Yazar "Kaya, Irem Ersoz" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • [ X ]
    Öğe
    Artificial Neural Networks as a Decision Support Tool in Curriculum Development
    (World Scientific Publ Co Pte Ltd, 2019) Kaya, Irem Ersoz
    In the context of learning analytics, machine learning techniques have been commonly used in order to shed a light on solving educational problems. The studies can be associated with the curriculum design, mostly at course level, which target to improve the learning and teaching processes. In this study, the relationship between the courses was analyzed via artificial neural network (ANN) to provide a support for curriculum development process at a program level. Extracting the dependence among courses within a program plays a key role in placing them coherently to ensure the success of curriculum. For this purpose, it was investigated if the performance of students in a subsequent course is influenced from the performance in some previous basic courses. The results demonstrated that the performance relations could be used to describe information for prioritization and sequencing of courses within a program. In addition, ANN can successfully predict the student performances leading to find out the relationships between these courses. The application of the multilayer feedforward neural network resulted in an achievement of a prospering prediction performance based on the grades of prerequisite courses with 87% accuracy rate without sacrificing a significant sensitivity.
  • [ X ]
    Öğe
    Predictive equivalent consumption minimization strategy for power-split hybrid electric vehicles using Monte Carlo algorithm
    (Gazi Univ, Fac Engineering Architecture, 2023) Gul, Merve Nur; Yazar, Ozan; Coskun, Serdar; Zhang, Fengqi; Li, Lin; Kaya, Irem Ersoz
    Purpose: The underlying research goal of this article is to put forward a reliable fuel saving performance based on the forecasted velocities of drive cycles for a power-split hybrid electric vehicle. Theory and Methods: The power distribution between energy sources is devised by utilizing the P-ECMS for the power-split hybrid electric vehicle using the uncertain drive cycle velocity estimation based on MC algorithm. Results: The effectiveness and accuracy of the method are evaluated under seven drive cycles. The MC provides good prediction results of the velocities. On the basis of it, the P-ECMS method decreases fuel consumption up to 6.01% under NEDC, up to 9.09% under WLTP, up to 6.33% under UDDS, up to 5.14% under HWFET, up to 1.96% under NYCC, up to 11.47% under LA-92, and up to 7.92% under ALL-CYC compared to a standard ECMS method. Conclusion: It is seen from the analysis results that battery SOC decreases slightly using the P-ECMS since the electric motor is actively used to meet power demand instead of the engine over the predicted speed profiles. In the end, the MC algorithm-based P-ECMS strategy can verify the optimal power distribution based on fuel-saving potentials as compared to the baseline ECMS strategy while keeping the battery SOC at a reasonable interval.