Yazar "Yüceer M." seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • Küçük Resim Yok
    Öğe
    River water quality model verification through a GIS based software
    (IFAC Secretariat, 2009) Yetik M.K.; Yüceer M.; Berber R.; Karadurmuş E.
    Research and development attempts on water quality models created valuable resources in the sense of model calibration and verification techniques. Recognizing the current degree of pollution in rivers and the importance of the sustainable water resources management, the interactive river monitoring appears to be at the center of recent focus. However the available information in this area is still far from expectations. On one side, the Geographical Information Systems (GIS) are gaining widespread acceptance and on the other side fast and reliable water quality models and parameter estimation techniques are becoming available. However, previous work on integrating water quality models and GIS is very limited. This work brings an integrated platform on which ArcMap as a GIS and a water quality model in Matlab™ are brought together in an interactive and user-friendly manner. The software developed allows the user to enter the data collected from the river, runs the dynamic model in the Matlab™ environment, predicts the values of pollution constituents along the river, extracts the results and displays the water quality on the map in different forms. The software thus provides a considerable ease in future real time application for on site river monitoring and environmental pollution assessment.
  • Küçük Resim Yok
    Öğe
    River water quality modelling using artificial intelligence techniques
    (Nova Science Publishers, Inc., 2020) Göz E.; Karadurmuş E.; Yüceer M.
    Water pollution has become a major issue in rivers. The possibility of a pollutant to be discharged to the river as municipal and industrial waste is an important problem for those using water from rivers. Nevertheless, due to the rapid population growth in the world and the irresponsible use of water resources, the world will face a serious lack of water in the near future. Therefore, the water resources of the future must be preserved very well to leave healthy and enough water for next generations. In order to prevent river water pollution, river water quality should be constantly monitored and evaluated. This way, information on the status of water quality may be obtained, and river basin management planning may be carried out. For this purpose, measurement at points can be made, or online monitoring stations can be established on river basins. According to the collected data, management actions may be created for how waterways function and how pollutants affect evaluation. In addition to this effect, seasonal changes and long-term trends must be taken into consideration. Artificial intelligence (AI) techniques have been used recently in many engineering fields. The most widely used ones among AI techniques are artificial neural networks (ANN). These are followed by support vector regression (SVR), least squares support vector regression (LS-SVR), least squares support vector machine (LS-SVM) and fuzzy logic. In the past 15 years, extreme learning machine (ELM) and its types have been used in development of many forecasting models. The statistical accuracy of classical models is commonly poor because natural systems tend to be complex and nonlinear for deterministic modelling methods. AI techniques provide a fast and flexible means of creating models for estimation of river water quality. In recent years, AI techniques have shown exceptional performance as regression tools, especially when used for pattern recognition and function estimation. In this study, AI techniques will be applied to river water quality data, and AI models will be developed. The data were collected from an on-line measurement station that was established on the Yeşilirmak River in Amasya/Turkey. In the selected region, two different measurement stations were built at a distance of about 28 km. Twelve parameters as luminescent dissolved oxygen (LDO), pH, conductivity, nitrate nitrogen (NO3-N), ammonium nitrogen (NH4-N), total organic carbon (TOC), chloride, orthophosphate, temperature, turbidity, suspended solid and flow rate were measured at five-minute intervals at these stations. Specifically, two different models as DO prediction and TOC prediction models were developed with five different approaches. These approaches were artificial neural network (ANN), support vector regression, least squares support vector regression, extreme learning machine and kernel extreme learning machine. Model performances were evaluated with some performance indices. This study is a state-of-the-art study due to the fact that parameters that are expensive to measure can be predicted from parameters that are cheaper to measure. For this reason, it will contribute significantly to reducing the cost of on-line measurement stations planned to be established in the future. © 2020 Nova Science Publishers, Inc.