예측은 필수적이다. 하지만 오늘날 해운항만물류산업 환경의 급격한 변화로 인해 기존 시계열 예측 방법으로는 예측 정확도 향상에 어려움이 있다. 본 연구에서는 부산항 컨테이너 물동량 예측 정확도 향상을 위해 딥러닝 모형 중 LSTM 모형을 활용하여 컨테이너 물동량을 예측한다. 모형의 성능 평가를 위해서 SARIMA 모형과 LSTM 모형의 예측 정확도를 비교한다. 그 결과 LSTM 모형이 SARIMA 모형보다 예측 정확도가 높게 나타났으며, 예측치가 실측치의 특성을 반영하여 잘 나타나고 있음을 확인하였다.

• 대한전기학회논문지:전력기술부문A
• /
• 제53권6호
• /
• pp.358-363
• /
• 2004

Due to the weather sensitivity of the power load, it is difficult to forecast accurately the peak power load of summer season. We improve the accuracy of the load forecasting considering weather condition. We introduced the sensitivity of temperature and proposed an improved forecasting algorithm. The proposed algorithm shows that the error of the load forecasting is 1.5%.

• 물과 미래
• /
• 제27권3호
• /
• pp.55-69
• /
• 1994

홍수시 한강 인도교에 대한 단기간 예보의 정확도를 제고하기 위한 통계학적 홍수예보모형으로 '인도교수위와 영향인자간의 다중회귀분석에 의한 다변수 모형(MM 모형)'과 '수위구간별 다중회귀분석에 의한 다수준 다변수 모형(MMP 모형)' 그리고 '수위의 증감추세에 따른 2 수준 다변수 모형(2MP 모형)'을 제시하였다. 연구대상으로는 분석된 세가지 모형 중, 'MM 모형'은 4시간예측시 평균오차가 35cm 이내의 정도를 나타내며 'MMP 모형'은 모형개발시에 구분한 각 수위구간에 대해서는 매우 작은 평균오차를 나타내지만 실제 홍수사상에 적용시에는 뚜렷한 정도의 향상을 나타내지 못하는 것으로 보인다. 이것은 실제홍수시 수위가 각 구간내에만 머물지 않기 때문인 것으로 보인다. 한편 '2MP 모형'은 예측정도가 가장 높으나 드물게 발산현상이 나타나고 있어 안정도가 떨어지며, 'MMP 모형'은 '2MP 모형'과 비교하여 예측정도는 약간 떨어지나 안정된 예측결과를 보여준다.

• Asia pacific journal of information systems
• /
• 제11권4호
• /
• pp.175-186
• /
• 2001

Though people frequently rely on intuition in managing activities, they rarely use it in developing effective decision-making support systems. In this report, we investigate the correlations between characteristics of cognition and emotion and judgmental time-series forecasting accuracy, and compare their strengths by using a self-supervised adaptive neural network. Through the experiments, we hope to help find a desirable atmosphere for decision-making. Our experiments showed that both cognition characteristics and emotion characteristics had correlations with the time-series forecasting accuracy, and that cognition characteristics had larger correlation than emotion characteristics. We also found that conceptual style had larger correlation than behavioral or analytical styles with the accuracy.

• Journal of Electrical Engineering and Technology
• /
• 제12권6호
• /
• pp.2166-2176
• /
• 2017

Electricity price prediction plays a crucial part in making the schedule and managing the risk to the competitive electricity market participants. However, it is a difficult and challenging task owing to the characteristics of the nonlinearity, non-stationarity and uncertainty of the price series. This study proposes a hybrid improved strategy which incorporates data preprocessor components and a forecasting engine component to enhance the forecasting accuracy of the electricity price. In the developed forecasting procedure, the Seasonal Adjustment (SA) method and the Ensemble Empirical Mode Decomposition (EEMD) technique are synthesized as the data preprocessing component; the Coupled Simulated Annealing (CSA) optimization method and the Least Square Support Vector Regression (LSSVR) algorithm construct the prediction engine. The proposed hybrid approach is verified with electricity price data sampled from the power market of New South Wales in Australia. The simulation outcome manifests that the proposed hybrid approach obtains the observable improvement in the forecasting accuracy compared with other approaches, which suggests that the proposed combinational approach occupies preferable predication ability and enough precision.

• 한국농공학회논문집
• /
• 제49권3호
• /
• pp.21-31
• /
• 2007

This study is aimed at the development of a runoff forecasting model by using the Fuzzy inference system and Neural Network model to solve the uncertainties occurring in the process of rainfall-runoff modeling and improve the modeling accuracy of the stream runoff forecasting. The Neuro-Fuzzy (NF) model were used in this study. The NF model, recently received a great deal of attention, improve the existing Neural Networks by the aid of the Fuzzy theory applied to each node. The study area is the downstreams of Naeseung-chun. Therefore, time-dependent data was obtained from the Wolpo water level gauging station. 11 and 2 out of total 13 flood events were selected for the training and testing set of model respectively. The schematic diagram method and the statistical analysis are conducted to evaluate the feasibility of rainfall-runoff modeling. The model accuracy was rapidly decreased as the forecasting time became longer. The NF model can give accurate runoff forecasts up to 4 hours ahead in standard above the Determination coefficient $(R^2)$ 0.7. In the comparison of the runoff forecasting using the NF and TANK models, characteristics of peak runoff in the TANK model was higher than ones in the NF models, but peak values of hydrograph in the NF models were similar.

• 한국정보시스템학회지:정보시스템연구
• /
• 제27권3호
• /
• pp.123-140
• /
• 2018

Purpose The purpose of this study is to explore the optimal trading frequency which is useful for stock price prediction by using deep learning for charting image data. We also want to identify the appropriate time for accurate forecasting of stock price when performing pattern analysis. Design/methodology/approach In order to find the optimal trading frequency patterns and forecast timings, this study is performed as follows. First, stock price data is collected using OpenAPI provided by Daishin Securities, and candle chart images are created by data frequency and forecasting time. Second, the patterns are generated by the charting images and the learning is performed using the CNN. Finally, we find the optimal trading frequency patterns and forecasting timings. Findings According to the experiment results, this study confirmed that when the 10 minute frequency data is judged to be a decline pattern at previous 1 tick, the accuracy of predicting the market frequency pattern at which the market decreasing is 76%, which is determined by the optimal frequency pattern. In addition, we confirmed that forecasting of the sales frequency pattern at previous 1 tick shows higher accuracy than previous 2 tick and 3 tick.