2017 Cilt 22 Sayı 2

Permanent URI for this collectionhttps://hdl.handle.net/11452/12031

Browse

Browsing by Language "en"

Filter results by typing the first few letters
Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Medium-range low flow forecasts in the lobith, River Rhine
    (Uludağ Üniversitesi, 2017-07-30) Bouwma, Pieter; Demirel, Mehmet C.
    The aim of this study is to predict low flows 14 days in advance using a data-driven model. First, we apply correlation analysis to select appropriate temporal scales of pre-selected inputs that are precipitation, potential evapotranspiration, discharge, groundwater, snow height and lake levels. The forecasted rainfall has also been used as model input to forecast low flows in the River Rhine at Lobith. The correlation analysis analysis between low flows and basin indicators show stronger correlations for the Alpine sub-basins than the rainfed sub-basins. The Middle and Lower Rhine are downstream channel areas and they do not contribute to the discharge. Therefore, they are excluded from the entire analysis. The low flow predictions for the Alpine sub-basins and the Mosel are reasonable during the validation period, whereas the ANN for Lobith shows low performance for a different test period. The results for the training and the validation period are more encouraging than the test period for Lobith, i.e. Nash-Sutcliffe (NS) efficiency of 0.75 and 0.73 respectively.
  • Thumbnail Image
    Item
    Performance study of bat algorithm and clonal selection algorithm for optimization tasks
    (Uludağ Üniversitesi, 2017-06-12) Ülker, Ezgi Deniz
    Evolutionary algorithms are preferred by many researchers in different areas for optimization tasks. It is quite important to find optimum points of problems with less number of iterations. In this paper, performance analysis of two powerful optimization algorithms; bat algorithm and clonal selection algorithm are studied using well-known benchmark functions. The experimental results show that bat algorithm outperforms clonal selection algorithm on most of the selected problems. It is also seen that bat algorithm can produce high quality results even at the first stages of iterations. This paper can be used as guidance of performance comparisons for future studies.
  • Thumbnail Image
    Item
    Simulation of vehicles’ gap acceptance decisions using reinforcement learning
    (Uludağ Üniversitesi, 2017-08-01) Bartın, Bekir Oğuz
    This paper presents the use of reinforcement learning approach for modeling vehicles' gap acceptance decisions at a stop-controlled intersection. The proposed formulation translates a simple gap acceptance decision into a reinforcement learning problem, assuming that drivers' ultimate objective in a traffic network is to optimize wait-time and safety. Using an off-the-shelf simulation tool, drivers are simulated without any notion of the outcome of their decisions. From multiple episodes of gap acceptance decisions, they learn from the outcome of their actions, i.e., wait-time and safety. A real-world traffic circle simulation network developed in Paramics simulation software is used to conduct experimental analyses. The results show that drivers' gap acceptance behavior in microscopic traffic simulation models can easily be validated with a high level of accuracy using Q-learning reinforcement-learning algorithm.