• Journal of the Korean Solar Energy Society
• /
• v.39 no.5
• /
• pp.41-52
• /
• 2019

The purpose of the work is to develop a simple solar irradiance prediction model using a deep learning method, the LSTM (long term short term memory). Other than existing prediction models, the proposed one uses only the cloudiness among the information forecasted from the national meterological forecast center. The future cloudiness is generally announced with four categories and for three-hour intervals. In this work, a daily irradiance pattern is used as an input vector to the LSTM together with that cloudiness information. The proposed model showed an error of 5% for learning and 30% for prediction. This level of error has lower influence on the load prediction in typical building cases.

• Atmosphere
• /
• v.33 no.2
• /
• pp.73-104
• /
• 2023

Over the past 60 years, Korean numerical weather prediction (NWP) has advanced rapidly with the collaborative effort between the science community and the operational modelling center. With an improved scientific understanding and the growth of information technology infrastructure, Korea is able to provide reliable and seamless weather forecast service, which can predict beyond a 10 days period. The application of NWP has expanded to support decision making in weather-sensitive sectors of society, exploiting both storm-scale high-impact weather forecasts in a very short range, and sub-seasonal climate predictions in an extended range. This article gives an approximate chronological account of the NWP over three periods separated by breakpoints in 1990 and 2005, in terms of dynamical core, physics, data assimilation, operational system, and forecast application. Challenges for future development of NWP are briefly discussed.

• International Journal of Control, Automation, and Systems
• /
• v.3 no.4
• /
• pp.509-523
• /
• 2005

We are developing a novel framework, PRIDE (PRediction In Dynamic Environments), to perform moving object prediction (MOP) for autonomous ground vehicles. The underlying concept is based upon a multi-resolutional, hierarchical approach which incorporates multiple prediction algorithms into a single, unifying framework. The lower levels of the framework utilize estimation-theoretic short-term predictions while the upper levels utilize a probabilistic prediction approach based on situation recognition with an underlying cost model. The estimation-theoretic short-term prediction is via an extended Kalman filter-based algorithm using sensor data to predict the future location of moving objects with an associated confidence measure. The proposed estimation-theoretic approach does not incorporate a priori knowledge such as road networks and traffic signage and assumes uninfluenced constant trajectory and is thus suited for short-term prediction in both on-road and off-road driving. In this article, we analyze the complementary role played by vehicle kinematic models in such short-term prediction of moving objects. In particular, the importance of vehicle process models and their effect on predicting the positions and orientations of moving objects for autonomous ground vehicle navigation are examined. We present results using field data obtained from different autonomous ground vehicles operating in outdoor environments.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.31 no.1D
• /
• pp.43-50
• /
• 2011

The monitoring system for link travel speed using taxi probe is one of key sub-systems of ITS. Link travel speed collected by taxi probe has been widely employed for both monitoring the traffic states of urban road network and providing real-time travel time information. When sample size of taxi probe is small and link travel time is longer than a length of time interval to collect travel speed data, and in turn the missing state is inevitable. Under this missing state, link travel speed data is real-timely not collected. This missing state changes from single to multiple time intervals. Existing single interval prediction techniques can not generate multiple future states. For this reason, it is necessary to replace multiple missing states with the estimations generated by multi-interval prediction method. In this study, a multi-interval prediction method to generate the speed estimations of single and multiple future time step is introduced overcoming the shortcomings of short-term techniques. The model is developed based on Non-Parametric Regression (NPR), and outperformed single-interval prediction methods in terms of prediction accuracy in spite of multi-interval prediction scheme.

• Ocean Systems Engineering
• /
• v.14 no.1
• /
• pp.85-99
• /
• 2024

Ocean surface currents have an essential role in the Earth's climate system and significantly impact the marine ecosystem, weather patterns, and human activities. However, predicting ocean surface currents remains challenging due to the complexity and variability of the oceanic processes involved. This review article provides an overview of the current research status, challenges, and opportunities in the prediction of ocean surface currents. We discuss the various observational and modelling approaches used to study ocean surface currents, including satellite remote sensing, in situ measurements, and numerical models. We also highlight the major challenges facing the prediction of ocean surface currents, such as data assimilation, model-observation integration, and the representation of sub-grid scale processes. In this article, we suggest that future research should focus on developing advanced modeling techniques, such as machine learning, and the integration of multiple observational platforms to improve the accuracy and skill of ocean surface current predictions. We also emphasize the need to address the limitations of observing instruments, such as delays in receiving data, versioning errors, missing data, and undocumented data processing techniques. Improving data availability and quality will be essential for enhancing the accuracy of predictions. The future research should focus on developing methods for effective bias correction, a series of data preprocessing procedures, and utilizing combined models and xAI models to incorporate data from various sources. Advancements in predicting ocean surface currents will benefit various applications such as maritime operations, climate studies, and ecosystem management.

• Water for future
• /
• v.18 no.1
• /
• pp.4-8
• /
• 1985

• Water for future
• /
• v.13 no.3
• /
• pp.63-70
• /
• 1980

• Water for future
• /
• v.47 no.5
• /
• pp.41-49
• /
• 2014

• Bulletin of Food Technology
• /
• v.24 no.3
• /
• pp.325-332
• /
• 2011

• Journal of the Korean Operations Research and Management Science Society
• /
• v.39 no.3
• /
• pp.63-82
• /
• 2014

The main objective of this research is to investigate whether the RFM (recency-frequency-monetary value) information of a customer's redemption behavior of loyalty points can improve the prediction of future value of the customer. The conventional measurement of customer value has been primarily based on purchase transactions behavior although a customer's future behavior can be also influenced by other interactions between the customer and the firm such as redemption of rewards in a loyalty program. We theorize why a customer's redemption behavior can influence her future purchases and thereby the customer's total value based on operant learning theory, goal gradient hypothesis, and lock-in effect. Using a dataset from a major book store in Korea spanning three years between 2008 and 2010, we analyze both purchase transactions and redemption records of over 10,000 customers. The results show that the redemption-based RFM information does improve the prediction accuracy of the customer's future purchases. Based on this result, we also propose an improved estimate of customer lifetime value (CLV) by combining purchase transactions and loyalty points redemption data. Managerial implications will be also discussed for firms managing loyalty programs to maximize the total value customers.